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 +4 -4
- checkpoints/checkpoint-1000/model.safetensors +2 -2
- checkpoints/checkpoint-1000/optimizer.pt +2 -2
- checkpoints/checkpoint-1000/rng_state.pth +1 -1
- checkpoints/checkpoint-1000/trainer_state.json +41 -41
- checkpoints/checkpoint-1000/training_args.bin +1 -1
- checkpoints/checkpoint-1500/config.json +4 -4
- checkpoints/checkpoint-1500/model.safetensors +2 -2
- checkpoints/checkpoint-1500/optimizer.pt +2 -2
- checkpoints/checkpoint-1500/rng_state.pth +1 -1
- checkpoints/checkpoint-1500/trainer_state.json +61 -61
- checkpoints/checkpoint-1500/training_args.bin +1 -1
- checkpoints/checkpoint-2000/config.json +4 -4
- checkpoints/checkpoint-2000/model.safetensors +2 -2
- checkpoints/checkpoint-2000/optimizer.pt +2 -2
- checkpoints/checkpoint-2000/rng_state.pth +1 -1
- checkpoints/checkpoint-2000/trainer_state.json +81 -81
- checkpoints/checkpoint-2000/training_args.bin +1 -1
- checkpoints/checkpoint-2500/config.json +4 -4
- checkpoints/checkpoint-2500/model.safetensors +2 -2
- checkpoints/checkpoint-2500/optimizer.pt +2 -2
- checkpoints/checkpoint-2500/rng_state.pth +1 -1
- checkpoints/checkpoint-2500/trainer_state.json +101 -101
- checkpoints/checkpoint-2500/training_args.bin +1 -1
- checkpoints/checkpoint-3000/config.json +4 -4
- checkpoints/checkpoint-3000/model.safetensors +2 -2
- checkpoints/checkpoint-3000/optimizer.pt +2 -2
- checkpoints/checkpoint-3000/rng_state.pth +1 -1
- checkpoints/checkpoint-3000/trainer_state.json +121 -121
- checkpoints/checkpoint-3000/training_args.bin +1 -1
- checkpoints/checkpoint-3500/config.json +4 -4
- checkpoints/checkpoint-3500/model.safetensors +2 -2
- checkpoints/checkpoint-3500/optimizer.pt +2 -2
- checkpoints/checkpoint-3500/rng_state.pth +1 -1
- checkpoints/checkpoint-3500/trainer_state.json +141 -141
- checkpoints/checkpoint-3500/training_args.bin +1 -1
- checkpoints/checkpoint-4000/config.json +4 -4
- checkpoints/checkpoint-4000/model.safetensors +2 -2
- checkpoints/checkpoint-4000/optimizer.pt +2 -2
- checkpoints/checkpoint-4000/rng_state.pth +1 -1
- checkpoints/checkpoint-4000/trainer_state.json +161 -161
- checkpoints/checkpoint-4000/training_args.bin +1 -1
- checkpoints/checkpoint-4500/config.json +4 -4
- checkpoints/checkpoint-4500/model.safetensors +2 -2
- checkpoints/checkpoint-4500/optimizer.pt +2 -2
- checkpoints/checkpoint-4500/rng_state.pth +1 -1
- checkpoints/checkpoint-4500/trainer_state.json +181 -181
- checkpoints/checkpoint-4500/training_args.bin +1 -1
- checkpoints/checkpoint-500/config.json +4 -4
- checkpoints/checkpoint-500/model.safetensors +2 -2
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":
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
-
"n_layer":
|
| 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": 2048,
|
| 14 |
+
"n_embd": 1024,
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
+
"n_layer": 12,
|
| 18 |
+
"n_positions": 2048,
|
| 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:ecddd7806bbce858d7a9c8f4a49aa06b7376eafb0955ae9642f0388121a2a689
|
| 3 |
+
size 617130824
|
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 sha256:05384fd919d4331c4f62eb3e56ba3d26561bf6ef9ab915ef8c1ddcd165540b84
|
| 3 |
+
size 1234355130
|
checkpoints/checkpoint-1000/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3404e1aa2fa72c55a7ea6b1c303c4b57bf41922fdf8d7c880092c6ac1ab5b6b9
|
| 3 |
size 14244
|
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":
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
-
"loss": 5.
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
-
"grad_norm":
|
| 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": 6.
|
| 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":
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
-
"loss": 3.
|
| 150 |
"step": 1000
|
| 151 |
}
|
| 152 |
],
|
|
@@ -167,7 +167,7 @@
|
|
| 167 |
"attributes": {}
|
| 168 |
}
|
| 169 |
},
|
| 170 |
-
"total_flos":
|
| 171 |
"train_batch_size": 4,
|
| 172 |
"trial_name": null,
|
| 173 |
"trial_params": null
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.24271844660194175,
|
| 14 |
+
"grad_norm": 2.8362834453582764,
|
| 15 |
"learning_rate": 4.960355987055016e-05,
|
| 16 |
+
"loss": 6.0742,
|
| 17 |
"step": 50
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4854368932038835,
|
| 21 |
+
"grad_norm": 2.9484496116638184,
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
+
"loss": 5.7218,
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
+
"grad_norm": 3.542572259902954,
|
| 29 |
"learning_rate": 4.879449838187702e-05,
|
| 30 |
+
"loss": 5.4227,
|
| 31 |
"step": 150
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.970873786407767,
|
| 35 |
+
"grad_norm": 4.634098052978516,
|
| 36 |
"learning_rate": 4.8389967637540455e-05,
|
| 37 |
+
"loss": 5.1655,
|
| 38 |
"step": 200
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.2135922330097086,
|
| 42 |
+
"grad_norm": 5.699330806732178,
|
| 43 |
"learning_rate": 4.798543689320388e-05,
|
| 44 |
+
"loss": 4.8009,
|
| 45 |
"step": 250
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 1.4563106796116505,
|
| 49 |
+
"grad_norm": 5.711933135986328,
|
| 50 |
"learning_rate": 4.7580906148867315e-05,
|
| 51 |
+
"loss": 4.607,
|
| 52 |
"step": 300
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 1.6990291262135924,
|
| 56 |
+
"grad_norm": 5.113691806793213,
|
| 57 |
"learning_rate": 4.717637540453075e-05,
|
| 58 |
+
"loss": 4.462,
|
| 59 |
"step": 350
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 1.941747572815534,
|
| 63 |
+
"grad_norm": 5.632521152496338,
|
| 64 |
"learning_rate": 4.6771844660194174e-05,
|
| 65 |
+
"loss": 4.3695,
|
| 66 |
"step": 400
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 2.1844660194174756,
|
| 70 |
+
"grad_norm": 5.428906440734863,
|
| 71 |
"learning_rate": 4.636731391585761e-05,
|
| 72 |
+
"loss": 4.1549,
|
| 73 |
"step": 450
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 2.4271844660194173,
|
| 77 |
+
"grad_norm": 5.037013530731201,
|
| 78 |
"learning_rate": 4.596278317152104e-05,
|
| 79 |
+
"loss": 4.0921,
|
| 80 |
"step": 500
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 2.6699029126213594,
|
| 84 |
+
"grad_norm": 5.231846809387207,
|
| 85 |
"learning_rate": 4.555825242718447e-05,
|
| 86 |
+
"loss": 4.0662,
|
| 87 |
"step": 550
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 2.912621359223301,
|
| 91 |
+
"grad_norm": 5.506278038024902,
|
| 92 |
"learning_rate": 4.51537216828479e-05,
|
| 93 |
+
"loss": 4.0403,
|
| 94 |
"step": 600
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 3.1553398058252426,
|
| 98 |
+
"grad_norm": 5.410216331481934,
|
| 99 |
"learning_rate": 4.4749190938511334e-05,
|
| 100 |
+
"loss": 3.8375,
|
| 101 |
"step": 650
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 3.3980582524271843,
|
| 105 |
+
"grad_norm": 5.637699127197266,
|
| 106 |
"learning_rate": 4.434466019417476e-05,
|
| 107 |
+
"loss": 3.7473,
|
| 108 |
"step": 700
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 3.6407766990291264,
|
| 112 |
+
"grad_norm": 5.575273513793945,
|
| 113 |
"learning_rate": 4.3940129449838194e-05,
|
| 114 |
+
"loss": 3.7835,
|
| 115 |
"step": 750
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 3.883495145631068,
|
| 119 |
+
"grad_norm": 5.737198352813721,
|
| 120 |
"learning_rate": 4.353559870550162e-05,
|
| 121 |
+
"loss": 3.7919,
|
| 122 |
"step": 800
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 4.12621359223301,
|
| 126 |
+
"grad_norm": 5.645532608032227,
|
| 127 |
"learning_rate": 4.313106796116505e-05,
|
| 128 |
+
"loss": 3.6425,
|
| 129 |
"step": 850
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 4.368932038834951,
|
| 133 |
+
"grad_norm": 6.308927059173584,
|
| 134 |
"learning_rate": 4.272653721682848e-05,
|
| 135 |
+
"loss": 3.4785,
|
| 136 |
"step": 900
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 4.611650485436893,
|
| 140 |
+
"grad_norm": 5.863094329833984,
|
| 141 |
"learning_rate": 4.232200647249191e-05,
|
| 142 |
+
"loss": 3.5268,
|
| 143 |
"step": 950
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 4.854368932038835,
|
| 147 |
+
"grad_norm": 6.120258331298828,
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
+
"loss": 3.5296,
|
| 150 |
"step": 1000
|
| 151 |
}
|
| 152 |
],
|
|
|
|
| 167 |
"attributes": {}
|
| 168 |
}
|
| 169 |
},
|
| 170 |
+
"total_flos": 464353492992000.0,
|
| 171 |
"train_batch_size": 4,
|
| 172 |
"trial_name": null,
|
| 173 |
"trial_params": null
|
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:d5c04ce53d557cd762ff77607aaadbd1248aaa177697662d9a21b90d444993b7
|
| 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":
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
-
"n_layer":
|
| 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": 2048,
|
| 14 |
+
"n_embd": 1024,
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
+
"n_layer": 12,
|
| 18 |
+
"n_positions": 2048,
|
| 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 sha256:3e9d84a36c4519bf72764d9744d791a4693b054535689bf3648c7d0833ae5356
|
| 3 |
+
size 617130824
|
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:e76d49ca329f71b0d0da9d3ae2c4624ac37ed78688ebf52ff5dee93474eb6e7b
|
| 3 |
+
size 1234355130
|
checkpoints/checkpoint-1500/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4f95e2cc3df656ac2c5f1f4c905d818b69371ad7deb474b77aec082625d08cce
|
| 3 |
size 14244
|
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":
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
-
"loss": 5.
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
-
"grad_norm":
|
| 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": 6.
|
| 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":
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
-
"loss": 3.
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
-
"grad_norm":
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
-
"loss": 3.
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
-
"grad_norm":
|
| 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":
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
-
"loss":
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
-
"grad_norm":
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
-
"loss":
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
-
"grad_norm":
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
-
"loss":
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
-
"grad_norm":
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
-
"loss":
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
-
"grad_norm": 7.
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
-
"loss":
|
| 220 |
"step": 1500
|
| 221 |
}
|
| 222 |
],
|
|
@@ -237,7 +237,7 @@
|
|
| 237 |
"attributes": {}
|
| 238 |
}
|
| 239 |
},
|
| 240 |
-
"total_flos":
|
| 241 |
"train_batch_size": 4,
|
| 242 |
"trial_name": null,
|
| 243 |
"trial_params": null
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.24271844660194175,
|
| 14 |
+
"grad_norm": 2.8362834453582764,
|
| 15 |
"learning_rate": 4.960355987055016e-05,
|
| 16 |
+
"loss": 6.0742,
|
| 17 |
"step": 50
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4854368932038835,
|
| 21 |
+
"grad_norm": 2.9484496116638184,
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
+
"loss": 5.7218,
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
+
"grad_norm": 3.542572259902954,
|
| 29 |
"learning_rate": 4.879449838187702e-05,
|
| 30 |
+
"loss": 5.4227,
|
| 31 |
"step": 150
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.970873786407767,
|
| 35 |
+
"grad_norm": 4.634098052978516,
|
| 36 |
"learning_rate": 4.8389967637540455e-05,
|
| 37 |
+
"loss": 5.1655,
|
| 38 |
"step": 200
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.2135922330097086,
|
| 42 |
+
"grad_norm": 5.699330806732178,
|
| 43 |
"learning_rate": 4.798543689320388e-05,
|
| 44 |
+
"loss": 4.8009,
|
| 45 |
"step": 250
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 1.4563106796116505,
|
| 49 |
+
"grad_norm": 5.711933135986328,
|
| 50 |
"learning_rate": 4.7580906148867315e-05,
|
| 51 |
+
"loss": 4.607,
|
| 52 |
"step": 300
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 1.6990291262135924,
|
| 56 |
+
"grad_norm": 5.113691806793213,
|
| 57 |
"learning_rate": 4.717637540453075e-05,
|
| 58 |
+
"loss": 4.462,
|
| 59 |
"step": 350
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 1.941747572815534,
|
| 63 |
+
"grad_norm": 5.632521152496338,
|
| 64 |
"learning_rate": 4.6771844660194174e-05,
|
| 65 |
+
"loss": 4.3695,
|
| 66 |
"step": 400
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 2.1844660194174756,
|
| 70 |
+
"grad_norm": 5.428906440734863,
|
| 71 |
"learning_rate": 4.636731391585761e-05,
|
| 72 |
+
"loss": 4.1549,
|
| 73 |
"step": 450
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 2.4271844660194173,
|
| 77 |
+
"grad_norm": 5.037013530731201,
|
| 78 |
"learning_rate": 4.596278317152104e-05,
|
| 79 |
+
"loss": 4.0921,
|
| 80 |
"step": 500
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 2.6699029126213594,
|
| 84 |
+
"grad_norm": 5.231846809387207,
|
| 85 |
"learning_rate": 4.555825242718447e-05,
|
| 86 |
+
"loss": 4.0662,
|
| 87 |
"step": 550
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 2.912621359223301,
|
| 91 |
+
"grad_norm": 5.506278038024902,
|
| 92 |
"learning_rate": 4.51537216828479e-05,
|
| 93 |
+
"loss": 4.0403,
|
| 94 |
"step": 600
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 3.1553398058252426,
|
| 98 |
+
"grad_norm": 5.410216331481934,
|
| 99 |
"learning_rate": 4.4749190938511334e-05,
|
| 100 |
+
"loss": 3.8375,
|
| 101 |
"step": 650
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 3.3980582524271843,
|
| 105 |
+
"grad_norm": 5.637699127197266,
|
| 106 |
"learning_rate": 4.434466019417476e-05,
|
| 107 |
+
"loss": 3.7473,
|
| 108 |
"step": 700
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 3.6407766990291264,
|
| 112 |
+
"grad_norm": 5.575273513793945,
|
| 113 |
"learning_rate": 4.3940129449838194e-05,
|
| 114 |
+
"loss": 3.7835,
|
| 115 |
"step": 750
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 3.883495145631068,
|
| 119 |
+
"grad_norm": 5.737198352813721,
|
| 120 |
"learning_rate": 4.353559870550162e-05,
|
| 121 |
+
"loss": 3.7919,
|
| 122 |
"step": 800
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 4.12621359223301,
|
| 126 |
+
"grad_norm": 5.645532608032227,
|
| 127 |
"learning_rate": 4.313106796116505e-05,
|
| 128 |
+
"loss": 3.6425,
|
| 129 |
"step": 850
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 4.368932038834951,
|
| 133 |
+
"grad_norm": 6.308927059173584,
|
| 134 |
"learning_rate": 4.272653721682848e-05,
|
| 135 |
+
"loss": 3.4785,
|
| 136 |
"step": 900
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 4.611650485436893,
|
| 140 |
+
"grad_norm": 5.863094329833984,
|
| 141 |
"learning_rate": 4.232200647249191e-05,
|
| 142 |
+
"loss": 3.5268,
|
| 143 |
"step": 950
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 4.854368932038835,
|
| 147 |
+
"grad_norm": 6.120258331298828,
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
+
"loss": 3.5296,
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
+
"grad_norm": 6.231515884399414,
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
+
"loss": 3.3802,
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
+
"grad_norm": 6.211188316345215,
|
| 162 |
"learning_rate": 4.11084142394822e-05,
|
| 163 |
+
"loss": 3.2148,
|
| 164 |
"step": 1100
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 5.58252427184466,
|
| 168 |
+
"grad_norm": 6.8676934242248535,
|
| 169 |
"learning_rate": 4.0703883495145634e-05,
|
| 170 |
+
"loss": 3.2111,
|
| 171 |
"step": 1150
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 5.825242718446602,
|
| 175 |
+
"grad_norm": 6.851133346557617,
|
| 176 |
"learning_rate": 4.029935275080906e-05,
|
| 177 |
+
"loss": 3.2771,
|
| 178 |
"step": 1200
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 6.067961165048544,
|
| 182 |
+
"grad_norm": 6.708765506744385,
|
| 183 |
"learning_rate": 3.9894822006472494e-05,
|
| 184 |
+
"loss": 3.1587,
|
| 185 |
"step": 1250
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 6.310679611650485,
|
| 189 |
+
"grad_norm": 7.136553764343262,
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
+
"loss": 2.9192,
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
+
"grad_norm": 7.370823383331299,
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
+
"loss": 2.922,
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
+
"grad_norm": 7.630361557006836,
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
+
"loss": 2.9417,
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
+
"grad_norm": 7.765511989593506,
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
+
"loss": 2.8669,
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
+
"grad_norm": 7.85439920425415,
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
+
"loss": 2.5412,
|
| 220 |
"step": 1500
|
| 221 |
}
|
| 222 |
],
|
|
|
|
| 237 |
"attributes": {}
|
| 238 |
}
|
| 239 |
},
|
| 240 |
+
"total_flos": 696530239488000.0,
|
| 241 |
"train_batch_size": 4,
|
| 242 |
"trial_name": null,
|
| 243 |
"trial_params": null
|
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:d5c04ce53d557cd762ff77607aaadbd1248aaa177697662d9a21b90d444993b7
|
| 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":
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
-
"n_layer":
|
| 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": 2048,
|
| 14 |
+
"n_embd": 1024,
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
+
"n_layer": 12,
|
| 18 |
+
"n_positions": 2048,
|
| 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 sha256:3139c52e149dceb7821da5478bb894919e6cc2a81a11d980add682dc884a5b76
|
| 3 |
+
size 617130824
|
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:a5931b4e56f412f06449e4f288e088bb30ed3fc4bf706aa5894a87df734d5d19
|
| 3 |
+
size 1234355130
|
checkpoints/checkpoint-2000/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ace39524c0ea246b461dfdc96e06376945ad766256fe88d13b43d6b4c7c3988e
|
| 3 |
size 14244
|
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":
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
-
"loss": 5.
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
-
"grad_norm":
|
| 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": 6.
|
| 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":
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
-
"loss": 3.
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
-
"grad_norm":
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
-
"loss": 3.
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
-
"grad_norm":
|
| 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":
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
-
"loss":
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
-
"grad_norm":
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
-
"loss":
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
-
"grad_norm":
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
-
"loss":
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
-
"grad_norm":
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
-
"loss":
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
-
"grad_norm": 7.
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
-
"loss":
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 7.524271844660194,
|
| 224 |
-
"grad_norm":
|
| 225 |
"learning_rate": 3.746763754045307e-05,
|
| 226 |
-
"loss":
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 7.766990291262136,
|
| 231 |
-
"grad_norm":
|
| 232 |
"learning_rate": 3.7063106796116507e-05,
|
| 233 |
-
"loss":
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 8.009708737864077,
|
| 238 |
-
"grad_norm":
|
| 239 |
"learning_rate": 3.665857605177993e-05,
|
| 240 |
-
"loss":
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 8.25242718446602,
|
| 245 |
-
"grad_norm":
|
| 246 |
"learning_rate": 3.6254045307443366e-05,
|
| 247 |
-
"loss": 2.
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 8.495145631067961,
|
| 252 |
-
"grad_norm":
|
| 253 |
"learning_rate": 3.584951456310679e-05,
|
| 254 |
-
"loss": 2.
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 8.737864077669903,
|
| 259 |
-
"grad_norm":
|
| 260 |
"learning_rate": 3.5444983818770226e-05,
|
| 261 |
-
"loss":
|
| 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":
|
| 274 |
"learning_rate": 3.4635922330097086e-05,
|
| 275 |
-
"loss":
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 9.466019417475728,
|
| 280 |
-
"grad_norm":
|
| 281 |
"learning_rate": 3.423139158576052e-05,
|
| 282 |
-
"loss":
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 9.70873786407767,
|
| 287 |
-
"grad_norm":
|
| 288 |
"learning_rate": 3.382686084142395e-05,
|
| 289 |
-
"loss":
|
| 290 |
"step": 2000
|
| 291 |
}
|
| 292 |
],
|
|
@@ -307,7 +307,7 @@
|
|
| 307 |
"attributes": {}
|
| 308 |
}
|
| 309 |
},
|
| 310 |
-
"total_flos":
|
| 311 |
"train_batch_size": 4,
|
| 312 |
"trial_name": null,
|
| 313 |
"trial_params": null
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.24271844660194175,
|
| 14 |
+
"grad_norm": 2.8362834453582764,
|
| 15 |
"learning_rate": 4.960355987055016e-05,
|
| 16 |
+
"loss": 6.0742,
|
| 17 |
"step": 50
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4854368932038835,
|
| 21 |
+
"grad_norm": 2.9484496116638184,
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
+
"loss": 5.7218,
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
+
"grad_norm": 3.542572259902954,
|
| 29 |
"learning_rate": 4.879449838187702e-05,
|
| 30 |
+
"loss": 5.4227,
|
| 31 |
"step": 150
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.970873786407767,
|
| 35 |
+
"grad_norm": 4.634098052978516,
|
| 36 |
"learning_rate": 4.8389967637540455e-05,
|
| 37 |
+
"loss": 5.1655,
|
| 38 |
"step": 200
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.2135922330097086,
|
| 42 |
+
"grad_norm": 5.699330806732178,
|
| 43 |
"learning_rate": 4.798543689320388e-05,
|
| 44 |
+
"loss": 4.8009,
|
| 45 |
"step": 250
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 1.4563106796116505,
|
| 49 |
+
"grad_norm": 5.711933135986328,
|
| 50 |
"learning_rate": 4.7580906148867315e-05,
|
| 51 |
+
"loss": 4.607,
|
| 52 |
"step": 300
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 1.6990291262135924,
|
| 56 |
+
"grad_norm": 5.113691806793213,
|
| 57 |
"learning_rate": 4.717637540453075e-05,
|
| 58 |
+
"loss": 4.462,
|
| 59 |
"step": 350
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 1.941747572815534,
|
| 63 |
+
"grad_norm": 5.632521152496338,
|
| 64 |
"learning_rate": 4.6771844660194174e-05,
|
| 65 |
+
"loss": 4.3695,
|
| 66 |
"step": 400
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 2.1844660194174756,
|
| 70 |
+
"grad_norm": 5.428906440734863,
|
| 71 |
"learning_rate": 4.636731391585761e-05,
|
| 72 |
+
"loss": 4.1549,
|
| 73 |
"step": 450
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 2.4271844660194173,
|
| 77 |
+
"grad_norm": 5.037013530731201,
|
| 78 |
"learning_rate": 4.596278317152104e-05,
|
| 79 |
+
"loss": 4.0921,
|
| 80 |
"step": 500
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 2.6699029126213594,
|
| 84 |
+
"grad_norm": 5.231846809387207,
|
| 85 |
"learning_rate": 4.555825242718447e-05,
|
| 86 |
+
"loss": 4.0662,
|
| 87 |
"step": 550
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 2.912621359223301,
|
| 91 |
+
"grad_norm": 5.506278038024902,
|
| 92 |
"learning_rate": 4.51537216828479e-05,
|
| 93 |
+
"loss": 4.0403,
|
| 94 |
"step": 600
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 3.1553398058252426,
|
| 98 |
+
"grad_norm": 5.410216331481934,
|
| 99 |
"learning_rate": 4.4749190938511334e-05,
|
| 100 |
+
"loss": 3.8375,
|
| 101 |
"step": 650
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 3.3980582524271843,
|
| 105 |
+
"grad_norm": 5.637699127197266,
|
| 106 |
"learning_rate": 4.434466019417476e-05,
|
| 107 |
+
"loss": 3.7473,
|
| 108 |
"step": 700
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 3.6407766990291264,
|
| 112 |
+
"grad_norm": 5.575273513793945,
|
| 113 |
"learning_rate": 4.3940129449838194e-05,
|
| 114 |
+
"loss": 3.7835,
|
| 115 |
"step": 750
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 3.883495145631068,
|
| 119 |
+
"grad_norm": 5.737198352813721,
|
| 120 |
"learning_rate": 4.353559870550162e-05,
|
| 121 |
+
"loss": 3.7919,
|
| 122 |
"step": 800
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 4.12621359223301,
|
| 126 |
+
"grad_norm": 5.645532608032227,
|
| 127 |
"learning_rate": 4.313106796116505e-05,
|
| 128 |
+
"loss": 3.6425,
|
| 129 |
"step": 850
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 4.368932038834951,
|
| 133 |
+
"grad_norm": 6.308927059173584,
|
| 134 |
"learning_rate": 4.272653721682848e-05,
|
| 135 |
+
"loss": 3.4785,
|
| 136 |
"step": 900
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 4.611650485436893,
|
| 140 |
+
"grad_norm": 5.863094329833984,
|
| 141 |
"learning_rate": 4.232200647249191e-05,
|
| 142 |
+
"loss": 3.5268,
|
| 143 |
"step": 950
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 4.854368932038835,
|
| 147 |
+
"grad_norm": 6.120258331298828,
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
+
"loss": 3.5296,
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
+
"grad_norm": 6.231515884399414,
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
+
"loss": 3.3802,
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
+
"grad_norm": 6.211188316345215,
|
| 162 |
"learning_rate": 4.11084142394822e-05,
|
| 163 |
+
"loss": 3.2148,
|
| 164 |
"step": 1100
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 5.58252427184466,
|
| 168 |
+
"grad_norm": 6.8676934242248535,
|
| 169 |
"learning_rate": 4.0703883495145634e-05,
|
| 170 |
+
"loss": 3.2111,
|
| 171 |
"step": 1150
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 5.825242718446602,
|
| 175 |
+
"grad_norm": 6.851133346557617,
|
| 176 |
"learning_rate": 4.029935275080906e-05,
|
| 177 |
+
"loss": 3.2771,
|
| 178 |
"step": 1200
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 6.067961165048544,
|
| 182 |
+
"grad_norm": 6.708765506744385,
|
| 183 |
"learning_rate": 3.9894822006472494e-05,
|
| 184 |
+
"loss": 3.1587,
|
| 185 |
"step": 1250
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 6.310679611650485,
|
| 189 |
+
"grad_norm": 7.136553764343262,
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
+
"loss": 2.9192,
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
+
"grad_norm": 7.370823383331299,
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
+
"loss": 2.922,
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
+
"grad_norm": 7.630361557006836,
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
+
"loss": 2.9417,
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
+
"grad_norm": 7.765511989593506,
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
+
"loss": 2.8669,
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
+
"grad_norm": 7.85439920425415,
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
+
"loss": 2.5412,
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 7.524271844660194,
|
| 224 |
+
"grad_norm": 8.064071655273438,
|
| 225 |
"learning_rate": 3.746763754045307e-05,
|
| 226 |
+
"loss": 2.6131,
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 7.766990291262136,
|
| 231 |
+
"grad_norm": 8.692522048950195,
|
| 232 |
"learning_rate": 3.7063106796116507e-05,
|
| 233 |
+
"loss": 2.6356,
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 8.009708737864077,
|
| 238 |
+
"grad_norm": 8.012410163879395,
|
| 239 |
"learning_rate": 3.665857605177993e-05,
|
| 240 |
+
"loss": 2.599,
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 8.25242718446602,
|
| 245 |
+
"grad_norm": 8.6799898147583,
|
| 246 |
"learning_rate": 3.6254045307443366e-05,
|
| 247 |
+
"loss": 2.1992,
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 8.495145631067961,
|
| 252 |
+
"grad_norm": 8.963507652282715,
|
| 253 |
"learning_rate": 3.584951456310679e-05,
|
| 254 |
+
"loss": 2.2701,
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 8.737864077669903,
|
| 259 |
+
"grad_norm": 9.833822250366211,
|
| 260 |
"learning_rate": 3.5444983818770226e-05,
|
| 261 |
+
"loss": 2.2837,
|
| 262 |
"step": 1800
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 8.980582524271846,
|
| 266 |
+
"grad_norm": 9.47242259979248,
|
| 267 |
"learning_rate": 3.504045307443366e-05,
|
| 268 |
+
"loss": 2.2854,
|
| 269 |
"step": 1850
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 9.223300970873787,
|
| 273 |
+
"grad_norm": 8.856402397155762,
|
| 274 |
"learning_rate": 3.4635922330097086e-05,
|
| 275 |
+
"loss": 1.8862,
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 9.466019417475728,
|
| 280 |
+
"grad_norm": 9.948341369628906,
|
| 281 |
"learning_rate": 3.423139158576052e-05,
|
| 282 |
+
"loss": 1.922,
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 9.70873786407767,
|
| 287 |
+
"grad_norm": 10.28939151763916,
|
| 288 |
"learning_rate": 3.382686084142395e-05,
|
| 289 |
+
"loss": 1.9622,
|
| 290 |
"step": 2000
|
| 291 |
}
|
| 292 |
],
|
|
|
|
| 307 |
"attributes": {}
|
| 308 |
}
|
| 309 |
},
|
| 310 |
+
"total_flos": 928706985984000.0,
|
| 311 |
"train_batch_size": 4,
|
| 312 |
"trial_name": null,
|
| 313 |
"trial_params": null
|
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:d5c04ce53d557cd762ff77607aaadbd1248aaa177697662d9a21b90d444993b7
|
| 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":
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
-
"n_layer":
|
| 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": 2048,
|
| 14 |
+
"n_embd": 1024,
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
+
"n_layer": 12,
|
| 18 |
+
"n_positions": 2048,
|
| 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 sha256:2a16b607fac318ea7e4614190d1d857c6fc3ed432f70935eac08139817139d27
|
| 3 |
+
size 617130824
|
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 sha256:1df103a8189689ad21192707d0a5df531adec3994d42b7d06db9ffe8efac9c0b
|
| 3 |
+
size 1234355130
|
checkpoints/checkpoint-2500/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6220bdac54f7d614e1573ac8dd4440b0a8ffb03b06787f2282bd064aa9d0431e
|
| 3 |
size 14244
|
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":
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
-
"loss": 5.
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
-
"grad_norm":
|
| 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": 6.
|
| 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":
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
-
"loss": 3.
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
-
"grad_norm":
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
-
"loss": 3.
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
-
"grad_norm":
|
| 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":
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
-
"loss":
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
-
"grad_norm":
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
-
"loss":
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
-
"grad_norm":
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
-
"loss":
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
-
"grad_norm":
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
-
"loss":
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
-
"grad_norm": 7.
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
-
"loss":
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 7.524271844660194,
|
| 224 |
-
"grad_norm":
|
| 225 |
"learning_rate": 3.746763754045307e-05,
|
| 226 |
-
"loss":
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 7.766990291262136,
|
| 231 |
-
"grad_norm":
|
| 232 |
"learning_rate": 3.7063106796116507e-05,
|
| 233 |
-
"loss":
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 8.009708737864077,
|
| 238 |
-
"grad_norm":
|
| 239 |
"learning_rate": 3.665857605177993e-05,
|
| 240 |
-
"loss":
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 8.25242718446602,
|
| 245 |
-
"grad_norm":
|
| 246 |
"learning_rate": 3.6254045307443366e-05,
|
| 247 |
-
"loss": 2.
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 8.495145631067961,
|
| 252 |
-
"grad_norm":
|
| 253 |
"learning_rate": 3.584951456310679e-05,
|
| 254 |
-
"loss": 2.
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 8.737864077669903,
|
| 259 |
-
"grad_norm":
|
| 260 |
"learning_rate": 3.5444983818770226e-05,
|
| 261 |
-
"loss":
|
| 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":
|
| 274 |
"learning_rate": 3.4635922330097086e-05,
|
| 275 |
-
"loss":
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 9.466019417475728,
|
| 280 |
-
"grad_norm":
|
| 281 |
"learning_rate": 3.423139158576052e-05,
|
| 282 |
-
"loss":
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 9.70873786407767,
|
| 287 |
-
"grad_norm":
|
| 288 |
"learning_rate": 3.382686084142395e-05,
|
| 289 |
-
"loss":
|
| 290 |
"step": 2000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 9.951456310679612,
|
| 294 |
-
"grad_norm":
|
| 295 |
"learning_rate": 3.342233009708738e-05,
|
| 296 |
-
"loss":
|
| 297 |
"step": 2050
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 10.194174757281553,
|
| 301 |
-
"grad_norm":
|
| 302 |
"learning_rate": 3.301779935275081e-05,
|
| 303 |
-
"loss":
|
| 304 |
"step": 2100
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 10.436893203883495,
|
| 308 |
-
"grad_norm":
|
| 309 |
"learning_rate": 3.2613268608414246e-05,
|
| 310 |
-
"loss":
|
| 311 |
"step": 2150
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 10.679611650485437,
|
| 315 |
-
"grad_norm":
|
| 316 |
"learning_rate": 3.220873786407767e-05,
|
| 317 |
-
"loss":
|
| 318 |
"step": 2200
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 10.922330097087379,
|
| 322 |
-
"grad_norm":
|
| 323 |
"learning_rate": 3.1804207119741106e-05,
|
| 324 |
-
"loss":
|
| 325 |
"step": 2250
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 11.16504854368932,
|
| 329 |
-
"grad_norm":
|
| 330 |
"learning_rate": 3.139967637540453e-05,
|
| 331 |
-
"loss":
|
| 332 |
"step": 2300
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 11.407766990291263,
|
| 336 |
-
"grad_norm":
|
| 337 |
"learning_rate": 3.099514563106796e-05,
|
| 338 |
-
"loss":
|
| 339 |
"step": 2350
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 11.650485436893204,
|
| 343 |
-
"grad_norm":
|
| 344 |
"learning_rate": 3.059061488673139e-05,
|
| 345 |
-
"loss":
|
| 346 |
"step": 2400
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 11.893203883495145,
|
| 350 |
-
"grad_norm":
|
| 351 |
"learning_rate": 3.0186084142394822e-05,
|
| 352 |
-
"loss":
|
| 353 |
"step": 2450
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 12.135922330097088,
|
| 357 |
-
"grad_norm":
|
| 358 |
"learning_rate": 2.9781553398058252e-05,
|
| 359 |
-
"loss":
|
| 360 |
"step": 2500
|
| 361 |
}
|
| 362 |
],
|
|
@@ -377,7 +377,7 @@
|
|
| 377 |
"attributes": {}
|
| 378 |
}
|
| 379 |
},
|
| 380 |
-
"total_flos":
|
| 381 |
"train_batch_size": 4,
|
| 382 |
"trial_name": null,
|
| 383 |
"trial_params": null
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.24271844660194175,
|
| 14 |
+
"grad_norm": 2.8362834453582764,
|
| 15 |
"learning_rate": 4.960355987055016e-05,
|
| 16 |
+
"loss": 6.0742,
|
| 17 |
"step": 50
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4854368932038835,
|
| 21 |
+
"grad_norm": 2.9484496116638184,
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
+
"loss": 5.7218,
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
+
"grad_norm": 3.542572259902954,
|
| 29 |
"learning_rate": 4.879449838187702e-05,
|
| 30 |
+
"loss": 5.4227,
|
| 31 |
"step": 150
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.970873786407767,
|
| 35 |
+
"grad_norm": 4.634098052978516,
|
| 36 |
"learning_rate": 4.8389967637540455e-05,
|
| 37 |
+
"loss": 5.1655,
|
| 38 |
"step": 200
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.2135922330097086,
|
| 42 |
+
"grad_norm": 5.699330806732178,
|
| 43 |
"learning_rate": 4.798543689320388e-05,
|
| 44 |
+
"loss": 4.8009,
|
| 45 |
"step": 250
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 1.4563106796116505,
|
| 49 |
+
"grad_norm": 5.711933135986328,
|
| 50 |
"learning_rate": 4.7580906148867315e-05,
|
| 51 |
+
"loss": 4.607,
|
| 52 |
"step": 300
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 1.6990291262135924,
|
| 56 |
+
"grad_norm": 5.113691806793213,
|
| 57 |
"learning_rate": 4.717637540453075e-05,
|
| 58 |
+
"loss": 4.462,
|
| 59 |
"step": 350
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 1.941747572815534,
|
| 63 |
+
"grad_norm": 5.632521152496338,
|
| 64 |
"learning_rate": 4.6771844660194174e-05,
|
| 65 |
+
"loss": 4.3695,
|
| 66 |
"step": 400
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 2.1844660194174756,
|
| 70 |
+
"grad_norm": 5.428906440734863,
|
| 71 |
"learning_rate": 4.636731391585761e-05,
|
| 72 |
+
"loss": 4.1549,
|
| 73 |
"step": 450
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 2.4271844660194173,
|
| 77 |
+
"grad_norm": 5.037013530731201,
|
| 78 |
"learning_rate": 4.596278317152104e-05,
|
| 79 |
+
"loss": 4.0921,
|
| 80 |
"step": 500
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 2.6699029126213594,
|
| 84 |
+
"grad_norm": 5.231846809387207,
|
| 85 |
"learning_rate": 4.555825242718447e-05,
|
| 86 |
+
"loss": 4.0662,
|
| 87 |
"step": 550
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 2.912621359223301,
|
| 91 |
+
"grad_norm": 5.506278038024902,
|
| 92 |
"learning_rate": 4.51537216828479e-05,
|
| 93 |
+
"loss": 4.0403,
|
| 94 |
"step": 600
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 3.1553398058252426,
|
| 98 |
+
"grad_norm": 5.410216331481934,
|
| 99 |
"learning_rate": 4.4749190938511334e-05,
|
| 100 |
+
"loss": 3.8375,
|
| 101 |
"step": 650
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 3.3980582524271843,
|
| 105 |
+
"grad_norm": 5.637699127197266,
|
| 106 |
"learning_rate": 4.434466019417476e-05,
|
| 107 |
+
"loss": 3.7473,
|
| 108 |
"step": 700
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 3.6407766990291264,
|
| 112 |
+
"grad_norm": 5.575273513793945,
|
| 113 |
"learning_rate": 4.3940129449838194e-05,
|
| 114 |
+
"loss": 3.7835,
|
| 115 |
"step": 750
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 3.883495145631068,
|
| 119 |
+
"grad_norm": 5.737198352813721,
|
| 120 |
"learning_rate": 4.353559870550162e-05,
|
| 121 |
+
"loss": 3.7919,
|
| 122 |
"step": 800
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 4.12621359223301,
|
| 126 |
+
"grad_norm": 5.645532608032227,
|
| 127 |
"learning_rate": 4.313106796116505e-05,
|
| 128 |
+
"loss": 3.6425,
|
| 129 |
"step": 850
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 4.368932038834951,
|
| 133 |
+
"grad_norm": 6.308927059173584,
|
| 134 |
"learning_rate": 4.272653721682848e-05,
|
| 135 |
+
"loss": 3.4785,
|
| 136 |
"step": 900
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 4.611650485436893,
|
| 140 |
+
"grad_norm": 5.863094329833984,
|
| 141 |
"learning_rate": 4.232200647249191e-05,
|
| 142 |
+
"loss": 3.5268,
|
| 143 |
"step": 950
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 4.854368932038835,
|
| 147 |
+
"grad_norm": 6.120258331298828,
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
+
"loss": 3.5296,
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
+
"grad_norm": 6.231515884399414,
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
+
"loss": 3.3802,
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
+
"grad_norm": 6.211188316345215,
|
| 162 |
"learning_rate": 4.11084142394822e-05,
|
| 163 |
+
"loss": 3.2148,
|
| 164 |
"step": 1100
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 5.58252427184466,
|
| 168 |
+
"grad_norm": 6.8676934242248535,
|
| 169 |
"learning_rate": 4.0703883495145634e-05,
|
| 170 |
+
"loss": 3.2111,
|
| 171 |
"step": 1150
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 5.825242718446602,
|
| 175 |
+
"grad_norm": 6.851133346557617,
|
| 176 |
"learning_rate": 4.029935275080906e-05,
|
| 177 |
+
"loss": 3.2771,
|
| 178 |
"step": 1200
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 6.067961165048544,
|
| 182 |
+
"grad_norm": 6.708765506744385,
|
| 183 |
"learning_rate": 3.9894822006472494e-05,
|
| 184 |
+
"loss": 3.1587,
|
| 185 |
"step": 1250
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 6.310679611650485,
|
| 189 |
+
"grad_norm": 7.136553764343262,
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
+
"loss": 2.9192,
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
+
"grad_norm": 7.370823383331299,
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
+
"loss": 2.922,
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
+
"grad_norm": 7.630361557006836,
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
+
"loss": 2.9417,
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
+
"grad_norm": 7.765511989593506,
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
+
"loss": 2.8669,
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
+
"grad_norm": 7.85439920425415,
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
+
"loss": 2.5412,
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 7.524271844660194,
|
| 224 |
+
"grad_norm": 8.064071655273438,
|
| 225 |
"learning_rate": 3.746763754045307e-05,
|
| 226 |
+
"loss": 2.6131,
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 7.766990291262136,
|
| 231 |
+
"grad_norm": 8.692522048950195,
|
| 232 |
"learning_rate": 3.7063106796116507e-05,
|
| 233 |
+
"loss": 2.6356,
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 8.009708737864077,
|
| 238 |
+
"grad_norm": 8.012410163879395,
|
| 239 |
"learning_rate": 3.665857605177993e-05,
|
| 240 |
+
"loss": 2.599,
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 8.25242718446602,
|
| 245 |
+
"grad_norm": 8.6799898147583,
|
| 246 |
"learning_rate": 3.6254045307443366e-05,
|
| 247 |
+
"loss": 2.1992,
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 8.495145631067961,
|
| 252 |
+
"grad_norm": 8.963507652282715,
|
| 253 |
"learning_rate": 3.584951456310679e-05,
|
| 254 |
+
"loss": 2.2701,
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 8.737864077669903,
|
| 259 |
+
"grad_norm": 9.833822250366211,
|
| 260 |
"learning_rate": 3.5444983818770226e-05,
|
| 261 |
+
"loss": 2.2837,
|
| 262 |
"step": 1800
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 8.980582524271846,
|
| 266 |
+
"grad_norm": 9.47242259979248,
|
| 267 |
"learning_rate": 3.504045307443366e-05,
|
| 268 |
+
"loss": 2.2854,
|
| 269 |
"step": 1850
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 9.223300970873787,
|
| 273 |
+
"grad_norm": 8.856402397155762,
|
| 274 |
"learning_rate": 3.4635922330097086e-05,
|
| 275 |
+
"loss": 1.8862,
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 9.466019417475728,
|
| 280 |
+
"grad_norm": 9.948341369628906,
|
| 281 |
"learning_rate": 3.423139158576052e-05,
|
| 282 |
+
"loss": 1.922,
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 9.70873786407767,
|
| 287 |
+
"grad_norm": 10.28939151763916,
|
| 288 |
"learning_rate": 3.382686084142395e-05,
|
| 289 |
+
"loss": 1.9622,
|
| 290 |
"step": 2000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 9.951456310679612,
|
| 294 |
+
"grad_norm": 10.114766120910645,
|
| 295 |
"learning_rate": 3.342233009708738e-05,
|
| 296 |
+
"loss": 1.9741,
|
| 297 |
"step": 2050
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 10.194174757281553,
|
| 301 |
+
"grad_norm": 9.489270210266113,
|
| 302 |
"learning_rate": 3.301779935275081e-05,
|
| 303 |
+
"loss": 1.6509,
|
| 304 |
"step": 2100
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 10.436893203883495,
|
| 308 |
+
"grad_norm": 10.16482162475586,
|
| 309 |
"learning_rate": 3.2613268608414246e-05,
|
| 310 |
+
"loss": 1.5883,
|
| 311 |
"step": 2150
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 10.679611650485437,
|
| 315 |
+
"grad_norm": 11.778059005737305,
|
| 316 |
"learning_rate": 3.220873786407767e-05,
|
| 317 |
+
"loss": 1.6234,
|
| 318 |
"step": 2200
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 10.922330097087379,
|
| 322 |
+
"grad_norm": 10.798454284667969,
|
| 323 |
"learning_rate": 3.1804207119741106e-05,
|
| 324 |
+
"loss": 1.667,
|
| 325 |
"step": 2250
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 11.16504854368932,
|
| 329 |
+
"grad_norm": 9.751585006713867,
|
| 330 |
"learning_rate": 3.139967637540453e-05,
|
| 331 |
+
"loss": 1.3785,
|
| 332 |
"step": 2300
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 11.407766990291263,
|
| 336 |
+
"grad_norm": 10.266364097595215,
|
| 337 |
"learning_rate": 3.099514563106796e-05,
|
| 338 |
+
"loss": 1.2956,
|
| 339 |
"step": 2350
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 11.650485436893204,
|
| 343 |
+
"grad_norm": 10.660856246948242,
|
| 344 |
"learning_rate": 3.059061488673139e-05,
|
| 345 |
+
"loss": 1.3321,
|
| 346 |
"step": 2400
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 11.893203883495145,
|
| 350 |
+
"grad_norm": 10.896187782287598,
|
| 351 |
"learning_rate": 3.0186084142394822e-05,
|
| 352 |
+
"loss": 1.3533,
|
| 353 |
"step": 2450
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 12.135922330097088,
|
| 357 |
+
"grad_norm": 10.952502250671387,
|
| 358 |
"learning_rate": 2.9781553398058252e-05,
|
| 359 |
+
"loss": 1.1486,
|
| 360 |
"step": 2500
|
| 361 |
}
|
| 362 |
],
|
|
|
|
| 377 |
"attributes": {}
|
| 378 |
}
|
| 379 |
},
|
| 380 |
+
"total_flos": 1160883732480000.0,
|
| 381 |
"train_batch_size": 4,
|
| 382 |
"trial_name": null,
|
| 383 |
"trial_params": null
|
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:d5c04ce53d557cd762ff77607aaadbd1248aaa177697662d9a21b90d444993b7
|
| 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":
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
-
"n_layer":
|
| 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": 2048,
|
| 14 |
+
"n_embd": 1024,
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
+
"n_layer": 12,
|
| 18 |
+
"n_positions": 2048,
|
| 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:dca0c20fbe7c227e25d34c3ae029f189aba9c47fb81a6c624663bff3d5ea0555
|
| 3 |
+
size 617130824
|
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:ac34bfb31dd330051c1b732092c16075d893324bb0adbf25c9943f622855be0d
|
| 3 |
+
size 1234355130
|
checkpoints/checkpoint-3000/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:871442522375fc4d6ba225860b83f783120e0bc1b3bc165d21ae61e30b77d444
|
| 3 |
size 14244
|
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":
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
-
"loss": 5.
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
-
"grad_norm":
|
| 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": 6.
|
| 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":
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
-
"loss": 3.
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
-
"grad_norm":
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
-
"loss": 3.
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
-
"grad_norm":
|
| 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":
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
-
"loss":
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
-
"grad_norm":
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
-
"loss":
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
-
"grad_norm":
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
-
"loss":
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
-
"grad_norm":
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
-
"loss":
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
-
"grad_norm": 7.
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
-
"loss":
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 7.524271844660194,
|
| 224 |
-
"grad_norm":
|
| 225 |
"learning_rate": 3.746763754045307e-05,
|
| 226 |
-
"loss":
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 7.766990291262136,
|
| 231 |
-
"grad_norm":
|
| 232 |
"learning_rate": 3.7063106796116507e-05,
|
| 233 |
-
"loss":
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 8.009708737864077,
|
| 238 |
-
"grad_norm":
|
| 239 |
"learning_rate": 3.665857605177993e-05,
|
| 240 |
-
"loss":
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 8.25242718446602,
|
| 245 |
-
"grad_norm":
|
| 246 |
"learning_rate": 3.6254045307443366e-05,
|
| 247 |
-
"loss": 2.
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 8.495145631067961,
|
| 252 |
-
"grad_norm":
|
| 253 |
"learning_rate": 3.584951456310679e-05,
|
| 254 |
-
"loss": 2.
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 8.737864077669903,
|
| 259 |
-
"grad_norm":
|
| 260 |
"learning_rate": 3.5444983818770226e-05,
|
| 261 |
-
"loss":
|
| 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":
|
| 274 |
"learning_rate": 3.4635922330097086e-05,
|
| 275 |
-
"loss":
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 9.466019417475728,
|
| 280 |
-
"grad_norm":
|
| 281 |
"learning_rate": 3.423139158576052e-05,
|
| 282 |
-
"loss":
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 9.70873786407767,
|
| 287 |
-
"grad_norm":
|
| 288 |
"learning_rate": 3.382686084142395e-05,
|
| 289 |
-
"loss":
|
| 290 |
"step": 2000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 9.951456310679612,
|
| 294 |
-
"grad_norm":
|
| 295 |
"learning_rate": 3.342233009708738e-05,
|
| 296 |
-
"loss":
|
| 297 |
"step": 2050
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 10.194174757281553,
|
| 301 |
-
"grad_norm":
|
| 302 |
"learning_rate": 3.301779935275081e-05,
|
| 303 |
-
"loss":
|
| 304 |
"step": 2100
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 10.436893203883495,
|
| 308 |
-
"grad_norm":
|
| 309 |
"learning_rate": 3.2613268608414246e-05,
|
| 310 |
-
"loss":
|
| 311 |
"step": 2150
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 10.679611650485437,
|
| 315 |
-
"grad_norm":
|
| 316 |
"learning_rate": 3.220873786407767e-05,
|
| 317 |
-
"loss":
|
| 318 |
"step": 2200
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 10.922330097087379,
|
| 322 |
-
"grad_norm":
|
| 323 |
"learning_rate": 3.1804207119741106e-05,
|
| 324 |
-
"loss":
|
| 325 |
"step": 2250
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 11.16504854368932,
|
| 329 |
-
"grad_norm":
|
| 330 |
"learning_rate": 3.139967637540453e-05,
|
| 331 |
-
"loss":
|
| 332 |
"step": 2300
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 11.407766990291263,
|
| 336 |
-
"grad_norm":
|
| 337 |
"learning_rate": 3.099514563106796e-05,
|
| 338 |
-
"loss":
|
| 339 |
"step": 2350
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 11.650485436893204,
|
| 343 |
-
"grad_norm":
|
| 344 |
"learning_rate": 3.059061488673139e-05,
|
| 345 |
-
"loss":
|
| 346 |
"step": 2400
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 11.893203883495145,
|
| 350 |
-
"grad_norm":
|
| 351 |
"learning_rate": 3.0186084142394822e-05,
|
| 352 |
-
"loss":
|
| 353 |
"step": 2450
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 12.135922330097088,
|
| 357 |
-
"grad_norm":
|
| 358 |
"learning_rate": 2.9781553398058252e-05,
|
| 359 |
-
"loss":
|
| 360 |
"step": 2500
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 12.37864077669903,
|
| 364 |
-
"grad_norm":
|
| 365 |
"learning_rate": 2.9377022653721686e-05,
|
| 366 |
-
"loss":
|
| 367 |
"step": 2550
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 12.62135922330097,
|
| 371 |
-
"grad_norm": 9.
|
| 372 |
"learning_rate": 2.8972491909385112e-05,
|
| 373 |
-
"loss":
|
| 374 |
"step": 2600
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 12.864077669902912,
|
| 378 |
-
"grad_norm":
|
| 379 |
"learning_rate": 2.8567961165048546e-05,
|
| 380 |
-
"loss":
|
| 381 |
"step": 2650
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 13.106796116504855,
|
| 385 |
-
"grad_norm": 10.
|
| 386 |
"learning_rate": 2.816343042071198e-05,
|
| 387 |
-
"loss":
|
| 388 |
"step": 2700
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 13.349514563106796,
|
| 392 |
-
"grad_norm":
|
| 393 |
"learning_rate": 2.7758899676375405e-05,
|
| 394 |
-
"loss":
|
| 395 |
"step": 2750
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 13.592233009708737,
|
| 399 |
-
"grad_norm": 10.
|
| 400 |
"learning_rate": 2.735436893203884e-05,
|
| 401 |
-
"loss":
|
| 402 |
"step": 2800
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 13.83495145631068,
|
| 406 |
-
"grad_norm":
|
| 407 |
"learning_rate": 2.6949838187702265e-05,
|
| 408 |
-
"loss":
|
| 409 |
"step": 2850
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 14.077669902912621,
|
| 413 |
-
"grad_norm":
|
| 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":
|
| 423 |
"step": 2950
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 14.563106796116505,
|
| 427 |
-
"grad_norm": 10.
|
| 428 |
"learning_rate": 2.5736245954692555e-05,
|
| 429 |
-
"loss":
|
| 430 |
"step": 3000
|
| 431 |
}
|
| 432 |
],
|
|
@@ -447,7 +447,7 @@
|
|
| 447 |
"attributes": {}
|
| 448 |
}
|
| 449 |
},
|
| 450 |
-
"total_flos":
|
| 451 |
"train_batch_size": 4,
|
| 452 |
"trial_name": null,
|
| 453 |
"trial_params": null
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.24271844660194175,
|
| 14 |
+
"grad_norm": 2.8362834453582764,
|
| 15 |
"learning_rate": 4.960355987055016e-05,
|
| 16 |
+
"loss": 6.0742,
|
| 17 |
"step": 50
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4854368932038835,
|
| 21 |
+
"grad_norm": 2.9484496116638184,
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
+
"loss": 5.7218,
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
+
"grad_norm": 3.542572259902954,
|
| 29 |
"learning_rate": 4.879449838187702e-05,
|
| 30 |
+
"loss": 5.4227,
|
| 31 |
"step": 150
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.970873786407767,
|
| 35 |
+
"grad_norm": 4.634098052978516,
|
| 36 |
"learning_rate": 4.8389967637540455e-05,
|
| 37 |
+
"loss": 5.1655,
|
| 38 |
"step": 200
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.2135922330097086,
|
| 42 |
+
"grad_norm": 5.699330806732178,
|
| 43 |
"learning_rate": 4.798543689320388e-05,
|
| 44 |
+
"loss": 4.8009,
|
| 45 |
"step": 250
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 1.4563106796116505,
|
| 49 |
+
"grad_norm": 5.711933135986328,
|
| 50 |
"learning_rate": 4.7580906148867315e-05,
|
| 51 |
+
"loss": 4.607,
|
| 52 |
"step": 300
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 1.6990291262135924,
|
| 56 |
+
"grad_norm": 5.113691806793213,
|
| 57 |
"learning_rate": 4.717637540453075e-05,
|
| 58 |
+
"loss": 4.462,
|
| 59 |
"step": 350
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 1.941747572815534,
|
| 63 |
+
"grad_norm": 5.632521152496338,
|
| 64 |
"learning_rate": 4.6771844660194174e-05,
|
| 65 |
+
"loss": 4.3695,
|
| 66 |
"step": 400
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 2.1844660194174756,
|
| 70 |
+
"grad_norm": 5.428906440734863,
|
| 71 |
"learning_rate": 4.636731391585761e-05,
|
| 72 |
+
"loss": 4.1549,
|
| 73 |
"step": 450
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 2.4271844660194173,
|
| 77 |
+
"grad_norm": 5.037013530731201,
|
| 78 |
"learning_rate": 4.596278317152104e-05,
|
| 79 |
+
"loss": 4.0921,
|
| 80 |
"step": 500
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 2.6699029126213594,
|
| 84 |
+
"grad_norm": 5.231846809387207,
|
| 85 |
"learning_rate": 4.555825242718447e-05,
|
| 86 |
+
"loss": 4.0662,
|
| 87 |
"step": 550
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 2.912621359223301,
|
| 91 |
+
"grad_norm": 5.506278038024902,
|
| 92 |
"learning_rate": 4.51537216828479e-05,
|
| 93 |
+
"loss": 4.0403,
|
| 94 |
"step": 600
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 3.1553398058252426,
|
| 98 |
+
"grad_norm": 5.410216331481934,
|
| 99 |
"learning_rate": 4.4749190938511334e-05,
|
| 100 |
+
"loss": 3.8375,
|
| 101 |
"step": 650
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 3.3980582524271843,
|
| 105 |
+
"grad_norm": 5.637699127197266,
|
| 106 |
"learning_rate": 4.434466019417476e-05,
|
| 107 |
+
"loss": 3.7473,
|
| 108 |
"step": 700
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 3.6407766990291264,
|
| 112 |
+
"grad_norm": 5.575273513793945,
|
| 113 |
"learning_rate": 4.3940129449838194e-05,
|
| 114 |
+
"loss": 3.7835,
|
| 115 |
"step": 750
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 3.883495145631068,
|
| 119 |
+
"grad_norm": 5.737198352813721,
|
| 120 |
"learning_rate": 4.353559870550162e-05,
|
| 121 |
+
"loss": 3.7919,
|
| 122 |
"step": 800
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 4.12621359223301,
|
| 126 |
+
"grad_norm": 5.645532608032227,
|
| 127 |
"learning_rate": 4.313106796116505e-05,
|
| 128 |
+
"loss": 3.6425,
|
| 129 |
"step": 850
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 4.368932038834951,
|
| 133 |
+
"grad_norm": 6.308927059173584,
|
| 134 |
"learning_rate": 4.272653721682848e-05,
|
| 135 |
+
"loss": 3.4785,
|
| 136 |
"step": 900
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 4.611650485436893,
|
| 140 |
+
"grad_norm": 5.863094329833984,
|
| 141 |
"learning_rate": 4.232200647249191e-05,
|
| 142 |
+
"loss": 3.5268,
|
| 143 |
"step": 950
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 4.854368932038835,
|
| 147 |
+
"grad_norm": 6.120258331298828,
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
+
"loss": 3.5296,
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
+
"grad_norm": 6.231515884399414,
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
+
"loss": 3.3802,
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
+
"grad_norm": 6.211188316345215,
|
| 162 |
"learning_rate": 4.11084142394822e-05,
|
| 163 |
+
"loss": 3.2148,
|
| 164 |
"step": 1100
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 5.58252427184466,
|
| 168 |
+
"grad_norm": 6.8676934242248535,
|
| 169 |
"learning_rate": 4.0703883495145634e-05,
|
| 170 |
+
"loss": 3.2111,
|
| 171 |
"step": 1150
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 5.825242718446602,
|
| 175 |
+
"grad_norm": 6.851133346557617,
|
| 176 |
"learning_rate": 4.029935275080906e-05,
|
| 177 |
+
"loss": 3.2771,
|
| 178 |
"step": 1200
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 6.067961165048544,
|
| 182 |
+
"grad_norm": 6.708765506744385,
|
| 183 |
"learning_rate": 3.9894822006472494e-05,
|
| 184 |
+
"loss": 3.1587,
|
| 185 |
"step": 1250
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 6.310679611650485,
|
| 189 |
+
"grad_norm": 7.136553764343262,
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
+
"loss": 2.9192,
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
+
"grad_norm": 7.370823383331299,
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
+
"loss": 2.922,
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
+
"grad_norm": 7.630361557006836,
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
+
"loss": 2.9417,
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
+
"grad_norm": 7.765511989593506,
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
+
"loss": 2.8669,
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
+
"grad_norm": 7.85439920425415,
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
+
"loss": 2.5412,
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 7.524271844660194,
|
| 224 |
+
"grad_norm": 8.064071655273438,
|
| 225 |
"learning_rate": 3.746763754045307e-05,
|
| 226 |
+
"loss": 2.6131,
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 7.766990291262136,
|
| 231 |
+
"grad_norm": 8.692522048950195,
|
| 232 |
"learning_rate": 3.7063106796116507e-05,
|
| 233 |
+
"loss": 2.6356,
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 8.009708737864077,
|
| 238 |
+
"grad_norm": 8.012410163879395,
|
| 239 |
"learning_rate": 3.665857605177993e-05,
|
| 240 |
+
"loss": 2.599,
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 8.25242718446602,
|
| 245 |
+
"grad_norm": 8.6799898147583,
|
| 246 |
"learning_rate": 3.6254045307443366e-05,
|
| 247 |
+
"loss": 2.1992,
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 8.495145631067961,
|
| 252 |
+
"grad_norm": 8.963507652282715,
|
| 253 |
"learning_rate": 3.584951456310679e-05,
|
| 254 |
+
"loss": 2.2701,
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 8.737864077669903,
|
| 259 |
+
"grad_norm": 9.833822250366211,
|
| 260 |
"learning_rate": 3.5444983818770226e-05,
|
| 261 |
+
"loss": 2.2837,
|
| 262 |
"step": 1800
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 8.980582524271846,
|
| 266 |
+
"grad_norm": 9.47242259979248,
|
| 267 |
"learning_rate": 3.504045307443366e-05,
|
| 268 |
+
"loss": 2.2854,
|
| 269 |
"step": 1850
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 9.223300970873787,
|
| 273 |
+
"grad_norm": 8.856402397155762,
|
| 274 |
"learning_rate": 3.4635922330097086e-05,
|
| 275 |
+
"loss": 1.8862,
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 9.466019417475728,
|
| 280 |
+
"grad_norm": 9.948341369628906,
|
| 281 |
"learning_rate": 3.423139158576052e-05,
|
| 282 |
+
"loss": 1.922,
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 9.70873786407767,
|
| 287 |
+
"grad_norm": 10.28939151763916,
|
| 288 |
"learning_rate": 3.382686084142395e-05,
|
| 289 |
+
"loss": 1.9622,
|
| 290 |
"step": 2000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 9.951456310679612,
|
| 294 |
+
"grad_norm": 10.114766120910645,
|
| 295 |
"learning_rate": 3.342233009708738e-05,
|
| 296 |
+
"loss": 1.9741,
|
| 297 |
"step": 2050
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 10.194174757281553,
|
| 301 |
+
"grad_norm": 9.489270210266113,
|
| 302 |
"learning_rate": 3.301779935275081e-05,
|
| 303 |
+
"loss": 1.6509,
|
| 304 |
"step": 2100
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 10.436893203883495,
|
| 308 |
+
"grad_norm": 10.16482162475586,
|
| 309 |
"learning_rate": 3.2613268608414246e-05,
|
| 310 |
+
"loss": 1.5883,
|
| 311 |
"step": 2150
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 10.679611650485437,
|
| 315 |
+
"grad_norm": 11.778059005737305,
|
| 316 |
"learning_rate": 3.220873786407767e-05,
|
| 317 |
+
"loss": 1.6234,
|
| 318 |
"step": 2200
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 10.922330097087379,
|
| 322 |
+
"grad_norm": 10.798454284667969,
|
| 323 |
"learning_rate": 3.1804207119741106e-05,
|
| 324 |
+
"loss": 1.667,
|
| 325 |
"step": 2250
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 11.16504854368932,
|
| 329 |
+
"grad_norm": 9.751585006713867,
|
| 330 |
"learning_rate": 3.139967637540453e-05,
|
| 331 |
+
"loss": 1.3785,
|
| 332 |
"step": 2300
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 11.407766990291263,
|
| 336 |
+
"grad_norm": 10.266364097595215,
|
| 337 |
"learning_rate": 3.099514563106796e-05,
|
| 338 |
+
"loss": 1.2956,
|
| 339 |
"step": 2350
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 11.650485436893204,
|
| 343 |
+
"grad_norm": 10.660856246948242,
|
| 344 |
"learning_rate": 3.059061488673139e-05,
|
| 345 |
+
"loss": 1.3321,
|
| 346 |
"step": 2400
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 11.893203883495145,
|
| 350 |
+
"grad_norm": 10.896187782287598,
|
| 351 |
"learning_rate": 3.0186084142394822e-05,
|
| 352 |
+
"loss": 1.3533,
|
| 353 |
"step": 2450
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 12.135922330097088,
|
| 357 |
+
"grad_norm": 10.952502250671387,
|
| 358 |
"learning_rate": 2.9781553398058252e-05,
|
| 359 |
+
"loss": 1.1486,
|
| 360 |
"step": 2500
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 12.37864077669903,
|
| 364 |
+
"grad_norm": 10.766463279724121,
|
| 365 |
"learning_rate": 2.9377022653721686e-05,
|
| 366 |
+
"loss": 1.0227,
|
| 367 |
"step": 2550
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 12.62135922330097,
|
| 371 |
+
"grad_norm": 9.385764122009277,
|
| 372 |
"learning_rate": 2.8972491909385112e-05,
|
| 373 |
+
"loss": 1.0651,
|
| 374 |
"step": 2600
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 12.864077669902912,
|
| 378 |
+
"grad_norm": 11.573925018310547,
|
| 379 |
"learning_rate": 2.8567961165048546e-05,
|
| 380 |
+
"loss": 1.1013,
|
| 381 |
"step": 2650
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 13.106796116504855,
|
| 385 |
+
"grad_norm": 10.138091087341309,
|
| 386 |
"learning_rate": 2.816343042071198e-05,
|
| 387 |
+
"loss": 0.9545,
|
| 388 |
"step": 2700
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 13.349514563106796,
|
| 392 |
+
"grad_norm": 9.867119789123535,
|
| 393 |
"learning_rate": 2.7758899676375405e-05,
|
| 394 |
+
"loss": 0.7892,
|
| 395 |
"step": 2750
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 13.592233009708737,
|
| 399 |
+
"grad_norm": 10.443971633911133,
|
| 400 |
"learning_rate": 2.735436893203884e-05,
|
| 401 |
+
"loss": 0.8432,
|
| 402 |
"step": 2800
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 13.83495145631068,
|
| 406 |
+
"grad_norm": 10.633298873901367,
|
| 407 |
"learning_rate": 2.6949838187702265e-05,
|
| 408 |
+
"loss": 0.8543,
|
| 409 |
"step": 2850
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 14.077669902912621,
|
| 413 |
+
"grad_norm": 8.640830039978027,
|
| 414 |
"learning_rate": 2.6545307443365695e-05,
|
| 415 |
+
"loss": 0.7751,
|
| 416 |
"step": 2900
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 14.320388349514563,
|
| 420 |
+
"grad_norm": 10.461247444152832,
|
| 421 |
"learning_rate": 2.614077669902913e-05,
|
| 422 |
+
"loss": 0.6074,
|
| 423 |
"step": 2950
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 14.563106796116505,
|
| 427 |
+
"grad_norm": 10.757479667663574,
|
| 428 |
"learning_rate": 2.5736245954692555e-05,
|
| 429 |
+
"loss": 0.6354,
|
| 430 |
"step": 3000
|
| 431 |
}
|
| 432 |
],
|
|
|
|
| 447 |
"attributes": {}
|
| 448 |
}
|
| 449 |
},
|
| 450 |
+
"total_flos": 1393060478976000.0,
|
| 451 |
"train_batch_size": 4,
|
| 452 |
"trial_name": null,
|
| 453 |
"trial_params": null
|
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:d5c04ce53d557cd762ff77607aaadbd1248aaa177697662d9a21b90d444993b7
|
| 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":
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
-
"n_layer":
|
| 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": 2048,
|
| 14 |
+
"n_embd": 1024,
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
+
"n_layer": 12,
|
| 18 |
+
"n_positions": 2048,
|
| 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:ee4eddcf2434fc10bcb84be31fad14eaf9cb56d0657a2676caebdc4fa8006954
|
| 3 |
+
size 617130824
|
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:f0146bc12519d4396fc95dbc8a2c0e2b1eba55de6a678f2ca3f1bb831c8111e6
|
| 3 |
+
size 1234355130
|
checkpoints/checkpoint-3500/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:98ff60b9771555670f092dfde608075c59197b1a93a27a330f270d03668fc64e
|
| 3 |
size 14244
|
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":
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
-
"loss": 5.
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
-
"grad_norm":
|
| 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": 6.
|
| 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":
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
-
"loss": 3.
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
-
"grad_norm":
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
-
"loss": 3.
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
-
"grad_norm":
|
| 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":
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
-
"loss":
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
-
"grad_norm":
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
-
"loss":
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
-
"grad_norm":
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
-
"loss":
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
-
"grad_norm":
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
-
"loss":
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
-
"grad_norm": 7.
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
-
"loss":
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 7.524271844660194,
|
| 224 |
-
"grad_norm":
|
| 225 |
"learning_rate": 3.746763754045307e-05,
|
| 226 |
-
"loss":
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 7.766990291262136,
|
| 231 |
-
"grad_norm":
|
| 232 |
"learning_rate": 3.7063106796116507e-05,
|
| 233 |
-
"loss":
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 8.009708737864077,
|
| 238 |
-
"grad_norm":
|
| 239 |
"learning_rate": 3.665857605177993e-05,
|
| 240 |
-
"loss":
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 8.25242718446602,
|
| 245 |
-
"grad_norm":
|
| 246 |
"learning_rate": 3.6254045307443366e-05,
|
| 247 |
-
"loss": 2.
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 8.495145631067961,
|
| 252 |
-
"grad_norm":
|
| 253 |
"learning_rate": 3.584951456310679e-05,
|
| 254 |
-
"loss": 2.
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 8.737864077669903,
|
| 259 |
-
"grad_norm":
|
| 260 |
"learning_rate": 3.5444983818770226e-05,
|
| 261 |
-
"loss":
|
| 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":
|
| 274 |
"learning_rate": 3.4635922330097086e-05,
|
| 275 |
-
"loss":
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 9.466019417475728,
|
| 280 |
-
"grad_norm":
|
| 281 |
"learning_rate": 3.423139158576052e-05,
|
| 282 |
-
"loss":
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 9.70873786407767,
|
| 287 |
-
"grad_norm":
|
| 288 |
"learning_rate": 3.382686084142395e-05,
|
| 289 |
-
"loss":
|
| 290 |
"step": 2000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 9.951456310679612,
|
| 294 |
-
"grad_norm":
|
| 295 |
"learning_rate": 3.342233009708738e-05,
|
| 296 |
-
"loss":
|
| 297 |
"step": 2050
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 10.194174757281553,
|
| 301 |
-
"grad_norm":
|
| 302 |
"learning_rate": 3.301779935275081e-05,
|
| 303 |
-
"loss":
|
| 304 |
"step": 2100
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 10.436893203883495,
|
| 308 |
-
"grad_norm":
|
| 309 |
"learning_rate": 3.2613268608414246e-05,
|
| 310 |
-
"loss":
|
| 311 |
"step": 2150
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 10.679611650485437,
|
| 315 |
-
"grad_norm":
|
| 316 |
"learning_rate": 3.220873786407767e-05,
|
| 317 |
-
"loss":
|
| 318 |
"step": 2200
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 10.922330097087379,
|
| 322 |
-
"grad_norm":
|
| 323 |
"learning_rate": 3.1804207119741106e-05,
|
| 324 |
-
"loss":
|
| 325 |
"step": 2250
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 11.16504854368932,
|
| 329 |
-
"grad_norm":
|
| 330 |
"learning_rate": 3.139967637540453e-05,
|
| 331 |
-
"loss":
|
| 332 |
"step": 2300
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 11.407766990291263,
|
| 336 |
-
"grad_norm":
|
| 337 |
"learning_rate": 3.099514563106796e-05,
|
| 338 |
-
"loss":
|
| 339 |
"step": 2350
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 11.650485436893204,
|
| 343 |
-
"grad_norm":
|
| 344 |
"learning_rate": 3.059061488673139e-05,
|
| 345 |
-
"loss":
|
| 346 |
"step": 2400
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 11.893203883495145,
|
| 350 |
-
"grad_norm":
|
| 351 |
"learning_rate": 3.0186084142394822e-05,
|
| 352 |
-
"loss":
|
| 353 |
"step": 2450
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 12.135922330097088,
|
| 357 |
-
"grad_norm":
|
| 358 |
"learning_rate": 2.9781553398058252e-05,
|
| 359 |
-
"loss":
|
| 360 |
"step": 2500
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 12.37864077669903,
|
| 364 |
-
"grad_norm":
|
| 365 |
"learning_rate": 2.9377022653721686e-05,
|
| 366 |
-
"loss":
|
| 367 |
"step": 2550
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 12.62135922330097,
|
| 371 |
-
"grad_norm": 9.
|
| 372 |
"learning_rate": 2.8972491909385112e-05,
|
| 373 |
-
"loss":
|
| 374 |
"step": 2600
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 12.864077669902912,
|
| 378 |
-
"grad_norm":
|
| 379 |
"learning_rate": 2.8567961165048546e-05,
|
| 380 |
-
"loss":
|
| 381 |
"step": 2650
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 13.106796116504855,
|
| 385 |
-
"grad_norm": 10.
|
| 386 |
"learning_rate": 2.816343042071198e-05,
|
| 387 |
-
"loss":
|
| 388 |
"step": 2700
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 13.349514563106796,
|
| 392 |
-
"grad_norm":
|
| 393 |
"learning_rate": 2.7758899676375405e-05,
|
| 394 |
-
"loss":
|
| 395 |
"step": 2750
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 13.592233009708737,
|
| 399 |
-
"grad_norm": 10.
|
| 400 |
"learning_rate": 2.735436893203884e-05,
|
| 401 |
-
"loss":
|
| 402 |
"step": 2800
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 13.83495145631068,
|
| 406 |
-
"grad_norm":
|
| 407 |
"learning_rate": 2.6949838187702265e-05,
|
| 408 |
-
"loss":
|
| 409 |
"step": 2850
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 14.077669902912621,
|
| 413 |
-
"grad_norm":
|
| 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":
|
| 423 |
"step": 2950
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 14.563106796116505,
|
| 427 |
-
"grad_norm": 10.
|
| 428 |
"learning_rate": 2.5736245954692555e-05,
|
| 429 |
-
"loss":
|
| 430 |
"step": 3000
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 14.805825242718447,
|
| 434 |
-
"grad_norm": 11.
|
| 435 |
"learning_rate": 2.533171521035599e-05,
|
| 436 |
-
"loss":
|
| 437 |
"step": 3050
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 15.048543689320388,
|
| 441 |
-
"grad_norm":
|
| 442 |
"learning_rate": 2.492718446601942e-05,
|
| 443 |
-
"loss":
|
| 444 |
"step": 3100
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 15.29126213592233,
|
| 448 |
-
"grad_norm":
|
| 449 |
"learning_rate": 2.452265372168285e-05,
|
| 450 |
-
"loss":
|
| 451 |
"step": 3150
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 15.533980582524272,
|
| 455 |
-
"grad_norm":
|
| 456 |
"learning_rate": 2.411812297734628e-05,
|
| 457 |
-
"loss":
|
| 458 |
"step": 3200
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 15.776699029126213,
|
| 462 |
-
"grad_norm":
|
| 463 |
"learning_rate": 2.3713592233009708e-05,
|
| 464 |
-
"loss":
|
| 465 |
"step": 3250
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 16.019417475728154,
|
| 469 |
-
"grad_norm":
|
| 470 |
"learning_rate": 2.3309061488673138e-05,
|
| 471 |
-
"loss":
|
| 472 |
"step": 3300
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 16.262135922330096,
|
| 476 |
-
"grad_norm":
|
| 477 |
"learning_rate": 2.290453074433657e-05,
|
| 478 |
-
"loss":
|
| 479 |
"step": 3350
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 16.50485436893204,
|
| 483 |
-
"grad_norm":
|
| 484 |
"learning_rate": 2.25e-05,
|
| 485 |
-
"loss":
|
| 486 |
"step": 3400
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 16.74757281553398,
|
| 490 |
-
"grad_norm":
|
| 491 |
"learning_rate": 2.209546925566343e-05,
|
| 492 |
-
"loss":
|
| 493 |
"step": 3450
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 16.990291262135923,
|
| 497 |
-
"grad_norm":
|
| 498 |
"learning_rate": 2.169093851132686e-05,
|
| 499 |
-
"loss":
|
| 500 |
"step": 3500
|
| 501 |
}
|
| 502 |
],
|
|
@@ -517,7 +517,7 @@
|
|
| 517 |
"attributes": {}
|
| 518 |
}
|
| 519 |
},
|
| 520 |
-
"total_flos":
|
| 521 |
"train_batch_size": 4,
|
| 522 |
"trial_name": null,
|
| 523 |
"trial_params": null
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.24271844660194175,
|
| 14 |
+
"grad_norm": 2.8362834453582764,
|
| 15 |
"learning_rate": 4.960355987055016e-05,
|
| 16 |
+
"loss": 6.0742,
|
| 17 |
"step": 50
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4854368932038835,
|
| 21 |
+
"grad_norm": 2.9484496116638184,
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
+
"loss": 5.7218,
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
+
"grad_norm": 3.542572259902954,
|
| 29 |
"learning_rate": 4.879449838187702e-05,
|
| 30 |
+
"loss": 5.4227,
|
| 31 |
"step": 150
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.970873786407767,
|
| 35 |
+
"grad_norm": 4.634098052978516,
|
| 36 |
"learning_rate": 4.8389967637540455e-05,
|
| 37 |
+
"loss": 5.1655,
|
| 38 |
"step": 200
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.2135922330097086,
|
| 42 |
+
"grad_norm": 5.699330806732178,
|
| 43 |
"learning_rate": 4.798543689320388e-05,
|
| 44 |
+
"loss": 4.8009,
|
| 45 |
"step": 250
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 1.4563106796116505,
|
| 49 |
+
"grad_norm": 5.711933135986328,
|
| 50 |
"learning_rate": 4.7580906148867315e-05,
|
| 51 |
+
"loss": 4.607,
|
| 52 |
"step": 300
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 1.6990291262135924,
|
| 56 |
+
"grad_norm": 5.113691806793213,
|
| 57 |
"learning_rate": 4.717637540453075e-05,
|
| 58 |
+
"loss": 4.462,
|
| 59 |
"step": 350
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 1.941747572815534,
|
| 63 |
+
"grad_norm": 5.632521152496338,
|
| 64 |
"learning_rate": 4.6771844660194174e-05,
|
| 65 |
+
"loss": 4.3695,
|
| 66 |
"step": 400
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 2.1844660194174756,
|
| 70 |
+
"grad_norm": 5.428906440734863,
|
| 71 |
"learning_rate": 4.636731391585761e-05,
|
| 72 |
+
"loss": 4.1549,
|
| 73 |
"step": 450
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 2.4271844660194173,
|
| 77 |
+
"grad_norm": 5.037013530731201,
|
| 78 |
"learning_rate": 4.596278317152104e-05,
|
| 79 |
+
"loss": 4.0921,
|
| 80 |
"step": 500
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 2.6699029126213594,
|
| 84 |
+
"grad_norm": 5.231846809387207,
|
| 85 |
"learning_rate": 4.555825242718447e-05,
|
| 86 |
+
"loss": 4.0662,
|
| 87 |
"step": 550
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 2.912621359223301,
|
| 91 |
+
"grad_norm": 5.506278038024902,
|
| 92 |
"learning_rate": 4.51537216828479e-05,
|
| 93 |
+
"loss": 4.0403,
|
| 94 |
"step": 600
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 3.1553398058252426,
|
| 98 |
+
"grad_norm": 5.410216331481934,
|
| 99 |
"learning_rate": 4.4749190938511334e-05,
|
| 100 |
+
"loss": 3.8375,
|
| 101 |
"step": 650
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 3.3980582524271843,
|
| 105 |
+
"grad_norm": 5.637699127197266,
|
| 106 |
"learning_rate": 4.434466019417476e-05,
|
| 107 |
+
"loss": 3.7473,
|
| 108 |
"step": 700
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 3.6407766990291264,
|
| 112 |
+
"grad_norm": 5.575273513793945,
|
| 113 |
"learning_rate": 4.3940129449838194e-05,
|
| 114 |
+
"loss": 3.7835,
|
| 115 |
"step": 750
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 3.883495145631068,
|
| 119 |
+
"grad_norm": 5.737198352813721,
|
| 120 |
"learning_rate": 4.353559870550162e-05,
|
| 121 |
+
"loss": 3.7919,
|
| 122 |
"step": 800
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 4.12621359223301,
|
| 126 |
+
"grad_norm": 5.645532608032227,
|
| 127 |
"learning_rate": 4.313106796116505e-05,
|
| 128 |
+
"loss": 3.6425,
|
| 129 |
"step": 850
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 4.368932038834951,
|
| 133 |
+
"grad_norm": 6.308927059173584,
|
| 134 |
"learning_rate": 4.272653721682848e-05,
|
| 135 |
+
"loss": 3.4785,
|
| 136 |
"step": 900
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 4.611650485436893,
|
| 140 |
+
"grad_norm": 5.863094329833984,
|
| 141 |
"learning_rate": 4.232200647249191e-05,
|
| 142 |
+
"loss": 3.5268,
|
| 143 |
"step": 950
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 4.854368932038835,
|
| 147 |
+
"grad_norm": 6.120258331298828,
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
+
"loss": 3.5296,
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
+
"grad_norm": 6.231515884399414,
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
+
"loss": 3.3802,
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
+
"grad_norm": 6.211188316345215,
|
| 162 |
"learning_rate": 4.11084142394822e-05,
|
| 163 |
+
"loss": 3.2148,
|
| 164 |
"step": 1100
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 5.58252427184466,
|
| 168 |
+
"grad_norm": 6.8676934242248535,
|
| 169 |
"learning_rate": 4.0703883495145634e-05,
|
| 170 |
+
"loss": 3.2111,
|
| 171 |
"step": 1150
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 5.825242718446602,
|
| 175 |
+
"grad_norm": 6.851133346557617,
|
| 176 |
"learning_rate": 4.029935275080906e-05,
|
| 177 |
+
"loss": 3.2771,
|
| 178 |
"step": 1200
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 6.067961165048544,
|
| 182 |
+
"grad_norm": 6.708765506744385,
|
| 183 |
"learning_rate": 3.9894822006472494e-05,
|
| 184 |
+
"loss": 3.1587,
|
| 185 |
"step": 1250
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 6.310679611650485,
|
| 189 |
+
"grad_norm": 7.136553764343262,
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
+
"loss": 2.9192,
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
+
"grad_norm": 7.370823383331299,
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
+
"loss": 2.922,
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
+
"grad_norm": 7.630361557006836,
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
+
"loss": 2.9417,
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
+
"grad_norm": 7.765511989593506,
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
+
"loss": 2.8669,
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
+
"grad_norm": 7.85439920425415,
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
+
"loss": 2.5412,
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 7.524271844660194,
|
| 224 |
+
"grad_norm": 8.064071655273438,
|
| 225 |
"learning_rate": 3.746763754045307e-05,
|
| 226 |
+
"loss": 2.6131,
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 7.766990291262136,
|
| 231 |
+
"grad_norm": 8.692522048950195,
|
| 232 |
"learning_rate": 3.7063106796116507e-05,
|
| 233 |
+
"loss": 2.6356,
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 8.009708737864077,
|
| 238 |
+
"grad_norm": 8.012410163879395,
|
| 239 |
"learning_rate": 3.665857605177993e-05,
|
| 240 |
+
"loss": 2.599,
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 8.25242718446602,
|
| 245 |
+
"grad_norm": 8.6799898147583,
|
| 246 |
"learning_rate": 3.6254045307443366e-05,
|
| 247 |
+
"loss": 2.1992,
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 8.495145631067961,
|
| 252 |
+
"grad_norm": 8.963507652282715,
|
| 253 |
"learning_rate": 3.584951456310679e-05,
|
| 254 |
+
"loss": 2.2701,
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 8.737864077669903,
|
| 259 |
+
"grad_norm": 9.833822250366211,
|
| 260 |
"learning_rate": 3.5444983818770226e-05,
|
| 261 |
+
"loss": 2.2837,
|
| 262 |
"step": 1800
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 8.980582524271846,
|
| 266 |
+
"grad_norm": 9.47242259979248,
|
| 267 |
"learning_rate": 3.504045307443366e-05,
|
| 268 |
+
"loss": 2.2854,
|
| 269 |
"step": 1850
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 9.223300970873787,
|
| 273 |
+
"grad_norm": 8.856402397155762,
|
| 274 |
"learning_rate": 3.4635922330097086e-05,
|
| 275 |
+
"loss": 1.8862,
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 9.466019417475728,
|
| 280 |
+
"grad_norm": 9.948341369628906,
|
| 281 |
"learning_rate": 3.423139158576052e-05,
|
| 282 |
+
"loss": 1.922,
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 9.70873786407767,
|
| 287 |
+
"grad_norm": 10.28939151763916,
|
| 288 |
"learning_rate": 3.382686084142395e-05,
|
| 289 |
+
"loss": 1.9622,
|
| 290 |
"step": 2000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 9.951456310679612,
|
| 294 |
+
"grad_norm": 10.114766120910645,
|
| 295 |
"learning_rate": 3.342233009708738e-05,
|
| 296 |
+
"loss": 1.9741,
|
| 297 |
"step": 2050
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 10.194174757281553,
|
| 301 |
+
"grad_norm": 9.489270210266113,
|
| 302 |
"learning_rate": 3.301779935275081e-05,
|
| 303 |
+
"loss": 1.6509,
|
| 304 |
"step": 2100
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 10.436893203883495,
|
| 308 |
+
"grad_norm": 10.16482162475586,
|
| 309 |
"learning_rate": 3.2613268608414246e-05,
|
| 310 |
+
"loss": 1.5883,
|
| 311 |
"step": 2150
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 10.679611650485437,
|
| 315 |
+
"grad_norm": 11.778059005737305,
|
| 316 |
"learning_rate": 3.220873786407767e-05,
|
| 317 |
+
"loss": 1.6234,
|
| 318 |
"step": 2200
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 10.922330097087379,
|
| 322 |
+
"grad_norm": 10.798454284667969,
|
| 323 |
"learning_rate": 3.1804207119741106e-05,
|
| 324 |
+
"loss": 1.667,
|
| 325 |
"step": 2250
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 11.16504854368932,
|
| 329 |
+
"grad_norm": 9.751585006713867,
|
| 330 |
"learning_rate": 3.139967637540453e-05,
|
| 331 |
+
"loss": 1.3785,
|
| 332 |
"step": 2300
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 11.407766990291263,
|
| 336 |
+
"grad_norm": 10.266364097595215,
|
| 337 |
"learning_rate": 3.099514563106796e-05,
|
| 338 |
+
"loss": 1.2956,
|
| 339 |
"step": 2350
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 11.650485436893204,
|
| 343 |
+
"grad_norm": 10.660856246948242,
|
| 344 |
"learning_rate": 3.059061488673139e-05,
|
| 345 |
+
"loss": 1.3321,
|
| 346 |
"step": 2400
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 11.893203883495145,
|
| 350 |
+
"grad_norm": 10.896187782287598,
|
| 351 |
"learning_rate": 3.0186084142394822e-05,
|
| 352 |
+
"loss": 1.3533,
|
| 353 |
"step": 2450
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 12.135922330097088,
|
| 357 |
+
"grad_norm": 10.952502250671387,
|
| 358 |
"learning_rate": 2.9781553398058252e-05,
|
| 359 |
+
"loss": 1.1486,
|
| 360 |
"step": 2500
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 12.37864077669903,
|
| 364 |
+
"grad_norm": 10.766463279724121,
|
| 365 |
"learning_rate": 2.9377022653721686e-05,
|
| 366 |
+
"loss": 1.0227,
|
| 367 |
"step": 2550
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 12.62135922330097,
|
| 371 |
+
"grad_norm": 9.385764122009277,
|
| 372 |
"learning_rate": 2.8972491909385112e-05,
|
| 373 |
+
"loss": 1.0651,
|
| 374 |
"step": 2600
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 12.864077669902912,
|
| 378 |
+
"grad_norm": 11.573925018310547,
|
| 379 |
"learning_rate": 2.8567961165048546e-05,
|
| 380 |
+
"loss": 1.1013,
|
| 381 |
"step": 2650
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 13.106796116504855,
|
| 385 |
+
"grad_norm": 10.138091087341309,
|
| 386 |
"learning_rate": 2.816343042071198e-05,
|
| 387 |
+
"loss": 0.9545,
|
| 388 |
"step": 2700
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 13.349514563106796,
|
| 392 |
+
"grad_norm": 9.867119789123535,
|
| 393 |
"learning_rate": 2.7758899676375405e-05,
|
| 394 |
+
"loss": 0.7892,
|
| 395 |
"step": 2750
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 13.592233009708737,
|
| 399 |
+
"grad_norm": 10.443971633911133,
|
| 400 |
"learning_rate": 2.735436893203884e-05,
|
| 401 |
+
"loss": 0.8432,
|
| 402 |
"step": 2800
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 13.83495145631068,
|
| 406 |
+
"grad_norm": 10.633298873901367,
|
| 407 |
"learning_rate": 2.6949838187702265e-05,
|
| 408 |
+
"loss": 0.8543,
|
| 409 |
"step": 2850
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 14.077669902912621,
|
| 413 |
+
"grad_norm": 8.640830039978027,
|
| 414 |
"learning_rate": 2.6545307443365695e-05,
|
| 415 |
+
"loss": 0.7751,
|
| 416 |
"step": 2900
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 14.320388349514563,
|
| 420 |
+
"grad_norm": 10.461247444152832,
|
| 421 |
"learning_rate": 2.614077669902913e-05,
|
| 422 |
+
"loss": 0.6074,
|
| 423 |
"step": 2950
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 14.563106796116505,
|
| 427 |
+
"grad_norm": 10.757479667663574,
|
| 428 |
"learning_rate": 2.5736245954692555e-05,
|
| 429 |
+
"loss": 0.6354,
|
| 430 |
"step": 3000
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 14.805825242718447,
|
| 434 |
+
"grad_norm": 11.599132537841797,
|
| 435 |
"learning_rate": 2.533171521035599e-05,
|
| 436 |
+
"loss": 0.6694,
|
| 437 |
"step": 3050
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 15.048543689320388,
|
| 441 |
+
"grad_norm": 8.085358619689941,
|
| 442 |
"learning_rate": 2.492718446601942e-05,
|
| 443 |
+
"loss": 0.6306,
|
| 444 |
"step": 3100
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 15.29126213592233,
|
| 448 |
+
"grad_norm": 8.327988624572754,
|
| 449 |
"learning_rate": 2.452265372168285e-05,
|
| 450 |
+
"loss": 0.4628,
|
| 451 |
"step": 3150
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 15.533980582524272,
|
| 455 |
+
"grad_norm": 8.545391082763672,
|
| 456 |
"learning_rate": 2.411812297734628e-05,
|
| 457 |
+
"loss": 0.4915,
|
| 458 |
"step": 3200
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 15.776699029126213,
|
| 462 |
+
"grad_norm": 10.176375389099121,
|
| 463 |
"learning_rate": 2.3713592233009708e-05,
|
| 464 |
+
"loss": 0.5105,
|
| 465 |
"step": 3250
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 16.019417475728154,
|
| 469 |
+
"grad_norm": 7.821159362792969,
|
| 470 |
"learning_rate": 2.3309061488673138e-05,
|
| 471 |
+
"loss": 0.5055,
|
| 472 |
"step": 3300
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 16.262135922330096,
|
| 476 |
+
"grad_norm": 7.797800064086914,
|
| 477 |
"learning_rate": 2.290453074433657e-05,
|
| 478 |
+
"loss": 0.3486,
|
| 479 |
"step": 3350
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 16.50485436893204,
|
| 483 |
+
"grad_norm": 8.497370719909668,
|
| 484 |
"learning_rate": 2.25e-05,
|
| 485 |
+
"loss": 0.3627,
|
| 486 |
"step": 3400
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 16.74757281553398,
|
| 490 |
+
"grad_norm": 9.842249870300293,
|
| 491 |
"learning_rate": 2.209546925566343e-05,
|
| 492 |
+
"loss": 0.3969,
|
| 493 |
"step": 3450
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 16.990291262135923,
|
| 497 |
+
"grad_norm": 10.705739974975586,
|
| 498 |
"learning_rate": 2.169093851132686e-05,
|
| 499 |
+
"loss": 0.4028,
|
| 500 |
"step": 3500
|
| 501 |
}
|
| 502 |
],
|
|
|
|
| 517 |
"attributes": {}
|
| 518 |
}
|
| 519 |
},
|
| 520 |
+
"total_flos": 1625237225472000.0,
|
| 521 |
"train_batch_size": 4,
|
| 522 |
"trial_name": null,
|
| 523 |
"trial_params": null
|
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:d5c04ce53d557cd762ff77607aaadbd1248aaa177697662d9a21b90d444993b7
|
| 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":
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
-
"n_layer":
|
| 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": 2048,
|
| 14 |
+
"n_embd": 1024,
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
+
"n_layer": 12,
|
| 18 |
+
"n_positions": 2048,
|
| 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 sha256:1c104ff20a7d8e0674d2df872410a78a14262bdb171fe8770ebf6ad6cb47abe3
|
| 3 |
+
size 617130824
|
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 sha256:62e74ed7e1c6f81b24fe69e4ef050494e5a8e9227ed6983c8647dc5fbd0d0ade
|
| 3 |
+
size 1234355130
|
checkpoints/checkpoint-4000/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2b2997b95ac6c0d6711e7d18cba7213a8b49deca7976b4a4ecbf3d8da67023b1
|
| 3 |
size 14244
|
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":
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
-
"loss": 5.
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
-
"grad_norm":
|
| 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": 6.
|
| 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":
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
-
"loss": 3.
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
-
"grad_norm":
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
-
"loss": 3.
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
-
"grad_norm":
|
| 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":
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
-
"loss":
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
-
"grad_norm":
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
-
"loss":
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
-
"grad_norm":
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
-
"loss":
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
-
"grad_norm":
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
-
"loss":
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
-
"grad_norm": 7.
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
-
"loss":
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 7.524271844660194,
|
| 224 |
-
"grad_norm":
|
| 225 |
"learning_rate": 3.746763754045307e-05,
|
| 226 |
-
"loss":
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 7.766990291262136,
|
| 231 |
-
"grad_norm":
|
| 232 |
"learning_rate": 3.7063106796116507e-05,
|
| 233 |
-
"loss":
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 8.009708737864077,
|
| 238 |
-
"grad_norm":
|
| 239 |
"learning_rate": 3.665857605177993e-05,
|
| 240 |
-
"loss":
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 8.25242718446602,
|
| 245 |
-
"grad_norm":
|
| 246 |
"learning_rate": 3.6254045307443366e-05,
|
| 247 |
-
"loss": 2.
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 8.495145631067961,
|
| 252 |
-
"grad_norm":
|
| 253 |
"learning_rate": 3.584951456310679e-05,
|
| 254 |
-
"loss": 2.
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 8.737864077669903,
|
| 259 |
-
"grad_norm":
|
| 260 |
"learning_rate": 3.5444983818770226e-05,
|
| 261 |
-
"loss":
|
| 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":
|
| 274 |
"learning_rate": 3.4635922330097086e-05,
|
| 275 |
-
"loss":
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 9.466019417475728,
|
| 280 |
-
"grad_norm":
|
| 281 |
"learning_rate": 3.423139158576052e-05,
|
| 282 |
-
"loss":
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 9.70873786407767,
|
| 287 |
-
"grad_norm":
|
| 288 |
"learning_rate": 3.382686084142395e-05,
|
| 289 |
-
"loss":
|
| 290 |
"step": 2000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 9.951456310679612,
|
| 294 |
-
"grad_norm":
|
| 295 |
"learning_rate": 3.342233009708738e-05,
|
| 296 |
-
"loss":
|
| 297 |
"step": 2050
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 10.194174757281553,
|
| 301 |
-
"grad_norm":
|
| 302 |
"learning_rate": 3.301779935275081e-05,
|
| 303 |
-
"loss":
|
| 304 |
"step": 2100
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 10.436893203883495,
|
| 308 |
-
"grad_norm":
|
| 309 |
"learning_rate": 3.2613268608414246e-05,
|
| 310 |
-
"loss":
|
| 311 |
"step": 2150
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 10.679611650485437,
|
| 315 |
-
"grad_norm":
|
| 316 |
"learning_rate": 3.220873786407767e-05,
|
| 317 |
-
"loss":
|
| 318 |
"step": 2200
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 10.922330097087379,
|
| 322 |
-
"grad_norm":
|
| 323 |
"learning_rate": 3.1804207119741106e-05,
|
| 324 |
-
"loss":
|
| 325 |
"step": 2250
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 11.16504854368932,
|
| 329 |
-
"grad_norm":
|
| 330 |
"learning_rate": 3.139967637540453e-05,
|
| 331 |
-
"loss":
|
| 332 |
"step": 2300
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 11.407766990291263,
|
| 336 |
-
"grad_norm":
|
| 337 |
"learning_rate": 3.099514563106796e-05,
|
| 338 |
-
"loss":
|
| 339 |
"step": 2350
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 11.650485436893204,
|
| 343 |
-
"grad_norm":
|
| 344 |
"learning_rate": 3.059061488673139e-05,
|
| 345 |
-
"loss":
|
| 346 |
"step": 2400
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 11.893203883495145,
|
| 350 |
-
"grad_norm":
|
| 351 |
"learning_rate": 3.0186084142394822e-05,
|
| 352 |
-
"loss":
|
| 353 |
"step": 2450
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 12.135922330097088,
|
| 357 |
-
"grad_norm":
|
| 358 |
"learning_rate": 2.9781553398058252e-05,
|
| 359 |
-
"loss":
|
| 360 |
"step": 2500
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 12.37864077669903,
|
| 364 |
-
"grad_norm":
|
| 365 |
"learning_rate": 2.9377022653721686e-05,
|
| 366 |
-
"loss":
|
| 367 |
"step": 2550
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 12.62135922330097,
|
| 371 |
-
"grad_norm": 9.
|
| 372 |
"learning_rate": 2.8972491909385112e-05,
|
| 373 |
-
"loss":
|
| 374 |
"step": 2600
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 12.864077669902912,
|
| 378 |
-
"grad_norm":
|
| 379 |
"learning_rate": 2.8567961165048546e-05,
|
| 380 |
-
"loss":
|
| 381 |
"step": 2650
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 13.106796116504855,
|
| 385 |
-
"grad_norm": 10.
|
| 386 |
"learning_rate": 2.816343042071198e-05,
|
| 387 |
-
"loss":
|
| 388 |
"step": 2700
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 13.349514563106796,
|
| 392 |
-
"grad_norm":
|
| 393 |
"learning_rate": 2.7758899676375405e-05,
|
| 394 |
-
"loss":
|
| 395 |
"step": 2750
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 13.592233009708737,
|
| 399 |
-
"grad_norm": 10.
|
| 400 |
"learning_rate": 2.735436893203884e-05,
|
| 401 |
-
"loss":
|
| 402 |
"step": 2800
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 13.83495145631068,
|
| 406 |
-
"grad_norm":
|
| 407 |
"learning_rate": 2.6949838187702265e-05,
|
| 408 |
-
"loss":
|
| 409 |
"step": 2850
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 14.077669902912621,
|
| 413 |
-
"grad_norm":
|
| 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":
|
| 423 |
"step": 2950
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 14.563106796116505,
|
| 427 |
-
"grad_norm": 10.
|
| 428 |
"learning_rate": 2.5736245954692555e-05,
|
| 429 |
-
"loss":
|
| 430 |
"step": 3000
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 14.805825242718447,
|
| 434 |
-
"grad_norm": 11.
|
| 435 |
"learning_rate": 2.533171521035599e-05,
|
| 436 |
-
"loss":
|
| 437 |
"step": 3050
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 15.048543689320388,
|
| 441 |
-
"grad_norm":
|
| 442 |
"learning_rate": 2.492718446601942e-05,
|
| 443 |
-
"loss":
|
| 444 |
"step": 3100
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 15.29126213592233,
|
| 448 |
-
"grad_norm":
|
| 449 |
"learning_rate": 2.452265372168285e-05,
|
| 450 |
-
"loss":
|
| 451 |
"step": 3150
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 15.533980582524272,
|
| 455 |
-
"grad_norm":
|
| 456 |
"learning_rate": 2.411812297734628e-05,
|
| 457 |
-
"loss":
|
| 458 |
"step": 3200
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 15.776699029126213,
|
| 462 |
-
"grad_norm":
|
| 463 |
"learning_rate": 2.3713592233009708e-05,
|
| 464 |
-
"loss":
|
| 465 |
"step": 3250
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 16.019417475728154,
|
| 469 |
-
"grad_norm":
|
| 470 |
"learning_rate": 2.3309061488673138e-05,
|
| 471 |
-
"loss":
|
| 472 |
"step": 3300
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 16.262135922330096,
|
| 476 |
-
"grad_norm":
|
| 477 |
"learning_rate": 2.290453074433657e-05,
|
| 478 |
-
"loss":
|
| 479 |
"step": 3350
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 16.50485436893204,
|
| 483 |
-
"grad_norm":
|
| 484 |
"learning_rate": 2.25e-05,
|
| 485 |
-
"loss":
|
| 486 |
"step": 3400
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 16.74757281553398,
|
| 490 |
-
"grad_norm":
|
| 491 |
"learning_rate": 2.209546925566343e-05,
|
| 492 |
-
"loss":
|
| 493 |
"step": 3450
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 16.990291262135923,
|
| 497 |
-
"grad_norm":
|
| 498 |
"learning_rate": 2.169093851132686e-05,
|
| 499 |
-
"loss":
|
| 500 |
"step": 3500
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 17.233009708737864,
|
| 504 |
-
"grad_norm":
|
| 505 |
"learning_rate": 2.1286407766990295e-05,
|
| 506 |
-
"loss":
|
| 507 |
"step": 3550
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 17.475728155339805,
|
| 511 |
-
"grad_norm":
|
| 512 |
"learning_rate": 2.0881877022653725e-05,
|
| 513 |
-
"loss":
|
| 514 |
"step": 3600
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 17.718446601941746,
|
| 518 |
-
"grad_norm":
|
| 519 |
"learning_rate": 2.047734627831715e-05,
|
| 520 |
-
"loss":
|
| 521 |
"step": 3650
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 17.96116504854369,
|
| 525 |
-
"grad_norm":
|
| 526 |
"learning_rate": 2.007281553398058e-05,
|
| 527 |
-
"loss":
|
| 528 |
"step": 3700
|
| 529 |
},
|
| 530 |
{
|
| 531 |
"epoch": 18.203883495145632,
|
| 532 |
-
"grad_norm":
|
| 533 |
"learning_rate": 1.9668284789644014e-05,
|
| 534 |
-
"loss":
|
| 535 |
"step": 3750
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"epoch": 18.446601941747574,
|
| 539 |
-
"grad_norm":
|
| 540 |
"learning_rate": 1.9263754045307444e-05,
|
| 541 |
-
"loss":
|
| 542 |
"step": 3800
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 18.689320388349515,
|
| 546 |
-
"grad_norm":
|
| 547 |
"learning_rate": 1.8859223300970874e-05,
|
| 548 |
-
"loss":
|
| 549 |
"step": 3850
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 18.932038834951456,
|
| 553 |
-
"grad_norm":
|
| 554 |
"learning_rate": 1.8454692556634304e-05,
|
| 555 |
-
"loss":
|
| 556 |
"step": 3900
|
| 557 |
},
|
| 558 |
{
|
| 559 |
"epoch": 19.174757281553397,
|
| 560 |
-
"grad_norm":
|
| 561 |
"learning_rate": 1.8050161812297738e-05,
|
| 562 |
-
"loss":
|
| 563 |
"step": 3950
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 19.41747572815534,
|
| 567 |
-
"grad_norm":
|
| 568 |
"learning_rate": 1.7645631067961167e-05,
|
| 569 |
-
"loss":
|
| 570 |
"step": 4000
|
| 571 |
}
|
| 572 |
],
|
|
@@ -587,7 +587,7 @@
|
|
| 587 |
"attributes": {}
|
| 588 |
}
|
| 589 |
},
|
| 590 |
-
"total_flos":
|
| 591 |
"train_batch_size": 4,
|
| 592 |
"trial_name": null,
|
| 593 |
"trial_params": null
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.24271844660194175,
|
| 14 |
+
"grad_norm": 2.8362834453582764,
|
| 15 |
"learning_rate": 4.960355987055016e-05,
|
| 16 |
+
"loss": 6.0742,
|
| 17 |
"step": 50
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4854368932038835,
|
| 21 |
+
"grad_norm": 2.9484496116638184,
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
+
"loss": 5.7218,
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
+
"grad_norm": 3.542572259902954,
|
| 29 |
"learning_rate": 4.879449838187702e-05,
|
| 30 |
+
"loss": 5.4227,
|
| 31 |
"step": 150
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.970873786407767,
|
| 35 |
+
"grad_norm": 4.634098052978516,
|
| 36 |
"learning_rate": 4.8389967637540455e-05,
|
| 37 |
+
"loss": 5.1655,
|
| 38 |
"step": 200
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.2135922330097086,
|
| 42 |
+
"grad_norm": 5.699330806732178,
|
| 43 |
"learning_rate": 4.798543689320388e-05,
|
| 44 |
+
"loss": 4.8009,
|
| 45 |
"step": 250
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 1.4563106796116505,
|
| 49 |
+
"grad_norm": 5.711933135986328,
|
| 50 |
"learning_rate": 4.7580906148867315e-05,
|
| 51 |
+
"loss": 4.607,
|
| 52 |
"step": 300
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 1.6990291262135924,
|
| 56 |
+
"grad_norm": 5.113691806793213,
|
| 57 |
"learning_rate": 4.717637540453075e-05,
|
| 58 |
+
"loss": 4.462,
|
| 59 |
"step": 350
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 1.941747572815534,
|
| 63 |
+
"grad_norm": 5.632521152496338,
|
| 64 |
"learning_rate": 4.6771844660194174e-05,
|
| 65 |
+
"loss": 4.3695,
|
| 66 |
"step": 400
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 2.1844660194174756,
|
| 70 |
+
"grad_norm": 5.428906440734863,
|
| 71 |
"learning_rate": 4.636731391585761e-05,
|
| 72 |
+
"loss": 4.1549,
|
| 73 |
"step": 450
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 2.4271844660194173,
|
| 77 |
+
"grad_norm": 5.037013530731201,
|
| 78 |
"learning_rate": 4.596278317152104e-05,
|
| 79 |
+
"loss": 4.0921,
|
| 80 |
"step": 500
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 2.6699029126213594,
|
| 84 |
+
"grad_norm": 5.231846809387207,
|
| 85 |
"learning_rate": 4.555825242718447e-05,
|
| 86 |
+
"loss": 4.0662,
|
| 87 |
"step": 550
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 2.912621359223301,
|
| 91 |
+
"grad_norm": 5.506278038024902,
|
| 92 |
"learning_rate": 4.51537216828479e-05,
|
| 93 |
+
"loss": 4.0403,
|
| 94 |
"step": 600
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 3.1553398058252426,
|
| 98 |
+
"grad_norm": 5.410216331481934,
|
| 99 |
"learning_rate": 4.4749190938511334e-05,
|
| 100 |
+
"loss": 3.8375,
|
| 101 |
"step": 650
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 3.3980582524271843,
|
| 105 |
+
"grad_norm": 5.637699127197266,
|
| 106 |
"learning_rate": 4.434466019417476e-05,
|
| 107 |
+
"loss": 3.7473,
|
| 108 |
"step": 700
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 3.6407766990291264,
|
| 112 |
+
"grad_norm": 5.575273513793945,
|
| 113 |
"learning_rate": 4.3940129449838194e-05,
|
| 114 |
+
"loss": 3.7835,
|
| 115 |
"step": 750
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 3.883495145631068,
|
| 119 |
+
"grad_norm": 5.737198352813721,
|
| 120 |
"learning_rate": 4.353559870550162e-05,
|
| 121 |
+
"loss": 3.7919,
|
| 122 |
"step": 800
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 4.12621359223301,
|
| 126 |
+
"grad_norm": 5.645532608032227,
|
| 127 |
"learning_rate": 4.313106796116505e-05,
|
| 128 |
+
"loss": 3.6425,
|
| 129 |
"step": 850
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 4.368932038834951,
|
| 133 |
+
"grad_norm": 6.308927059173584,
|
| 134 |
"learning_rate": 4.272653721682848e-05,
|
| 135 |
+
"loss": 3.4785,
|
| 136 |
"step": 900
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 4.611650485436893,
|
| 140 |
+
"grad_norm": 5.863094329833984,
|
| 141 |
"learning_rate": 4.232200647249191e-05,
|
| 142 |
+
"loss": 3.5268,
|
| 143 |
"step": 950
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 4.854368932038835,
|
| 147 |
+
"grad_norm": 6.120258331298828,
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
+
"loss": 3.5296,
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
+
"grad_norm": 6.231515884399414,
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
+
"loss": 3.3802,
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
+
"grad_norm": 6.211188316345215,
|
| 162 |
"learning_rate": 4.11084142394822e-05,
|
| 163 |
+
"loss": 3.2148,
|
| 164 |
"step": 1100
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 5.58252427184466,
|
| 168 |
+
"grad_norm": 6.8676934242248535,
|
| 169 |
"learning_rate": 4.0703883495145634e-05,
|
| 170 |
+
"loss": 3.2111,
|
| 171 |
"step": 1150
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 5.825242718446602,
|
| 175 |
+
"grad_norm": 6.851133346557617,
|
| 176 |
"learning_rate": 4.029935275080906e-05,
|
| 177 |
+
"loss": 3.2771,
|
| 178 |
"step": 1200
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 6.067961165048544,
|
| 182 |
+
"grad_norm": 6.708765506744385,
|
| 183 |
"learning_rate": 3.9894822006472494e-05,
|
| 184 |
+
"loss": 3.1587,
|
| 185 |
"step": 1250
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 6.310679611650485,
|
| 189 |
+
"grad_norm": 7.136553764343262,
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
+
"loss": 2.9192,
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
+
"grad_norm": 7.370823383331299,
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
+
"loss": 2.922,
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
+
"grad_norm": 7.630361557006836,
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
+
"loss": 2.9417,
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
+
"grad_norm": 7.765511989593506,
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
+
"loss": 2.8669,
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
+
"grad_norm": 7.85439920425415,
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
+
"loss": 2.5412,
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 7.524271844660194,
|
| 224 |
+
"grad_norm": 8.064071655273438,
|
| 225 |
"learning_rate": 3.746763754045307e-05,
|
| 226 |
+
"loss": 2.6131,
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 7.766990291262136,
|
| 231 |
+
"grad_norm": 8.692522048950195,
|
| 232 |
"learning_rate": 3.7063106796116507e-05,
|
| 233 |
+
"loss": 2.6356,
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 8.009708737864077,
|
| 238 |
+
"grad_norm": 8.012410163879395,
|
| 239 |
"learning_rate": 3.665857605177993e-05,
|
| 240 |
+
"loss": 2.599,
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 8.25242718446602,
|
| 245 |
+
"grad_norm": 8.6799898147583,
|
| 246 |
"learning_rate": 3.6254045307443366e-05,
|
| 247 |
+
"loss": 2.1992,
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 8.495145631067961,
|
| 252 |
+
"grad_norm": 8.963507652282715,
|
| 253 |
"learning_rate": 3.584951456310679e-05,
|
| 254 |
+
"loss": 2.2701,
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 8.737864077669903,
|
| 259 |
+
"grad_norm": 9.833822250366211,
|
| 260 |
"learning_rate": 3.5444983818770226e-05,
|
| 261 |
+
"loss": 2.2837,
|
| 262 |
"step": 1800
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 8.980582524271846,
|
| 266 |
+
"grad_norm": 9.47242259979248,
|
| 267 |
"learning_rate": 3.504045307443366e-05,
|
| 268 |
+
"loss": 2.2854,
|
| 269 |
"step": 1850
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 9.223300970873787,
|
| 273 |
+
"grad_norm": 8.856402397155762,
|
| 274 |
"learning_rate": 3.4635922330097086e-05,
|
| 275 |
+
"loss": 1.8862,
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 9.466019417475728,
|
| 280 |
+
"grad_norm": 9.948341369628906,
|
| 281 |
"learning_rate": 3.423139158576052e-05,
|
| 282 |
+
"loss": 1.922,
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 9.70873786407767,
|
| 287 |
+
"grad_norm": 10.28939151763916,
|
| 288 |
"learning_rate": 3.382686084142395e-05,
|
| 289 |
+
"loss": 1.9622,
|
| 290 |
"step": 2000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 9.951456310679612,
|
| 294 |
+
"grad_norm": 10.114766120910645,
|
| 295 |
"learning_rate": 3.342233009708738e-05,
|
| 296 |
+
"loss": 1.9741,
|
| 297 |
"step": 2050
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 10.194174757281553,
|
| 301 |
+
"grad_norm": 9.489270210266113,
|
| 302 |
"learning_rate": 3.301779935275081e-05,
|
| 303 |
+
"loss": 1.6509,
|
| 304 |
"step": 2100
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 10.436893203883495,
|
| 308 |
+
"grad_norm": 10.16482162475586,
|
| 309 |
"learning_rate": 3.2613268608414246e-05,
|
| 310 |
+
"loss": 1.5883,
|
| 311 |
"step": 2150
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 10.679611650485437,
|
| 315 |
+
"grad_norm": 11.778059005737305,
|
| 316 |
"learning_rate": 3.220873786407767e-05,
|
| 317 |
+
"loss": 1.6234,
|
| 318 |
"step": 2200
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 10.922330097087379,
|
| 322 |
+
"grad_norm": 10.798454284667969,
|
| 323 |
"learning_rate": 3.1804207119741106e-05,
|
| 324 |
+
"loss": 1.667,
|
| 325 |
"step": 2250
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 11.16504854368932,
|
| 329 |
+
"grad_norm": 9.751585006713867,
|
| 330 |
"learning_rate": 3.139967637540453e-05,
|
| 331 |
+
"loss": 1.3785,
|
| 332 |
"step": 2300
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 11.407766990291263,
|
| 336 |
+
"grad_norm": 10.266364097595215,
|
| 337 |
"learning_rate": 3.099514563106796e-05,
|
| 338 |
+
"loss": 1.2956,
|
| 339 |
"step": 2350
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 11.650485436893204,
|
| 343 |
+
"grad_norm": 10.660856246948242,
|
| 344 |
"learning_rate": 3.059061488673139e-05,
|
| 345 |
+
"loss": 1.3321,
|
| 346 |
"step": 2400
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 11.893203883495145,
|
| 350 |
+
"grad_norm": 10.896187782287598,
|
| 351 |
"learning_rate": 3.0186084142394822e-05,
|
| 352 |
+
"loss": 1.3533,
|
| 353 |
"step": 2450
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 12.135922330097088,
|
| 357 |
+
"grad_norm": 10.952502250671387,
|
| 358 |
"learning_rate": 2.9781553398058252e-05,
|
| 359 |
+
"loss": 1.1486,
|
| 360 |
"step": 2500
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 12.37864077669903,
|
| 364 |
+
"grad_norm": 10.766463279724121,
|
| 365 |
"learning_rate": 2.9377022653721686e-05,
|
| 366 |
+
"loss": 1.0227,
|
| 367 |
"step": 2550
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 12.62135922330097,
|
| 371 |
+
"grad_norm": 9.385764122009277,
|
| 372 |
"learning_rate": 2.8972491909385112e-05,
|
| 373 |
+
"loss": 1.0651,
|
| 374 |
"step": 2600
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 12.864077669902912,
|
| 378 |
+
"grad_norm": 11.573925018310547,
|
| 379 |
"learning_rate": 2.8567961165048546e-05,
|
| 380 |
+
"loss": 1.1013,
|
| 381 |
"step": 2650
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 13.106796116504855,
|
| 385 |
+
"grad_norm": 10.138091087341309,
|
| 386 |
"learning_rate": 2.816343042071198e-05,
|
| 387 |
+
"loss": 0.9545,
|
| 388 |
"step": 2700
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 13.349514563106796,
|
| 392 |
+
"grad_norm": 9.867119789123535,
|
| 393 |
"learning_rate": 2.7758899676375405e-05,
|
| 394 |
+
"loss": 0.7892,
|
| 395 |
"step": 2750
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 13.592233009708737,
|
| 399 |
+
"grad_norm": 10.443971633911133,
|
| 400 |
"learning_rate": 2.735436893203884e-05,
|
| 401 |
+
"loss": 0.8432,
|
| 402 |
"step": 2800
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 13.83495145631068,
|
| 406 |
+
"grad_norm": 10.633298873901367,
|
| 407 |
"learning_rate": 2.6949838187702265e-05,
|
| 408 |
+
"loss": 0.8543,
|
| 409 |
"step": 2850
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 14.077669902912621,
|
| 413 |
+
"grad_norm": 8.640830039978027,
|
| 414 |
"learning_rate": 2.6545307443365695e-05,
|
| 415 |
+
"loss": 0.7751,
|
| 416 |
"step": 2900
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 14.320388349514563,
|
| 420 |
+
"grad_norm": 10.461247444152832,
|
| 421 |
"learning_rate": 2.614077669902913e-05,
|
| 422 |
+
"loss": 0.6074,
|
| 423 |
"step": 2950
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 14.563106796116505,
|
| 427 |
+
"grad_norm": 10.757479667663574,
|
| 428 |
"learning_rate": 2.5736245954692555e-05,
|
| 429 |
+
"loss": 0.6354,
|
| 430 |
"step": 3000
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 14.805825242718447,
|
| 434 |
+
"grad_norm": 11.599132537841797,
|
| 435 |
"learning_rate": 2.533171521035599e-05,
|
| 436 |
+
"loss": 0.6694,
|
| 437 |
"step": 3050
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 15.048543689320388,
|
| 441 |
+
"grad_norm": 8.085358619689941,
|
| 442 |
"learning_rate": 2.492718446601942e-05,
|
| 443 |
+
"loss": 0.6306,
|
| 444 |
"step": 3100
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 15.29126213592233,
|
| 448 |
+
"grad_norm": 8.327988624572754,
|
| 449 |
"learning_rate": 2.452265372168285e-05,
|
| 450 |
+
"loss": 0.4628,
|
| 451 |
"step": 3150
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 15.533980582524272,
|
| 455 |
+
"grad_norm": 8.545391082763672,
|
| 456 |
"learning_rate": 2.411812297734628e-05,
|
| 457 |
+
"loss": 0.4915,
|
| 458 |
"step": 3200
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 15.776699029126213,
|
| 462 |
+
"grad_norm": 10.176375389099121,
|
| 463 |
"learning_rate": 2.3713592233009708e-05,
|
| 464 |
+
"loss": 0.5105,
|
| 465 |
"step": 3250
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 16.019417475728154,
|
| 469 |
+
"grad_norm": 7.821159362792969,
|
| 470 |
"learning_rate": 2.3309061488673138e-05,
|
| 471 |
+
"loss": 0.5055,
|
| 472 |
"step": 3300
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 16.262135922330096,
|
| 476 |
+
"grad_norm": 7.797800064086914,
|
| 477 |
"learning_rate": 2.290453074433657e-05,
|
| 478 |
+
"loss": 0.3486,
|
| 479 |
"step": 3350
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 16.50485436893204,
|
| 483 |
+
"grad_norm": 8.497370719909668,
|
| 484 |
"learning_rate": 2.25e-05,
|
| 485 |
+
"loss": 0.3627,
|
| 486 |
"step": 3400
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 16.74757281553398,
|
| 490 |
+
"grad_norm": 9.842249870300293,
|
| 491 |
"learning_rate": 2.209546925566343e-05,
|
| 492 |
+
"loss": 0.3969,
|
| 493 |
"step": 3450
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 16.990291262135923,
|
| 497 |
+
"grad_norm": 10.705739974975586,
|
| 498 |
"learning_rate": 2.169093851132686e-05,
|
| 499 |
+
"loss": 0.4028,
|
| 500 |
"step": 3500
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 17.233009708737864,
|
| 504 |
+
"grad_norm": 7.950681686401367,
|
| 505 |
"learning_rate": 2.1286407766990295e-05,
|
| 506 |
+
"loss": 0.284,
|
| 507 |
"step": 3550
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 17.475728155339805,
|
| 511 |
+
"grad_norm": 8.526535034179688,
|
| 512 |
"learning_rate": 2.0881877022653725e-05,
|
| 513 |
+
"loss": 0.2934,
|
| 514 |
"step": 3600
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 17.718446601941746,
|
| 518 |
+
"grad_norm": 8.999290466308594,
|
| 519 |
"learning_rate": 2.047734627831715e-05,
|
| 520 |
+
"loss": 0.2964,
|
| 521 |
"step": 3650
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 17.96116504854369,
|
| 525 |
+
"grad_norm": 9.228137969970703,
|
| 526 |
"learning_rate": 2.007281553398058e-05,
|
| 527 |
+
"loss": 0.311,
|
| 528 |
"step": 3700
|
| 529 |
},
|
| 530 |
{
|
| 531 |
"epoch": 18.203883495145632,
|
| 532 |
+
"grad_norm": 6.715802192687988,
|
| 533 |
"learning_rate": 1.9668284789644014e-05,
|
| 534 |
+
"loss": 0.2263,
|
| 535 |
"step": 3750
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"epoch": 18.446601941747574,
|
| 539 |
+
"grad_norm": 7.834264278411865,
|
| 540 |
"learning_rate": 1.9263754045307444e-05,
|
| 541 |
+
"loss": 0.2293,
|
| 542 |
"step": 3800
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 18.689320388349515,
|
| 546 |
+
"grad_norm": 6.83228874206543,
|
| 547 |
"learning_rate": 1.8859223300970874e-05,
|
| 548 |
+
"loss": 0.2392,
|
| 549 |
"step": 3850
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 18.932038834951456,
|
| 553 |
+
"grad_norm": 8.507209777832031,
|
| 554 |
"learning_rate": 1.8454692556634304e-05,
|
| 555 |
+
"loss": 0.2377,
|
| 556 |
"step": 3900
|
| 557 |
},
|
| 558 |
{
|
| 559 |
"epoch": 19.174757281553397,
|
| 560 |
+
"grad_norm": 6.565855026245117,
|
| 561 |
"learning_rate": 1.8050161812297738e-05,
|
| 562 |
+
"loss": 0.1905,
|
| 563 |
"step": 3950
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 19.41747572815534,
|
| 567 |
+
"grad_norm": 7.470388889312744,
|
| 568 |
"learning_rate": 1.7645631067961167e-05,
|
| 569 |
+
"loss": 0.1846,
|
| 570 |
"step": 4000
|
| 571 |
}
|
| 572 |
],
|
|
|
|
| 587 |
"attributes": {}
|
| 588 |
}
|
| 589 |
},
|
| 590 |
+
"total_flos": 1857413971968000.0,
|
| 591 |
"train_batch_size": 4,
|
| 592 |
"trial_name": null,
|
| 593 |
"trial_params": null
|
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:d5c04ce53d557cd762ff77607aaadbd1248aaa177697662d9a21b90d444993b7
|
| 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":
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
-
"n_layer":
|
| 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": 2048,
|
| 14 |
+
"n_embd": 1024,
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
+
"n_layer": 12,
|
| 18 |
+
"n_positions": 2048,
|
| 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:ebb0541da09e5cfb609d483adeef00bc2ad0501c3ae4444273ec13e5ec19556b
|
| 3 |
+
size 617130824
|
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:dc325834365b49ed70544206aee634b00dc3bc67405ea69b0c181cd595db3143
|
| 3 |
+
size 1234355130
|
checkpoints/checkpoint-4500/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6ba93584c0da2ef2f3f0c556dbddf6f03b4d82613be3aa5b8fb3c23d16b4d966
|
| 3 |
size 14244
|
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":
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
-
"loss": 5.
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
-
"grad_norm":
|
| 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": 6.
|
| 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":
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
-
"loss": 3.
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
-
"grad_norm":
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
-
"loss": 3.
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
-
"grad_norm":
|
| 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":
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
-
"loss":
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
-
"grad_norm":
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
-
"loss":
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
-
"grad_norm":
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
-
"loss":
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
-
"grad_norm":
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
-
"loss":
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
-
"grad_norm": 7.
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
-
"loss":
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 7.524271844660194,
|
| 224 |
-
"grad_norm":
|
| 225 |
"learning_rate": 3.746763754045307e-05,
|
| 226 |
-
"loss":
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 7.766990291262136,
|
| 231 |
-
"grad_norm":
|
| 232 |
"learning_rate": 3.7063106796116507e-05,
|
| 233 |
-
"loss":
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 8.009708737864077,
|
| 238 |
-
"grad_norm":
|
| 239 |
"learning_rate": 3.665857605177993e-05,
|
| 240 |
-
"loss":
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 8.25242718446602,
|
| 245 |
-
"grad_norm":
|
| 246 |
"learning_rate": 3.6254045307443366e-05,
|
| 247 |
-
"loss": 2.
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 8.495145631067961,
|
| 252 |
-
"grad_norm":
|
| 253 |
"learning_rate": 3.584951456310679e-05,
|
| 254 |
-
"loss": 2.
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 8.737864077669903,
|
| 259 |
-
"grad_norm":
|
| 260 |
"learning_rate": 3.5444983818770226e-05,
|
| 261 |
-
"loss":
|
| 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":
|
| 274 |
"learning_rate": 3.4635922330097086e-05,
|
| 275 |
-
"loss":
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 9.466019417475728,
|
| 280 |
-
"grad_norm":
|
| 281 |
"learning_rate": 3.423139158576052e-05,
|
| 282 |
-
"loss":
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 9.70873786407767,
|
| 287 |
-
"grad_norm":
|
| 288 |
"learning_rate": 3.382686084142395e-05,
|
| 289 |
-
"loss":
|
| 290 |
"step": 2000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 9.951456310679612,
|
| 294 |
-
"grad_norm":
|
| 295 |
"learning_rate": 3.342233009708738e-05,
|
| 296 |
-
"loss":
|
| 297 |
"step": 2050
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 10.194174757281553,
|
| 301 |
-
"grad_norm":
|
| 302 |
"learning_rate": 3.301779935275081e-05,
|
| 303 |
-
"loss":
|
| 304 |
"step": 2100
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 10.436893203883495,
|
| 308 |
-
"grad_norm":
|
| 309 |
"learning_rate": 3.2613268608414246e-05,
|
| 310 |
-
"loss":
|
| 311 |
"step": 2150
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 10.679611650485437,
|
| 315 |
-
"grad_norm":
|
| 316 |
"learning_rate": 3.220873786407767e-05,
|
| 317 |
-
"loss":
|
| 318 |
"step": 2200
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 10.922330097087379,
|
| 322 |
-
"grad_norm":
|
| 323 |
"learning_rate": 3.1804207119741106e-05,
|
| 324 |
-
"loss":
|
| 325 |
"step": 2250
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 11.16504854368932,
|
| 329 |
-
"grad_norm":
|
| 330 |
"learning_rate": 3.139967637540453e-05,
|
| 331 |
-
"loss":
|
| 332 |
"step": 2300
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 11.407766990291263,
|
| 336 |
-
"grad_norm":
|
| 337 |
"learning_rate": 3.099514563106796e-05,
|
| 338 |
-
"loss":
|
| 339 |
"step": 2350
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 11.650485436893204,
|
| 343 |
-
"grad_norm":
|
| 344 |
"learning_rate": 3.059061488673139e-05,
|
| 345 |
-
"loss":
|
| 346 |
"step": 2400
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 11.893203883495145,
|
| 350 |
-
"grad_norm":
|
| 351 |
"learning_rate": 3.0186084142394822e-05,
|
| 352 |
-
"loss":
|
| 353 |
"step": 2450
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 12.135922330097088,
|
| 357 |
-
"grad_norm":
|
| 358 |
"learning_rate": 2.9781553398058252e-05,
|
| 359 |
-
"loss":
|
| 360 |
"step": 2500
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 12.37864077669903,
|
| 364 |
-
"grad_norm":
|
| 365 |
"learning_rate": 2.9377022653721686e-05,
|
| 366 |
-
"loss":
|
| 367 |
"step": 2550
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 12.62135922330097,
|
| 371 |
-
"grad_norm": 9.
|
| 372 |
"learning_rate": 2.8972491909385112e-05,
|
| 373 |
-
"loss":
|
| 374 |
"step": 2600
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 12.864077669902912,
|
| 378 |
-
"grad_norm":
|
| 379 |
"learning_rate": 2.8567961165048546e-05,
|
| 380 |
-
"loss":
|
| 381 |
"step": 2650
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 13.106796116504855,
|
| 385 |
-
"grad_norm": 10.
|
| 386 |
"learning_rate": 2.816343042071198e-05,
|
| 387 |
-
"loss":
|
| 388 |
"step": 2700
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 13.349514563106796,
|
| 392 |
-
"grad_norm":
|
| 393 |
"learning_rate": 2.7758899676375405e-05,
|
| 394 |
-
"loss":
|
| 395 |
"step": 2750
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 13.592233009708737,
|
| 399 |
-
"grad_norm": 10.
|
| 400 |
"learning_rate": 2.735436893203884e-05,
|
| 401 |
-
"loss":
|
| 402 |
"step": 2800
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 13.83495145631068,
|
| 406 |
-
"grad_norm":
|
| 407 |
"learning_rate": 2.6949838187702265e-05,
|
| 408 |
-
"loss":
|
| 409 |
"step": 2850
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 14.077669902912621,
|
| 413 |
-
"grad_norm":
|
| 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":
|
| 423 |
"step": 2950
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 14.563106796116505,
|
| 427 |
-
"grad_norm": 10.
|
| 428 |
"learning_rate": 2.5736245954692555e-05,
|
| 429 |
-
"loss":
|
| 430 |
"step": 3000
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 14.805825242718447,
|
| 434 |
-
"grad_norm": 11.
|
| 435 |
"learning_rate": 2.533171521035599e-05,
|
| 436 |
-
"loss":
|
| 437 |
"step": 3050
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 15.048543689320388,
|
| 441 |
-
"grad_norm":
|
| 442 |
"learning_rate": 2.492718446601942e-05,
|
| 443 |
-
"loss":
|
| 444 |
"step": 3100
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 15.29126213592233,
|
| 448 |
-
"grad_norm":
|
| 449 |
"learning_rate": 2.452265372168285e-05,
|
| 450 |
-
"loss":
|
| 451 |
"step": 3150
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 15.533980582524272,
|
| 455 |
-
"grad_norm":
|
| 456 |
"learning_rate": 2.411812297734628e-05,
|
| 457 |
-
"loss":
|
| 458 |
"step": 3200
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 15.776699029126213,
|
| 462 |
-
"grad_norm":
|
| 463 |
"learning_rate": 2.3713592233009708e-05,
|
| 464 |
-
"loss":
|
| 465 |
"step": 3250
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 16.019417475728154,
|
| 469 |
-
"grad_norm":
|
| 470 |
"learning_rate": 2.3309061488673138e-05,
|
| 471 |
-
"loss":
|
| 472 |
"step": 3300
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 16.262135922330096,
|
| 476 |
-
"grad_norm":
|
| 477 |
"learning_rate": 2.290453074433657e-05,
|
| 478 |
-
"loss":
|
| 479 |
"step": 3350
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 16.50485436893204,
|
| 483 |
-
"grad_norm":
|
| 484 |
"learning_rate": 2.25e-05,
|
| 485 |
-
"loss":
|
| 486 |
"step": 3400
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 16.74757281553398,
|
| 490 |
-
"grad_norm":
|
| 491 |
"learning_rate": 2.209546925566343e-05,
|
| 492 |
-
"loss":
|
| 493 |
"step": 3450
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 16.990291262135923,
|
| 497 |
-
"grad_norm":
|
| 498 |
"learning_rate": 2.169093851132686e-05,
|
| 499 |
-
"loss":
|
| 500 |
"step": 3500
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 17.233009708737864,
|
| 504 |
-
"grad_norm":
|
| 505 |
"learning_rate": 2.1286407766990295e-05,
|
| 506 |
-
"loss":
|
| 507 |
"step": 3550
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 17.475728155339805,
|
| 511 |
-
"grad_norm":
|
| 512 |
"learning_rate": 2.0881877022653725e-05,
|
| 513 |
-
"loss":
|
| 514 |
"step": 3600
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 17.718446601941746,
|
| 518 |
-
"grad_norm":
|
| 519 |
"learning_rate": 2.047734627831715e-05,
|
| 520 |
-
"loss":
|
| 521 |
"step": 3650
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 17.96116504854369,
|
| 525 |
-
"grad_norm":
|
| 526 |
"learning_rate": 2.007281553398058e-05,
|
| 527 |
-
"loss":
|
| 528 |
"step": 3700
|
| 529 |
},
|
| 530 |
{
|
| 531 |
"epoch": 18.203883495145632,
|
| 532 |
-
"grad_norm":
|
| 533 |
"learning_rate": 1.9668284789644014e-05,
|
| 534 |
-
"loss":
|
| 535 |
"step": 3750
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"epoch": 18.446601941747574,
|
| 539 |
-
"grad_norm":
|
| 540 |
"learning_rate": 1.9263754045307444e-05,
|
| 541 |
-
"loss":
|
| 542 |
"step": 3800
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 18.689320388349515,
|
| 546 |
-
"grad_norm":
|
| 547 |
"learning_rate": 1.8859223300970874e-05,
|
| 548 |
-
"loss":
|
| 549 |
"step": 3850
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 18.932038834951456,
|
| 553 |
-
"grad_norm":
|
| 554 |
"learning_rate": 1.8454692556634304e-05,
|
| 555 |
-
"loss":
|
| 556 |
"step": 3900
|
| 557 |
},
|
| 558 |
{
|
| 559 |
"epoch": 19.174757281553397,
|
| 560 |
-
"grad_norm":
|
| 561 |
"learning_rate": 1.8050161812297738e-05,
|
| 562 |
-
"loss":
|
| 563 |
"step": 3950
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 19.41747572815534,
|
| 567 |
-
"grad_norm":
|
| 568 |
"learning_rate": 1.7645631067961167e-05,
|
| 569 |
-
"loss":
|
| 570 |
"step": 4000
|
| 571 |
},
|
| 572 |
{
|
| 573 |
"epoch": 19.660194174757283,
|
| 574 |
-
"grad_norm":
|
| 575 |
"learning_rate": 1.7241100323624594e-05,
|
| 576 |
-
"loss":
|
| 577 |
"step": 4050
|
| 578 |
},
|
| 579 |
{
|
| 580 |
"epoch": 19.902912621359224,
|
| 581 |
-
"grad_norm":
|
| 582 |
"learning_rate": 1.6836569579288027e-05,
|
| 583 |
-
"loss":
|
| 584 |
"step": 4100
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 20.145631067961165,
|
| 588 |
-
"grad_norm":
|
| 589 |
"learning_rate": 1.6432038834951457e-05,
|
| 590 |
-
"loss":
|
| 591 |
"step": 4150
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 20.388349514563107,
|
| 595 |
-
"grad_norm":
|
| 596 |
"learning_rate": 1.6027508090614887e-05,
|
| 597 |
-
"loss":
|
| 598 |
"step": 4200
|
| 599 |
},
|
| 600 |
{
|
| 601 |
"epoch": 20.631067961165048,
|
| 602 |
-
"grad_norm":
|
| 603 |
"learning_rate": 1.5622977346278317e-05,
|
| 604 |
-
"loss":
|
| 605 |
"step": 4250
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"epoch": 20.87378640776699,
|
| 609 |
-
"grad_norm":
|
| 610 |
"learning_rate": 1.5218446601941749e-05,
|
| 611 |
-
"loss":
|
| 612 |
"step": 4300
|
| 613 |
},
|
| 614 |
{
|
| 615 |
"epoch": 21.116504854368934,
|
| 616 |
-
"grad_norm":
|
| 617 |
"learning_rate": 1.4813915857605179e-05,
|
| 618 |
-
"loss":
|
| 619 |
"step": 4350
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"epoch": 21.359223300970875,
|
| 623 |
-
"grad_norm":
|
| 624 |
"learning_rate": 1.4409385113268609e-05,
|
| 625 |
-
"loss":
|
| 626 |
"step": 4400
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 21.601941747572816,
|
| 630 |
-
"grad_norm":
|
| 631 |
"learning_rate": 1.4004854368932039e-05,
|
| 632 |
-
"loss":
|
| 633 |
"step": 4450
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 21.844660194174757,
|
| 637 |
-
"grad_norm":
|
| 638 |
"learning_rate": 1.360032362459547e-05,
|
| 639 |
-
"loss":
|
| 640 |
"step": 4500
|
| 641 |
}
|
| 642 |
],
|
|
@@ -657,7 +657,7 @@
|
|
| 657 |
"attributes": {}
|
| 658 |
}
|
| 659 |
},
|
| 660 |
-
"total_flos":
|
| 661 |
"train_batch_size": 4,
|
| 662 |
"trial_name": null,
|
| 663 |
"trial_params": null
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.24271844660194175,
|
| 14 |
+
"grad_norm": 2.8362834453582764,
|
| 15 |
"learning_rate": 4.960355987055016e-05,
|
| 16 |
+
"loss": 6.0742,
|
| 17 |
"step": 50
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4854368932038835,
|
| 21 |
+
"grad_norm": 2.9484496116638184,
|
| 22 |
"learning_rate": 4.9199029126213595e-05,
|
| 23 |
+
"loss": 5.7218,
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.7281553398058253,
|
| 28 |
+
"grad_norm": 3.542572259902954,
|
| 29 |
"learning_rate": 4.879449838187702e-05,
|
| 30 |
+
"loss": 5.4227,
|
| 31 |
"step": 150
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.970873786407767,
|
| 35 |
+
"grad_norm": 4.634098052978516,
|
| 36 |
"learning_rate": 4.8389967637540455e-05,
|
| 37 |
+
"loss": 5.1655,
|
| 38 |
"step": 200
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.2135922330097086,
|
| 42 |
+
"grad_norm": 5.699330806732178,
|
| 43 |
"learning_rate": 4.798543689320388e-05,
|
| 44 |
+
"loss": 4.8009,
|
| 45 |
"step": 250
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 1.4563106796116505,
|
| 49 |
+
"grad_norm": 5.711933135986328,
|
| 50 |
"learning_rate": 4.7580906148867315e-05,
|
| 51 |
+
"loss": 4.607,
|
| 52 |
"step": 300
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 1.6990291262135924,
|
| 56 |
+
"grad_norm": 5.113691806793213,
|
| 57 |
"learning_rate": 4.717637540453075e-05,
|
| 58 |
+
"loss": 4.462,
|
| 59 |
"step": 350
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 1.941747572815534,
|
| 63 |
+
"grad_norm": 5.632521152496338,
|
| 64 |
"learning_rate": 4.6771844660194174e-05,
|
| 65 |
+
"loss": 4.3695,
|
| 66 |
"step": 400
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 2.1844660194174756,
|
| 70 |
+
"grad_norm": 5.428906440734863,
|
| 71 |
"learning_rate": 4.636731391585761e-05,
|
| 72 |
+
"loss": 4.1549,
|
| 73 |
"step": 450
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 2.4271844660194173,
|
| 77 |
+
"grad_norm": 5.037013530731201,
|
| 78 |
"learning_rate": 4.596278317152104e-05,
|
| 79 |
+
"loss": 4.0921,
|
| 80 |
"step": 500
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 2.6699029126213594,
|
| 84 |
+
"grad_norm": 5.231846809387207,
|
| 85 |
"learning_rate": 4.555825242718447e-05,
|
| 86 |
+
"loss": 4.0662,
|
| 87 |
"step": 550
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 2.912621359223301,
|
| 91 |
+
"grad_norm": 5.506278038024902,
|
| 92 |
"learning_rate": 4.51537216828479e-05,
|
| 93 |
+
"loss": 4.0403,
|
| 94 |
"step": 600
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 3.1553398058252426,
|
| 98 |
+
"grad_norm": 5.410216331481934,
|
| 99 |
"learning_rate": 4.4749190938511334e-05,
|
| 100 |
+
"loss": 3.8375,
|
| 101 |
"step": 650
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 3.3980582524271843,
|
| 105 |
+
"grad_norm": 5.637699127197266,
|
| 106 |
"learning_rate": 4.434466019417476e-05,
|
| 107 |
+
"loss": 3.7473,
|
| 108 |
"step": 700
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 3.6407766990291264,
|
| 112 |
+
"grad_norm": 5.575273513793945,
|
| 113 |
"learning_rate": 4.3940129449838194e-05,
|
| 114 |
+
"loss": 3.7835,
|
| 115 |
"step": 750
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 3.883495145631068,
|
| 119 |
+
"grad_norm": 5.737198352813721,
|
| 120 |
"learning_rate": 4.353559870550162e-05,
|
| 121 |
+
"loss": 3.7919,
|
| 122 |
"step": 800
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 4.12621359223301,
|
| 126 |
+
"grad_norm": 5.645532608032227,
|
| 127 |
"learning_rate": 4.313106796116505e-05,
|
| 128 |
+
"loss": 3.6425,
|
| 129 |
"step": 850
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 4.368932038834951,
|
| 133 |
+
"grad_norm": 6.308927059173584,
|
| 134 |
"learning_rate": 4.272653721682848e-05,
|
| 135 |
+
"loss": 3.4785,
|
| 136 |
"step": 900
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 4.611650485436893,
|
| 140 |
+
"grad_norm": 5.863094329833984,
|
| 141 |
"learning_rate": 4.232200647249191e-05,
|
| 142 |
+
"loss": 3.5268,
|
| 143 |
"step": 950
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 4.854368932038835,
|
| 147 |
+
"grad_norm": 6.120258331298828,
|
| 148 |
"learning_rate": 4.191747572815534e-05,
|
| 149 |
+
"loss": 3.5296,
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 5.097087378640777,
|
| 154 |
+
"grad_norm": 6.231515884399414,
|
| 155 |
"learning_rate": 4.1512944983818774e-05,
|
| 156 |
+
"loss": 3.3802,
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 5.339805825242719,
|
| 161 |
+
"grad_norm": 6.211188316345215,
|
| 162 |
"learning_rate": 4.11084142394822e-05,
|
| 163 |
+
"loss": 3.2148,
|
| 164 |
"step": 1100
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 5.58252427184466,
|
| 168 |
+
"grad_norm": 6.8676934242248535,
|
| 169 |
"learning_rate": 4.0703883495145634e-05,
|
| 170 |
+
"loss": 3.2111,
|
| 171 |
"step": 1150
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 5.825242718446602,
|
| 175 |
+
"grad_norm": 6.851133346557617,
|
| 176 |
"learning_rate": 4.029935275080906e-05,
|
| 177 |
+
"loss": 3.2771,
|
| 178 |
"step": 1200
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 6.067961165048544,
|
| 182 |
+
"grad_norm": 6.708765506744385,
|
| 183 |
"learning_rate": 3.9894822006472494e-05,
|
| 184 |
+
"loss": 3.1587,
|
| 185 |
"step": 1250
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 6.310679611650485,
|
| 189 |
+
"grad_norm": 7.136553764343262,
|
| 190 |
"learning_rate": 3.949029126213593e-05,
|
| 191 |
+
"loss": 2.9192,
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 6.553398058252427,
|
| 196 |
+
"grad_norm": 7.370823383331299,
|
| 197 |
"learning_rate": 3.9085760517799354e-05,
|
| 198 |
+
"loss": 2.922,
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 6.796116504854369,
|
| 203 |
+
"grad_norm": 7.630361557006836,
|
| 204 |
"learning_rate": 3.868122977346279e-05,
|
| 205 |
+
"loss": 2.9417,
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 7.038834951456311,
|
| 210 |
+
"grad_norm": 7.765511989593506,
|
| 211 |
"learning_rate": 3.827669902912622e-05,
|
| 212 |
+
"loss": 2.8669,
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 7.281553398058253,
|
| 217 |
+
"grad_norm": 7.85439920425415,
|
| 218 |
"learning_rate": 3.787216828478965e-05,
|
| 219 |
+
"loss": 2.5412,
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 7.524271844660194,
|
| 224 |
+
"grad_norm": 8.064071655273438,
|
| 225 |
"learning_rate": 3.746763754045307e-05,
|
| 226 |
+
"loss": 2.6131,
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 7.766990291262136,
|
| 231 |
+
"grad_norm": 8.692522048950195,
|
| 232 |
"learning_rate": 3.7063106796116507e-05,
|
| 233 |
+
"loss": 2.6356,
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 8.009708737864077,
|
| 238 |
+
"grad_norm": 8.012410163879395,
|
| 239 |
"learning_rate": 3.665857605177993e-05,
|
| 240 |
+
"loss": 2.599,
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 8.25242718446602,
|
| 245 |
+
"grad_norm": 8.6799898147583,
|
| 246 |
"learning_rate": 3.6254045307443366e-05,
|
| 247 |
+
"loss": 2.1992,
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 8.495145631067961,
|
| 252 |
+
"grad_norm": 8.963507652282715,
|
| 253 |
"learning_rate": 3.584951456310679e-05,
|
| 254 |
+
"loss": 2.2701,
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 8.737864077669903,
|
| 259 |
+
"grad_norm": 9.833822250366211,
|
| 260 |
"learning_rate": 3.5444983818770226e-05,
|
| 261 |
+
"loss": 2.2837,
|
| 262 |
"step": 1800
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 8.980582524271846,
|
| 266 |
+
"grad_norm": 9.47242259979248,
|
| 267 |
"learning_rate": 3.504045307443366e-05,
|
| 268 |
+
"loss": 2.2854,
|
| 269 |
"step": 1850
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 9.223300970873787,
|
| 273 |
+
"grad_norm": 8.856402397155762,
|
| 274 |
"learning_rate": 3.4635922330097086e-05,
|
| 275 |
+
"loss": 1.8862,
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 9.466019417475728,
|
| 280 |
+
"grad_norm": 9.948341369628906,
|
| 281 |
"learning_rate": 3.423139158576052e-05,
|
| 282 |
+
"loss": 1.922,
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 9.70873786407767,
|
| 287 |
+
"grad_norm": 10.28939151763916,
|
| 288 |
"learning_rate": 3.382686084142395e-05,
|
| 289 |
+
"loss": 1.9622,
|
| 290 |
"step": 2000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 9.951456310679612,
|
| 294 |
+
"grad_norm": 10.114766120910645,
|
| 295 |
"learning_rate": 3.342233009708738e-05,
|
| 296 |
+
"loss": 1.9741,
|
| 297 |
"step": 2050
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 10.194174757281553,
|
| 301 |
+
"grad_norm": 9.489270210266113,
|
| 302 |
"learning_rate": 3.301779935275081e-05,
|
| 303 |
+
"loss": 1.6509,
|
| 304 |
"step": 2100
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 10.436893203883495,
|
| 308 |
+
"grad_norm": 10.16482162475586,
|
| 309 |
"learning_rate": 3.2613268608414246e-05,
|
| 310 |
+
"loss": 1.5883,
|
| 311 |
"step": 2150
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 10.679611650485437,
|
| 315 |
+
"grad_norm": 11.778059005737305,
|
| 316 |
"learning_rate": 3.220873786407767e-05,
|
| 317 |
+
"loss": 1.6234,
|
| 318 |
"step": 2200
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 10.922330097087379,
|
| 322 |
+
"grad_norm": 10.798454284667969,
|
| 323 |
"learning_rate": 3.1804207119741106e-05,
|
| 324 |
+
"loss": 1.667,
|
| 325 |
"step": 2250
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 11.16504854368932,
|
| 329 |
+
"grad_norm": 9.751585006713867,
|
| 330 |
"learning_rate": 3.139967637540453e-05,
|
| 331 |
+
"loss": 1.3785,
|
| 332 |
"step": 2300
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 11.407766990291263,
|
| 336 |
+
"grad_norm": 10.266364097595215,
|
| 337 |
"learning_rate": 3.099514563106796e-05,
|
| 338 |
+
"loss": 1.2956,
|
| 339 |
"step": 2350
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 11.650485436893204,
|
| 343 |
+
"grad_norm": 10.660856246948242,
|
| 344 |
"learning_rate": 3.059061488673139e-05,
|
| 345 |
+
"loss": 1.3321,
|
| 346 |
"step": 2400
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 11.893203883495145,
|
| 350 |
+
"grad_norm": 10.896187782287598,
|
| 351 |
"learning_rate": 3.0186084142394822e-05,
|
| 352 |
+
"loss": 1.3533,
|
| 353 |
"step": 2450
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 12.135922330097088,
|
| 357 |
+
"grad_norm": 10.952502250671387,
|
| 358 |
"learning_rate": 2.9781553398058252e-05,
|
| 359 |
+
"loss": 1.1486,
|
| 360 |
"step": 2500
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 12.37864077669903,
|
| 364 |
+
"grad_norm": 10.766463279724121,
|
| 365 |
"learning_rate": 2.9377022653721686e-05,
|
| 366 |
+
"loss": 1.0227,
|
| 367 |
"step": 2550
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 12.62135922330097,
|
| 371 |
+
"grad_norm": 9.385764122009277,
|
| 372 |
"learning_rate": 2.8972491909385112e-05,
|
| 373 |
+
"loss": 1.0651,
|
| 374 |
"step": 2600
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 12.864077669902912,
|
| 378 |
+
"grad_norm": 11.573925018310547,
|
| 379 |
"learning_rate": 2.8567961165048546e-05,
|
| 380 |
+
"loss": 1.1013,
|
| 381 |
"step": 2650
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 13.106796116504855,
|
| 385 |
+
"grad_norm": 10.138091087341309,
|
| 386 |
"learning_rate": 2.816343042071198e-05,
|
| 387 |
+
"loss": 0.9545,
|
| 388 |
"step": 2700
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 13.349514563106796,
|
| 392 |
+
"grad_norm": 9.867119789123535,
|
| 393 |
"learning_rate": 2.7758899676375405e-05,
|
| 394 |
+
"loss": 0.7892,
|
| 395 |
"step": 2750
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 13.592233009708737,
|
| 399 |
+
"grad_norm": 10.443971633911133,
|
| 400 |
"learning_rate": 2.735436893203884e-05,
|
| 401 |
+
"loss": 0.8432,
|
| 402 |
"step": 2800
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 13.83495145631068,
|
| 406 |
+
"grad_norm": 10.633298873901367,
|
| 407 |
"learning_rate": 2.6949838187702265e-05,
|
| 408 |
+
"loss": 0.8543,
|
| 409 |
"step": 2850
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 14.077669902912621,
|
| 413 |
+
"grad_norm": 8.640830039978027,
|
| 414 |
"learning_rate": 2.6545307443365695e-05,
|
| 415 |
+
"loss": 0.7751,
|
| 416 |
"step": 2900
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 14.320388349514563,
|
| 420 |
+
"grad_norm": 10.461247444152832,
|
| 421 |
"learning_rate": 2.614077669902913e-05,
|
| 422 |
+
"loss": 0.6074,
|
| 423 |
"step": 2950
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 14.563106796116505,
|
| 427 |
+
"grad_norm": 10.757479667663574,
|
| 428 |
"learning_rate": 2.5736245954692555e-05,
|
| 429 |
+
"loss": 0.6354,
|
| 430 |
"step": 3000
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 14.805825242718447,
|
| 434 |
+
"grad_norm": 11.599132537841797,
|
| 435 |
"learning_rate": 2.533171521035599e-05,
|
| 436 |
+
"loss": 0.6694,
|
| 437 |
"step": 3050
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 15.048543689320388,
|
| 441 |
+
"grad_norm": 8.085358619689941,
|
| 442 |
"learning_rate": 2.492718446601942e-05,
|
| 443 |
+
"loss": 0.6306,
|
| 444 |
"step": 3100
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 15.29126213592233,
|
| 448 |
+
"grad_norm": 8.327988624572754,
|
| 449 |
"learning_rate": 2.452265372168285e-05,
|
| 450 |
+
"loss": 0.4628,
|
| 451 |
"step": 3150
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 15.533980582524272,
|
| 455 |
+
"grad_norm": 8.545391082763672,
|
| 456 |
"learning_rate": 2.411812297734628e-05,
|
| 457 |
+
"loss": 0.4915,
|
| 458 |
"step": 3200
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 15.776699029126213,
|
| 462 |
+
"grad_norm": 10.176375389099121,
|
| 463 |
"learning_rate": 2.3713592233009708e-05,
|
| 464 |
+
"loss": 0.5105,
|
| 465 |
"step": 3250
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 16.019417475728154,
|
| 469 |
+
"grad_norm": 7.821159362792969,
|
| 470 |
"learning_rate": 2.3309061488673138e-05,
|
| 471 |
+
"loss": 0.5055,
|
| 472 |
"step": 3300
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 16.262135922330096,
|
| 476 |
+
"grad_norm": 7.797800064086914,
|
| 477 |
"learning_rate": 2.290453074433657e-05,
|
| 478 |
+
"loss": 0.3486,
|
| 479 |
"step": 3350
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 16.50485436893204,
|
| 483 |
+
"grad_norm": 8.497370719909668,
|
| 484 |
"learning_rate": 2.25e-05,
|
| 485 |
+
"loss": 0.3627,
|
| 486 |
"step": 3400
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 16.74757281553398,
|
| 490 |
+
"grad_norm": 9.842249870300293,
|
| 491 |
"learning_rate": 2.209546925566343e-05,
|
| 492 |
+
"loss": 0.3969,
|
| 493 |
"step": 3450
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 16.990291262135923,
|
| 497 |
+
"grad_norm": 10.705739974975586,
|
| 498 |
"learning_rate": 2.169093851132686e-05,
|
| 499 |
+
"loss": 0.4028,
|
| 500 |
"step": 3500
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 17.233009708737864,
|
| 504 |
+
"grad_norm": 7.950681686401367,
|
| 505 |
"learning_rate": 2.1286407766990295e-05,
|
| 506 |
+
"loss": 0.284,
|
| 507 |
"step": 3550
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 17.475728155339805,
|
| 511 |
+
"grad_norm": 8.526535034179688,
|
| 512 |
"learning_rate": 2.0881877022653725e-05,
|
| 513 |
+
"loss": 0.2934,
|
| 514 |
"step": 3600
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 17.718446601941746,
|
| 518 |
+
"grad_norm": 8.999290466308594,
|
| 519 |
"learning_rate": 2.047734627831715e-05,
|
| 520 |
+
"loss": 0.2964,
|
| 521 |
"step": 3650
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 17.96116504854369,
|
| 525 |
+
"grad_norm": 9.228137969970703,
|
| 526 |
"learning_rate": 2.007281553398058e-05,
|
| 527 |
+
"loss": 0.311,
|
| 528 |
"step": 3700
|
| 529 |
},
|
| 530 |
{
|
| 531 |
"epoch": 18.203883495145632,
|
| 532 |
+
"grad_norm": 6.715802192687988,
|
| 533 |
"learning_rate": 1.9668284789644014e-05,
|
| 534 |
+
"loss": 0.2263,
|
| 535 |
"step": 3750
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"epoch": 18.446601941747574,
|
| 539 |
+
"grad_norm": 7.834264278411865,
|
| 540 |
"learning_rate": 1.9263754045307444e-05,
|
| 541 |
+
"loss": 0.2293,
|
| 542 |
"step": 3800
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 18.689320388349515,
|
| 546 |
+
"grad_norm": 6.83228874206543,
|
| 547 |
"learning_rate": 1.8859223300970874e-05,
|
| 548 |
+
"loss": 0.2392,
|
| 549 |
"step": 3850
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 18.932038834951456,
|
| 553 |
+
"grad_norm": 8.507209777832031,
|
| 554 |
"learning_rate": 1.8454692556634304e-05,
|
| 555 |
+
"loss": 0.2377,
|
| 556 |
"step": 3900
|
| 557 |
},
|
| 558 |
{
|
| 559 |
"epoch": 19.174757281553397,
|
| 560 |
+
"grad_norm": 6.565855026245117,
|
| 561 |
"learning_rate": 1.8050161812297738e-05,
|
| 562 |
+
"loss": 0.1905,
|
| 563 |
"step": 3950
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 19.41747572815534,
|
| 567 |
+
"grad_norm": 7.470388889312744,
|
| 568 |
"learning_rate": 1.7645631067961167e-05,
|
| 569 |
+
"loss": 0.1846,
|
| 570 |
"step": 4000
|
| 571 |
},
|
| 572 |
{
|
| 573 |
"epoch": 19.660194174757283,
|
| 574 |
+
"grad_norm": 7.033578395843506,
|
| 575 |
"learning_rate": 1.7241100323624594e-05,
|
| 576 |
+
"loss": 0.1911,
|
| 577 |
"step": 4050
|
| 578 |
},
|
| 579 |
{
|
| 580 |
"epoch": 19.902912621359224,
|
| 581 |
+
"grad_norm": 8.110376358032227,
|
| 582 |
"learning_rate": 1.6836569579288027e-05,
|
| 583 |
+
"loss": 0.1954,
|
| 584 |
"step": 4100
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 20.145631067961165,
|
| 588 |
+
"grad_norm": 7.335451602935791,
|
| 589 |
"learning_rate": 1.6432038834951457e-05,
|
| 590 |
+
"loss": 0.1619,
|
| 591 |
"step": 4150
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 20.388349514563107,
|
| 595 |
+
"grad_norm": 7.478648662567139,
|
| 596 |
"learning_rate": 1.6027508090614887e-05,
|
| 597 |
+
"loss": 0.1497,
|
| 598 |
"step": 4200
|
| 599 |
},
|
| 600 |
{
|
| 601 |
"epoch": 20.631067961165048,
|
| 602 |
+
"grad_norm": 7.764925479888916,
|
| 603 |
"learning_rate": 1.5622977346278317e-05,
|
| 604 |
+
"loss": 0.1559,
|
| 605 |
"step": 4250
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"epoch": 20.87378640776699,
|
| 609 |
+
"grad_norm": 6.432894706726074,
|
| 610 |
"learning_rate": 1.5218446601941749e-05,
|
| 611 |
+
"loss": 0.1601,
|
| 612 |
"step": 4300
|
| 613 |
},
|
| 614 |
{
|
| 615 |
"epoch": 21.116504854368934,
|
| 616 |
+
"grad_norm": 7.728542327880859,
|
| 617 |
"learning_rate": 1.4813915857605179e-05,
|
| 618 |
+
"loss": 0.1416,
|
| 619 |
"step": 4350
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"epoch": 21.359223300970875,
|
| 623 |
+
"grad_norm": 6.370733737945557,
|
| 624 |
"learning_rate": 1.4409385113268609e-05,
|
| 625 |
+
"loss": 0.1211,
|
| 626 |
"step": 4400
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 21.601941747572816,
|
| 630 |
+
"grad_norm": 5.939419746398926,
|
| 631 |
"learning_rate": 1.4004854368932039e-05,
|
| 632 |
+
"loss": 0.1278,
|
| 633 |
"step": 4450
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 21.844660194174757,
|
| 637 |
+
"grad_norm": 6.0013957023620605,
|
| 638 |
"learning_rate": 1.360032362459547e-05,
|
| 639 |
+
"loss": 0.1274,
|
| 640 |
"step": 4500
|
| 641 |
}
|
| 642 |
],
|
|
|
|
| 657 |
"attributes": {}
|
| 658 |
}
|
| 659 |
},
|
| 660 |
+
"total_flos": 2089590718464000.0,
|
| 661 |
"train_batch_size": 4,
|
| 662 |
"trial_name": null,
|
| 663 |
"trial_params": null
|
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:d5c04ce53d557cd762ff77607aaadbd1248aaa177697662d9a21b90d444993b7
|
| 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":
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
-
"n_layer":
|
| 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": 2048,
|
| 14 |
+
"n_embd": 1024,
|
| 15 |
"n_head": 2,
|
| 16 |
"n_inner": null,
|
| 17 |
+
"n_layer": 12,
|
| 18 |
+
"n_positions": 2048,
|
| 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 sha256:3d034447d0bc7b5f683ac7a16c807a7ff485dd3b1b8422db07d941ad747c0e7e
|
| 3 |
+
size 617130824
|