Upload folder using huggingface_hub
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- checkpoints/checkpoint-1000/config.json +2 -2
- checkpoints/checkpoint-1000/model.safetensors +2 -2
- checkpoints/checkpoint-1000/optimizer.pt +2 -2
- checkpoints/checkpoint-1000/trainer_state.json +40 -40
- checkpoints/checkpoint-1000/training_args.bin +1 -1
- checkpoints/checkpoint-1500/config.json +2 -2
- checkpoints/checkpoint-1500/model.safetensors +2 -2
- checkpoints/checkpoint-1500/optimizer.pt +2 -2
- checkpoints/checkpoint-1500/trainer_state.json +60 -60
- checkpoints/checkpoint-1500/training_args.bin +1 -1
- checkpoints/checkpoint-2000/config.json +2 -2
- checkpoints/checkpoint-2000/model.safetensors +2 -2
- checkpoints/checkpoint-2000/optimizer.pt +2 -2
- checkpoints/checkpoint-2000/trainer_state.json +80 -80
- checkpoints/checkpoint-2000/training_args.bin +1 -1
- checkpoints/checkpoint-2500/config.json +2 -2
- checkpoints/checkpoint-2500/model.safetensors +2 -2
- checkpoints/checkpoint-2500/optimizer.pt +2 -2
- checkpoints/checkpoint-2500/trainer_state.json +100 -100
- checkpoints/checkpoint-2500/training_args.bin +1 -1
- checkpoints/checkpoint-3000/config.json +2 -2
- checkpoints/checkpoint-3000/model.safetensors +2 -2
- checkpoints/checkpoint-3000/optimizer.pt +2 -2
- checkpoints/checkpoint-3000/trainer_state.json +120 -120
- checkpoints/checkpoint-3000/training_args.bin +1 -1
- checkpoints/checkpoint-3500/config.json +2 -2
- checkpoints/checkpoint-3500/model.safetensors +2 -2
- checkpoints/checkpoint-3500/optimizer.pt +2 -2
- checkpoints/checkpoint-3500/trainer_state.json +140 -140
- checkpoints/checkpoint-3500/training_args.bin +1 -1
- checkpoints/checkpoint-4000/config.json +2 -2
- checkpoints/checkpoint-4000/model.safetensors +2 -2
- checkpoints/checkpoint-4000/optimizer.pt +2 -2
- checkpoints/checkpoint-4000/trainer_state.json +160 -160
- checkpoints/checkpoint-4000/training_args.bin +1 -1
- checkpoints/checkpoint-4500/config.json +2 -2
- checkpoints/checkpoint-4500/model.safetensors +2 -2
- checkpoints/checkpoint-4500/optimizer.pt +2 -2
- checkpoints/checkpoint-4500/trainer_state.json +180 -180
- checkpoints/checkpoint-4500/training_args.bin +1 -1
- checkpoints/checkpoint-500/config.json +2 -2
- checkpoints/checkpoint-500/model.safetensors +2 -2
- checkpoints/checkpoint-500/optimizer.pt +2 -2
- checkpoints/checkpoint-500/trainer_state.json +20 -20
- checkpoints/checkpoint-500/training_args.bin +1 -1
- checkpoints/checkpoint-5000/config.json +2 -2
- checkpoints/checkpoint-5000/model.safetensors +2 -2
- checkpoints/checkpoint-5000/optimizer.pt +2 -2
- checkpoints/checkpoint-5000/trainer_state.json +200 -200
- checkpoints/checkpoint-5000/training_args.bin +1 -1
checkpoints/checkpoint-1000/config.json
CHANGED
@@ -10,12 +10,12 @@
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
-
"n_ctx":
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
-
"n_positions":
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
+
"n_ctx": 1024,
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
+
"n_positions": 1024,
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
checkpoints/checkpoint-1000/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c5ae997f2cf9eadb57922b94260884615421d73c0edf124fcadd583baa9960a6
|
3 |
+
size 176343496
|
checkpoints/checkpoint-1000/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b82b36c45fe5fc732e4631b0473258f619c92959baa3cbb2820eb493ff496b54
|
3 |
+
size 352735610
|
checkpoints/checkpoint-1000/trainer_state.json
CHANGED
@@ -11,142 +11,142 @@
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
-
"grad_norm": 2.
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
-
"loss": 6.
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
-
"grad_norm": 3.
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
-
"loss": 5.
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
-
"grad_norm": 4.
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
-
"loss": 5.
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
-
"grad_norm":
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
-
"loss": 5.
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
-
"grad_norm": 5.
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
-
"loss": 4.
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
-
"grad_norm": 5.
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
-
"loss": 4.
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
-
"grad_norm": 5.
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
-
"loss": 4.
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
-
"grad_norm": 5.
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
-
"loss": 4.
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
-
"grad_norm": 5.
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
-
"loss": 4.
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
-
"grad_norm": 5.
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
-
"loss": 4.
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
-
"grad_norm": 5.
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
-
"loss": 4.
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
-
"grad_norm": 5.
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
-
"loss": 4.
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
-
"grad_norm": 5.
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
-
"loss": 3.
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
-
"grad_norm": 5.
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
-
"loss": 3.
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
-
"grad_norm": 5.
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
-
"loss": 3.
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
-
"grad_norm": 5.
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
-
"loss": 3.
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
-
"grad_norm": 5.
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
-
"loss": 3.
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
-
"grad_norm":
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
-
"loss": 3.
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
-
"grad_norm": 5.
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
-
"loss": 3.
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
-
"grad_norm": 5.
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
-
"loss": 3.
|
150 |
"step": 1000
|
151 |
}
|
152 |
],
|
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
+
"grad_norm": 2.239194869995117,
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
+
"loss": 6.099,
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
+
"grad_norm": 3.8653194904327393,
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
+
"loss": 5.6954,
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
+
"grad_norm": 4.652886867523193,
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
+
"loss": 5.3598,
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
+
"grad_norm": 5.048999309539795,
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
+
"loss": 5.0982,
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
+
"grad_norm": 5.21944522857666,
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
+
"loss": 4.7456,
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
+
"grad_norm": 5.471007347106934,
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
+
"loss": 4.6254,
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
+
"grad_norm": 5.345245838165283,
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
+
"loss": 4.5216,
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
+
"grad_norm": 5.54531192779541,
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
+
"loss": 4.4353,
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
+
"grad_norm": 5.244582176208496,
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
+
"loss": 4.2444,
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
+
"grad_norm": 5.338261127471924,
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
+
"loss": 4.1915,
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
+
"grad_norm": 5.35228157043457,
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
+
"loss": 4.1612,
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
+
"grad_norm": 5.287161827087402,
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
+
"loss": 4.1322,
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
+
"grad_norm": 5.2932209968566895,
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
+
"loss": 3.9684,
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
+
"grad_norm": 5.4360833168029785,
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
+
"loss": 3.8933,
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
+
"grad_norm": 5.5595221519470215,
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
+
"loss": 3.9306,
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
+
"grad_norm": 5.540086269378662,
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
+
"loss": 3.9317,
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
+
"grad_norm": 5.567137718200684,
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
+
"loss": 3.8135,
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
+
"grad_norm": 6.00346040725708,
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
+
"loss": 3.6924,
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
+
"grad_norm": 5.635782718658447,
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
+
"loss": 3.7304,
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
+
"grad_norm": 5.9545440673828125,
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
+
"loss": 3.7471,
|
150 |
"step": 1000
|
151 |
}
|
152 |
],
|
checkpoints/checkpoint-1000/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7935e3b7629485cf26f624afe618b58a390d4c3c9c6dcf878b3ba9099ef1027b
|
3 |
size 5304
|
checkpoints/checkpoint-1500/config.json
CHANGED
@@ -10,12 +10,12 @@
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
-
"n_ctx":
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
-
"n_positions":
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
+
"n_ctx": 1024,
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
+
"n_positions": 1024,
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
checkpoints/checkpoint-1500/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:be299c4059db11d4c5f71f16a3deddb85120abd5b479bf8e8320316f7aae7e16
|
3 |
+
size 176343496
|
checkpoints/checkpoint-1500/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:295f43bb7dbf683a742daa0d0fbb0499cfe10fbb60f183a5953d842c37ce5de5
|
3 |
+
size 352735610
|
checkpoints/checkpoint-1500/trainer_state.json
CHANGED
@@ -11,212 +11,212 @@
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
-
"grad_norm": 2.
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
-
"loss": 6.
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
-
"grad_norm": 3.
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
-
"loss": 5.
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
-
"grad_norm": 4.
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
-
"loss": 5.
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
-
"grad_norm":
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
-
"loss": 5.
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
-
"grad_norm": 5.
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
-
"loss": 4.
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
-
"grad_norm": 5.
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
-
"loss": 4.
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
-
"grad_norm": 5.
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
-
"loss": 4.
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
-
"grad_norm": 5.
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
-
"loss": 4.
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
-
"grad_norm": 5.
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
-
"loss": 4.
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
-
"grad_norm": 5.
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
-
"loss": 4.
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
-
"grad_norm": 5.
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
-
"loss": 4.
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
-
"grad_norm": 5.
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
-
"loss": 4.
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
-
"grad_norm": 5.
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
-
"loss": 3.
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
-
"grad_norm": 5.
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
-
"loss": 3.
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
-
"grad_norm": 5.
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
-
"loss": 3.
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
-
"grad_norm": 5.
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
-
"loss": 3.
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
-
"grad_norm": 5.
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
-
"loss": 3.
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
-
"grad_norm":
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
-
"loss": 3.
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
-
"grad_norm": 5.
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
-
"loss": 3.
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
-
"grad_norm": 5.
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
-
"loss": 3.
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
-
"grad_norm": 5.
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
-
"loss": 3.
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
-
"grad_norm": 5.
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
-
"loss": 3.
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
-
"grad_norm": 6.
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
-
"loss": 3.
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
-
"grad_norm": 6.
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
-
"loss": 3.
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
-
"grad_norm": 6.
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
-
"loss": 3.
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
-
"grad_norm": 6.
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
-
"loss": 3.
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
-
"grad_norm": 6.
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
-
"loss": 3.
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
-
"grad_norm": 6.
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
-
"loss": 3.
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
-
"grad_norm": 6.
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
-
"loss": 3.
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
-
"grad_norm": 7.
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
-
"loss": 3.
|
220 |
"step": 1500
|
221 |
}
|
222 |
],
|
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
+
"grad_norm": 2.239194869995117,
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
+
"loss": 6.099,
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
+
"grad_norm": 3.8653194904327393,
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
+
"loss": 5.6954,
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
+
"grad_norm": 4.652886867523193,
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
+
"loss": 5.3598,
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
+
"grad_norm": 5.048999309539795,
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
+
"loss": 5.0982,
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
+
"grad_norm": 5.21944522857666,
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
+
"loss": 4.7456,
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
+
"grad_norm": 5.471007347106934,
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
+
"loss": 4.6254,
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
+
"grad_norm": 5.345245838165283,
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
+
"loss": 4.5216,
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
+
"grad_norm": 5.54531192779541,
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
+
"loss": 4.4353,
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
+
"grad_norm": 5.244582176208496,
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
+
"loss": 4.2444,
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
+
"grad_norm": 5.338261127471924,
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
+
"loss": 4.1915,
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
+
"grad_norm": 5.35228157043457,
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
+
"loss": 4.1612,
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
+
"grad_norm": 5.287161827087402,
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
+
"loss": 4.1322,
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
+
"grad_norm": 5.2932209968566895,
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
+
"loss": 3.9684,
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
+
"grad_norm": 5.4360833168029785,
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
+
"loss": 3.8933,
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
+
"grad_norm": 5.5595221519470215,
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
+
"loss": 3.9306,
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
+
"grad_norm": 5.540086269378662,
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
+
"loss": 3.9317,
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
+
"grad_norm": 5.567137718200684,
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
+
"loss": 3.8135,
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
+
"grad_norm": 6.00346040725708,
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
+
"loss": 3.6924,
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
+
"grad_norm": 5.635782718658447,
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
+
"loss": 3.7304,
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
+
"grad_norm": 5.9545440673828125,
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
+
"loss": 3.7471,
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
+
"grad_norm": 5.696166038513184,
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
+
"loss": 3.6287,
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
+
"grad_norm": 5.844620704650879,
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
+
"loss": 3.5299,
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
+
"grad_norm": 6.373577117919922,
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
+
"loss": 3.5285,
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
+
"grad_norm": 6.488651275634766,
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
+
"loss": 3.5789,
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
+
"grad_norm": 6.170980930328369,
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
+
"loss": 3.494,
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
+
"grad_norm": 6.387782573699951,
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
+
"loss": 3.353,
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
+
"grad_norm": 6.718175888061523,
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
+
"loss": 3.3521,
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
+
"grad_norm": 6.7870025634765625,
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
+
"loss": 3.3658,
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
+
"grad_norm": 6.513721466064453,
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
+
"loss": 3.3301,
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
+
"grad_norm": 7.0680036544799805,
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
+
"loss": 3.1289,
|
220 |
"step": 1500
|
221 |
}
|
222 |
],
|
checkpoints/checkpoint-1500/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7935e3b7629485cf26f624afe618b58a390d4c3c9c6dcf878b3ba9099ef1027b
|
3 |
size 5304
|
checkpoints/checkpoint-2000/config.json
CHANGED
@@ -10,12 +10,12 @@
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
-
"n_ctx":
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
-
"n_positions":
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
+
"n_ctx": 1024,
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
+
"n_positions": 1024,
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
checkpoints/checkpoint-2000/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:31f153b730bb5ccd728c40ab63c17b596f9e9ae82c15275165a9eeb69cc79bf0
|
3 |
+
size 176343496
|
checkpoints/checkpoint-2000/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7d2690494d465e208982d5145f8448519f3fe6f6f92a6f6306bf06891d5d5b7c
|
3 |
+
size 352735610
|
checkpoints/checkpoint-2000/trainer_state.json
CHANGED
@@ -11,282 +11,282 @@
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
-
"grad_norm": 2.
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
-
"loss": 6.
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
-
"grad_norm": 3.
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
-
"loss": 5.
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
-
"grad_norm": 4.
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
-
"loss": 5.
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
-
"grad_norm":
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
-
"loss": 5.
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
-
"grad_norm": 5.
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
-
"loss": 4.
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
-
"grad_norm": 5.
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
-
"loss": 4.
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
-
"grad_norm": 5.
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
-
"loss": 4.
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
-
"grad_norm": 5.
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
-
"loss": 4.
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
-
"grad_norm": 5.
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
-
"loss": 4.
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
-
"grad_norm": 5.
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
-
"loss": 4.
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
-
"grad_norm": 5.
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
-
"loss": 4.
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
-
"grad_norm": 5.
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
-
"loss": 4.
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
-
"grad_norm": 5.
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
-
"loss": 3.
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
-
"grad_norm": 5.
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
-
"loss": 3.
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
-
"grad_norm": 5.
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
-
"loss": 3.
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
-
"grad_norm": 5.
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
-
"loss": 3.
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
-
"grad_norm": 5.
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
-
"loss": 3.
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
-
"grad_norm":
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
-
"loss": 3.
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
-
"grad_norm": 5.
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
-
"loss": 3.
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
-
"grad_norm": 5.
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
-
"loss": 3.
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
-
"grad_norm": 5.
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
-
"loss": 3.
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
-
"grad_norm": 5.
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
-
"loss": 3.
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
-
"grad_norm": 6.
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
-
"loss": 3.
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
-
"grad_norm": 6.
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
-
"loss": 3.
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
-
"grad_norm": 6.
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
-
"loss": 3.
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
-
"grad_norm": 6.
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
-
"loss": 3.
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
-
"grad_norm": 6.
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
-
"loss": 3.
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
-
"grad_norm": 6.
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
-
"loss": 3.
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
-
"grad_norm": 6.
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
-
"loss": 3.
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
-
"grad_norm": 7.
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
-
"loss": 3.
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
-
"grad_norm": 7.
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
-
"loss": 3.
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
-
"grad_norm": 7.
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
-
"loss": 3.
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
-
"grad_norm":
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
-
"loss": 3.
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
-
"grad_norm": 7.
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
-
"loss": 2.
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
-
"grad_norm": 7.
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
-
"loss":
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
-
"grad_norm": 8.
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
-
"loss": 3.
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
-
"grad_norm":
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
-
"loss": 2.
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
-
"grad_norm": 7.
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
-
"loss": 2.
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
-
"grad_norm":
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
-
"loss": 2.
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
-
"grad_norm": 8.
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
-
"loss": 2.
|
290 |
"step": 2000
|
291 |
}
|
292 |
],
|
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
+
"grad_norm": 2.239194869995117,
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
+
"loss": 6.099,
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
+
"grad_norm": 3.8653194904327393,
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
+
"loss": 5.6954,
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
+
"grad_norm": 4.652886867523193,
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
+
"loss": 5.3598,
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
+
"grad_norm": 5.048999309539795,
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
+
"loss": 5.0982,
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
+
"grad_norm": 5.21944522857666,
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
+
"loss": 4.7456,
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
+
"grad_norm": 5.471007347106934,
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
+
"loss": 4.6254,
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
+
"grad_norm": 5.345245838165283,
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
+
"loss": 4.5216,
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
+
"grad_norm": 5.54531192779541,
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
+
"loss": 4.4353,
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
+
"grad_norm": 5.244582176208496,
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
+
"loss": 4.2444,
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
+
"grad_norm": 5.338261127471924,
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
+
"loss": 4.1915,
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
+
"grad_norm": 5.35228157043457,
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
+
"loss": 4.1612,
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
+
"grad_norm": 5.287161827087402,
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
+
"loss": 4.1322,
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
+
"grad_norm": 5.2932209968566895,
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
+
"loss": 3.9684,
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
+
"grad_norm": 5.4360833168029785,
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
+
"loss": 3.8933,
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
+
"grad_norm": 5.5595221519470215,
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
+
"loss": 3.9306,
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
+
"grad_norm": 5.540086269378662,
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
+
"loss": 3.9317,
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
+
"grad_norm": 5.567137718200684,
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
+
"loss": 3.8135,
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
+
"grad_norm": 6.00346040725708,
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
+
"loss": 3.6924,
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
+
"grad_norm": 5.635782718658447,
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
+
"loss": 3.7304,
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
+
"grad_norm": 5.9545440673828125,
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
+
"loss": 3.7471,
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
+
"grad_norm": 5.696166038513184,
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
+
"loss": 3.6287,
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
+
"grad_norm": 5.844620704650879,
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
+
"loss": 3.5299,
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
+
"grad_norm": 6.373577117919922,
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
+
"loss": 3.5285,
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
+
"grad_norm": 6.488651275634766,
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
+
"loss": 3.5789,
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
+
"grad_norm": 6.170980930328369,
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
+
"loss": 3.494,
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
+
"grad_norm": 6.387782573699951,
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
+
"loss": 3.353,
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
+
"grad_norm": 6.718175888061523,
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
+
"loss": 3.3521,
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
+
"grad_norm": 6.7870025634765625,
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
+
"loss": 3.3658,
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
+
"grad_norm": 6.513721466064453,
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
+
"loss": 3.3301,
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
+
"grad_norm": 7.0680036544799805,
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
+
"loss": 3.1289,
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
+
"grad_norm": 7.019896507263184,
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
+
"loss": 3.1878,
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
+
"grad_norm": 7.3232197761535645,
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
+
"loss": 3.2006,
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
+
"grad_norm": 7.003912925720215,
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
+
"loss": 3.1737,
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
+
"grad_norm": 7.592433929443359,
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
+
"loss": 2.9258,
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
+
"grad_norm": 7.629278182983398,
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
+
"loss": 2.9999,
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
+
"grad_norm": 8.137636184692383,
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
+
"loss": 3.0155,
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
+
"grad_norm": 7.967227935791016,
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
+
"loss": 2.9945,
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
+
"grad_norm": 7.9910078048706055,
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
+
"loss": 2.7456,
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
+
"grad_norm": 8.552720069885254,
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
+
"loss": 2.8071,
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
+
"grad_norm": 8.336930274963379,
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
+
"loss": 2.8117,
|
290 |
"step": 2000
|
291 |
}
|
292 |
],
|
checkpoints/checkpoint-2000/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7935e3b7629485cf26f624afe618b58a390d4c3c9c6dcf878b3ba9099ef1027b
|
3 |
size 5304
|
checkpoints/checkpoint-2500/config.json
CHANGED
@@ -10,12 +10,12 @@
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
-
"n_ctx":
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
-
"n_positions":
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
+
"n_ctx": 1024,
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
+
"n_positions": 1024,
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
checkpoints/checkpoint-2500/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1cf1426568d31687412e315e1e4bfca12c7d5f958e2e1e9165ffa0db03a7554b
|
3 |
+
size 176343496
|
checkpoints/checkpoint-2500/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4d131c0a3b1896514cda171b1698f8ad491e48fe541af53ae5dc2ed7bff06444
|
3 |
+
size 352735610
|
checkpoints/checkpoint-2500/trainer_state.json
CHANGED
@@ -11,352 +11,352 @@
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
-
"grad_norm": 2.
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
-
"loss": 6.
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
-
"grad_norm": 3.
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
-
"loss": 5.
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
-
"grad_norm": 4.
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
-
"loss": 5.
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
-
"grad_norm":
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
-
"loss": 5.
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
-
"grad_norm": 5.
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
-
"loss": 4.
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
-
"grad_norm": 5.
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
-
"loss": 4.
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
-
"grad_norm": 5.
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
-
"loss": 4.
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
-
"grad_norm": 5.
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
-
"loss": 4.
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
-
"grad_norm": 5.
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
-
"loss": 4.
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
-
"grad_norm": 5.
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
-
"loss": 4.
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
-
"grad_norm": 5.
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
-
"loss": 4.
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
-
"grad_norm": 5.
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
-
"loss": 4.
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
-
"grad_norm": 5.
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
-
"loss": 3.
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
-
"grad_norm": 5.
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
-
"loss": 3.
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
-
"grad_norm": 5.
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
-
"loss": 3.
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
-
"grad_norm": 5.
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
-
"loss": 3.
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
-
"grad_norm": 5.
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
-
"loss": 3.
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
-
"grad_norm":
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
-
"loss": 3.
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
-
"grad_norm": 5.
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
-
"loss": 3.
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
-
"grad_norm": 5.
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
-
"loss": 3.
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
-
"grad_norm": 5.
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
-
"loss": 3.
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
-
"grad_norm": 5.
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
-
"loss": 3.
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
-
"grad_norm": 6.
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
-
"loss": 3.
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
-
"grad_norm": 6.
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
-
"loss": 3.
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
-
"grad_norm": 6.
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
-
"loss": 3.
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
-
"grad_norm": 6.
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
-
"loss": 3.
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
-
"grad_norm": 6.
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
-
"loss": 3.
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
-
"grad_norm": 6.
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
-
"loss": 3.
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
-
"grad_norm": 6.
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
-
"loss": 3.
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
-
"grad_norm": 7.
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
-
"loss": 3.
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
-
"grad_norm": 7.
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
-
"loss": 3.
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
-
"grad_norm": 7.
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
-
"loss": 3.
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
-
"grad_norm":
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
-
"loss": 3.
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
-
"grad_norm": 7.
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
-
"loss": 2.
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
-
"grad_norm": 7.
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
-
"loss":
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
-
"grad_norm": 8.
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
-
"loss": 3.
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
-
"grad_norm":
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
-
"loss": 2.
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
-
"grad_norm": 7.
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
-
"loss": 2.
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
-
"grad_norm":
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
-
"loss": 2.
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
-
"grad_norm": 8.
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
-
"loss": 2.
|
290 |
"step": 2000
|
291 |
},
|
292 |
{
|
293 |
"epoch": 9.951456310679612,
|
294 |
-
"grad_norm": 8.
|
295 |
"learning_rate": 3.342233009708738e-05,
|
296 |
-
"loss": 2.
|
297 |
"step": 2050
|
298 |
},
|
299 |
{
|
300 |
"epoch": 10.194174757281553,
|
301 |
-
"grad_norm":
|
302 |
"learning_rate": 3.301779935275081e-05,
|
303 |
-
"loss": 2.
|
304 |
"step": 2100
|
305 |
},
|
306 |
{
|
307 |
"epoch": 10.436893203883495,
|
308 |
-
"grad_norm":
|
309 |
"learning_rate": 3.2613268608414246e-05,
|
310 |
-
"loss": 2.
|
311 |
"step": 2150
|
312 |
},
|
313 |
{
|
314 |
"epoch": 10.679611650485437,
|
315 |
-
"grad_norm": 9.
|
316 |
"learning_rate": 3.220873786407767e-05,
|
317 |
-
"loss": 2.
|
318 |
"step": 2200
|
319 |
},
|
320 |
{
|
321 |
"epoch": 10.922330097087379,
|
322 |
-
"grad_norm": 9.
|
323 |
"learning_rate": 3.1804207119741106e-05,
|
324 |
-
"loss": 2.
|
325 |
"step": 2250
|
326 |
},
|
327 |
{
|
328 |
"epoch": 11.16504854368932,
|
329 |
-
"grad_norm": 8.
|
330 |
"learning_rate": 3.139967637540453e-05,
|
331 |
-
"loss": 2.
|
332 |
"step": 2300
|
333 |
},
|
334 |
{
|
335 |
"epoch": 11.407766990291263,
|
336 |
-
"grad_norm": 9.
|
337 |
"learning_rate": 3.099514563106796e-05,
|
338 |
-
"loss": 2.
|
339 |
"step": 2350
|
340 |
},
|
341 |
{
|
342 |
"epoch": 11.650485436893204,
|
343 |
-
"grad_norm": 9.
|
344 |
"learning_rate": 3.059061488673139e-05,
|
345 |
-
"loss": 2.
|
346 |
"step": 2400
|
347 |
},
|
348 |
{
|
349 |
"epoch": 11.893203883495145,
|
350 |
-
"grad_norm": 9.
|
351 |
"learning_rate": 3.0186084142394822e-05,
|
352 |
-
"loss": 2.
|
353 |
"step": 2450
|
354 |
},
|
355 |
{
|
356 |
"epoch": 12.135922330097088,
|
357 |
-
"grad_norm": 9.
|
358 |
"learning_rate": 2.9781553398058252e-05,
|
359 |
-
"loss": 2.
|
360 |
"step": 2500
|
361 |
}
|
362 |
],
|
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
+
"grad_norm": 2.239194869995117,
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
+
"loss": 6.099,
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
+
"grad_norm": 3.8653194904327393,
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
+
"loss": 5.6954,
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
+
"grad_norm": 4.652886867523193,
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
+
"loss": 5.3598,
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
+
"grad_norm": 5.048999309539795,
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
+
"loss": 5.0982,
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
+
"grad_norm": 5.21944522857666,
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
+
"loss": 4.7456,
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
+
"grad_norm": 5.471007347106934,
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
+
"loss": 4.6254,
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
+
"grad_norm": 5.345245838165283,
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
+
"loss": 4.5216,
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
+
"grad_norm": 5.54531192779541,
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
+
"loss": 4.4353,
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
+
"grad_norm": 5.244582176208496,
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
+
"loss": 4.2444,
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
+
"grad_norm": 5.338261127471924,
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
+
"loss": 4.1915,
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
+
"grad_norm": 5.35228157043457,
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
+
"loss": 4.1612,
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
+
"grad_norm": 5.287161827087402,
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
+
"loss": 4.1322,
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
+
"grad_norm": 5.2932209968566895,
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
+
"loss": 3.9684,
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
+
"grad_norm": 5.4360833168029785,
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
+
"loss": 3.8933,
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
+
"grad_norm": 5.5595221519470215,
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
+
"loss": 3.9306,
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
+
"grad_norm": 5.540086269378662,
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
+
"loss": 3.9317,
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
+
"grad_norm": 5.567137718200684,
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
+
"loss": 3.8135,
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
+
"grad_norm": 6.00346040725708,
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
+
"loss": 3.6924,
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
+
"grad_norm": 5.635782718658447,
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
+
"loss": 3.7304,
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
+
"grad_norm": 5.9545440673828125,
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
+
"loss": 3.7471,
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
+
"grad_norm": 5.696166038513184,
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
+
"loss": 3.6287,
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
+
"grad_norm": 5.844620704650879,
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
+
"loss": 3.5299,
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
+
"grad_norm": 6.373577117919922,
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
+
"loss": 3.5285,
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
+
"grad_norm": 6.488651275634766,
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
+
"loss": 3.5789,
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
+
"grad_norm": 6.170980930328369,
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
+
"loss": 3.494,
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
+
"grad_norm": 6.387782573699951,
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
+
"loss": 3.353,
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
+
"grad_norm": 6.718175888061523,
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
+
"loss": 3.3521,
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
+
"grad_norm": 6.7870025634765625,
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
+
"loss": 3.3658,
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
+
"grad_norm": 6.513721466064453,
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
+
"loss": 3.3301,
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
+
"grad_norm": 7.0680036544799805,
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
+
"loss": 3.1289,
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
+
"grad_norm": 7.019896507263184,
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
+
"loss": 3.1878,
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
+
"grad_norm": 7.3232197761535645,
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
+
"loss": 3.2006,
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
+
"grad_norm": 7.003912925720215,
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
+
"loss": 3.1737,
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
+
"grad_norm": 7.592433929443359,
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
+
"loss": 2.9258,
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
+
"grad_norm": 7.629278182983398,
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
+
"loss": 2.9999,
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
+
"grad_norm": 8.137636184692383,
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
+
"loss": 3.0155,
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
+
"grad_norm": 7.967227935791016,
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
+
"loss": 2.9945,
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
+
"grad_norm": 7.9910078048706055,
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
+
"loss": 2.7456,
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
+
"grad_norm": 8.552720069885254,
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
+
"loss": 2.8071,
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
+
"grad_norm": 8.336930274963379,
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
+
"loss": 2.8117,
|
290 |
"step": 2000
|
291 |
},
|
292 |
{
|
293 |
"epoch": 9.951456310679612,
|
294 |
+
"grad_norm": 8.636338233947754,
|
295 |
"learning_rate": 3.342233009708738e-05,
|
296 |
+
"loss": 2.8197,
|
297 |
"step": 2050
|
298 |
},
|
299 |
{
|
300 |
"epoch": 10.194174757281553,
|
301 |
+
"grad_norm": 7.955165863037109,
|
302 |
"learning_rate": 3.301779935275081e-05,
|
303 |
+
"loss": 2.6245,
|
304 |
"step": 2100
|
305 |
},
|
306 |
{
|
307 |
"epoch": 10.436893203883495,
|
308 |
+
"grad_norm": 8.679149627685547,
|
309 |
"learning_rate": 3.2613268608414246e-05,
|
310 |
+
"loss": 2.565,
|
311 |
"step": 2150
|
312 |
},
|
313 |
{
|
314 |
"epoch": 10.679611650485437,
|
315 |
+
"grad_norm": 9.150839805603027,
|
316 |
"learning_rate": 3.220873786407767e-05,
|
317 |
+
"loss": 2.6258,
|
318 |
"step": 2200
|
319 |
},
|
320 |
{
|
321 |
"epoch": 10.922330097087379,
|
322 |
+
"grad_norm": 9.2735013961792,
|
323 |
"learning_rate": 3.1804207119741106e-05,
|
324 |
+
"loss": 2.6439,
|
325 |
"step": 2250
|
326 |
},
|
327 |
{
|
328 |
"epoch": 11.16504854368932,
|
329 |
+
"grad_norm": 8.620939254760742,
|
330 |
"learning_rate": 3.139967637540453e-05,
|
331 |
+
"loss": 2.4371,
|
332 |
"step": 2300
|
333 |
},
|
334 |
{
|
335 |
"epoch": 11.407766990291263,
|
336 |
+
"grad_norm": 9.722040176391602,
|
337 |
"learning_rate": 3.099514563106796e-05,
|
338 |
+
"loss": 2.4074,
|
339 |
"step": 2350
|
340 |
},
|
341 |
{
|
342 |
"epoch": 11.650485436893204,
|
343 |
+
"grad_norm": 9.200730323791504,
|
344 |
"learning_rate": 3.059061488673139e-05,
|
345 |
+
"loss": 2.4307,
|
346 |
"step": 2400
|
347 |
},
|
348 |
{
|
349 |
"epoch": 11.893203883495145,
|
350 |
+
"grad_norm": 9.692273139953613,
|
351 |
"learning_rate": 3.0186084142394822e-05,
|
352 |
+
"loss": 2.424,
|
353 |
"step": 2450
|
354 |
},
|
355 |
{
|
356 |
"epoch": 12.135922330097088,
|
357 |
+
"grad_norm": 9.822249412536621,
|
358 |
"learning_rate": 2.9781553398058252e-05,
|
359 |
+
"loss": 2.2862,
|
360 |
"step": 2500
|
361 |
}
|
362 |
],
|
checkpoints/checkpoint-2500/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7935e3b7629485cf26f624afe618b58a390d4c3c9c6dcf878b3ba9099ef1027b
|
3 |
size 5304
|
checkpoints/checkpoint-3000/config.json
CHANGED
@@ -10,12 +10,12 @@
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
-
"n_ctx":
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
-
"n_positions":
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
+
"n_ctx": 1024,
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
+
"n_positions": 1024,
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
checkpoints/checkpoint-3000/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ee9950fb5c86bae073ca52771961c408c692f4811dffc7d19b51bc4ce3680cc7
|
3 |
+
size 176343496
|
checkpoints/checkpoint-3000/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e038dba515e8f39fe285658df720ade045fb28c8083d1b685733be9fc873618d
|
3 |
+
size 352735610
|
checkpoints/checkpoint-3000/trainer_state.json
CHANGED
@@ -11,422 +11,422 @@
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
-
"grad_norm": 2.
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
-
"loss": 6.
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
-
"grad_norm": 3.
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
-
"loss": 5.
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
-
"grad_norm": 4.
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
-
"loss": 5.
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
-
"grad_norm":
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
-
"loss": 5.
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
-
"grad_norm": 5.
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
-
"loss": 4.
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
-
"grad_norm": 5.
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
-
"loss": 4.
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
-
"grad_norm": 5.
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
-
"loss": 4.
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
-
"grad_norm": 5.
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
-
"loss": 4.
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
-
"grad_norm": 5.
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
-
"loss": 4.
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
-
"grad_norm": 5.
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
-
"loss": 4.
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
-
"grad_norm": 5.
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
-
"loss": 4.
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
-
"grad_norm": 5.
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
-
"loss": 4.
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
-
"grad_norm": 5.
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
-
"loss": 3.
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
-
"grad_norm": 5.
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
-
"loss": 3.
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
-
"grad_norm": 5.
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
-
"loss": 3.
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
-
"grad_norm": 5.
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
-
"loss": 3.
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
-
"grad_norm": 5.
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
-
"loss": 3.
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
-
"grad_norm":
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
-
"loss": 3.
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
-
"grad_norm": 5.
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
-
"loss": 3.
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
-
"grad_norm": 5.
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
-
"loss": 3.
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
-
"grad_norm": 5.
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
-
"loss": 3.
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
-
"grad_norm": 5.
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
-
"loss": 3.
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
-
"grad_norm": 6.
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
-
"loss": 3.
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
-
"grad_norm": 6.
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
-
"loss": 3.
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
-
"grad_norm": 6.
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
-
"loss": 3.
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
-
"grad_norm": 6.
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
-
"loss": 3.
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
-
"grad_norm": 6.
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
-
"loss": 3.
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
-
"grad_norm": 6.
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
-
"loss": 3.
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
-
"grad_norm": 6.
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
-
"loss": 3.
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
-
"grad_norm": 7.
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
-
"loss": 3.
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
-
"grad_norm": 7.
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
-
"loss": 3.
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
-
"grad_norm": 7.
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
-
"loss": 3.
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
-
"grad_norm":
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
-
"loss": 3.
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
-
"grad_norm": 7.
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
-
"loss": 2.
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
-
"grad_norm": 7.
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
-
"loss":
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
-
"grad_norm": 8.
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
-
"loss": 3.
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
-
"grad_norm":
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
-
"loss": 2.
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
-
"grad_norm": 7.
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
-
"loss": 2.
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
-
"grad_norm":
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
-
"loss": 2.
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
-
"grad_norm": 8.
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
-
"loss": 2.
|
290 |
"step": 2000
|
291 |
},
|
292 |
{
|
293 |
"epoch": 9.951456310679612,
|
294 |
-
"grad_norm": 8.
|
295 |
"learning_rate": 3.342233009708738e-05,
|
296 |
-
"loss": 2.
|
297 |
"step": 2050
|
298 |
},
|
299 |
{
|
300 |
"epoch": 10.194174757281553,
|
301 |
-
"grad_norm":
|
302 |
"learning_rate": 3.301779935275081e-05,
|
303 |
-
"loss": 2.
|
304 |
"step": 2100
|
305 |
},
|
306 |
{
|
307 |
"epoch": 10.436893203883495,
|
308 |
-
"grad_norm":
|
309 |
"learning_rate": 3.2613268608414246e-05,
|
310 |
-
"loss": 2.
|
311 |
"step": 2150
|
312 |
},
|
313 |
{
|
314 |
"epoch": 10.679611650485437,
|
315 |
-
"grad_norm": 9.
|
316 |
"learning_rate": 3.220873786407767e-05,
|
317 |
-
"loss": 2.
|
318 |
"step": 2200
|
319 |
},
|
320 |
{
|
321 |
"epoch": 10.922330097087379,
|
322 |
-
"grad_norm": 9.
|
323 |
"learning_rate": 3.1804207119741106e-05,
|
324 |
-
"loss": 2.
|
325 |
"step": 2250
|
326 |
},
|
327 |
{
|
328 |
"epoch": 11.16504854368932,
|
329 |
-
"grad_norm": 8.
|
330 |
"learning_rate": 3.139967637540453e-05,
|
331 |
-
"loss": 2.
|
332 |
"step": 2300
|
333 |
},
|
334 |
{
|
335 |
"epoch": 11.407766990291263,
|
336 |
-
"grad_norm": 9.
|
337 |
"learning_rate": 3.099514563106796e-05,
|
338 |
-
"loss": 2.
|
339 |
"step": 2350
|
340 |
},
|
341 |
{
|
342 |
"epoch": 11.650485436893204,
|
343 |
-
"grad_norm": 9.
|
344 |
"learning_rate": 3.059061488673139e-05,
|
345 |
-
"loss": 2.
|
346 |
"step": 2400
|
347 |
},
|
348 |
{
|
349 |
"epoch": 11.893203883495145,
|
350 |
-
"grad_norm": 9.
|
351 |
"learning_rate": 3.0186084142394822e-05,
|
352 |
-
"loss": 2.
|
353 |
"step": 2450
|
354 |
},
|
355 |
{
|
356 |
"epoch": 12.135922330097088,
|
357 |
-
"grad_norm": 9.
|
358 |
"learning_rate": 2.9781553398058252e-05,
|
359 |
-
"loss": 2.
|
360 |
"step": 2500
|
361 |
},
|
362 |
{
|
363 |
"epoch": 12.37864077669903,
|
364 |
-
"grad_norm":
|
365 |
"learning_rate": 2.9377022653721686e-05,
|
366 |
-
"loss": 2.
|
367 |
"step": 2550
|
368 |
},
|
369 |
{
|
370 |
"epoch": 12.62135922330097,
|
371 |
-
"grad_norm": 9.
|
372 |
"learning_rate": 2.8972491909385112e-05,
|
373 |
-
"loss": 2.
|
374 |
"step": 2600
|
375 |
},
|
376 |
{
|
377 |
"epoch": 12.864077669902912,
|
378 |
-
"grad_norm": 10.
|
379 |
"learning_rate": 2.8567961165048546e-05,
|
380 |
-
"loss": 2.
|
381 |
"step": 2650
|
382 |
},
|
383 |
{
|
384 |
"epoch": 13.106796116504855,
|
385 |
-
"grad_norm": 10.
|
386 |
"learning_rate": 2.816343042071198e-05,
|
387 |
-
"loss": 2.
|
388 |
"step": 2700
|
389 |
},
|
390 |
{
|
391 |
"epoch": 13.349514563106796,
|
392 |
-
"grad_norm":
|
393 |
"learning_rate": 2.7758899676375405e-05,
|
394 |
-
"loss": 2.
|
395 |
"step": 2750
|
396 |
},
|
397 |
{
|
398 |
"epoch": 13.592233009708737,
|
399 |
-
"grad_norm":
|
400 |
"learning_rate": 2.735436893203884e-05,
|
401 |
-
"loss": 2.
|
402 |
"step": 2800
|
403 |
},
|
404 |
{
|
405 |
"epoch": 13.83495145631068,
|
406 |
-
"grad_norm":
|
407 |
"learning_rate": 2.6949838187702265e-05,
|
408 |
-
"loss": 2.
|
409 |
"step": 2850
|
410 |
},
|
411 |
{
|
412 |
"epoch": 14.077669902912621,
|
413 |
-
"grad_norm": 9.
|
414 |
"learning_rate": 2.6545307443365695e-05,
|
415 |
-
"loss":
|
416 |
"step": 2900
|
417 |
},
|
418 |
{
|
419 |
"epoch": 14.320388349514563,
|
420 |
-
"grad_norm": 10.
|
421 |
"learning_rate": 2.614077669902913e-05,
|
422 |
-
"loss": 1.
|
423 |
"step": 2950
|
424 |
},
|
425 |
{
|
426 |
"epoch": 14.563106796116505,
|
427 |
-
"grad_norm": 10.
|
428 |
"learning_rate": 2.5736245954692555e-05,
|
429 |
-
"loss": 1.
|
430 |
"step": 3000
|
431 |
}
|
432 |
],
|
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
+
"grad_norm": 2.239194869995117,
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
+
"loss": 6.099,
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
+
"grad_norm": 3.8653194904327393,
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
+
"loss": 5.6954,
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
+
"grad_norm": 4.652886867523193,
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
+
"loss": 5.3598,
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
+
"grad_norm": 5.048999309539795,
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
+
"loss": 5.0982,
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
+
"grad_norm": 5.21944522857666,
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
+
"loss": 4.7456,
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
+
"grad_norm": 5.471007347106934,
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
+
"loss": 4.6254,
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
+
"grad_norm": 5.345245838165283,
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
+
"loss": 4.5216,
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
+
"grad_norm": 5.54531192779541,
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
+
"loss": 4.4353,
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
+
"grad_norm": 5.244582176208496,
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
+
"loss": 4.2444,
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
+
"grad_norm": 5.338261127471924,
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
+
"loss": 4.1915,
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
+
"grad_norm": 5.35228157043457,
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
+
"loss": 4.1612,
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
+
"grad_norm": 5.287161827087402,
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
+
"loss": 4.1322,
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
+
"grad_norm": 5.2932209968566895,
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
+
"loss": 3.9684,
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
+
"grad_norm": 5.4360833168029785,
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
+
"loss": 3.8933,
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
+
"grad_norm": 5.5595221519470215,
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
+
"loss": 3.9306,
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
+
"grad_norm": 5.540086269378662,
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
+
"loss": 3.9317,
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
+
"grad_norm": 5.567137718200684,
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
+
"loss": 3.8135,
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
+
"grad_norm": 6.00346040725708,
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
+
"loss": 3.6924,
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
+
"grad_norm": 5.635782718658447,
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
+
"loss": 3.7304,
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
+
"grad_norm": 5.9545440673828125,
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
+
"loss": 3.7471,
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
+
"grad_norm": 5.696166038513184,
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
+
"loss": 3.6287,
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
+
"grad_norm": 5.844620704650879,
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
+
"loss": 3.5299,
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
+
"grad_norm": 6.373577117919922,
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
+
"loss": 3.5285,
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
+
"grad_norm": 6.488651275634766,
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
+
"loss": 3.5789,
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
+
"grad_norm": 6.170980930328369,
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
+
"loss": 3.494,
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
+
"grad_norm": 6.387782573699951,
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
+
"loss": 3.353,
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
+
"grad_norm": 6.718175888061523,
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
+
"loss": 3.3521,
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
+
"grad_norm": 6.7870025634765625,
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
+
"loss": 3.3658,
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
+
"grad_norm": 6.513721466064453,
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
+
"loss": 3.3301,
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
+
"grad_norm": 7.0680036544799805,
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
+
"loss": 3.1289,
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
+
"grad_norm": 7.019896507263184,
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
+
"loss": 3.1878,
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
+
"grad_norm": 7.3232197761535645,
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
+
"loss": 3.2006,
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
+
"grad_norm": 7.003912925720215,
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
+
"loss": 3.1737,
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
+
"grad_norm": 7.592433929443359,
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
+
"loss": 2.9258,
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
+
"grad_norm": 7.629278182983398,
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
+
"loss": 2.9999,
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
+
"grad_norm": 8.137636184692383,
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
+
"loss": 3.0155,
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
+
"grad_norm": 7.967227935791016,
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
+
"loss": 2.9945,
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
+
"grad_norm": 7.9910078048706055,
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
+
"loss": 2.7456,
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
+
"grad_norm": 8.552720069885254,
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
+
"loss": 2.8071,
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
+
"grad_norm": 8.336930274963379,
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
+
"loss": 2.8117,
|
290 |
"step": 2000
|
291 |
},
|
292 |
{
|
293 |
"epoch": 9.951456310679612,
|
294 |
+
"grad_norm": 8.636338233947754,
|
295 |
"learning_rate": 3.342233009708738e-05,
|
296 |
+
"loss": 2.8197,
|
297 |
"step": 2050
|
298 |
},
|
299 |
{
|
300 |
"epoch": 10.194174757281553,
|
301 |
+
"grad_norm": 7.955165863037109,
|
302 |
"learning_rate": 3.301779935275081e-05,
|
303 |
+
"loss": 2.6245,
|
304 |
"step": 2100
|
305 |
},
|
306 |
{
|
307 |
"epoch": 10.436893203883495,
|
308 |
+
"grad_norm": 8.679149627685547,
|
309 |
"learning_rate": 3.2613268608414246e-05,
|
310 |
+
"loss": 2.565,
|
311 |
"step": 2150
|
312 |
},
|
313 |
{
|
314 |
"epoch": 10.679611650485437,
|
315 |
+
"grad_norm": 9.150839805603027,
|
316 |
"learning_rate": 3.220873786407767e-05,
|
317 |
+
"loss": 2.6258,
|
318 |
"step": 2200
|
319 |
},
|
320 |
{
|
321 |
"epoch": 10.922330097087379,
|
322 |
+
"grad_norm": 9.2735013961792,
|
323 |
"learning_rate": 3.1804207119741106e-05,
|
324 |
+
"loss": 2.6439,
|
325 |
"step": 2250
|
326 |
},
|
327 |
{
|
328 |
"epoch": 11.16504854368932,
|
329 |
+
"grad_norm": 8.620939254760742,
|
330 |
"learning_rate": 3.139967637540453e-05,
|
331 |
+
"loss": 2.4371,
|
332 |
"step": 2300
|
333 |
},
|
334 |
{
|
335 |
"epoch": 11.407766990291263,
|
336 |
+
"grad_norm": 9.722040176391602,
|
337 |
"learning_rate": 3.099514563106796e-05,
|
338 |
+
"loss": 2.4074,
|
339 |
"step": 2350
|
340 |
},
|
341 |
{
|
342 |
"epoch": 11.650485436893204,
|
343 |
+
"grad_norm": 9.200730323791504,
|
344 |
"learning_rate": 3.059061488673139e-05,
|
345 |
+
"loss": 2.4307,
|
346 |
"step": 2400
|
347 |
},
|
348 |
{
|
349 |
"epoch": 11.893203883495145,
|
350 |
+
"grad_norm": 9.692273139953613,
|
351 |
"learning_rate": 3.0186084142394822e-05,
|
352 |
+
"loss": 2.424,
|
353 |
"step": 2450
|
354 |
},
|
355 |
{
|
356 |
"epoch": 12.135922330097088,
|
357 |
+
"grad_norm": 9.822249412536621,
|
358 |
"learning_rate": 2.9781553398058252e-05,
|
359 |
+
"loss": 2.2862,
|
360 |
"step": 2500
|
361 |
},
|
362 |
{
|
363 |
"epoch": 12.37864077669903,
|
364 |
+
"grad_norm": 9.919060707092285,
|
365 |
"learning_rate": 2.9377022653721686e-05,
|
366 |
+
"loss": 2.1913,
|
367 |
"step": 2550
|
368 |
},
|
369 |
{
|
370 |
"epoch": 12.62135922330097,
|
371 |
+
"grad_norm": 9.181950569152832,
|
372 |
"learning_rate": 2.8972491909385112e-05,
|
373 |
+
"loss": 2.241,
|
374 |
"step": 2600
|
375 |
},
|
376 |
{
|
377 |
"epoch": 12.864077669902912,
|
378 |
+
"grad_norm": 10.82077407836914,
|
379 |
"learning_rate": 2.8567961165048546e-05,
|
380 |
+
"loss": 2.2653,
|
381 |
"step": 2650
|
382 |
},
|
383 |
{
|
384 |
"epoch": 13.106796116504855,
|
385 |
+
"grad_norm": 10.365830421447754,
|
386 |
"learning_rate": 2.816343042071198e-05,
|
387 |
+
"loss": 2.1352,
|
388 |
"step": 2700
|
389 |
},
|
390 |
{
|
391 |
"epoch": 13.349514563106796,
|
392 |
+
"grad_norm": 10.073640823364258,
|
393 |
"learning_rate": 2.7758899676375405e-05,
|
394 |
+
"loss": 2.0227,
|
395 |
"step": 2750
|
396 |
},
|
397 |
{
|
398 |
"epoch": 13.592233009708737,
|
399 |
+
"grad_norm": 10.520242691040039,
|
400 |
"learning_rate": 2.735436893203884e-05,
|
401 |
+
"loss": 2.0579,
|
402 |
"step": 2800
|
403 |
},
|
404 |
{
|
405 |
"epoch": 13.83495145631068,
|
406 |
+
"grad_norm": 11.176478385925293,
|
407 |
"learning_rate": 2.6949838187702265e-05,
|
408 |
+
"loss": 2.0934,
|
409 |
"step": 2850
|
410 |
},
|
411 |
{
|
412 |
"epoch": 14.077669902912621,
|
413 |
+
"grad_norm": 9.541868209838867,
|
414 |
"learning_rate": 2.6545307443365695e-05,
|
415 |
+
"loss": 2.0118,
|
416 |
"step": 2900
|
417 |
},
|
418 |
{
|
419 |
"epoch": 14.320388349514563,
|
420 |
+
"grad_norm": 10.162981986999512,
|
421 |
"learning_rate": 2.614077669902913e-05,
|
422 |
+
"loss": 1.8624,
|
423 |
"step": 2950
|
424 |
},
|
425 |
{
|
426 |
"epoch": 14.563106796116505,
|
427 |
+
"grad_norm": 10.51291275024414,
|
428 |
"learning_rate": 2.5736245954692555e-05,
|
429 |
+
"loss": 1.8979,
|
430 |
"step": 3000
|
431 |
}
|
432 |
],
|
checkpoints/checkpoint-3000/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7935e3b7629485cf26f624afe618b58a390d4c3c9c6dcf878b3ba9099ef1027b
|
3 |
size 5304
|
checkpoints/checkpoint-3500/config.json
CHANGED
@@ -10,12 +10,12 @@
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
-
"n_ctx":
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
-
"n_positions":
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
+
"n_ctx": 1024,
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
+
"n_positions": 1024,
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
checkpoints/checkpoint-3500/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c86a085cb6768ba7ee32df0f83847ca1776cf0b157c33a638d029df90f359293
|
3 |
+
size 176343496
|
checkpoints/checkpoint-3500/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:41d7bf321a4eb94e1fe13d18a0e7d4c754699c0e788055516e6543148a689329
|
3 |
+
size 352735610
|
checkpoints/checkpoint-3500/trainer_state.json
CHANGED
@@ -11,492 +11,492 @@
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
-
"grad_norm": 2.
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
-
"loss": 6.
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
-
"grad_norm": 3.
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
-
"loss": 5.
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
-
"grad_norm": 4.
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
-
"loss": 5.
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
-
"grad_norm":
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
-
"loss": 5.
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
-
"grad_norm": 5.
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
-
"loss": 4.
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
-
"grad_norm": 5.
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
-
"loss": 4.
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
-
"grad_norm": 5.
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
-
"loss": 4.
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
-
"grad_norm": 5.
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
-
"loss": 4.
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
-
"grad_norm": 5.
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
-
"loss": 4.
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
-
"grad_norm": 5.
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
-
"loss": 4.
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
-
"grad_norm": 5.
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
-
"loss": 4.
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
-
"grad_norm": 5.
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
-
"loss": 4.
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
-
"grad_norm": 5.
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
-
"loss": 3.
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
-
"grad_norm": 5.
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
-
"loss": 3.
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
-
"grad_norm": 5.
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
-
"loss": 3.
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
-
"grad_norm": 5.
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
-
"loss": 3.
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
-
"grad_norm": 5.
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
-
"loss": 3.
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
-
"grad_norm":
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
-
"loss": 3.
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
-
"grad_norm": 5.
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
-
"loss": 3.
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
-
"grad_norm": 5.
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
-
"loss": 3.
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
-
"grad_norm": 5.
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
-
"loss": 3.
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
-
"grad_norm": 5.
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
-
"loss": 3.
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
-
"grad_norm": 6.
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
-
"loss": 3.
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
-
"grad_norm": 6.
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
-
"loss": 3.
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
-
"grad_norm": 6.
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
-
"loss": 3.
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
-
"grad_norm": 6.
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
-
"loss": 3.
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
-
"grad_norm": 6.
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
-
"loss": 3.
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
-
"grad_norm": 6.
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
-
"loss": 3.
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
-
"grad_norm": 6.
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
-
"loss": 3.
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
-
"grad_norm": 7.
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
-
"loss": 3.
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
-
"grad_norm": 7.
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
-
"loss": 3.
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
-
"grad_norm": 7.
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
-
"loss": 3.
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
-
"grad_norm":
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
-
"loss": 3.
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
-
"grad_norm": 7.
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
-
"loss": 2.
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
-
"grad_norm": 7.
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
-
"loss":
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
-
"grad_norm": 8.
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
-
"loss": 3.
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
-
"grad_norm":
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
-
"loss": 2.
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
-
"grad_norm": 7.
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
-
"loss": 2.
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
-
"grad_norm":
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
-
"loss": 2.
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
-
"grad_norm": 8.
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
-
"loss": 2.
|
290 |
"step": 2000
|
291 |
},
|
292 |
{
|
293 |
"epoch": 9.951456310679612,
|
294 |
-
"grad_norm": 8.
|
295 |
"learning_rate": 3.342233009708738e-05,
|
296 |
-
"loss": 2.
|
297 |
"step": 2050
|
298 |
},
|
299 |
{
|
300 |
"epoch": 10.194174757281553,
|
301 |
-
"grad_norm":
|
302 |
"learning_rate": 3.301779935275081e-05,
|
303 |
-
"loss": 2.
|
304 |
"step": 2100
|
305 |
},
|
306 |
{
|
307 |
"epoch": 10.436893203883495,
|
308 |
-
"grad_norm":
|
309 |
"learning_rate": 3.2613268608414246e-05,
|
310 |
-
"loss": 2.
|
311 |
"step": 2150
|
312 |
},
|
313 |
{
|
314 |
"epoch": 10.679611650485437,
|
315 |
-
"grad_norm": 9.
|
316 |
"learning_rate": 3.220873786407767e-05,
|
317 |
-
"loss": 2.
|
318 |
"step": 2200
|
319 |
},
|
320 |
{
|
321 |
"epoch": 10.922330097087379,
|
322 |
-
"grad_norm": 9.
|
323 |
"learning_rate": 3.1804207119741106e-05,
|
324 |
-
"loss": 2.
|
325 |
"step": 2250
|
326 |
},
|
327 |
{
|
328 |
"epoch": 11.16504854368932,
|
329 |
-
"grad_norm": 8.
|
330 |
"learning_rate": 3.139967637540453e-05,
|
331 |
-
"loss": 2.
|
332 |
"step": 2300
|
333 |
},
|
334 |
{
|
335 |
"epoch": 11.407766990291263,
|
336 |
-
"grad_norm": 9.
|
337 |
"learning_rate": 3.099514563106796e-05,
|
338 |
-
"loss": 2.
|
339 |
"step": 2350
|
340 |
},
|
341 |
{
|
342 |
"epoch": 11.650485436893204,
|
343 |
-
"grad_norm": 9.
|
344 |
"learning_rate": 3.059061488673139e-05,
|
345 |
-
"loss": 2.
|
346 |
"step": 2400
|
347 |
},
|
348 |
{
|
349 |
"epoch": 11.893203883495145,
|
350 |
-
"grad_norm": 9.
|
351 |
"learning_rate": 3.0186084142394822e-05,
|
352 |
-
"loss": 2.
|
353 |
"step": 2450
|
354 |
},
|
355 |
{
|
356 |
"epoch": 12.135922330097088,
|
357 |
-
"grad_norm": 9.
|
358 |
"learning_rate": 2.9781553398058252e-05,
|
359 |
-
"loss": 2.
|
360 |
"step": 2500
|
361 |
},
|
362 |
{
|
363 |
"epoch": 12.37864077669903,
|
364 |
-
"grad_norm":
|
365 |
"learning_rate": 2.9377022653721686e-05,
|
366 |
-
"loss": 2.
|
367 |
"step": 2550
|
368 |
},
|
369 |
{
|
370 |
"epoch": 12.62135922330097,
|
371 |
-
"grad_norm": 9.
|
372 |
"learning_rate": 2.8972491909385112e-05,
|
373 |
-
"loss": 2.
|
374 |
"step": 2600
|
375 |
},
|
376 |
{
|
377 |
"epoch": 12.864077669902912,
|
378 |
-
"grad_norm": 10.
|
379 |
"learning_rate": 2.8567961165048546e-05,
|
380 |
-
"loss": 2.
|
381 |
"step": 2650
|
382 |
},
|
383 |
{
|
384 |
"epoch": 13.106796116504855,
|
385 |
-
"grad_norm": 10.
|
386 |
"learning_rate": 2.816343042071198e-05,
|
387 |
-
"loss": 2.
|
388 |
"step": 2700
|
389 |
},
|
390 |
{
|
391 |
"epoch": 13.349514563106796,
|
392 |
-
"grad_norm":
|
393 |
"learning_rate": 2.7758899676375405e-05,
|
394 |
-
"loss": 2.
|
395 |
"step": 2750
|
396 |
},
|
397 |
{
|
398 |
"epoch": 13.592233009708737,
|
399 |
-
"grad_norm":
|
400 |
"learning_rate": 2.735436893203884e-05,
|
401 |
-
"loss": 2.
|
402 |
"step": 2800
|
403 |
},
|
404 |
{
|
405 |
"epoch": 13.83495145631068,
|
406 |
-
"grad_norm":
|
407 |
"learning_rate": 2.6949838187702265e-05,
|
408 |
-
"loss": 2.
|
409 |
"step": 2850
|
410 |
},
|
411 |
{
|
412 |
"epoch": 14.077669902912621,
|
413 |
-
"grad_norm": 9.
|
414 |
"learning_rate": 2.6545307443365695e-05,
|
415 |
-
"loss":
|
416 |
"step": 2900
|
417 |
},
|
418 |
{
|
419 |
"epoch": 14.320388349514563,
|
420 |
-
"grad_norm": 10.
|
421 |
"learning_rate": 2.614077669902913e-05,
|
422 |
-
"loss": 1.
|
423 |
"step": 2950
|
424 |
},
|
425 |
{
|
426 |
"epoch": 14.563106796116505,
|
427 |
-
"grad_norm": 10.
|
428 |
"learning_rate": 2.5736245954692555e-05,
|
429 |
-
"loss": 1.
|
430 |
"step": 3000
|
431 |
},
|
432 |
{
|
433 |
"epoch": 14.805825242718447,
|
434 |
-
"grad_norm":
|
435 |
"learning_rate": 2.533171521035599e-05,
|
436 |
-
"loss": 1.
|
437 |
"step": 3050
|
438 |
},
|
439 |
{
|
440 |
"epoch": 15.048543689320388,
|
441 |
-
"grad_norm": 9.
|
442 |
"learning_rate": 2.492718446601942e-05,
|
443 |
-
"loss": 1.
|
444 |
"step": 3100
|
445 |
},
|
446 |
{
|
447 |
"epoch": 15.29126213592233,
|
448 |
-
"grad_norm": 10.
|
449 |
"learning_rate": 2.452265372168285e-05,
|
450 |
-
"loss": 1.
|
451 |
"step": 3150
|
452 |
},
|
453 |
{
|
454 |
"epoch": 15.533980582524272,
|
455 |
-
"grad_norm": 10.
|
456 |
"learning_rate": 2.411812297734628e-05,
|
457 |
-
"loss": 1.
|
458 |
"step": 3200
|
459 |
},
|
460 |
{
|
461 |
"epoch": 15.776699029126213,
|
462 |
-
"grad_norm": 11.
|
463 |
"learning_rate": 2.3713592233009708e-05,
|
464 |
-
"loss": 1.
|
465 |
"step": 3250
|
466 |
},
|
467 |
{
|
468 |
"epoch": 16.019417475728154,
|
469 |
-
"grad_norm":
|
470 |
"learning_rate": 2.3309061488673138e-05,
|
471 |
-
"loss": 1.
|
472 |
"step": 3300
|
473 |
},
|
474 |
{
|
475 |
"epoch": 16.262135922330096,
|
476 |
-
"grad_norm": 9.
|
477 |
"learning_rate": 2.290453074433657e-05,
|
478 |
-
"loss": 1.
|
479 |
"step": 3350
|
480 |
},
|
481 |
{
|
482 |
"epoch": 16.50485436893204,
|
483 |
-
"grad_norm": 11.
|
484 |
"learning_rate": 2.25e-05,
|
485 |
-
"loss": 1.
|
486 |
"step": 3400
|
487 |
},
|
488 |
{
|
489 |
"epoch": 16.74757281553398,
|
490 |
-
"grad_norm": 11.
|
491 |
"learning_rate": 2.209546925566343e-05,
|
492 |
-
"loss": 1.
|
493 |
"step": 3450
|
494 |
},
|
495 |
{
|
496 |
"epoch": 16.990291262135923,
|
497 |
-
"grad_norm": 11.
|
498 |
"learning_rate": 2.169093851132686e-05,
|
499 |
-
"loss": 1.
|
500 |
"step": 3500
|
501 |
}
|
502 |
],
|
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
+
"grad_norm": 2.239194869995117,
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
+
"loss": 6.099,
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
+
"grad_norm": 3.8653194904327393,
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
+
"loss": 5.6954,
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
+
"grad_norm": 4.652886867523193,
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
+
"loss": 5.3598,
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
+
"grad_norm": 5.048999309539795,
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
+
"loss": 5.0982,
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
+
"grad_norm": 5.21944522857666,
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
+
"loss": 4.7456,
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
+
"grad_norm": 5.471007347106934,
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
+
"loss": 4.6254,
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
+
"grad_norm": 5.345245838165283,
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
+
"loss": 4.5216,
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
+
"grad_norm": 5.54531192779541,
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
+
"loss": 4.4353,
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
+
"grad_norm": 5.244582176208496,
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
+
"loss": 4.2444,
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
+
"grad_norm": 5.338261127471924,
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
+
"loss": 4.1915,
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
+
"grad_norm": 5.35228157043457,
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
+
"loss": 4.1612,
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
+
"grad_norm": 5.287161827087402,
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
+
"loss": 4.1322,
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
+
"grad_norm": 5.2932209968566895,
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
+
"loss": 3.9684,
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
+
"grad_norm": 5.4360833168029785,
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
+
"loss": 3.8933,
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
+
"grad_norm": 5.5595221519470215,
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
+
"loss": 3.9306,
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
+
"grad_norm": 5.540086269378662,
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
+
"loss": 3.9317,
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
+
"grad_norm": 5.567137718200684,
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
+
"loss": 3.8135,
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
+
"grad_norm": 6.00346040725708,
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
+
"loss": 3.6924,
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
+
"grad_norm": 5.635782718658447,
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
+
"loss": 3.7304,
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
+
"grad_norm": 5.9545440673828125,
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
+
"loss": 3.7471,
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
+
"grad_norm": 5.696166038513184,
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
+
"loss": 3.6287,
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
+
"grad_norm": 5.844620704650879,
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
+
"loss": 3.5299,
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
+
"grad_norm": 6.373577117919922,
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
+
"loss": 3.5285,
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
+
"grad_norm": 6.488651275634766,
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
+
"loss": 3.5789,
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
+
"grad_norm": 6.170980930328369,
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
+
"loss": 3.494,
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
+
"grad_norm": 6.387782573699951,
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
+
"loss": 3.353,
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
+
"grad_norm": 6.718175888061523,
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
+
"loss": 3.3521,
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
+
"grad_norm": 6.7870025634765625,
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
+
"loss": 3.3658,
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
+
"grad_norm": 6.513721466064453,
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
+
"loss": 3.3301,
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
+
"grad_norm": 7.0680036544799805,
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
+
"loss": 3.1289,
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
+
"grad_norm": 7.019896507263184,
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
+
"loss": 3.1878,
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
+
"grad_norm": 7.3232197761535645,
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
+
"loss": 3.2006,
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
+
"grad_norm": 7.003912925720215,
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
+
"loss": 3.1737,
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
+
"grad_norm": 7.592433929443359,
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
+
"loss": 2.9258,
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
+
"grad_norm": 7.629278182983398,
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
+
"loss": 2.9999,
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
+
"grad_norm": 8.137636184692383,
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
+
"loss": 3.0155,
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
+
"grad_norm": 7.967227935791016,
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
+
"loss": 2.9945,
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
+
"grad_norm": 7.9910078048706055,
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
+
"loss": 2.7456,
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
+
"grad_norm": 8.552720069885254,
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
+
"loss": 2.8071,
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
+
"grad_norm": 8.336930274963379,
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
+
"loss": 2.8117,
|
290 |
"step": 2000
|
291 |
},
|
292 |
{
|
293 |
"epoch": 9.951456310679612,
|
294 |
+
"grad_norm": 8.636338233947754,
|
295 |
"learning_rate": 3.342233009708738e-05,
|
296 |
+
"loss": 2.8197,
|
297 |
"step": 2050
|
298 |
},
|
299 |
{
|
300 |
"epoch": 10.194174757281553,
|
301 |
+
"grad_norm": 7.955165863037109,
|
302 |
"learning_rate": 3.301779935275081e-05,
|
303 |
+
"loss": 2.6245,
|
304 |
"step": 2100
|
305 |
},
|
306 |
{
|
307 |
"epoch": 10.436893203883495,
|
308 |
+
"grad_norm": 8.679149627685547,
|
309 |
"learning_rate": 3.2613268608414246e-05,
|
310 |
+
"loss": 2.565,
|
311 |
"step": 2150
|
312 |
},
|
313 |
{
|
314 |
"epoch": 10.679611650485437,
|
315 |
+
"grad_norm": 9.150839805603027,
|
316 |
"learning_rate": 3.220873786407767e-05,
|
317 |
+
"loss": 2.6258,
|
318 |
"step": 2200
|
319 |
},
|
320 |
{
|
321 |
"epoch": 10.922330097087379,
|
322 |
+
"grad_norm": 9.2735013961792,
|
323 |
"learning_rate": 3.1804207119741106e-05,
|
324 |
+
"loss": 2.6439,
|
325 |
"step": 2250
|
326 |
},
|
327 |
{
|
328 |
"epoch": 11.16504854368932,
|
329 |
+
"grad_norm": 8.620939254760742,
|
330 |
"learning_rate": 3.139967637540453e-05,
|
331 |
+
"loss": 2.4371,
|
332 |
"step": 2300
|
333 |
},
|
334 |
{
|
335 |
"epoch": 11.407766990291263,
|
336 |
+
"grad_norm": 9.722040176391602,
|
337 |
"learning_rate": 3.099514563106796e-05,
|
338 |
+
"loss": 2.4074,
|
339 |
"step": 2350
|
340 |
},
|
341 |
{
|
342 |
"epoch": 11.650485436893204,
|
343 |
+
"grad_norm": 9.200730323791504,
|
344 |
"learning_rate": 3.059061488673139e-05,
|
345 |
+
"loss": 2.4307,
|
346 |
"step": 2400
|
347 |
},
|
348 |
{
|
349 |
"epoch": 11.893203883495145,
|
350 |
+
"grad_norm": 9.692273139953613,
|
351 |
"learning_rate": 3.0186084142394822e-05,
|
352 |
+
"loss": 2.424,
|
353 |
"step": 2450
|
354 |
},
|
355 |
{
|
356 |
"epoch": 12.135922330097088,
|
357 |
+
"grad_norm": 9.822249412536621,
|
358 |
"learning_rate": 2.9781553398058252e-05,
|
359 |
+
"loss": 2.2862,
|
360 |
"step": 2500
|
361 |
},
|
362 |
{
|
363 |
"epoch": 12.37864077669903,
|
364 |
+
"grad_norm": 9.919060707092285,
|
365 |
"learning_rate": 2.9377022653721686e-05,
|
366 |
+
"loss": 2.1913,
|
367 |
"step": 2550
|
368 |
},
|
369 |
{
|
370 |
"epoch": 12.62135922330097,
|
371 |
+
"grad_norm": 9.181950569152832,
|
372 |
"learning_rate": 2.8972491909385112e-05,
|
373 |
+
"loss": 2.241,
|
374 |
"step": 2600
|
375 |
},
|
376 |
{
|
377 |
"epoch": 12.864077669902912,
|
378 |
+
"grad_norm": 10.82077407836914,
|
379 |
"learning_rate": 2.8567961165048546e-05,
|
380 |
+
"loss": 2.2653,
|
381 |
"step": 2650
|
382 |
},
|
383 |
{
|
384 |
"epoch": 13.106796116504855,
|
385 |
+
"grad_norm": 10.365830421447754,
|
386 |
"learning_rate": 2.816343042071198e-05,
|
387 |
+
"loss": 2.1352,
|
388 |
"step": 2700
|
389 |
},
|
390 |
{
|
391 |
"epoch": 13.349514563106796,
|
392 |
+
"grad_norm": 10.073640823364258,
|
393 |
"learning_rate": 2.7758899676375405e-05,
|
394 |
+
"loss": 2.0227,
|
395 |
"step": 2750
|
396 |
},
|
397 |
{
|
398 |
"epoch": 13.592233009708737,
|
399 |
+
"grad_norm": 10.520242691040039,
|
400 |
"learning_rate": 2.735436893203884e-05,
|
401 |
+
"loss": 2.0579,
|
402 |
"step": 2800
|
403 |
},
|
404 |
{
|
405 |
"epoch": 13.83495145631068,
|
406 |
+
"grad_norm": 11.176478385925293,
|
407 |
"learning_rate": 2.6949838187702265e-05,
|
408 |
+
"loss": 2.0934,
|
409 |
"step": 2850
|
410 |
},
|
411 |
{
|
412 |
"epoch": 14.077669902912621,
|
413 |
+
"grad_norm": 9.541868209838867,
|
414 |
"learning_rate": 2.6545307443365695e-05,
|
415 |
+
"loss": 2.0118,
|
416 |
"step": 2900
|
417 |
},
|
418 |
{
|
419 |
"epoch": 14.320388349514563,
|
420 |
+
"grad_norm": 10.162981986999512,
|
421 |
"learning_rate": 2.614077669902913e-05,
|
422 |
+
"loss": 1.8624,
|
423 |
"step": 2950
|
424 |
},
|
425 |
{
|
426 |
"epoch": 14.563106796116505,
|
427 |
+
"grad_norm": 10.51291275024414,
|
428 |
"learning_rate": 2.5736245954692555e-05,
|
429 |
+
"loss": 1.8979,
|
430 |
"step": 3000
|
431 |
},
|
432 |
{
|
433 |
"epoch": 14.805825242718447,
|
434 |
+
"grad_norm": 11.316717147827148,
|
435 |
"learning_rate": 2.533171521035599e-05,
|
436 |
+
"loss": 1.915,
|
437 |
"step": 3050
|
438 |
},
|
439 |
{
|
440 |
"epoch": 15.048543689320388,
|
441 |
+
"grad_norm": 9.184353828430176,
|
442 |
"learning_rate": 2.492718446601942e-05,
|
443 |
+
"loss": 1.885,
|
444 |
"step": 3100
|
445 |
},
|
446 |
{
|
447 |
"epoch": 15.29126213592233,
|
448 |
+
"grad_norm": 10.70876693725586,
|
449 |
"learning_rate": 2.452265372168285e-05,
|
450 |
+
"loss": 1.7024,
|
451 |
"step": 3150
|
452 |
},
|
453 |
{
|
454 |
"epoch": 15.533980582524272,
|
455 |
+
"grad_norm": 10.767953872680664,
|
456 |
"learning_rate": 2.411812297734628e-05,
|
457 |
+
"loss": 1.729,
|
458 |
"step": 3200
|
459 |
},
|
460 |
{
|
461 |
"epoch": 15.776699029126213,
|
462 |
+
"grad_norm": 11.83687973022461,
|
463 |
"learning_rate": 2.3713592233009708e-05,
|
464 |
+
"loss": 1.7697,
|
465 |
"step": 3250
|
466 |
},
|
467 |
{
|
468 |
"epoch": 16.019417475728154,
|
469 |
+
"grad_norm": 9.888689994812012,
|
470 |
"learning_rate": 2.3309061488673138e-05,
|
471 |
+
"loss": 1.7605,
|
472 |
"step": 3300
|
473 |
},
|
474 |
{
|
475 |
"epoch": 16.262135922330096,
|
476 |
+
"grad_norm": 9.394012451171875,
|
477 |
"learning_rate": 2.290453074433657e-05,
|
478 |
+
"loss": 1.5389,
|
479 |
"step": 3350
|
480 |
},
|
481 |
{
|
482 |
"epoch": 16.50485436893204,
|
483 |
+
"grad_norm": 11.3172025680542,
|
484 |
"learning_rate": 2.25e-05,
|
485 |
+
"loss": 1.5836,
|
486 |
"step": 3400
|
487 |
},
|
488 |
{
|
489 |
"epoch": 16.74757281553398,
|
490 |
+
"grad_norm": 11.000115394592285,
|
491 |
"learning_rate": 2.209546925566343e-05,
|
492 |
+
"loss": 1.6398,
|
493 |
"step": 3450
|
494 |
},
|
495 |
{
|
496 |
"epoch": 16.990291262135923,
|
497 |
+
"grad_norm": 11.453404426574707,
|
498 |
"learning_rate": 2.169093851132686e-05,
|
499 |
+
"loss": 1.6466,
|
500 |
"step": 3500
|
501 |
}
|
502 |
],
|
checkpoints/checkpoint-3500/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7935e3b7629485cf26f624afe618b58a390d4c3c9c6dcf878b3ba9099ef1027b
|
3 |
size 5304
|
checkpoints/checkpoint-4000/config.json
CHANGED
@@ -10,12 +10,12 @@
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
-
"n_ctx":
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
-
"n_positions":
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
+
"n_ctx": 1024,
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
+
"n_positions": 1024,
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
checkpoints/checkpoint-4000/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b6dec913fb019f72d2b4e5ba61a3e03faf3f77c63613f93c6c44239205901dc9
|
3 |
+
size 176343496
|
checkpoints/checkpoint-4000/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3e578e56f38303973d77fb3e27d42c6872cb40d601b37756b09c847fa9d0eb4a
|
3 |
+
size 352735610
|
checkpoints/checkpoint-4000/trainer_state.json
CHANGED
@@ -11,562 +11,562 @@
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
-
"grad_norm": 2.
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
-
"loss": 6.
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
-
"grad_norm": 3.
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
-
"loss": 5.
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
-
"grad_norm": 4.
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
-
"loss": 5.
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
-
"grad_norm":
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
-
"loss": 5.
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
-
"grad_norm": 5.
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
-
"loss": 4.
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
-
"grad_norm": 5.
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
-
"loss": 4.
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
-
"grad_norm": 5.
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
-
"loss": 4.
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
-
"grad_norm": 5.
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
-
"loss": 4.
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
-
"grad_norm": 5.
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
-
"loss": 4.
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
-
"grad_norm": 5.
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
-
"loss": 4.
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
-
"grad_norm": 5.
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
-
"loss": 4.
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
-
"grad_norm": 5.
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
-
"loss": 4.
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
-
"grad_norm": 5.
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
-
"loss": 3.
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
-
"grad_norm": 5.
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
-
"loss": 3.
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
-
"grad_norm": 5.
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
-
"loss": 3.
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
-
"grad_norm": 5.
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
-
"loss": 3.
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
-
"grad_norm": 5.
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
-
"loss": 3.
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
-
"grad_norm":
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
-
"loss": 3.
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
-
"grad_norm": 5.
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
-
"loss": 3.
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
-
"grad_norm": 5.
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
-
"loss": 3.
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
-
"grad_norm": 5.
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
-
"loss": 3.
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
-
"grad_norm": 5.
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
-
"loss": 3.
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
-
"grad_norm": 6.
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
-
"loss": 3.
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
-
"grad_norm": 6.
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
-
"loss": 3.
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
-
"grad_norm": 6.
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
-
"loss": 3.
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
-
"grad_norm": 6.
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
-
"loss": 3.
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
-
"grad_norm": 6.
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
-
"loss": 3.
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
-
"grad_norm": 6.
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
-
"loss": 3.
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
-
"grad_norm": 6.
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
-
"loss": 3.
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
-
"grad_norm": 7.
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
-
"loss": 3.
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
-
"grad_norm": 7.
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
-
"loss": 3.
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
-
"grad_norm": 7.
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
-
"loss": 3.
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
-
"grad_norm":
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
-
"loss": 3.
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
-
"grad_norm": 7.
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
-
"loss": 2.
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
-
"grad_norm": 7.
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
-
"loss":
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
-
"grad_norm": 8.
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
-
"loss": 3.
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
-
"grad_norm":
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
-
"loss": 2.
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
-
"grad_norm": 7.
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
-
"loss": 2.
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
-
"grad_norm":
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
-
"loss": 2.
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
-
"grad_norm": 8.
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
-
"loss": 2.
|
290 |
"step": 2000
|
291 |
},
|
292 |
{
|
293 |
"epoch": 9.951456310679612,
|
294 |
-
"grad_norm": 8.
|
295 |
"learning_rate": 3.342233009708738e-05,
|
296 |
-
"loss": 2.
|
297 |
"step": 2050
|
298 |
},
|
299 |
{
|
300 |
"epoch": 10.194174757281553,
|
301 |
-
"grad_norm":
|
302 |
"learning_rate": 3.301779935275081e-05,
|
303 |
-
"loss": 2.
|
304 |
"step": 2100
|
305 |
},
|
306 |
{
|
307 |
"epoch": 10.436893203883495,
|
308 |
-
"grad_norm":
|
309 |
"learning_rate": 3.2613268608414246e-05,
|
310 |
-
"loss": 2.
|
311 |
"step": 2150
|
312 |
},
|
313 |
{
|
314 |
"epoch": 10.679611650485437,
|
315 |
-
"grad_norm": 9.
|
316 |
"learning_rate": 3.220873786407767e-05,
|
317 |
-
"loss": 2.
|
318 |
"step": 2200
|
319 |
},
|
320 |
{
|
321 |
"epoch": 10.922330097087379,
|
322 |
-
"grad_norm": 9.
|
323 |
"learning_rate": 3.1804207119741106e-05,
|
324 |
-
"loss": 2.
|
325 |
"step": 2250
|
326 |
},
|
327 |
{
|
328 |
"epoch": 11.16504854368932,
|
329 |
-
"grad_norm": 8.
|
330 |
"learning_rate": 3.139967637540453e-05,
|
331 |
-
"loss": 2.
|
332 |
"step": 2300
|
333 |
},
|
334 |
{
|
335 |
"epoch": 11.407766990291263,
|
336 |
-
"grad_norm": 9.
|
337 |
"learning_rate": 3.099514563106796e-05,
|
338 |
-
"loss": 2.
|
339 |
"step": 2350
|
340 |
},
|
341 |
{
|
342 |
"epoch": 11.650485436893204,
|
343 |
-
"grad_norm": 9.
|
344 |
"learning_rate": 3.059061488673139e-05,
|
345 |
-
"loss": 2.
|
346 |
"step": 2400
|
347 |
},
|
348 |
{
|
349 |
"epoch": 11.893203883495145,
|
350 |
-
"grad_norm": 9.
|
351 |
"learning_rate": 3.0186084142394822e-05,
|
352 |
-
"loss": 2.
|
353 |
"step": 2450
|
354 |
},
|
355 |
{
|
356 |
"epoch": 12.135922330097088,
|
357 |
-
"grad_norm": 9.
|
358 |
"learning_rate": 2.9781553398058252e-05,
|
359 |
-
"loss": 2.
|
360 |
"step": 2500
|
361 |
},
|
362 |
{
|
363 |
"epoch": 12.37864077669903,
|
364 |
-
"grad_norm":
|
365 |
"learning_rate": 2.9377022653721686e-05,
|
366 |
-
"loss": 2.
|
367 |
"step": 2550
|
368 |
},
|
369 |
{
|
370 |
"epoch": 12.62135922330097,
|
371 |
-
"grad_norm": 9.
|
372 |
"learning_rate": 2.8972491909385112e-05,
|
373 |
-
"loss": 2.
|
374 |
"step": 2600
|
375 |
},
|
376 |
{
|
377 |
"epoch": 12.864077669902912,
|
378 |
-
"grad_norm": 10.
|
379 |
"learning_rate": 2.8567961165048546e-05,
|
380 |
-
"loss": 2.
|
381 |
"step": 2650
|
382 |
},
|
383 |
{
|
384 |
"epoch": 13.106796116504855,
|
385 |
-
"grad_norm": 10.
|
386 |
"learning_rate": 2.816343042071198e-05,
|
387 |
-
"loss": 2.
|
388 |
"step": 2700
|
389 |
},
|
390 |
{
|
391 |
"epoch": 13.349514563106796,
|
392 |
-
"grad_norm":
|
393 |
"learning_rate": 2.7758899676375405e-05,
|
394 |
-
"loss": 2.
|
395 |
"step": 2750
|
396 |
},
|
397 |
{
|
398 |
"epoch": 13.592233009708737,
|
399 |
-
"grad_norm":
|
400 |
"learning_rate": 2.735436893203884e-05,
|
401 |
-
"loss": 2.
|
402 |
"step": 2800
|
403 |
},
|
404 |
{
|
405 |
"epoch": 13.83495145631068,
|
406 |
-
"grad_norm":
|
407 |
"learning_rate": 2.6949838187702265e-05,
|
408 |
-
"loss": 2.
|
409 |
"step": 2850
|
410 |
},
|
411 |
{
|
412 |
"epoch": 14.077669902912621,
|
413 |
-
"grad_norm": 9.
|
414 |
"learning_rate": 2.6545307443365695e-05,
|
415 |
-
"loss":
|
416 |
"step": 2900
|
417 |
},
|
418 |
{
|
419 |
"epoch": 14.320388349514563,
|
420 |
-
"grad_norm": 10.
|
421 |
"learning_rate": 2.614077669902913e-05,
|
422 |
-
"loss": 1.
|
423 |
"step": 2950
|
424 |
},
|
425 |
{
|
426 |
"epoch": 14.563106796116505,
|
427 |
-
"grad_norm": 10.
|
428 |
"learning_rate": 2.5736245954692555e-05,
|
429 |
-
"loss": 1.
|
430 |
"step": 3000
|
431 |
},
|
432 |
{
|
433 |
"epoch": 14.805825242718447,
|
434 |
-
"grad_norm":
|
435 |
"learning_rate": 2.533171521035599e-05,
|
436 |
-
"loss": 1.
|
437 |
"step": 3050
|
438 |
},
|
439 |
{
|
440 |
"epoch": 15.048543689320388,
|
441 |
-
"grad_norm": 9.
|
442 |
"learning_rate": 2.492718446601942e-05,
|
443 |
-
"loss": 1.
|
444 |
"step": 3100
|
445 |
},
|
446 |
{
|
447 |
"epoch": 15.29126213592233,
|
448 |
-
"grad_norm": 10.
|
449 |
"learning_rate": 2.452265372168285e-05,
|
450 |
-
"loss": 1.
|
451 |
"step": 3150
|
452 |
},
|
453 |
{
|
454 |
"epoch": 15.533980582524272,
|
455 |
-
"grad_norm": 10.
|
456 |
"learning_rate": 2.411812297734628e-05,
|
457 |
-
"loss": 1.
|
458 |
"step": 3200
|
459 |
},
|
460 |
{
|
461 |
"epoch": 15.776699029126213,
|
462 |
-
"grad_norm": 11.
|
463 |
"learning_rate": 2.3713592233009708e-05,
|
464 |
-
"loss": 1.
|
465 |
"step": 3250
|
466 |
},
|
467 |
{
|
468 |
"epoch": 16.019417475728154,
|
469 |
-
"grad_norm":
|
470 |
"learning_rate": 2.3309061488673138e-05,
|
471 |
-
"loss": 1.
|
472 |
"step": 3300
|
473 |
},
|
474 |
{
|
475 |
"epoch": 16.262135922330096,
|
476 |
-
"grad_norm": 9.
|
477 |
"learning_rate": 2.290453074433657e-05,
|
478 |
-
"loss": 1.
|
479 |
"step": 3350
|
480 |
},
|
481 |
{
|
482 |
"epoch": 16.50485436893204,
|
483 |
-
"grad_norm": 11.
|
484 |
"learning_rate": 2.25e-05,
|
485 |
-
"loss": 1.
|
486 |
"step": 3400
|
487 |
},
|
488 |
{
|
489 |
"epoch": 16.74757281553398,
|
490 |
-
"grad_norm": 11.
|
491 |
"learning_rate": 2.209546925566343e-05,
|
492 |
-
"loss": 1.
|
493 |
"step": 3450
|
494 |
},
|
495 |
{
|
496 |
"epoch": 16.990291262135923,
|
497 |
-
"grad_norm": 11.
|
498 |
"learning_rate": 2.169093851132686e-05,
|
499 |
-
"loss": 1.
|
500 |
"step": 3500
|
501 |
},
|
502 |
{
|
503 |
"epoch": 17.233009708737864,
|
504 |
-
"grad_norm":
|
505 |
"learning_rate": 2.1286407766990295e-05,
|
506 |
-
"loss": 1.
|
507 |
"step": 3550
|
508 |
},
|
509 |
{
|
510 |
"epoch": 17.475728155339805,
|
511 |
-
"grad_norm": 11.
|
512 |
"learning_rate": 2.0881877022653725e-05,
|
513 |
-
"loss": 1.
|
514 |
"step": 3600
|
515 |
},
|
516 |
{
|
517 |
"epoch": 17.718446601941746,
|
518 |
-
"grad_norm":
|
519 |
"learning_rate": 2.047734627831715e-05,
|
520 |
-
"loss": 1.
|
521 |
"step": 3650
|
522 |
},
|
523 |
{
|
524 |
"epoch": 17.96116504854369,
|
525 |
-
"grad_norm":
|
526 |
"learning_rate": 2.007281553398058e-05,
|
527 |
-
"loss": 1.
|
528 |
"step": 3700
|
529 |
},
|
530 |
{
|
531 |
"epoch": 18.203883495145632,
|
532 |
-
"grad_norm": 9.
|
533 |
"learning_rate": 1.9668284789644014e-05,
|
534 |
-
"loss": 1.
|
535 |
"step": 3750
|
536 |
},
|
537 |
{
|
538 |
"epoch": 18.446601941747574,
|
539 |
-
"grad_norm":
|
540 |
"learning_rate": 1.9263754045307444e-05,
|
541 |
-
"loss": 1.
|
542 |
"step": 3800
|
543 |
},
|
544 |
{
|
545 |
"epoch": 18.689320388349515,
|
546 |
-
"grad_norm": 11.
|
547 |
"learning_rate": 1.8859223300970874e-05,
|
548 |
-
"loss": 1.
|
549 |
"step": 3850
|
550 |
},
|
551 |
{
|
552 |
"epoch": 18.932038834951456,
|
553 |
-
"grad_norm": 10.
|
554 |
"learning_rate": 1.8454692556634304e-05,
|
555 |
-
"loss": 1.
|
556 |
"step": 3900
|
557 |
},
|
558 |
{
|
559 |
"epoch": 19.174757281553397,
|
560 |
-
"grad_norm":
|
561 |
"learning_rate": 1.8050161812297738e-05,
|
562 |
-
"loss": 1.
|
563 |
"step": 3950
|
564 |
},
|
565 |
{
|
566 |
"epoch": 19.41747572815534,
|
567 |
-
"grad_norm":
|
568 |
"learning_rate": 1.7645631067961167e-05,
|
569 |
-
"loss": 1.
|
570 |
"step": 4000
|
571 |
}
|
572 |
],
|
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
+
"grad_norm": 2.239194869995117,
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
+
"loss": 6.099,
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
+
"grad_norm": 3.8653194904327393,
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
+
"loss": 5.6954,
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
+
"grad_norm": 4.652886867523193,
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
+
"loss": 5.3598,
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
+
"grad_norm": 5.048999309539795,
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
+
"loss": 5.0982,
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
+
"grad_norm": 5.21944522857666,
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
+
"loss": 4.7456,
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
+
"grad_norm": 5.471007347106934,
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
+
"loss": 4.6254,
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
+
"grad_norm": 5.345245838165283,
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
+
"loss": 4.5216,
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
+
"grad_norm": 5.54531192779541,
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
+
"loss": 4.4353,
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
+
"grad_norm": 5.244582176208496,
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
+
"loss": 4.2444,
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
+
"grad_norm": 5.338261127471924,
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
+
"loss": 4.1915,
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
+
"grad_norm": 5.35228157043457,
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
+
"loss": 4.1612,
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
+
"grad_norm": 5.287161827087402,
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
+
"loss": 4.1322,
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
+
"grad_norm": 5.2932209968566895,
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
+
"loss": 3.9684,
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
+
"grad_norm": 5.4360833168029785,
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
+
"loss": 3.8933,
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
+
"grad_norm": 5.5595221519470215,
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
+
"loss": 3.9306,
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
+
"grad_norm": 5.540086269378662,
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
+
"loss": 3.9317,
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
+
"grad_norm": 5.567137718200684,
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
+
"loss": 3.8135,
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
+
"grad_norm": 6.00346040725708,
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
+
"loss": 3.6924,
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
+
"grad_norm": 5.635782718658447,
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
+
"loss": 3.7304,
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
+
"grad_norm": 5.9545440673828125,
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
+
"loss": 3.7471,
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
+
"grad_norm": 5.696166038513184,
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
+
"loss": 3.6287,
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
+
"grad_norm": 5.844620704650879,
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
+
"loss": 3.5299,
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
+
"grad_norm": 6.373577117919922,
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
+
"loss": 3.5285,
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
+
"grad_norm": 6.488651275634766,
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
+
"loss": 3.5789,
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
+
"grad_norm": 6.170980930328369,
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
+
"loss": 3.494,
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
+
"grad_norm": 6.387782573699951,
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
+
"loss": 3.353,
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
+
"grad_norm": 6.718175888061523,
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
+
"loss": 3.3521,
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
+
"grad_norm": 6.7870025634765625,
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
+
"loss": 3.3658,
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
+
"grad_norm": 6.513721466064453,
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
+
"loss": 3.3301,
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
+
"grad_norm": 7.0680036544799805,
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
+
"loss": 3.1289,
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
+
"grad_norm": 7.019896507263184,
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
+
"loss": 3.1878,
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
+
"grad_norm": 7.3232197761535645,
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
+
"loss": 3.2006,
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
+
"grad_norm": 7.003912925720215,
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
+
"loss": 3.1737,
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
+
"grad_norm": 7.592433929443359,
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
+
"loss": 2.9258,
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
+
"grad_norm": 7.629278182983398,
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
+
"loss": 2.9999,
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
+
"grad_norm": 8.137636184692383,
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
+
"loss": 3.0155,
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
+
"grad_norm": 7.967227935791016,
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
+
"loss": 2.9945,
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
+
"grad_norm": 7.9910078048706055,
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
+
"loss": 2.7456,
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
+
"grad_norm": 8.552720069885254,
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
+
"loss": 2.8071,
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
+
"grad_norm": 8.336930274963379,
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
+
"loss": 2.8117,
|
290 |
"step": 2000
|
291 |
},
|
292 |
{
|
293 |
"epoch": 9.951456310679612,
|
294 |
+
"grad_norm": 8.636338233947754,
|
295 |
"learning_rate": 3.342233009708738e-05,
|
296 |
+
"loss": 2.8197,
|
297 |
"step": 2050
|
298 |
},
|
299 |
{
|
300 |
"epoch": 10.194174757281553,
|
301 |
+
"grad_norm": 7.955165863037109,
|
302 |
"learning_rate": 3.301779935275081e-05,
|
303 |
+
"loss": 2.6245,
|
304 |
"step": 2100
|
305 |
},
|
306 |
{
|
307 |
"epoch": 10.436893203883495,
|
308 |
+
"grad_norm": 8.679149627685547,
|
309 |
"learning_rate": 3.2613268608414246e-05,
|
310 |
+
"loss": 2.565,
|
311 |
"step": 2150
|
312 |
},
|
313 |
{
|
314 |
"epoch": 10.679611650485437,
|
315 |
+
"grad_norm": 9.150839805603027,
|
316 |
"learning_rate": 3.220873786407767e-05,
|
317 |
+
"loss": 2.6258,
|
318 |
"step": 2200
|
319 |
},
|
320 |
{
|
321 |
"epoch": 10.922330097087379,
|
322 |
+
"grad_norm": 9.2735013961792,
|
323 |
"learning_rate": 3.1804207119741106e-05,
|
324 |
+
"loss": 2.6439,
|
325 |
"step": 2250
|
326 |
},
|
327 |
{
|
328 |
"epoch": 11.16504854368932,
|
329 |
+
"grad_norm": 8.620939254760742,
|
330 |
"learning_rate": 3.139967637540453e-05,
|
331 |
+
"loss": 2.4371,
|
332 |
"step": 2300
|
333 |
},
|
334 |
{
|
335 |
"epoch": 11.407766990291263,
|
336 |
+
"grad_norm": 9.722040176391602,
|
337 |
"learning_rate": 3.099514563106796e-05,
|
338 |
+
"loss": 2.4074,
|
339 |
"step": 2350
|
340 |
},
|
341 |
{
|
342 |
"epoch": 11.650485436893204,
|
343 |
+
"grad_norm": 9.200730323791504,
|
344 |
"learning_rate": 3.059061488673139e-05,
|
345 |
+
"loss": 2.4307,
|
346 |
"step": 2400
|
347 |
},
|
348 |
{
|
349 |
"epoch": 11.893203883495145,
|
350 |
+
"grad_norm": 9.692273139953613,
|
351 |
"learning_rate": 3.0186084142394822e-05,
|
352 |
+
"loss": 2.424,
|
353 |
"step": 2450
|
354 |
},
|
355 |
{
|
356 |
"epoch": 12.135922330097088,
|
357 |
+
"grad_norm": 9.822249412536621,
|
358 |
"learning_rate": 2.9781553398058252e-05,
|
359 |
+
"loss": 2.2862,
|
360 |
"step": 2500
|
361 |
},
|
362 |
{
|
363 |
"epoch": 12.37864077669903,
|
364 |
+
"grad_norm": 9.919060707092285,
|
365 |
"learning_rate": 2.9377022653721686e-05,
|
366 |
+
"loss": 2.1913,
|
367 |
"step": 2550
|
368 |
},
|
369 |
{
|
370 |
"epoch": 12.62135922330097,
|
371 |
+
"grad_norm": 9.181950569152832,
|
372 |
"learning_rate": 2.8972491909385112e-05,
|
373 |
+
"loss": 2.241,
|
374 |
"step": 2600
|
375 |
},
|
376 |
{
|
377 |
"epoch": 12.864077669902912,
|
378 |
+
"grad_norm": 10.82077407836914,
|
379 |
"learning_rate": 2.8567961165048546e-05,
|
380 |
+
"loss": 2.2653,
|
381 |
"step": 2650
|
382 |
},
|
383 |
{
|
384 |
"epoch": 13.106796116504855,
|
385 |
+
"grad_norm": 10.365830421447754,
|
386 |
"learning_rate": 2.816343042071198e-05,
|
387 |
+
"loss": 2.1352,
|
388 |
"step": 2700
|
389 |
},
|
390 |
{
|
391 |
"epoch": 13.349514563106796,
|
392 |
+
"grad_norm": 10.073640823364258,
|
393 |
"learning_rate": 2.7758899676375405e-05,
|
394 |
+
"loss": 2.0227,
|
395 |
"step": 2750
|
396 |
},
|
397 |
{
|
398 |
"epoch": 13.592233009708737,
|
399 |
+
"grad_norm": 10.520242691040039,
|
400 |
"learning_rate": 2.735436893203884e-05,
|
401 |
+
"loss": 2.0579,
|
402 |
"step": 2800
|
403 |
},
|
404 |
{
|
405 |
"epoch": 13.83495145631068,
|
406 |
+
"grad_norm": 11.176478385925293,
|
407 |
"learning_rate": 2.6949838187702265e-05,
|
408 |
+
"loss": 2.0934,
|
409 |
"step": 2850
|
410 |
},
|
411 |
{
|
412 |
"epoch": 14.077669902912621,
|
413 |
+
"grad_norm": 9.541868209838867,
|
414 |
"learning_rate": 2.6545307443365695e-05,
|
415 |
+
"loss": 2.0118,
|
416 |
"step": 2900
|
417 |
},
|
418 |
{
|
419 |
"epoch": 14.320388349514563,
|
420 |
+
"grad_norm": 10.162981986999512,
|
421 |
"learning_rate": 2.614077669902913e-05,
|
422 |
+
"loss": 1.8624,
|
423 |
"step": 2950
|
424 |
},
|
425 |
{
|
426 |
"epoch": 14.563106796116505,
|
427 |
+
"grad_norm": 10.51291275024414,
|
428 |
"learning_rate": 2.5736245954692555e-05,
|
429 |
+
"loss": 1.8979,
|
430 |
"step": 3000
|
431 |
},
|
432 |
{
|
433 |
"epoch": 14.805825242718447,
|
434 |
+
"grad_norm": 11.316717147827148,
|
435 |
"learning_rate": 2.533171521035599e-05,
|
436 |
+
"loss": 1.915,
|
437 |
"step": 3050
|
438 |
},
|
439 |
{
|
440 |
"epoch": 15.048543689320388,
|
441 |
+
"grad_norm": 9.184353828430176,
|
442 |
"learning_rate": 2.492718446601942e-05,
|
443 |
+
"loss": 1.885,
|
444 |
"step": 3100
|
445 |
},
|
446 |
{
|
447 |
"epoch": 15.29126213592233,
|
448 |
+
"grad_norm": 10.70876693725586,
|
449 |
"learning_rate": 2.452265372168285e-05,
|
450 |
+
"loss": 1.7024,
|
451 |
"step": 3150
|
452 |
},
|
453 |
{
|
454 |
"epoch": 15.533980582524272,
|
455 |
+
"grad_norm": 10.767953872680664,
|
456 |
"learning_rate": 2.411812297734628e-05,
|
457 |
+
"loss": 1.729,
|
458 |
"step": 3200
|
459 |
},
|
460 |
{
|
461 |
"epoch": 15.776699029126213,
|
462 |
+
"grad_norm": 11.83687973022461,
|
463 |
"learning_rate": 2.3713592233009708e-05,
|
464 |
+
"loss": 1.7697,
|
465 |
"step": 3250
|
466 |
},
|
467 |
{
|
468 |
"epoch": 16.019417475728154,
|
469 |
+
"grad_norm": 9.888689994812012,
|
470 |
"learning_rate": 2.3309061488673138e-05,
|
471 |
+
"loss": 1.7605,
|
472 |
"step": 3300
|
473 |
},
|
474 |
{
|
475 |
"epoch": 16.262135922330096,
|
476 |
+
"grad_norm": 9.394012451171875,
|
477 |
"learning_rate": 2.290453074433657e-05,
|
478 |
+
"loss": 1.5389,
|
479 |
"step": 3350
|
480 |
},
|
481 |
{
|
482 |
"epoch": 16.50485436893204,
|
483 |
+
"grad_norm": 11.3172025680542,
|
484 |
"learning_rate": 2.25e-05,
|
485 |
+
"loss": 1.5836,
|
486 |
"step": 3400
|
487 |
},
|
488 |
{
|
489 |
"epoch": 16.74757281553398,
|
490 |
+
"grad_norm": 11.000115394592285,
|
491 |
"learning_rate": 2.209546925566343e-05,
|
492 |
+
"loss": 1.6398,
|
493 |
"step": 3450
|
494 |
},
|
495 |
{
|
496 |
"epoch": 16.990291262135923,
|
497 |
+
"grad_norm": 11.453404426574707,
|
498 |
"learning_rate": 2.169093851132686e-05,
|
499 |
+
"loss": 1.6466,
|
500 |
"step": 3500
|
501 |
},
|
502 |
{
|
503 |
"epoch": 17.233009708737864,
|
504 |
+
"grad_norm": 11.012279510498047,
|
505 |
"learning_rate": 2.1286407766990295e-05,
|
506 |
+
"loss": 1.4434,
|
507 |
"step": 3550
|
508 |
},
|
509 |
{
|
510 |
"epoch": 17.475728155339805,
|
511 |
+
"grad_norm": 11.989383697509766,
|
512 |
"learning_rate": 2.0881877022653725e-05,
|
513 |
+
"loss": 1.4708,
|
514 |
"step": 3600
|
515 |
},
|
516 |
{
|
517 |
"epoch": 17.718446601941746,
|
518 |
+
"grad_norm": 12.47008991241455,
|
519 |
"learning_rate": 2.047734627831715e-05,
|
520 |
+
"loss": 1.4783,
|
521 |
"step": 3650
|
522 |
},
|
523 |
{
|
524 |
"epoch": 17.96116504854369,
|
525 |
+
"grad_norm": 10.983756065368652,
|
526 |
"learning_rate": 2.007281553398058e-05,
|
527 |
+
"loss": 1.4839,
|
528 |
"step": 3700
|
529 |
},
|
530 |
{
|
531 |
"epoch": 18.203883495145632,
|
532 |
+
"grad_norm": 9.582510948181152,
|
533 |
"learning_rate": 1.9668284789644014e-05,
|
534 |
+
"loss": 1.3226,
|
535 |
"step": 3750
|
536 |
},
|
537 |
{
|
538 |
"epoch": 18.446601941747574,
|
539 |
+
"grad_norm": 11.663228034973145,
|
540 |
"learning_rate": 1.9263754045307444e-05,
|
541 |
+
"loss": 1.3522,
|
542 |
"step": 3800
|
543 |
},
|
544 |
{
|
545 |
"epoch": 18.689320388349515,
|
546 |
+
"grad_norm": 11.278139114379883,
|
547 |
"learning_rate": 1.8859223300970874e-05,
|
548 |
+
"loss": 1.3753,
|
549 |
"step": 3850
|
550 |
},
|
551 |
{
|
552 |
"epoch": 18.932038834951456,
|
553 |
+
"grad_norm": 10.586488723754883,
|
554 |
"learning_rate": 1.8454692556634304e-05,
|
555 |
+
"loss": 1.3774,
|
556 |
"step": 3900
|
557 |
},
|
558 |
{
|
559 |
"epoch": 19.174757281553397,
|
560 |
+
"grad_norm": 11.243274688720703,
|
561 |
"learning_rate": 1.8050161812297738e-05,
|
562 |
+
"loss": 1.2506,
|
563 |
"step": 3950
|
564 |
},
|
565 |
{
|
566 |
"epoch": 19.41747572815534,
|
567 |
+
"grad_norm": 10.601967811584473,
|
568 |
"learning_rate": 1.7645631067961167e-05,
|
569 |
+
"loss": 1.2276,
|
570 |
"step": 4000
|
571 |
}
|
572 |
],
|
checkpoints/checkpoint-4000/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7935e3b7629485cf26f624afe618b58a390d4c3c9c6dcf878b3ba9099ef1027b
|
3 |
size 5304
|
checkpoints/checkpoint-4500/config.json
CHANGED
@@ -10,12 +10,12 @@
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
-
"n_ctx":
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
-
"n_positions":
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
+
"n_ctx": 1024,
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
+
"n_positions": 1024,
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
checkpoints/checkpoint-4500/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b51a1bf42be97dd74f6362135ec6a10b42f4ea2ee64da3e8045792e198567c0d
|
3 |
+
size 176343496
|
checkpoints/checkpoint-4500/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c5627ce537e783ab06ced7a600ccfa407d8fc413d26712840d8afd6f239fb207
|
3 |
+
size 352735610
|
checkpoints/checkpoint-4500/trainer_state.json
CHANGED
@@ -11,632 +11,632 @@
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
-
"grad_norm": 2.
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
-
"loss": 6.
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
-
"grad_norm": 3.
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
-
"loss": 5.
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
-
"grad_norm": 4.
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
-
"loss": 5.
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
-
"grad_norm":
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
-
"loss": 5.
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
-
"grad_norm": 5.
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
-
"loss": 4.
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
-
"grad_norm": 5.
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
-
"loss": 4.
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
-
"grad_norm": 5.
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
-
"loss": 4.
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
-
"grad_norm": 5.
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
-
"loss": 4.
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
-
"grad_norm": 5.
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
-
"loss": 4.
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
-
"grad_norm": 5.
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
-
"loss": 4.
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
-
"grad_norm": 5.
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
-
"loss": 4.
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
-
"grad_norm": 5.
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
-
"loss": 4.
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
-
"grad_norm": 5.
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
-
"loss": 3.
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
-
"grad_norm": 5.
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
-
"loss": 3.
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
-
"grad_norm": 5.
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
-
"loss": 3.
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
-
"grad_norm": 5.
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
-
"loss": 3.
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
-
"grad_norm": 5.
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
-
"loss": 3.
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
-
"grad_norm":
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
-
"loss": 3.
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
-
"grad_norm": 5.
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
-
"loss": 3.
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
-
"grad_norm": 5.
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
-
"loss": 3.
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
-
"grad_norm": 5.
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
-
"loss": 3.
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
-
"grad_norm": 5.
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
-
"loss": 3.
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
-
"grad_norm": 6.
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
-
"loss": 3.
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
-
"grad_norm": 6.
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
-
"loss": 3.
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
-
"grad_norm": 6.
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
-
"loss": 3.
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
-
"grad_norm": 6.
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
-
"loss": 3.
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
-
"grad_norm": 6.
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
-
"loss": 3.
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
-
"grad_norm": 6.
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
-
"loss": 3.
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
-
"grad_norm": 6.
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
-
"loss": 3.
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
-
"grad_norm": 7.
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
-
"loss": 3.
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
-
"grad_norm": 7.
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
-
"loss": 3.
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
-
"grad_norm": 7.
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
-
"loss": 3.
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
-
"grad_norm":
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
-
"loss": 3.
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
-
"grad_norm": 7.
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
-
"loss": 2.
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
-
"grad_norm": 7.
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
-
"loss":
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
-
"grad_norm": 8.
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
-
"loss": 3.
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
-
"grad_norm":
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
-
"loss": 2.
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
-
"grad_norm": 7.
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
-
"loss": 2.
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
-
"grad_norm":
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
-
"loss": 2.
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
-
"grad_norm": 8.
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
-
"loss": 2.
|
290 |
"step": 2000
|
291 |
},
|
292 |
{
|
293 |
"epoch": 9.951456310679612,
|
294 |
-
"grad_norm": 8.
|
295 |
"learning_rate": 3.342233009708738e-05,
|
296 |
-
"loss": 2.
|
297 |
"step": 2050
|
298 |
},
|
299 |
{
|
300 |
"epoch": 10.194174757281553,
|
301 |
-
"grad_norm":
|
302 |
"learning_rate": 3.301779935275081e-05,
|
303 |
-
"loss": 2.
|
304 |
"step": 2100
|
305 |
},
|
306 |
{
|
307 |
"epoch": 10.436893203883495,
|
308 |
-
"grad_norm":
|
309 |
"learning_rate": 3.2613268608414246e-05,
|
310 |
-
"loss": 2.
|
311 |
"step": 2150
|
312 |
},
|
313 |
{
|
314 |
"epoch": 10.679611650485437,
|
315 |
-
"grad_norm": 9.
|
316 |
"learning_rate": 3.220873786407767e-05,
|
317 |
-
"loss": 2.
|
318 |
"step": 2200
|
319 |
},
|
320 |
{
|
321 |
"epoch": 10.922330097087379,
|
322 |
-
"grad_norm": 9.
|
323 |
"learning_rate": 3.1804207119741106e-05,
|
324 |
-
"loss": 2.
|
325 |
"step": 2250
|
326 |
},
|
327 |
{
|
328 |
"epoch": 11.16504854368932,
|
329 |
-
"grad_norm": 8.
|
330 |
"learning_rate": 3.139967637540453e-05,
|
331 |
-
"loss": 2.
|
332 |
"step": 2300
|
333 |
},
|
334 |
{
|
335 |
"epoch": 11.407766990291263,
|
336 |
-
"grad_norm": 9.
|
337 |
"learning_rate": 3.099514563106796e-05,
|
338 |
-
"loss": 2.
|
339 |
"step": 2350
|
340 |
},
|
341 |
{
|
342 |
"epoch": 11.650485436893204,
|
343 |
-
"grad_norm": 9.
|
344 |
"learning_rate": 3.059061488673139e-05,
|
345 |
-
"loss": 2.
|
346 |
"step": 2400
|
347 |
},
|
348 |
{
|
349 |
"epoch": 11.893203883495145,
|
350 |
-
"grad_norm": 9.
|
351 |
"learning_rate": 3.0186084142394822e-05,
|
352 |
-
"loss": 2.
|
353 |
"step": 2450
|
354 |
},
|
355 |
{
|
356 |
"epoch": 12.135922330097088,
|
357 |
-
"grad_norm": 9.
|
358 |
"learning_rate": 2.9781553398058252e-05,
|
359 |
-
"loss": 2.
|
360 |
"step": 2500
|
361 |
},
|
362 |
{
|
363 |
"epoch": 12.37864077669903,
|
364 |
-
"grad_norm":
|
365 |
"learning_rate": 2.9377022653721686e-05,
|
366 |
-
"loss": 2.
|
367 |
"step": 2550
|
368 |
},
|
369 |
{
|
370 |
"epoch": 12.62135922330097,
|
371 |
-
"grad_norm": 9.
|
372 |
"learning_rate": 2.8972491909385112e-05,
|
373 |
-
"loss": 2.
|
374 |
"step": 2600
|
375 |
},
|
376 |
{
|
377 |
"epoch": 12.864077669902912,
|
378 |
-
"grad_norm": 10.
|
379 |
"learning_rate": 2.8567961165048546e-05,
|
380 |
-
"loss": 2.
|
381 |
"step": 2650
|
382 |
},
|
383 |
{
|
384 |
"epoch": 13.106796116504855,
|
385 |
-
"grad_norm": 10.
|
386 |
"learning_rate": 2.816343042071198e-05,
|
387 |
-
"loss": 2.
|
388 |
"step": 2700
|
389 |
},
|
390 |
{
|
391 |
"epoch": 13.349514563106796,
|
392 |
-
"grad_norm":
|
393 |
"learning_rate": 2.7758899676375405e-05,
|
394 |
-
"loss": 2.
|
395 |
"step": 2750
|
396 |
},
|
397 |
{
|
398 |
"epoch": 13.592233009708737,
|
399 |
-
"grad_norm":
|
400 |
"learning_rate": 2.735436893203884e-05,
|
401 |
-
"loss": 2.
|
402 |
"step": 2800
|
403 |
},
|
404 |
{
|
405 |
"epoch": 13.83495145631068,
|
406 |
-
"grad_norm":
|
407 |
"learning_rate": 2.6949838187702265e-05,
|
408 |
-
"loss": 2.
|
409 |
"step": 2850
|
410 |
},
|
411 |
{
|
412 |
"epoch": 14.077669902912621,
|
413 |
-
"grad_norm": 9.
|
414 |
"learning_rate": 2.6545307443365695e-05,
|
415 |
-
"loss":
|
416 |
"step": 2900
|
417 |
},
|
418 |
{
|
419 |
"epoch": 14.320388349514563,
|
420 |
-
"grad_norm": 10.
|
421 |
"learning_rate": 2.614077669902913e-05,
|
422 |
-
"loss": 1.
|
423 |
"step": 2950
|
424 |
},
|
425 |
{
|
426 |
"epoch": 14.563106796116505,
|
427 |
-
"grad_norm": 10.
|
428 |
"learning_rate": 2.5736245954692555e-05,
|
429 |
-
"loss": 1.
|
430 |
"step": 3000
|
431 |
},
|
432 |
{
|
433 |
"epoch": 14.805825242718447,
|
434 |
-
"grad_norm":
|
435 |
"learning_rate": 2.533171521035599e-05,
|
436 |
-
"loss": 1.
|
437 |
"step": 3050
|
438 |
},
|
439 |
{
|
440 |
"epoch": 15.048543689320388,
|
441 |
-
"grad_norm": 9.
|
442 |
"learning_rate": 2.492718446601942e-05,
|
443 |
-
"loss": 1.
|
444 |
"step": 3100
|
445 |
},
|
446 |
{
|
447 |
"epoch": 15.29126213592233,
|
448 |
-
"grad_norm": 10.
|
449 |
"learning_rate": 2.452265372168285e-05,
|
450 |
-
"loss": 1.
|
451 |
"step": 3150
|
452 |
},
|
453 |
{
|
454 |
"epoch": 15.533980582524272,
|
455 |
-
"grad_norm": 10.
|
456 |
"learning_rate": 2.411812297734628e-05,
|
457 |
-
"loss": 1.
|
458 |
"step": 3200
|
459 |
},
|
460 |
{
|
461 |
"epoch": 15.776699029126213,
|
462 |
-
"grad_norm": 11.
|
463 |
"learning_rate": 2.3713592233009708e-05,
|
464 |
-
"loss": 1.
|
465 |
"step": 3250
|
466 |
},
|
467 |
{
|
468 |
"epoch": 16.019417475728154,
|
469 |
-
"grad_norm":
|
470 |
"learning_rate": 2.3309061488673138e-05,
|
471 |
-
"loss": 1.
|
472 |
"step": 3300
|
473 |
},
|
474 |
{
|
475 |
"epoch": 16.262135922330096,
|
476 |
-
"grad_norm": 9.
|
477 |
"learning_rate": 2.290453074433657e-05,
|
478 |
-
"loss": 1.
|
479 |
"step": 3350
|
480 |
},
|
481 |
{
|
482 |
"epoch": 16.50485436893204,
|
483 |
-
"grad_norm": 11.
|
484 |
"learning_rate": 2.25e-05,
|
485 |
-
"loss": 1.
|
486 |
"step": 3400
|
487 |
},
|
488 |
{
|
489 |
"epoch": 16.74757281553398,
|
490 |
-
"grad_norm": 11.
|
491 |
"learning_rate": 2.209546925566343e-05,
|
492 |
-
"loss": 1.
|
493 |
"step": 3450
|
494 |
},
|
495 |
{
|
496 |
"epoch": 16.990291262135923,
|
497 |
-
"grad_norm": 11.
|
498 |
"learning_rate": 2.169093851132686e-05,
|
499 |
-
"loss": 1.
|
500 |
"step": 3500
|
501 |
},
|
502 |
{
|
503 |
"epoch": 17.233009708737864,
|
504 |
-
"grad_norm":
|
505 |
"learning_rate": 2.1286407766990295e-05,
|
506 |
-
"loss": 1.
|
507 |
"step": 3550
|
508 |
},
|
509 |
{
|
510 |
"epoch": 17.475728155339805,
|
511 |
-
"grad_norm": 11.
|
512 |
"learning_rate": 2.0881877022653725e-05,
|
513 |
-
"loss": 1.
|
514 |
"step": 3600
|
515 |
},
|
516 |
{
|
517 |
"epoch": 17.718446601941746,
|
518 |
-
"grad_norm":
|
519 |
"learning_rate": 2.047734627831715e-05,
|
520 |
-
"loss": 1.
|
521 |
"step": 3650
|
522 |
},
|
523 |
{
|
524 |
"epoch": 17.96116504854369,
|
525 |
-
"grad_norm":
|
526 |
"learning_rate": 2.007281553398058e-05,
|
527 |
-
"loss": 1.
|
528 |
"step": 3700
|
529 |
},
|
530 |
{
|
531 |
"epoch": 18.203883495145632,
|
532 |
-
"grad_norm": 9.
|
533 |
"learning_rate": 1.9668284789644014e-05,
|
534 |
-
"loss": 1.
|
535 |
"step": 3750
|
536 |
},
|
537 |
{
|
538 |
"epoch": 18.446601941747574,
|
539 |
-
"grad_norm":
|
540 |
"learning_rate": 1.9263754045307444e-05,
|
541 |
-
"loss": 1.
|
542 |
"step": 3800
|
543 |
},
|
544 |
{
|
545 |
"epoch": 18.689320388349515,
|
546 |
-
"grad_norm": 11.
|
547 |
"learning_rate": 1.8859223300970874e-05,
|
548 |
-
"loss": 1.
|
549 |
"step": 3850
|
550 |
},
|
551 |
{
|
552 |
"epoch": 18.932038834951456,
|
553 |
-
"grad_norm": 10.
|
554 |
"learning_rate": 1.8454692556634304e-05,
|
555 |
-
"loss": 1.
|
556 |
"step": 3900
|
557 |
},
|
558 |
{
|
559 |
"epoch": 19.174757281553397,
|
560 |
-
"grad_norm":
|
561 |
"learning_rate": 1.8050161812297738e-05,
|
562 |
-
"loss": 1.
|
563 |
"step": 3950
|
564 |
},
|
565 |
{
|
566 |
"epoch": 19.41747572815534,
|
567 |
-
"grad_norm":
|
568 |
"learning_rate": 1.7645631067961167e-05,
|
569 |
-
"loss": 1.
|
570 |
"step": 4000
|
571 |
},
|
572 |
{
|
573 |
"epoch": 19.660194174757283,
|
574 |
-
"grad_norm": 10.
|
575 |
"learning_rate": 1.7241100323624594e-05,
|
576 |
-
"loss": 1.
|
577 |
"step": 4050
|
578 |
},
|
579 |
{
|
580 |
"epoch": 19.902912621359224,
|
581 |
-
"grad_norm": 11.
|
582 |
"learning_rate": 1.6836569579288027e-05,
|
583 |
-
"loss": 1.
|
584 |
"step": 4100
|
585 |
},
|
586 |
{
|
587 |
"epoch": 20.145631067961165,
|
588 |
-
"grad_norm": 10.
|
589 |
"learning_rate": 1.6432038834951457e-05,
|
590 |
-
"loss": 1.
|
591 |
"step": 4150
|
592 |
},
|
593 |
{
|
594 |
"epoch": 20.388349514563107,
|
595 |
-
"grad_norm":
|
596 |
"learning_rate": 1.6027508090614887e-05,
|
597 |
-
"loss": 1.
|
598 |
"step": 4200
|
599 |
},
|
600 |
{
|
601 |
"epoch": 20.631067961165048,
|
602 |
-
"grad_norm": 10.
|
603 |
"learning_rate": 1.5622977346278317e-05,
|
604 |
-
"loss": 1.
|
605 |
"step": 4250
|
606 |
},
|
607 |
{
|
608 |
"epoch": 20.87378640776699,
|
609 |
-
"grad_norm": 11.
|
610 |
"learning_rate": 1.5218446601941749e-05,
|
611 |
-
"loss": 1.
|
612 |
"step": 4300
|
613 |
},
|
614 |
{
|
615 |
"epoch": 21.116504854368934,
|
616 |
-
"grad_norm": 11.
|
617 |
"learning_rate": 1.4813915857605179e-05,
|
618 |
-
"loss": 1.
|
619 |
"step": 4350
|
620 |
},
|
621 |
{
|
622 |
"epoch": 21.359223300970875,
|
623 |
-
"grad_norm": 10.
|
624 |
"learning_rate": 1.4409385113268609e-05,
|
625 |
-
"loss": 1.
|
626 |
"step": 4400
|
627 |
},
|
628 |
{
|
629 |
"epoch": 21.601941747572816,
|
630 |
-
"grad_norm": 11.
|
631 |
"learning_rate": 1.4004854368932039e-05,
|
632 |
-
"loss": 1.
|
633 |
"step": 4450
|
634 |
},
|
635 |
{
|
636 |
"epoch": 21.844660194174757,
|
637 |
-
"grad_norm":
|
638 |
"learning_rate": 1.360032362459547e-05,
|
639 |
-
"loss": 1.
|
640 |
"step": 4500
|
641 |
}
|
642 |
],
|
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
+
"grad_norm": 2.239194869995117,
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
+
"loss": 6.099,
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
+
"grad_norm": 3.8653194904327393,
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
+
"loss": 5.6954,
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
+
"grad_norm": 4.652886867523193,
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
+
"loss": 5.3598,
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
+
"grad_norm": 5.048999309539795,
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
+
"loss": 5.0982,
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
+
"grad_norm": 5.21944522857666,
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
+
"loss": 4.7456,
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
+
"grad_norm": 5.471007347106934,
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
+
"loss": 4.6254,
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
+
"grad_norm": 5.345245838165283,
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
+
"loss": 4.5216,
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
+
"grad_norm": 5.54531192779541,
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
+
"loss": 4.4353,
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
+
"grad_norm": 5.244582176208496,
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
+
"loss": 4.2444,
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
+
"grad_norm": 5.338261127471924,
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
+
"loss": 4.1915,
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
+
"grad_norm": 5.35228157043457,
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
+
"loss": 4.1612,
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
+
"grad_norm": 5.287161827087402,
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
+
"loss": 4.1322,
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
+
"grad_norm": 5.2932209968566895,
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
+
"loss": 3.9684,
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
+
"grad_norm": 5.4360833168029785,
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
+
"loss": 3.8933,
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
+
"grad_norm": 5.5595221519470215,
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
+
"loss": 3.9306,
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
+
"grad_norm": 5.540086269378662,
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
+
"loss": 3.9317,
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
+
"grad_norm": 5.567137718200684,
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
+
"loss": 3.8135,
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
+
"grad_norm": 6.00346040725708,
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
+
"loss": 3.6924,
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
+
"grad_norm": 5.635782718658447,
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
+
"loss": 3.7304,
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
+
"grad_norm": 5.9545440673828125,
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
+
"loss": 3.7471,
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
+
"grad_norm": 5.696166038513184,
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
+
"loss": 3.6287,
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
+
"grad_norm": 5.844620704650879,
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
+
"loss": 3.5299,
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
+
"grad_norm": 6.373577117919922,
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
+
"loss": 3.5285,
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
+
"grad_norm": 6.488651275634766,
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
+
"loss": 3.5789,
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
+
"grad_norm": 6.170980930328369,
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
+
"loss": 3.494,
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
+
"grad_norm": 6.387782573699951,
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
+
"loss": 3.353,
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
+
"grad_norm": 6.718175888061523,
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
+
"loss": 3.3521,
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
+
"grad_norm": 6.7870025634765625,
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
+
"loss": 3.3658,
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
+
"grad_norm": 6.513721466064453,
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
+
"loss": 3.3301,
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
+
"grad_norm": 7.0680036544799805,
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
+
"loss": 3.1289,
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
+
"grad_norm": 7.019896507263184,
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
+
"loss": 3.1878,
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
+
"grad_norm": 7.3232197761535645,
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
+
"loss": 3.2006,
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
+
"grad_norm": 7.003912925720215,
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
+
"loss": 3.1737,
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
+
"grad_norm": 7.592433929443359,
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
+
"loss": 2.9258,
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
+
"grad_norm": 7.629278182983398,
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
+
"loss": 2.9999,
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
+
"grad_norm": 8.137636184692383,
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
+
"loss": 3.0155,
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
+
"grad_norm": 7.967227935791016,
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
+
"loss": 2.9945,
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
+
"grad_norm": 7.9910078048706055,
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
+
"loss": 2.7456,
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
+
"grad_norm": 8.552720069885254,
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
+
"loss": 2.8071,
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
+
"grad_norm": 8.336930274963379,
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
+
"loss": 2.8117,
|
290 |
"step": 2000
|
291 |
},
|
292 |
{
|
293 |
"epoch": 9.951456310679612,
|
294 |
+
"grad_norm": 8.636338233947754,
|
295 |
"learning_rate": 3.342233009708738e-05,
|
296 |
+
"loss": 2.8197,
|
297 |
"step": 2050
|
298 |
},
|
299 |
{
|
300 |
"epoch": 10.194174757281553,
|
301 |
+
"grad_norm": 7.955165863037109,
|
302 |
"learning_rate": 3.301779935275081e-05,
|
303 |
+
"loss": 2.6245,
|
304 |
"step": 2100
|
305 |
},
|
306 |
{
|
307 |
"epoch": 10.436893203883495,
|
308 |
+
"grad_norm": 8.679149627685547,
|
309 |
"learning_rate": 3.2613268608414246e-05,
|
310 |
+
"loss": 2.565,
|
311 |
"step": 2150
|
312 |
},
|
313 |
{
|
314 |
"epoch": 10.679611650485437,
|
315 |
+
"grad_norm": 9.150839805603027,
|
316 |
"learning_rate": 3.220873786407767e-05,
|
317 |
+
"loss": 2.6258,
|
318 |
"step": 2200
|
319 |
},
|
320 |
{
|
321 |
"epoch": 10.922330097087379,
|
322 |
+
"grad_norm": 9.2735013961792,
|
323 |
"learning_rate": 3.1804207119741106e-05,
|
324 |
+
"loss": 2.6439,
|
325 |
"step": 2250
|
326 |
},
|
327 |
{
|
328 |
"epoch": 11.16504854368932,
|
329 |
+
"grad_norm": 8.620939254760742,
|
330 |
"learning_rate": 3.139967637540453e-05,
|
331 |
+
"loss": 2.4371,
|
332 |
"step": 2300
|
333 |
},
|
334 |
{
|
335 |
"epoch": 11.407766990291263,
|
336 |
+
"grad_norm": 9.722040176391602,
|
337 |
"learning_rate": 3.099514563106796e-05,
|
338 |
+
"loss": 2.4074,
|
339 |
"step": 2350
|
340 |
},
|
341 |
{
|
342 |
"epoch": 11.650485436893204,
|
343 |
+
"grad_norm": 9.200730323791504,
|
344 |
"learning_rate": 3.059061488673139e-05,
|
345 |
+
"loss": 2.4307,
|
346 |
"step": 2400
|
347 |
},
|
348 |
{
|
349 |
"epoch": 11.893203883495145,
|
350 |
+
"grad_norm": 9.692273139953613,
|
351 |
"learning_rate": 3.0186084142394822e-05,
|
352 |
+
"loss": 2.424,
|
353 |
"step": 2450
|
354 |
},
|
355 |
{
|
356 |
"epoch": 12.135922330097088,
|
357 |
+
"grad_norm": 9.822249412536621,
|
358 |
"learning_rate": 2.9781553398058252e-05,
|
359 |
+
"loss": 2.2862,
|
360 |
"step": 2500
|
361 |
},
|
362 |
{
|
363 |
"epoch": 12.37864077669903,
|
364 |
+
"grad_norm": 9.919060707092285,
|
365 |
"learning_rate": 2.9377022653721686e-05,
|
366 |
+
"loss": 2.1913,
|
367 |
"step": 2550
|
368 |
},
|
369 |
{
|
370 |
"epoch": 12.62135922330097,
|
371 |
+
"grad_norm": 9.181950569152832,
|
372 |
"learning_rate": 2.8972491909385112e-05,
|
373 |
+
"loss": 2.241,
|
374 |
"step": 2600
|
375 |
},
|
376 |
{
|
377 |
"epoch": 12.864077669902912,
|
378 |
+
"grad_norm": 10.82077407836914,
|
379 |
"learning_rate": 2.8567961165048546e-05,
|
380 |
+
"loss": 2.2653,
|
381 |
"step": 2650
|
382 |
},
|
383 |
{
|
384 |
"epoch": 13.106796116504855,
|
385 |
+
"grad_norm": 10.365830421447754,
|
386 |
"learning_rate": 2.816343042071198e-05,
|
387 |
+
"loss": 2.1352,
|
388 |
"step": 2700
|
389 |
},
|
390 |
{
|
391 |
"epoch": 13.349514563106796,
|
392 |
+
"grad_norm": 10.073640823364258,
|
393 |
"learning_rate": 2.7758899676375405e-05,
|
394 |
+
"loss": 2.0227,
|
395 |
"step": 2750
|
396 |
},
|
397 |
{
|
398 |
"epoch": 13.592233009708737,
|
399 |
+
"grad_norm": 10.520242691040039,
|
400 |
"learning_rate": 2.735436893203884e-05,
|
401 |
+
"loss": 2.0579,
|
402 |
"step": 2800
|
403 |
},
|
404 |
{
|
405 |
"epoch": 13.83495145631068,
|
406 |
+
"grad_norm": 11.176478385925293,
|
407 |
"learning_rate": 2.6949838187702265e-05,
|
408 |
+
"loss": 2.0934,
|
409 |
"step": 2850
|
410 |
},
|
411 |
{
|
412 |
"epoch": 14.077669902912621,
|
413 |
+
"grad_norm": 9.541868209838867,
|
414 |
"learning_rate": 2.6545307443365695e-05,
|
415 |
+
"loss": 2.0118,
|
416 |
"step": 2900
|
417 |
},
|
418 |
{
|
419 |
"epoch": 14.320388349514563,
|
420 |
+
"grad_norm": 10.162981986999512,
|
421 |
"learning_rate": 2.614077669902913e-05,
|
422 |
+
"loss": 1.8624,
|
423 |
"step": 2950
|
424 |
},
|
425 |
{
|
426 |
"epoch": 14.563106796116505,
|
427 |
+
"grad_norm": 10.51291275024414,
|
428 |
"learning_rate": 2.5736245954692555e-05,
|
429 |
+
"loss": 1.8979,
|
430 |
"step": 3000
|
431 |
},
|
432 |
{
|
433 |
"epoch": 14.805825242718447,
|
434 |
+
"grad_norm": 11.316717147827148,
|
435 |
"learning_rate": 2.533171521035599e-05,
|
436 |
+
"loss": 1.915,
|
437 |
"step": 3050
|
438 |
},
|
439 |
{
|
440 |
"epoch": 15.048543689320388,
|
441 |
+
"grad_norm": 9.184353828430176,
|
442 |
"learning_rate": 2.492718446601942e-05,
|
443 |
+
"loss": 1.885,
|
444 |
"step": 3100
|
445 |
},
|
446 |
{
|
447 |
"epoch": 15.29126213592233,
|
448 |
+
"grad_norm": 10.70876693725586,
|
449 |
"learning_rate": 2.452265372168285e-05,
|
450 |
+
"loss": 1.7024,
|
451 |
"step": 3150
|
452 |
},
|
453 |
{
|
454 |
"epoch": 15.533980582524272,
|
455 |
+
"grad_norm": 10.767953872680664,
|
456 |
"learning_rate": 2.411812297734628e-05,
|
457 |
+
"loss": 1.729,
|
458 |
"step": 3200
|
459 |
},
|
460 |
{
|
461 |
"epoch": 15.776699029126213,
|
462 |
+
"grad_norm": 11.83687973022461,
|
463 |
"learning_rate": 2.3713592233009708e-05,
|
464 |
+
"loss": 1.7697,
|
465 |
"step": 3250
|
466 |
},
|
467 |
{
|
468 |
"epoch": 16.019417475728154,
|
469 |
+
"grad_norm": 9.888689994812012,
|
470 |
"learning_rate": 2.3309061488673138e-05,
|
471 |
+
"loss": 1.7605,
|
472 |
"step": 3300
|
473 |
},
|
474 |
{
|
475 |
"epoch": 16.262135922330096,
|
476 |
+
"grad_norm": 9.394012451171875,
|
477 |
"learning_rate": 2.290453074433657e-05,
|
478 |
+
"loss": 1.5389,
|
479 |
"step": 3350
|
480 |
},
|
481 |
{
|
482 |
"epoch": 16.50485436893204,
|
483 |
+
"grad_norm": 11.3172025680542,
|
484 |
"learning_rate": 2.25e-05,
|
485 |
+
"loss": 1.5836,
|
486 |
"step": 3400
|
487 |
},
|
488 |
{
|
489 |
"epoch": 16.74757281553398,
|
490 |
+
"grad_norm": 11.000115394592285,
|
491 |
"learning_rate": 2.209546925566343e-05,
|
492 |
+
"loss": 1.6398,
|
493 |
"step": 3450
|
494 |
},
|
495 |
{
|
496 |
"epoch": 16.990291262135923,
|
497 |
+
"grad_norm": 11.453404426574707,
|
498 |
"learning_rate": 2.169093851132686e-05,
|
499 |
+
"loss": 1.6466,
|
500 |
"step": 3500
|
501 |
},
|
502 |
{
|
503 |
"epoch": 17.233009708737864,
|
504 |
+
"grad_norm": 11.012279510498047,
|
505 |
"learning_rate": 2.1286407766990295e-05,
|
506 |
+
"loss": 1.4434,
|
507 |
"step": 3550
|
508 |
},
|
509 |
{
|
510 |
"epoch": 17.475728155339805,
|
511 |
+
"grad_norm": 11.989383697509766,
|
512 |
"learning_rate": 2.0881877022653725e-05,
|
513 |
+
"loss": 1.4708,
|
514 |
"step": 3600
|
515 |
},
|
516 |
{
|
517 |
"epoch": 17.718446601941746,
|
518 |
+
"grad_norm": 12.47008991241455,
|
519 |
"learning_rate": 2.047734627831715e-05,
|
520 |
+
"loss": 1.4783,
|
521 |
"step": 3650
|
522 |
},
|
523 |
{
|
524 |
"epoch": 17.96116504854369,
|
525 |
+
"grad_norm": 10.983756065368652,
|
526 |
"learning_rate": 2.007281553398058e-05,
|
527 |
+
"loss": 1.4839,
|
528 |
"step": 3700
|
529 |
},
|
530 |
{
|
531 |
"epoch": 18.203883495145632,
|
532 |
+
"grad_norm": 9.582510948181152,
|
533 |
"learning_rate": 1.9668284789644014e-05,
|
534 |
+
"loss": 1.3226,
|
535 |
"step": 3750
|
536 |
},
|
537 |
{
|
538 |
"epoch": 18.446601941747574,
|
539 |
+
"grad_norm": 11.663228034973145,
|
540 |
"learning_rate": 1.9263754045307444e-05,
|
541 |
+
"loss": 1.3522,
|
542 |
"step": 3800
|
543 |
},
|
544 |
{
|
545 |
"epoch": 18.689320388349515,
|
546 |
+
"grad_norm": 11.278139114379883,
|
547 |
"learning_rate": 1.8859223300970874e-05,
|
548 |
+
"loss": 1.3753,
|
549 |
"step": 3850
|
550 |
},
|
551 |
{
|
552 |
"epoch": 18.932038834951456,
|
553 |
+
"grad_norm": 10.586488723754883,
|
554 |
"learning_rate": 1.8454692556634304e-05,
|
555 |
+
"loss": 1.3774,
|
556 |
"step": 3900
|
557 |
},
|
558 |
{
|
559 |
"epoch": 19.174757281553397,
|
560 |
+
"grad_norm": 11.243274688720703,
|
561 |
"learning_rate": 1.8050161812297738e-05,
|
562 |
+
"loss": 1.2506,
|
563 |
"step": 3950
|
564 |
},
|
565 |
{
|
566 |
"epoch": 19.41747572815534,
|
567 |
+
"grad_norm": 10.601967811584473,
|
568 |
"learning_rate": 1.7645631067961167e-05,
|
569 |
+
"loss": 1.2276,
|
570 |
"step": 4000
|
571 |
},
|
572 |
{
|
573 |
"epoch": 19.660194174757283,
|
574 |
+
"grad_norm": 10.919422149658203,
|
575 |
"learning_rate": 1.7241100323624594e-05,
|
576 |
+
"loss": 1.2518,
|
577 |
"step": 4050
|
578 |
},
|
579 |
{
|
580 |
"epoch": 19.902912621359224,
|
581 |
+
"grad_norm": 11.303253173828125,
|
582 |
"learning_rate": 1.6836569579288027e-05,
|
583 |
+
"loss": 1.279,
|
584 |
"step": 4100
|
585 |
},
|
586 |
{
|
587 |
"epoch": 20.145631067961165,
|
588 |
+
"grad_norm": 10.864278793334961,
|
589 |
"learning_rate": 1.6432038834951457e-05,
|
590 |
+
"loss": 1.1686,
|
591 |
"step": 4150
|
592 |
},
|
593 |
{
|
594 |
"epoch": 20.388349514563107,
|
595 |
+
"grad_norm": 11.179913520812988,
|
596 |
"learning_rate": 1.6027508090614887e-05,
|
597 |
+
"loss": 1.1252,
|
598 |
"step": 4200
|
599 |
},
|
600 |
{
|
601 |
"epoch": 20.631067961165048,
|
602 |
+
"grad_norm": 10.640514373779297,
|
603 |
"learning_rate": 1.5622977346278317e-05,
|
604 |
+
"loss": 1.1581,
|
605 |
"step": 4250
|
606 |
},
|
607 |
{
|
608 |
"epoch": 20.87378640776699,
|
609 |
+
"grad_norm": 11.186044692993164,
|
610 |
"learning_rate": 1.5218446601941749e-05,
|
611 |
+
"loss": 1.1782,
|
612 |
"step": 4300
|
613 |
},
|
614 |
{
|
615 |
"epoch": 21.116504854368934,
|
616 |
+
"grad_norm": 11.105583190917969,
|
617 |
"learning_rate": 1.4813915857605179e-05,
|
618 |
+
"loss": 1.0972,
|
619 |
"step": 4350
|
620 |
},
|
621 |
{
|
622 |
"epoch": 21.359223300970875,
|
623 |
+
"grad_norm": 10.809669494628906,
|
624 |
"learning_rate": 1.4409385113268609e-05,
|
625 |
+
"loss": 1.0487,
|
626 |
"step": 4400
|
627 |
},
|
628 |
{
|
629 |
"epoch": 21.601941747572816,
|
630 |
+
"grad_norm": 11.11020278930664,
|
631 |
"learning_rate": 1.4004854368932039e-05,
|
632 |
+
"loss": 1.0566,
|
633 |
"step": 4450
|
634 |
},
|
635 |
{
|
636 |
"epoch": 21.844660194174757,
|
637 |
+
"grad_norm": 11.062104225158691,
|
638 |
"learning_rate": 1.360032362459547e-05,
|
639 |
+
"loss": 1.078,
|
640 |
"step": 4500
|
641 |
}
|
642 |
],
|
checkpoints/checkpoint-4500/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7935e3b7629485cf26f624afe618b58a390d4c3c9c6dcf878b3ba9099ef1027b
|
3 |
size 5304
|
checkpoints/checkpoint-500/config.json
CHANGED
@@ -10,12 +10,12 @@
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
-
"n_ctx":
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
-
"n_positions":
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
+
"n_ctx": 1024,
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
+
"n_positions": 1024,
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
checkpoints/checkpoint-500/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e4c1afbb39df3411cf84f25ad19241038d8abaac41edf1c0a677dd67a17d89ed
|
3 |
+
size 176343496
|
checkpoints/checkpoint-500/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8da3e16f41a201489b2992d544d1b99e830f95e73f551e3c0c5458655519c05e
|
3 |
+
size 352735610
|
checkpoints/checkpoint-500/trainer_state.json
CHANGED
@@ -11,72 +11,72 @@
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
-
"grad_norm": 2.
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
-
"loss": 6.
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
-
"grad_norm": 3.
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
-
"loss": 5.
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
-
"grad_norm": 4.
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
-
"loss": 5.
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
-
"grad_norm":
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
-
"loss": 5.
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
-
"grad_norm": 5.
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
-
"loss": 4.
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
-
"grad_norm": 5.
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
-
"loss": 4.
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
-
"grad_norm": 5.
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
-
"loss": 4.
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
-
"grad_norm": 5.
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
-
"loss": 4.
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
-
"grad_norm": 5.
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
-
"loss": 4.
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
-
"grad_norm": 5.
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
-
"loss": 4.
|
80 |
"step": 500
|
81 |
}
|
82 |
],
|
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
+
"grad_norm": 2.239194869995117,
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
+
"loss": 6.099,
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
+
"grad_norm": 3.8653194904327393,
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
+
"loss": 5.6954,
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
+
"grad_norm": 4.652886867523193,
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
+
"loss": 5.3598,
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
+
"grad_norm": 5.048999309539795,
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
+
"loss": 5.0982,
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
+
"grad_norm": 5.21944522857666,
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
+
"loss": 4.7456,
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
+
"grad_norm": 5.471007347106934,
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
+
"loss": 4.6254,
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
+
"grad_norm": 5.345245838165283,
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
+
"loss": 4.5216,
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
+
"grad_norm": 5.54531192779541,
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
+
"loss": 4.4353,
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
+
"grad_norm": 5.244582176208496,
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
+
"loss": 4.2444,
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
+
"grad_norm": 5.338261127471924,
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
+
"loss": 4.1915,
|
80 |
"step": 500
|
81 |
}
|
82 |
],
|
checkpoints/checkpoint-500/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7935e3b7629485cf26f624afe618b58a390d4c3c9c6dcf878b3ba9099ef1027b
|
3 |
size 5304
|
checkpoints/checkpoint-5000/config.json
CHANGED
@@ -10,12 +10,12 @@
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
-
"n_ctx":
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
-
"n_positions":
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
|
|
10 |
"initializer_range": 0.02,
|
11 |
"layer_norm_epsilon": 1e-05,
|
12 |
"model_type": "gpt2",
|
13 |
+
"n_ctx": 1024,
|
14 |
"n_embd": 768,
|
15 |
"n_head": 2,
|
16 |
"n_inner": null,
|
17 |
"n_layer": 6,
|
18 |
+
"n_positions": 1024,
|
19 |
"pad_token_id": 1,
|
20 |
"reorder_and_upcast_attn": false,
|
21 |
"resid_pdrop": 0.1,
|
checkpoints/checkpoint-5000/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cca1bc49b67480ffa73ef99d3dd4267ece3cd384aa12183c64d2d5627624ab95
|
3 |
+
size 176343496
|
checkpoints/checkpoint-5000/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d2dde814f5f4a6369ec42b5837437fea636742525202277486f84437ae7a67cb
|
3 |
+
size 352735610
|
checkpoints/checkpoint-5000/trainer_state.json
CHANGED
@@ -11,702 +11,702 @@
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
-
"grad_norm": 2.
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
-
"loss": 6.
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
-
"grad_norm": 3.
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
-
"loss": 5.
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
-
"grad_norm": 4.
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
-
"loss": 5.
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
-
"grad_norm":
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
-
"loss": 5.
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
-
"grad_norm": 5.
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
-
"loss": 4.
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
-
"grad_norm": 5.
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
-
"loss": 4.
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
-
"grad_norm": 5.
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
-
"loss": 4.
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
-
"grad_norm": 5.
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
-
"loss": 4.
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
-
"grad_norm": 5.
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
-
"loss": 4.
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
-
"grad_norm": 5.
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
-
"loss": 4.
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
-
"grad_norm": 5.
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
-
"loss": 4.
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
-
"grad_norm": 5.
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
-
"loss": 4.
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
-
"grad_norm": 5.
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
-
"loss": 3.
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
-
"grad_norm": 5.
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
-
"loss": 3.
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
-
"grad_norm": 5.
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
-
"loss": 3.
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
-
"grad_norm": 5.
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
-
"loss": 3.
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
-
"grad_norm": 5.
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
-
"loss": 3.
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
-
"grad_norm":
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
-
"loss": 3.
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
-
"grad_norm": 5.
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
-
"loss": 3.
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
-
"grad_norm": 5.
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
-
"loss": 3.
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
-
"grad_norm": 5.
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
-
"loss": 3.
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
-
"grad_norm": 5.
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
-
"loss": 3.
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
-
"grad_norm": 6.
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
-
"loss": 3.
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
-
"grad_norm": 6.
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
-
"loss": 3.
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
-
"grad_norm": 6.
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
-
"loss": 3.
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
-
"grad_norm": 6.
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
-
"loss": 3.
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
-
"grad_norm": 6.
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
-
"loss": 3.
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
-
"grad_norm": 6.
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
-
"loss": 3.
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
-
"grad_norm": 6.
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
-
"loss": 3.
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
-
"grad_norm": 7.
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
-
"loss": 3.
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
-
"grad_norm": 7.
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
-
"loss": 3.
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
-
"grad_norm": 7.
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
-
"loss": 3.
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
-
"grad_norm":
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
-
"loss": 3.
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
-
"grad_norm": 7.
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
-
"loss": 2.
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
-
"grad_norm": 7.
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
-
"loss":
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
-
"grad_norm": 8.
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
-
"loss": 3.
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
-
"grad_norm":
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
-
"loss": 2.
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
-
"grad_norm": 7.
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
-
"loss": 2.
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
-
"grad_norm":
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
-
"loss": 2.
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
-
"grad_norm": 8.
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
-
"loss": 2.
|
290 |
"step": 2000
|
291 |
},
|
292 |
{
|
293 |
"epoch": 9.951456310679612,
|
294 |
-
"grad_norm": 8.
|
295 |
"learning_rate": 3.342233009708738e-05,
|
296 |
-
"loss": 2.
|
297 |
"step": 2050
|
298 |
},
|
299 |
{
|
300 |
"epoch": 10.194174757281553,
|
301 |
-
"grad_norm":
|
302 |
"learning_rate": 3.301779935275081e-05,
|
303 |
-
"loss": 2.
|
304 |
"step": 2100
|
305 |
},
|
306 |
{
|
307 |
"epoch": 10.436893203883495,
|
308 |
-
"grad_norm":
|
309 |
"learning_rate": 3.2613268608414246e-05,
|
310 |
-
"loss": 2.
|
311 |
"step": 2150
|
312 |
},
|
313 |
{
|
314 |
"epoch": 10.679611650485437,
|
315 |
-
"grad_norm": 9.
|
316 |
"learning_rate": 3.220873786407767e-05,
|
317 |
-
"loss": 2.
|
318 |
"step": 2200
|
319 |
},
|
320 |
{
|
321 |
"epoch": 10.922330097087379,
|
322 |
-
"grad_norm": 9.
|
323 |
"learning_rate": 3.1804207119741106e-05,
|
324 |
-
"loss": 2.
|
325 |
"step": 2250
|
326 |
},
|
327 |
{
|
328 |
"epoch": 11.16504854368932,
|
329 |
-
"grad_norm": 8.
|
330 |
"learning_rate": 3.139967637540453e-05,
|
331 |
-
"loss": 2.
|
332 |
"step": 2300
|
333 |
},
|
334 |
{
|
335 |
"epoch": 11.407766990291263,
|
336 |
-
"grad_norm": 9.
|
337 |
"learning_rate": 3.099514563106796e-05,
|
338 |
-
"loss": 2.
|
339 |
"step": 2350
|
340 |
},
|
341 |
{
|
342 |
"epoch": 11.650485436893204,
|
343 |
-
"grad_norm": 9.
|
344 |
"learning_rate": 3.059061488673139e-05,
|
345 |
-
"loss": 2.
|
346 |
"step": 2400
|
347 |
},
|
348 |
{
|
349 |
"epoch": 11.893203883495145,
|
350 |
-
"grad_norm": 9.
|
351 |
"learning_rate": 3.0186084142394822e-05,
|
352 |
-
"loss": 2.
|
353 |
"step": 2450
|
354 |
},
|
355 |
{
|
356 |
"epoch": 12.135922330097088,
|
357 |
-
"grad_norm": 9.
|
358 |
"learning_rate": 2.9781553398058252e-05,
|
359 |
-
"loss": 2.
|
360 |
"step": 2500
|
361 |
},
|
362 |
{
|
363 |
"epoch": 12.37864077669903,
|
364 |
-
"grad_norm":
|
365 |
"learning_rate": 2.9377022653721686e-05,
|
366 |
-
"loss": 2.
|
367 |
"step": 2550
|
368 |
},
|
369 |
{
|
370 |
"epoch": 12.62135922330097,
|
371 |
-
"grad_norm": 9.
|
372 |
"learning_rate": 2.8972491909385112e-05,
|
373 |
-
"loss": 2.
|
374 |
"step": 2600
|
375 |
},
|
376 |
{
|
377 |
"epoch": 12.864077669902912,
|
378 |
-
"grad_norm": 10.
|
379 |
"learning_rate": 2.8567961165048546e-05,
|
380 |
-
"loss": 2.
|
381 |
"step": 2650
|
382 |
},
|
383 |
{
|
384 |
"epoch": 13.106796116504855,
|
385 |
-
"grad_norm": 10.
|
386 |
"learning_rate": 2.816343042071198e-05,
|
387 |
-
"loss": 2.
|
388 |
"step": 2700
|
389 |
},
|
390 |
{
|
391 |
"epoch": 13.349514563106796,
|
392 |
-
"grad_norm":
|
393 |
"learning_rate": 2.7758899676375405e-05,
|
394 |
-
"loss": 2.
|
395 |
"step": 2750
|
396 |
},
|
397 |
{
|
398 |
"epoch": 13.592233009708737,
|
399 |
-
"grad_norm":
|
400 |
"learning_rate": 2.735436893203884e-05,
|
401 |
-
"loss": 2.
|
402 |
"step": 2800
|
403 |
},
|
404 |
{
|
405 |
"epoch": 13.83495145631068,
|
406 |
-
"grad_norm":
|
407 |
"learning_rate": 2.6949838187702265e-05,
|
408 |
-
"loss": 2.
|
409 |
"step": 2850
|
410 |
},
|
411 |
{
|
412 |
"epoch": 14.077669902912621,
|
413 |
-
"grad_norm": 9.
|
414 |
"learning_rate": 2.6545307443365695e-05,
|
415 |
-
"loss":
|
416 |
"step": 2900
|
417 |
},
|
418 |
{
|
419 |
"epoch": 14.320388349514563,
|
420 |
-
"grad_norm": 10.
|
421 |
"learning_rate": 2.614077669902913e-05,
|
422 |
-
"loss": 1.
|
423 |
"step": 2950
|
424 |
},
|
425 |
{
|
426 |
"epoch": 14.563106796116505,
|
427 |
-
"grad_norm": 10.
|
428 |
"learning_rate": 2.5736245954692555e-05,
|
429 |
-
"loss": 1.
|
430 |
"step": 3000
|
431 |
},
|
432 |
{
|
433 |
"epoch": 14.805825242718447,
|
434 |
-
"grad_norm":
|
435 |
"learning_rate": 2.533171521035599e-05,
|
436 |
-
"loss": 1.
|
437 |
"step": 3050
|
438 |
},
|
439 |
{
|
440 |
"epoch": 15.048543689320388,
|
441 |
-
"grad_norm": 9.
|
442 |
"learning_rate": 2.492718446601942e-05,
|
443 |
-
"loss": 1.
|
444 |
"step": 3100
|
445 |
},
|
446 |
{
|
447 |
"epoch": 15.29126213592233,
|
448 |
-
"grad_norm": 10.
|
449 |
"learning_rate": 2.452265372168285e-05,
|
450 |
-
"loss": 1.
|
451 |
"step": 3150
|
452 |
},
|
453 |
{
|
454 |
"epoch": 15.533980582524272,
|
455 |
-
"grad_norm": 10.
|
456 |
"learning_rate": 2.411812297734628e-05,
|
457 |
-
"loss": 1.
|
458 |
"step": 3200
|
459 |
},
|
460 |
{
|
461 |
"epoch": 15.776699029126213,
|
462 |
-
"grad_norm": 11.
|
463 |
"learning_rate": 2.3713592233009708e-05,
|
464 |
-
"loss": 1.
|
465 |
"step": 3250
|
466 |
},
|
467 |
{
|
468 |
"epoch": 16.019417475728154,
|
469 |
-
"grad_norm":
|
470 |
"learning_rate": 2.3309061488673138e-05,
|
471 |
-
"loss": 1.
|
472 |
"step": 3300
|
473 |
},
|
474 |
{
|
475 |
"epoch": 16.262135922330096,
|
476 |
-
"grad_norm": 9.
|
477 |
"learning_rate": 2.290453074433657e-05,
|
478 |
-
"loss": 1.
|
479 |
"step": 3350
|
480 |
},
|
481 |
{
|
482 |
"epoch": 16.50485436893204,
|
483 |
-
"grad_norm": 11.
|
484 |
"learning_rate": 2.25e-05,
|
485 |
-
"loss": 1.
|
486 |
"step": 3400
|
487 |
},
|
488 |
{
|
489 |
"epoch": 16.74757281553398,
|
490 |
-
"grad_norm": 11.
|
491 |
"learning_rate": 2.209546925566343e-05,
|
492 |
-
"loss": 1.
|
493 |
"step": 3450
|
494 |
},
|
495 |
{
|
496 |
"epoch": 16.990291262135923,
|
497 |
-
"grad_norm": 11.
|
498 |
"learning_rate": 2.169093851132686e-05,
|
499 |
-
"loss": 1.
|
500 |
"step": 3500
|
501 |
},
|
502 |
{
|
503 |
"epoch": 17.233009708737864,
|
504 |
-
"grad_norm":
|
505 |
"learning_rate": 2.1286407766990295e-05,
|
506 |
-
"loss": 1.
|
507 |
"step": 3550
|
508 |
},
|
509 |
{
|
510 |
"epoch": 17.475728155339805,
|
511 |
-
"grad_norm": 11.
|
512 |
"learning_rate": 2.0881877022653725e-05,
|
513 |
-
"loss": 1.
|
514 |
"step": 3600
|
515 |
},
|
516 |
{
|
517 |
"epoch": 17.718446601941746,
|
518 |
-
"grad_norm":
|
519 |
"learning_rate": 2.047734627831715e-05,
|
520 |
-
"loss": 1.
|
521 |
"step": 3650
|
522 |
},
|
523 |
{
|
524 |
"epoch": 17.96116504854369,
|
525 |
-
"grad_norm":
|
526 |
"learning_rate": 2.007281553398058e-05,
|
527 |
-
"loss": 1.
|
528 |
"step": 3700
|
529 |
},
|
530 |
{
|
531 |
"epoch": 18.203883495145632,
|
532 |
-
"grad_norm": 9.
|
533 |
"learning_rate": 1.9668284789644014e-05,
|
534 |
-
"loss": 1.
|
535 |
"step": 3750
|
536 |
},
|
537 |
{
|
538 |
"epoch": 18.446601941747574,
|
539 |
-
"grad_norm":
|
540 |
"learning_rate": 1.9263754045307444e-05,
|
541 |
-
"loss": 1.
|
542 |
"step": 3800
|
543 |
},
|
544 |
{
|
545 |
"epoch": 18.689320388349515,
|
546 |
-
"grad_norm": 11.
|
547 |
"learning_rate": 1.8859223300970874e-05,
|
548 |
-
"loss": 1.
|
549 |
"step": 3850
|
550 |
},
|
551 |
{
|
552 |
"epoch": 18.932038834951456,
|
553 |
-
"grad_norm": 10.
|
554 |
"learning_rate": 1.8454692556634304e-05,
|
555 |
-
"loss": 1.
|
556 |
"step": 3900
|
557 |
},
|
558 |
{
|
559 |
"epoch": 19.174757281553397,
|
560 |
-
"grad_norm":
|
561 |
"learning_rate": 1.8050161812297738e-05,
|
562 |
-
"loss": 1.
|
563 |
"step": 3950
|
564 |
},
|
565 |
{
|
566 |
"epoch": 19.41747572815534,
|
567 |
-
"grad_norm":
|
568 |
"learning_rate": 1.7645631067961167e-05,
|
569 |
-
"loss": 1.
|
570 |
"step": 4000
|
571 |
},
|
572 |
{
|
573 |
"epoch": 19.660194174757283,
|
574 |
-
"grad_norm": 10.
|
575 |
"learning_rate": 1.7241100323624594e-05,
|
576 |
-
"loss": 1.
|
577 |
"step": 4050
|
578 |
},
|
579 |
{
|
580 |
"epoch": 19.902912621359224,
|
581 |
-
"grad_norm": 11.
|
582 |
"learning_rate": 1.6836569579288027e-05,
|
583 |
-
"loss": 1.
|
584 |
"step": 4100
|
585 |
},
|
586 |
{
|
587 |
"epoch": 20.145631067961165,
|
588 |
-
"grad_norm": 10.
|
589 |
"learning_rate": 1.6432038834951457e-05,
|
590 |
-
"loss": 1.
|
591 |
"step": 4150
|
592 |
},
|
593 |
{
|
594 |
"epoch": 20.388349514563107,
|
595 |
-
"grad_norm":
|
596 |
"learning_rate": 1.6027508090614887e-05,
|
597 |
-
"loss": 1.
|
598 |
"step": 4200
|
599 |
},
|
600 |
{
|
601 |
"epoch": 20.631067961165048,
|
602 |
-
"grad_norm": 10.
|
603 |
"learning_rate": 1.5622977346278317e-05,
|
604 |
-
"loss": 1.
|
605 |
"step": 4250
|
606 |
},
|
607 |
{
|
608 |
"epoch": 20.87378640776699,
|
609 |
-
"grad_norm": 11.
|
610 |
"learning_rate": 1.5218446601941749e-05,
|
611 |
-
"loss": 1.
|
612 |
"step": 4300
|
613 |
},
|
614 |
{
|
615 |
"epoch": 21.116504854368934,
|
616 |
-
"grad_norm": 11.
|
617 |
"learning_rate": 1.4813915857605179e-05,
|
618 |
-
"loss": 1.
|
619 |
"step": 4350
|
620 |
},
|
621 |
{
|
622 |
"epoch": 21.359223300970875,
|
623 |
-
"grad_norm": 10.
|
624 |
"learning_rate": 1.4409385113268609e-05,
|
625 |
-
"loss": 1.
|
626 |
"step": 4400
|
627 |
},
|
628 |
{
|
629 |
"epoch": 21.601941747572816,
|
630 |
-
"grad_norm": 11.
|
631 |
"learning_rate": 1.4004854368932039e-05,
|
632 |
-
"loss": 1.
|
633 |
"step": 4450
|
634 |
},
|
635 |
{
|
636 |
"epoch": 21.844660194174757,
|
637 |
-
"grad_norm":
|
638 |
"learning_rate": 1.360032362459547e-05,
|
639 |
-
"loss": 1.
|
640 |
"step": 4500
|
641 |
},
|
642 |
{
|
643 |
"epoch": 22.0873786407767,
|
644 |
-
"grad_norm": 10.
|
645 |
"learning_rate": 1.31957928802589e-05,
|
646 |
-
"loss": 1.
|
647 |
"step": 4550
|
648 |
},
|
649 |
{
|
650 |
"epoch": 22.33009708737864,
|
651 |
-
"grad_norm":
|
652 |
"learning_rate": 1.279126213592233e-05,
|
653 |
-
"loss": 0.
|
654 |
"step": 4600
|
655 |
},
|
656 |
{
|
657 |
"epoch": 22.57281553398058,
|
658 |
-
"grad_norm": 10.
|
659 |
"learning_rate": 1.2386731391585762e-05,
|
660 |
-
"loss": 0.
|
661 |
"step": 4650
|
662 |
},
|
663 |
{
|
664 |
"epoch": 22.815533980582526,
|
665 |
-
"grad_norm": 11.
|
666 |
"learning_rate": 1.1982200647249192e-05,
|
667 |
-
"loss": 0.
|
668 |
"step": 4700
|
669 |
},
|
670 |
{
|
671 |
"epoch": 23.058252427184467,
|
672 |
-
"grad_norm": 10.
|
673 |
"learning_rate": 1.1577669902912622e-05,
|
674 |
-
"loss": 0.
|
675 |
"step": 4750
|
676 |
},
|
677 |
{
|
678 |
"epoch": 23.300970873786408,
|
679 |
-
"grad_norm": 11.
|
680 |
"learning_rate": 1.1173139158576053e-05,
|
681 |
-
"loss": 0.
|
682 |
"step": 4800
|
683 |
},
|
684 |
{
|
685 |
"epoch": 23.54368932038835,
|
686 |
-
"grad_norm": 11.
|
687 |
"learning_rate": 1.0768608414239483e-05,
|
688 |
-
"loss": 0.
|
689 |
"step": 4850
|
690 |
},
|
691 |
{
|
692 |
"epoch": 23.78640776699029,
|
693 |
-
"grad_norm":
|
694 |
"learning_rate": 1.0364077669902913e-05,
|
695 |
-
"loss": 0.
|
696 |
"step": 4900
|
697 |
},
|
698 |
{
|
699 |
"epoch": 24.02912621359223,
|
700 |
-
"grad_norm":
|
701 |
"learning_rate": 9.959546925566343e-06,
|
702 |
-
"loss": 0.
|
703 |
"step": 4950
|
704 |
},
|
705 |
{
|
706 |
"epoch": 24.271844660194176,
|
707 |
-
"grad_norm": 10.
|
708 |
"learning_rate": 9.555016181229775e-06,
|
709 |
-
"loss": 0.
|
710 |
"step": 5000
|
711 |
}
|
712 |
],
|
|
|
11 |
"log_history": [
|
12 |
{
|
13 |
"epoch": 0.24271844660194175,
|
14 |
+
"grad_norm": 2.239194869995117,
|
15 |
"learning_rate": 4.960355987055016e-05,
|
16 |
+
"loss": 6.099,
|
17 |
"step": 50
|
18 |
},
|
19 |
{
|
20 |
"epoch": 0.4854368932038835,
|
21 |
+
"grad_norm": 3.8653194904327393,
|
22 |
"learning_rate": 4.9199029126213595e-05,
|
23 |
+
"loss": 5.6954,
|
24 |
"step": 100
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.7281553398058253,
|
28 |
+
"grad_norm": 4.652886867523193,
|
29 |
"learning_rate": 4.879449838187702e-05,
|
30 |
+
"loss": 5.3598,
|
31 |
"step": 150
|
32 |
},
|
33 |
{
|
34 |
"epoch": 0.970873786407767,
|
35 |
+
"grad_norm": 5.048999309539795,
|
36 |
"learning_rate": 4.8389967637540455e-05,
|
37 |
+
"loss": 5.0982,
|
38 |
"step": 200
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.2135922330097086,
|
42 |
+
"grad_norm": 5.21944522857666,
|
43 |
"learning_rate": 4.798543689320388e-05,
|
44 |
+
"loss": 4.7456,
|
45 |
"step": 250
|
46 |
},
|
47 |
{
|
48 |
"epoch": 1.4563106796116505,
|
49 |
+
"grad_norm": 5.471007347106934,
|
50 |
"learning_rate": 4.7580906148867315e-05,
|
51 |
+
"loss": 4.6254,
|
52 |
"step": 300
|
53 |
},
|
54 |
{
|
55 |
"epoch": 1.6990291262135924,
|
56 |
+
"grad_norm": 5.345245838165283,
|
57 |
"learning_rate": 4.717637540453075e-05,
|
58 |
+
"loss": 4.5216,
|
59 |
"step": 350
|
60 |
},
|
61 |
{
|
62 |
"epoch": 1.941747572815534,
|
63 |
+
"grad_norm": 5.54531192779541,
|
64 |
"learning_rate": 4.6771844660194174e-05,
|
65 |
+
"loss": 4.4353,
|
66 |
"step": 400
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.1844660194174756,
|
70 |
+
"grad_norm": 5.244582176208496,
|
71 |
"learning_rate": 4.636731391585761e-05,
|
72 |
+
"loss": 4.2444,
|
73 |
"step": 450
|
74 |
},
|
75 |
{
|
76 |
"epoch": 2.4271844660194173,
|
77 |
+
"grad_norm": 5.338261127471924,
|
78 |
"learning_rate": 4.596278317152104e-05,
|
79 |
+
"loss": 4.1915,
|
80 |
"step": 500
|
81 |
},
|
82 |
{
|
83 |
"epoch": 2.6699029126213594,
|
84 |
+
"grad_norm": 5.35228157043457,
|
85 |
"learning_rate": 4.555825242718447e-05,
|
86 |
+
"loss": 4.1612,
|
87 |
"step": 550
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.912621359223301,
|
91 |
+
"grad_norm": 5.287161827087402,
|
92 |
"learning_rate": 4.51537216828479e-05,
|
93 |
+
"loss": 4.1322,
|
94 |
"step": 600
|
95 |
},
|
96 |
{
|
97 |
"epoch": 3.1553398058252426,
|
98 |
+
"grad_norm": 5.2932209968566895,
|
99 |
"learning_rate": 4.4749190938511334e-05,
|
100 |
+
"loss": 3.9684,
|
101 |
"step": 650
|
102 |
},
|
103 |
{
|
104 |
"epoch": 3.3980582524271843,
|
105 |
+
"grad_norm": 5.4360833168029785,
|
106 |
"learning_rate": 4.434466019417476e-05,
|
107 |
+
"loss": 3.8933,
|
108 |
"step": 700
|
109 |
},
|
110 |
{
|
111 |
"epoch": 3.6407766990291264,
|
112 |
+
"grad_norm": 5.5595221519470215,
|
113 |
"learning_rate": 4.3940129449838194e-05,
|
114 |
+
"loss": 3.9306,
|
115 |
"step": 750
|
116 |
},
|
117 |
{
|
118 |
"epoch": 3.883495145631068,
|
119 |
+
"grad_norm": 5.540086269378662,
|
120 |
"learning_rate": 4.353559870550162e-05,
|
121 |
+
"loss": 3.9317,
|
122 |
"step": 800
|
123 |
},
|
124 |
{
|
125 |
"epoch": 4.12621359223301,
|
126 |
+
"grad_norm": 5.567137718200684,
|
127 |
"learning_rate": 4.313106796116505e-05,
|
128 |
+
"loss": 3.8135,
|
129 |
"step": 850
|
130 |
},
|
131 |
{
|
132 |
"epoch": 4.368932038834951,
|
133 |
+
"grad_norm": 6.00346040725708,
|
134 |
"learning_rate": 4.272653721682848e-05,
|
135 |
+
"loss": 3.6924,
|
136 |
"step": 900
|
137 |
},
|
138 |
{
|
139 |
"epoch": 4.611650485436893,
|
140 |
+
"grad_norm": 5.635782718658447,
|
141 |
"learning_rate": 4.232200647249191e-05,
|
142 |
+
"loss": 3.7304,
|
143 |
"step": 950
|
144 |
},
|
145 |
{
|
146 |
"epoch": 4.854368932038835,
|
147 |
+
"grad_norm": 5.9545440673828125,
|
148 |
"learning_rate": 4.191747572815534e-05,
|
149 |
+
"loss": 3.7471,
|
150 |
"step": 1000
|
151 |
},
|
152 |
{
|
153 |
"epoch": 5.097087378640777,
|
154 |
+
"grad_norm": 5.696166038513184,
|
155 |
"learning_rate": 4.1512944983818774e-05,
|
156 |
+
"loss": 3.6287,
|
157 |
"step": 1050
|
158 |
},
|
159 |
{
|
160 |
"epoch": 5.339805825242719,
|
161 |
+
"grad_norm": 5.844620704650879,
|
162 |
"learning_rate": 4.11084142394822e-05,
|
163 |
+
"loss": 3.5299,
|
164 |
"step": 1100
|
165 |
},
|
166 |
{
|
167 |
"epoch": 5.58252427184466,
|
168 |
+
"grad_norm": 6.373577117919922,
|
169 |
"learning_rate": 4.0703883495145634e-05,
|
170 |
+
"loss": 3.5285,
|
171 |
"step": 1150
|
172 |
},
|
173 |
{
|
174 |
"epoch": 5.825242718446602,
|
175 |
+
"grad_norm": 6.488651275634766,
|
176 |
"learning_rate": 4.029935275080906e-05,
|
177 |
+
"loss": 3.5789,
|
178 |
"step": 1200
|
179 |
},
|
180 |
{
|
181 |
"epoch": 6.067961165048544,
|
182 |
+
"grad_norm": 6.170980930328369,
|
183 |
"learning_rate": 3.9894822006472494e-05,
|
184 |
+
"loss": 3.494,
|
185 |
"step": 1250
|
186 |
},
|
187 |
{
|
188 |
"epoch": 6.310679611650485,
|
189 |
+
"grad_norm": 6.387782573699951,
|
190 |
"learning_rate": 3.949029126213593e-05,
|
191 |
+
"loss": 3.353,
|
192 |
"step": 1300
|
193 |
},
|
194 |
{
|
195 |
"epoch": 6.553398058252427,
|
196 |
+
"grad_norm": 6.718175888061523,
|
197 |
"learning_rate": 3.9085760517799354e-05,
|
198 |
+
"loss": 3.3521,
|
199 |
"step": 1350
|
200 |
},
|
201 |
{
|
202 |
"epoch": 6.796116504854369,
|
203 |
+
"grad_norm": 6.7870025634765625,
|
204 |
"learning_rate": 3.868122977346279e-05,
|
205 |
+
"loss": 3.3658,
|
206 |
"step": 1400
|
207 |
},
|
208 |
{
|
209 |
"epoch": 7.038834951456311,
|
210 |
+
"grad_norm": 6.513721466064453,
|
211 |
"learning_rate": 3.827669902912622e-05,
|
212 |
+
"loss": 3.3301,
|
213 |
"step": 1450
|
214 |
},
|
215 |
{
|
216 |
"epoch": 7.281553398058253,
|
217 |
+
"grad_norm": 7.0680036544799805,
|
218 |
"learning_rate": 3.787216828478965e-05,
|
219 |
+
"loss": 3.1289,
|
220 |
"step": 1500
|
221 |
},
|
222 |
{
|
223 |
"epoch": 7.524271844660194,
|
224 |
+
"grad_norm": 7.019896507263184,
|
225 |
"learning_rate": 3.746763754045307e-05,
|
226 |
+
"loss": 3.1878,
|
227 |
"step": 1550
|
228 |
},
|
229 |
{
|
230 |
"epoch": 7.766990291262136,
|
231 |
+
"grad_norm": 7.3232197761535645,
|
232 |
"learning_rate": 3.7063106796116507e-05,
|
233 |
+
"loss": 3.2006,
|
234 |
"step": 1600
|
235 |
},
|
236 |
{
|
237 |
"epoch": 8.009708737864077,
|
238 |
+
"grad_norm": 7.003912925720215,
|
239 |
"learning_rate": 3.665857605177993e-05,
|
240 |
+
"loss": 3.1737,
|
241 |
"step": 1650
|
242 |
},
|
243 |
{
|
244 |
"epoch": 8.25242718446602,
|
245 |
+
"grad_norm": 7.592433929443359,
|
246 |
"learning_rate": 3.6254045307443366e-05,
|
247 |
+
"loss": 2.9258,
|
248 |
"step": 1700
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.495145631067961,
|
252 |
+
"grad_norm": 7.629278182983398,
|
253 |
"learning_rate": 3.584951456310679e-05,
|
254 |
+
"loss": 2.9999,
|
255 |
"step": 1750
|
256 |
},
|
257 |
{
|
258 |
"epoch": 8.737864077669903,
|
259 |
+
"grad_norm": 8.137636184692383,
|
260 |
"learning_rate": 3.5444983818770226e-05,
|
261 |
+
"loss": 3.0155,
|
262 |
"step": 1800
|
263 |
},
|
264 |
{
|
265 |
"epoch": 8.980582524271846,
|
266 |
+
"grad_norm": 7.967227935791016,
|
267 |
"learning_rate": 3.504045307443366e-05,
|
268 |
+
"loss": 2.9945,
|
269 |
"step": 1850
|
270 |
},
|
271 |
{
|
272 |
"epoch": 9.223300970873787,
|
273 |
+
"grad_norm": 7.9910078048706055,
|
274 |
"learning_rate": 3.4635922330097086e-05,
|
275 |
+
"loss": 2.7456,
|
276 |
"step": 1900
|
277 |
},
|
278 |
{
|
279 |
"epoch": 9.466019417475728,
|
280 |
+
"grad_norm": 8.552720069885254,
|
281 |
"learning_rate": 3.423139158576052e-05,
|
282 |
+
"loss": 2.8071,
|
283 |
"step": 1950
|
284 |
},
|
285 |
{
|
286 |
"epoch": 9.70873786407767,
|
287 |
+
"grad_norm": 8.336930274963379,
|
288 |
"learning_rate": 3.382686084142395e-05,
|
289 |
+
"loss": 2.8117,
|
290 |
"step": 2000
|
291 |
},
|
292 |
{
|
293 |
"epoch": 9.951456310679612,
|
294 |
+
"grad_norm": 8.636338233947754,
|
295 |
"learning_rate": 3.342233009708738e-05,
|
296 |
+
"loss": 2.8197,
|
297 |
"step": 2050
|
298 |
},
|
299 |
{
|
300 |
"epoch": 10.194174757281553,
|
301 |
+
"grad_norm": 7.955165863037109,
|
302 |
"learning_rate": 3.301779935275081e-05,
|
303 |
+
"loss": 2.6245,
|
304 |
"step": 2100
|
305 |
},
|
306 |
{
|
307 |
"epoch": 10.436893203883495,
|
308 |
+
"grad_norm": 8.679149627685547,
|
309 |
"learning_rate": 3.2613268608414246e-05,
|
310 |
+
"loss": 2.565,
|
311 |
"step": 2150
|
312 |
},
|
313 |
{
|
314 |
"epoch": 10.679611650485437,
|
315 |
+
"grad_norm": 9.150839805603027,
|
316 |
"learning_rate": 3.220873786407767e-05,
|
317 |
+
"loss": 2.6258,
|
318 |
"step": 2200
|
319 |
},
|
320 |
{
|
321 |
"epoch": 10.922330097087379,
|
322 |
+
"grad_norm": 9.2735013961792,
|
323 |
"learning_rate": 3.1804207119741106e-05,
|
324 |
+
"loss": 2.6439,
|
325 |
"step": 2250
|
326 |
},
|
327 |
{
|
328 |
"epoch": 11.16504854368932,
|
329 |
+
"grad_norm": 8.620939254760742,
|
330 |
"learning_rate": 3.139967637540453e-05,
|
331 |
+
"loss": 2.4371,
|
332 |
"step": 2300
|
333 |
},
|
334 |
{
|
335 |
"epoch": 11.407766990291263,
|
336 |
+
"grad_norm": 9.722040176391602,
|
337 |
"learning_rate": 3.099514563106796e-05,
|
338 |
+
"loss": 2.4074,
|
339 |
"step": 2350
|
340 |
},
|
341 |
{
|
342 |
"epoch": 11.650485436893204,
|
343 |
+
"grad_norm": 9.200730323791504,
|
344 |
"learning_rate": 3.059061488673139e-05,
|
345 |
+
"loss": 2.4307,
|
346 |
"step": 2400
|
347 |
},
|
348 |
{
|
349 |
"epoch": 11.893203883495145,
|
350 |
+
"grad_norm": 9.692273139953613,
|
351 |
"learning_rate": 3.0186084142394822e-05,
|
352 |
+
"loss": 2.424,
|
353 |
"step": 2450
|
354 |
},
|
355 |
{
|
356 |
"epoch": 12.135922330097088,
|
357 |
+
"grad_norm": 9.822249412536621,
|
358 |
"learning_rate": 2.9781553398058252e-05,
|
359 |
+
"loss": 2.2862,
|
360 |
"step": 2500
|
361 |
},
|
362 |
{
|
363 |
"epoch": 12.37864077669903,
|
364 |
+
"grad_norm": 9.919060707092285,
|
365 |
"learning_rate": 2.9377022653721686e-05,
|
366 |
+
"loss": 2.1913,
|
367 |
"step": 2550
|
368 |
},
|
369 |
{
|
370 |
"epoch": 12.62135922330097,
|
371 |
+
"grad_norm": 9.181950569152832,
|
372 |
"learning_rate": 2.8972491909385112e-05,
|
373 |
+
"loss": 2.241,
|
374 |
"step": 2600
|
375 |
},
|
376 |
{
|
377 |
"epoch": 12.864077669902912,
|
378 |
+
"grad_norm": 10.82077407836914,
|
379 |
"learning_rate": 2.8567961165048546e-05,
|
380 |
+
"loss": 2.2653,
|
381 |
"step": 2650
|
382 |
},
|
383 |
{
|
384 |
"epoch": 13.106796116504855,
|
385 |
+
"grad_norm": 10.365830421447754,
|
386 |
"learning_rate": 2.816343042071198e-05,
|
387 |
+
"loss": 2.1352,
|
388 |
"step": 2700
|
389 |
},
|
390 |
{
|
391 |
"epoch": 13.349514563106796,
|
392 |
+
"grad_norm": 10.073640823364258,
|
393 |
"learning_rate": 2.7758899676375405e-05,
|
394 |
+
"loss": 2.0227,
|
395 |
"step": 2750
|
396 |
},
|
397 |
{
|
398 |
"epoch": 13.592233009708737,
|
399 |
+
"grad_norm": 10.520242691040039,
|
400 |
"learning_rate": 2.735436893203884e-05,
|
401 |
+
"loss": 2.0579,
|
402 |
"step": 2800
|
403 |
},
|
404 |
{
|
405 |
"epoch": 13.83495145631068,
|
406 |
+
"grad_norm": 11.176478385925293,
|
407 |
"learning_rate": 2.6949838187702265e-05,
|
408 |
+
"loss": 2.0934,
|
409 |
"step": 2850
|
410 |
},
|
411 |
{
|
412 |
"epoch": 14.077669902912621,
|
413 |
+
"grad_norm": 9.541868209838867,
|
414 |
"learning_rate": 2.6545307443365695e-05,
|
415 |
+
"loss": 2.0118,
|
416 |
"step": 2900
|
417 |
},
|
418 |
{
|
419 |
"epoch": 14.320388349514563,
|
420 |
+
"grad_norm": 10.162981986999512,
|
421 |
"learning_rate": 2.614077669902913e-05,
|
422 |
+
"loss": 1.8624,
|
423 |
"step": 2950
|
424 |
},
|
425 |
{
|
426 |
"epoch": 14.563106796116505,
|
427 |
+
"grad_norm": 10.51291275024414,
|
428 |
"learning_rate": 2.5736245954692555e-05,
|
429 |
+
"loss": 1.8979,
|
430 |
"step": 3000
|
431 |
},
|
432 |
{
|
433 |
"epoch": 14.805825242718447,
|
434 |
+
"grad_norm": 11.316717147827148,
|
435 |
"learning_rate": 2.533171521035599e-05,
|
436 |
+
"loss": 1.915,
|
437 |
"step": 3050
|
438 |
},
|
439 |
{
|
440 |
"epoch": 15.048543689320388,
|
441 |
+
"grad_norm": 9.184353828430176,
|
442 |
"learning_rate": 2.492718446601942e-05,
|
443 |
+
"loss": 1.885,
|
444 |
"step": 3100
|
445 |
},
|
446 |
{
|
447 |
"epoch": 15.29126213592233,
|
448 |
+
"grad_norm": 10.70876693725586,
|
449 |
"learning_rate": 2.452265372168285e-05,
|
450 |
+
"loss": 1.7024,
|
451 |
"step": 3150
|
452 |
},
|
453 |
{
|
454 |
"epoch": 15.533980582524272,
|
455 |
+
"grad_norm": 10.767953872680664,
|
456 |
"learning_rate": 2.411812297734628e-05,
|
457 |
+
"loss": 1.729,
|
458 |
"step": 3200
|
459 |
},
|
460 |
{
|
461 |
"epoch": 15.776699029126213,
|
462 |
+
"grad_norm": 11.83687973022461,
|
463 |
"learning_rate": 2.3713592233009708e-05,
|
464 |
+
"loss": 1.7697,
|
465 |
"step": 3250
|
466 |
},
|
467 |
{
|
468 |
"epoch": 16.019417475728154,
|
469 |
+
"grad_norm": 9.888689994812012,
|
470 |
"learning_rate": 2.3309061488673138e-05,
|
471 |
+
"loss": 1.7605,
|
472 |
"step": 3300
|
473 |
},
|
474 |
{
|
475 |
"epoch": 16.262135922330096,
|
476 |
+
"grad_norm": 9.394012451171875,
|
477 |
"learning_rate": 2.290453074433657e-05,
|
478 |
+
"loss": 1.5389,
|
479 |
"step": 3350
|
480 |
},
|
481 |
{
|
482 |
"epoch": 16.50485436893204,
|
483 |
+
"grad_norm": 11.3172025680542,
|
484 |
"learning_rate": 2.25e-05,
|
485 |
+
"loss": 1.5836,
|
486 |
"step": 3400
|
487 |
},
|
488 |
{
|
489 |
"epoch": 16.74757281553398,
|
490 |
+
"grad_norm": 11.000115394592285,
|
491 |
"learning_rate": 2.209546925566343e-05,
|
492 |
+
"loss": 1.6398,
|
493 |
"step": 3450
|
494 |
},
|
495 |
{
|
496 |
"epoch": 16.990291262135923,
|
497 |
+
"grad_norm": 11.453404426574707,
|
498 |
"learning_rate": 2.169093851132686e-05,
|
499 |
+
"loss": 1.6466,
|
500 |
"step": 3500
|
501 |
},
|
502 |
{
|
503 |
"epoch": 17.233009708737864,
|
504 |
+
"grad_norm": 11.012279510498047,
|
505 |
"learning_rate": 2.1286407766990295e-05,
|
506 |
+
"loss": 1.4434,
|
507 |
"step": 3550
|
508 |
},
|
509 |
{
|
510 |
"epoch": 17.475728155339805,
|
511 |
+
"grad_norm": 11.989383697509766,
|
512 |
"learning_rate": 2.0881877022653725e-05,
|
513 |
+
"loss": 1.4708,
|
514 |
"step": 3600
|
515 |
},
|
516 |
{
|
517 |
"epoch": 17.718446601941746,
|
518 |
+
"grad_norm": 12.47008991241455,
|
519 |
"learning_rate": 2.047734627831715e-05,
|
520 |
+
"loss": 1.4783,
|
521 |
"step": 3650
|
522 |
},
|
523 |
{
|
524 |
"epoch": 17.96116504854369,
|
525 |
+
"grad_norm": 10.983756065368652,
|
526 |
"learning_rate": 2.007281553398058e-05,
|
527 |
+
"loss": 1.4839,
|
528 |
"step": 3700
|
529 |
},
|
530 |
{
|
531 |
"epoch": 18.203883495145632,
|
532 |
+
"grad_norm": 9.582510948181152,
|
533 |
"learning_rate": 1.9668284789644014e-05,
|
534 |
+
"loss": 1.3226,
|
535 |
"step": 3750
|
536 |
},
|
537 |
{
|
538 |
"epoch": 18.446601941747574,
|
539 |
+
"grad_norm": 11.663228034973145,
|
540 |
"learning_rate": 1.9263754045307444e-05,
|
541 |
+
"loss": 1.3522,
|
542 |
"step": 3800
|
543 |
},
|
544 |
{
|
545 |
"epoch": 18.689320388349515,
|
546 |
+
"grad_norm": 11.278139114379883,
|
547 |
"learning_rate": 1.8859223300970874e-05,
|
548 |
+
"loss": 1.3753,
|
549 |
"step": 3850
|
550 |
},
|
551 |
{
|
552 |
"epoch": 18.932038834951456,
|
553 |
+
"grad_norm": 10.586488723754883,
|
554 |
"learning_rate": 1.8454692556634304e-05,
|
555 |
+
"loss": 1.3774,
|
556 |
"step": 3900
|
557 |
},
|
558 |
{
|
559 |
"epoch": 19.174757281553397,
|
560 |
+
"grad_norm": 11.243274688720703,
|
561 |
"learning_rate": 1.8050161812297738e-05,
|
562 |
+
"loss": 1.2506,
|
563 |
"step": 3950
|
564 |
},
|
565 |
{
|
566 |
"epoch": 19.41747572815534,
|
567 |
+
"grad_norm": 10.601967811584473,
|
568 |
"learning_rate": 1.7645631067961167e-05,
|
569 |
+
"loss": 1.2276,
|
570 |
"step": 4000
|
571 |
},
|
572 |
{
|
573 |
"epoch": 19.660194174757283,
|
574 |
+
"grad_norm": 10.919422149658203,
|
575 |
"learning_rate": 1.7241100323624594e-05,
|
576 |
+
"loss": 1.2518,
|
577 |
"step": 4050
|
578 |
},
|
579 |
{
|
580 |
"epoch": 19.902912621359224,
|
581 |
+
"grad_norm": 11.303253173828125,
|
582 |
"learning_rate": 1.6836569579288027e-05,
|
583 |
+
"loss": 1.279,
|
584 |
"step": 4100
|
585 |
},
|
586 |
{
|
587 |
"epoch": 20.145631067961165,
|
588 |
+
"grad_norm": 10.864278793334961,
|
589 |
"learning_rate": 1.6432038834951457e-05,
|
590 |
+
"loss": 1.1686,
|
591 |
"step": 4150
|
592 |
},
|
593 |
{
|
594 |
"epoch": 20.388349514563107,
|
595 |
+
"grad_norm": 11.179913520812988,
|
596 |
"learning_rate": 1.6027508090614887e-05,
|
597 |
+
"loss": 1.1252,
|
598 |
"step": 4200
|
599 |
},
|
600 |
{
|
601 |
"epoch": 20.631067961165048,
|
602 |
+
"grad_norm": 10.640514373779297,
|
603 |
"learning_rate": 1.5622977346278317e-05,
|
604 |
+
"loss": 1.1581,
|
605 |
"step": 4250
|
606 |
},
|
607 |
{
|
608 |
"epoch": 20.87378640776699,
|
609 |
+
"grad_norm": 11.186044692993164,
|
610 |
"learning_rate": 1.5218446601941749e-05,
|
611 |
+
"loss": 1.1782,
|
612 |
"step": 4300
|
613 |
},
|
614 |
{
|
615 |
"epoch": 21.116504854368934,
|
616 |
+
"grad_norm": 11.105583190917969,
|
617 |
"learning_rate": 1.4813915857605179e-05,
|
618 |
+
"loss": 1.0972,
|
619 |
"step": 4350
|
620 |
},
|
621 |
{
|
622 |
"epoch": 21.359223300970875,
|
623 |
+
"grad_norm": 10.809669494628906,
|
624 |
"learning_rate": 1.4409385113268609e-05,
|
625 |
+
"loss": 1.0487,
|
626 |
"step": 4400
|
627 |
},
|
628 |
{
|
629 |
"epoch": 21.601941747572816,
|
630 |
+
"grad_norm": 11.11020278930664,
|
631 |
"learning_rate": 1.4004854368932039e-05,
|
632 |
+
"loss": 1.0566,
|
633 |
"step": 4450
|
634 |
},
|
635 |
{
|
636 |
"epoch": 21.844660194174757,
|
637 |
+
"grad_norm": 11.062104225158691,
|
638 |
"learning_rate": 1.360032362459547e-05,
|
639 |
+
"loss": 1.078,
|
640 |
"step": 4500
|
641 |
},
|
642 |
{
|
643 |
"epoch": 22.0873786407767,
|
644 |
+
"grad_norm": 10.393362998962402,
|
645 |
"learning_rate": 1.31957928802589e-05,
|
646 |
+
"loss": 1.0351,
|
647 |
"step": 4550
|
648 |
},
|
649 |
{
|
650 |
"epoch": 22.33009708737864,
|
651 |
+
"grad_norm": 9.539032936096191,
|
652 |
"learning_rate": 1.279126213592233e-05,
|
653 |
+
"loss": 0.9655,
|
654 |
"step": 4600
|
655 |
},
|
656 |
{
|
657 |
"epoch": 22.57281553398058,
|
658 |
+
"grad_norm": 10.375469207763672,
|
659 |
"learning_rate": 1.2386731391585762e-05,
|
660 |
+
"loss": 0.9857,
|
661 |
"step": 4650
|
662 |
},
|
663 |
{
|
664 |
"epoch": 22.815533980582526,
|
665 |
+
"grad_norm": 11.338729858398438,
|
666 |
"learning_rate": 1.1982200647249192e-05,
|
667 |
+
"loss": 0.9975,
|
668 |
"step": 4700
|
669 |
},
|
670 |
{
|
671 |
"epoch": 23.058252427184467,
|
672 |
+
"grad_norm": 10.976126670837402,
|
673 |
"learning_rate": 1.1577669902912622e-05,
|
674 |
+
"loss": 0.9785,
|
675 |
"step": 4750
|
676 |
},
|
677 |
{
|
678 |
"epoch": 23.300970873786408,
|
679 |
+
"grad_norm": 11.16224193572998,
|
680 |
"learning_rate": 1.1173139158576053e-05,
|
681 |
+
"loss": 0.9011,
|
682 |
"step": 4800
|
683 |
},
|
684 |
{
|
685 |
"epoch": 23.54368932038835,
|
686 |
+
"grad_norm": 11.640942573547363,
|
687 |
"learning_rate": 1.0768608414239483e-05,
|
688 |
+
"loss": 0.9056,
|
689 |
"step": 4850
|
690 |
},
|
691 |
{
|
692 |
"epoch": 23.78640776699029,
|
693 |
+
"grad_norm": 10.686304092407227,
|
694 |
"learning_rate": 1.0364077669902913e-05,
|
695 |
+
"loss": 0.9267,
|
696 |
"step": 4900
|
697 |
},
|
698 |
{
|
699 |
"epoch": 24.02912621359223,
|
700 |
+
"grad_norm": 9.798011779785156,
|
701 |
"learning_rate": 9.959546925566343e-06,
|
702 |
+
"loss": 0.9134,
|
703 |
"step": 4950
|
704 |
},
|
705 |
{
|
706 |
"epoch": 24.271844660194176,
|
707 |
+
"grad_norm": 10.638201713562012,
|
708 |
"learning_rate": 9.555016181229775e-06,
|
709 |
+
"loss": 0.8257,
|
710 |
"step": 5000
|
711 |
}
|
712 |
],
|
checkpoints/checkpoint-5000/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7935e3b7629485cf26f624afe618b58a390d4c3c9c6dcf878b3ba9099ef1027b
|
3 |
size 5304
|