ssu53 commited on
Commit
5071bc2
·
verified ·
1 Parent(s): 4294779

Upload model

Browse files
README.md ADDED
@@ -0,0 +1,61 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers
3
+ license: other
4
+ base_model: Qwen/Qwen2.5-32B-Instruct
5
+ tags:
6
+ - llama-factory
7
+ - full
8
+ - generated_from_trainer
9
+ model-index:
10
+ - name: no_pipeline_10k_32B
11
+ results: []
12
+ ---
13
+
14
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
15
+ should probably proofread and complete it, then remove this comment. -->
16
+
17
+ # no_pipeline_10k_32B
18
+
19
+ This model is a fine-tuned version of [Qwen/Qwen2.5-32B-Instruct](https://huggingface.co/Qwen/Qwen2.5-32B-Instruct) on the mlfoundations-dev/no_pipeline_10k dataset.
20
+
21
+ ## Model description
22
+
23
+ More information needed
24
+
25
+ ## Intended uses & limitations
26
+
27
+ More information needed
28
+
29
+ ## Training and evaluation data
30
+
31
+ More information needed
32
+
33
+ ## Training procedure
34
+
35
+ ### Training hyperparameters
36
+
37
+ The following hyperparameters were used during training:
38
+ - learning_rate: 4e-05
39
+ - train_batch_size: 1
40
+ - eval_batch_size: 8
41
+ - seed: 42
42
+ - distributed_type: multi-GPU
43
+ - num_devices: 64
44
+ - gradient_accumulation_steps: 2
45
+ - total_train_batch_size: 128
46
+ - total_eval_batch_size: 512
47
+ - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
48
+ - lr_scheduler_type: cosine
49
+ - lr_scheduler_warmup_ratio: 0.1
50
+ - num_epochs: 5.0
51
+
52
+ ### Training results
53
+
54
+
55
+
56
+ ### Framework versions
57
+
58
+ - Transformers 4.46.1
59
+ - Pytorch 2.6.0+cu124
60
+ - Datasets 3.1.0
61
+ - Tokenizers 0.20.3
all_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 4.968152866242038,
3
+ "total_flos": 4.6317288847048704e+17,
4
+ "train_loss": 0.30433785139750213,
5
+ "train_runtime": 26201.4476,
6
+ "train_samples_per_second": 1.908,
7
+ "train_steps_per_second": 0.015
8
+ }
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "Qwen/Qwen2.5-32B-Instruct",
3
  "architectures": [
4
  "Qwen2ForCausalLM"
5
  ],
 
1
  {
2
+ "_name_or_path": "/leonardo_work/EUHPC_E03_068/DCFT_shared/hub/models--Qwen--Qwen2.5-32B-Instruct/snapshots/5ede1c97bbab6ce5cda5812749b4c0bdf79b18dd",
3
  "architectures": [
4
  "Qwen2ForCausalLM"
5
  ],
generation_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 151643,
3
+ "do_sample": true,
4
+ "eos_token_id": [
5
+ 151645,
6
+ 151643
7
+ ],
8
+ "pad_token_id": 151643,
9
+ "repetition_penalty": 1.05,
10
+ "temperature": 0.7,
11
+ "top_k": 20,
12
+ "top_p": 0.8,
13
+ "transformers_version": "4.46.1"
14
+ }
model-00001-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4741f7e789cbe835ca5afd15db6e5b962c55615daf6b4c9ebcb3a9179593c3ca
3
  size 4891730992
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f1f636f4e5f229264ac499fd1b6498b4d634e0eca7417ad74ac4370ba27b60a
3
  size 4891730992
model-00002-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:636c0d6d1c1ca0e93e048dcf9dae2a0bc210be51a1a22a7999903ca254d69ff2
3
  size 4876059352
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3c8221dca74000d315fc1656d1c4366e245ba02391a02a1eba22e33ba0b2a8de
3
  size 4876059352
model-00003-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:771abf431c41751f47a76fffe3c2aaaba1984f98ef6a9d8eacc0eb68d40ad1db
3
  size 4876059384
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:201914ba32fcae76ad2cf974c3484207f19fe190408a0164379eebb8b86e1695
3
  size 4876059384
model-00004-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:84c13970f1431cfbd9ce6b14e1ce4e12e40eeb71777c3c26a61963e1df2876d4
3
  size 4876059416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:82c112a3a8451b88bf953f38513836d1adc0e1db14a2831475ac973d4fa641a9
3
  size 4876059416
model-00005-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fad9b29536808b8ff32ec5cd3bb06c2eb3cf80d12380dd358e5c251cf76099e1
3
  size 4876059416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3129e4cc601cb61a0814c6830fe510c2c9d300906baece2a426c43a1bbe113eb
3
  size 4876059416
model-00006-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d43232873b47fb3a984f4b6bc75a3894edc039326cb8f0009dbd64d086192469
3
  size 4876059416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e7e40333a2d4843e5b1f3623940d74f0e8252facca31ca43abd8dae0efd8ffbb
3
  size 4876059416
model-00007-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:836ac301a2e83ec6c108aaece9e4cc9526c24a6d7babf9e28b19d12538a55c5e
3
  size 4876059416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:404f102e5ec35134edaaa203203bb81cff320b58693184f957347d7c678e22ce
3
  size 4876059416
model-00008-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2e66b9ced4e2db6b7f64d8809d46aa2644813b48e8df01662eb36d0e1e9c8570
3
  size 4876059416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:34072aa1e01e6ea3b9e3e412d5865816e80bc8b65f51380cd2221e9b46d8eb1c
3
  size 4876059416
model-00009-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:cdbb35ea74ffb5d45461e74bfd99239919194910a748458792b3b87d8afb8dce
3
  size 4876059416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5e2dd006f483441af0076cd4b3ad572f7c10ef6ac456eff0a3284fc7221eab7d
3
  size 4876059416
model-00010-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3226df592c7c99b4e395ce9af5358d6b105d48a758a5990a8c4aa4d32269b354
3
  size 4876059416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c6d8c7e642ae0dfd55cf6e525a9f218392262122d6aa16f8ad480ea41e70b731
3
  size 4876059416
model-00011-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ea2d1abb15bfaa5325c584b16f73ea84da8b035c76af656c7d3ab9cfe540f707
3
  size 4876059416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6124bdf156fe7d99c7187e88e65d9715a8bd9746b4ba4402637a9ecfad91e43f
3
  size 4876059416
model-00012-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9e30fe19ec6aff4e03d68eb778aa957232542322c1f0ec3f93854f0e14945788
3
  size 4876059416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0cbcef5f9025a3c86cf248fc09be439280f6eb729db80ad661d71c8490489193
3
  size 4876059416
model-00013-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bf4db43150a3495621e5ba85a18c7b84e8dc9984dc04432cdc143d2d63a302a6
3
  size 4876059416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:faf8c95d5b8e086c30e1d58c808ab5c7b3dae6b49e36aaf7bbb443c959c3d569
3
  size 4876059416
model-00014-of-00014.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f1399acf53c0688d628995bfa59f8b32becce5f233e6c41ae54433f52ce7e3b7
3
  size 2123397800
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bb9704bfc526eda757df1be22bb99acd6990237efd892d4e8799d28cd08770ab
3
  size 2123397800
start_end.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"start_time": "2025-04-29 10:28:29", "end_time": "2025-04-29 17:52:34"}
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 4.968152866242038,
3
+ "total_flos": 4.6317288847048704e+17,
4
+ "train_loss": 0.30433785139750213,
5
+ "train_runtime": 26201.4476,
6
+ "train_samples_per_second": 1.908,
7
+ "train_steps_per_second": 0.015
8
+ }
trainer_log.jsonl CHANGED
The diff for this file is too large to render. See raw diff
 
trainer_state.json ADDED
@@ -0,0 +1,2772 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 4.968152866242038,
5
+ "eval_steps": 500,
6
+ "global_step": 390,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.012738853503184714,
13
+ "grad_norm": 3.7423072661703385,
14
+ "learning_rate": 1.0256410256410257e-06,
15
+ "loss": 0.972,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 0.025477707006369428,
20
+ "grad_norm": 4.014331193631274,
21
+ "learning_rate": 2.0512820512820513e-06,
22
+ "loss": 0.9843,
23
+ "step": 2
24
+ },
25
+ {
26
+ "epoch": 0.03821656050955414,
27
+ "grad_norm": 3.8392264390505235,
28
+ "learning_rate": 3.0769230769230774e-06,
29
+ "loss": 0.9844,
30
+ "step": 3
31
+ },
32
+ {
33
+ "epoch": 0.050955414012738856,
34
+ "grad_norm": 3.4206021200245513,
35
+ "learning_rate": 4.102564102564103e-06,
36
+ "loss": 0.9535,
37
+ "step": 4
38
+ },
39
+ {
40
+ "epoch": 0.06369426751592357,
41
+ "grad_norm": 2.844532001391515,
42
+ "learning_rate": 5.128205128205128e-06,
43
+ "loss": 0.9244,
44
+ "step": 5
45
+ },
46
+ {
47
+ "epoch": 0.07643312101910828,
48
+ "grad_norm": 2.019106639258673,
49
+ "learning_rate": 6.153846153846155e-06,
50
+ "loss": 0.8528,
51
+ "step": 6
52
+ },
53
+ {
54
+ "epoch": 0.08917197452229299,
55
+ "grad_norm": 1.57361042148724,
56
+ "learning_rate": 7.17948717948718e-06,
57
+ "loss": 0.8202,
58
+ "step": 7
59
+ },
60
+ {
61
+ "epoch": 0.10191082802547771,
62
+ "grad_norm": 1.3225717537131316,
63
+ "learning_rate": 8.205128205128205e-06,
64
+ "loss": 0.806,
65
+ "step": 8
66
+ },
67
+ {
68
+ "epoch": 0.11464968152866242,
69
+ "grad_norm": 1.297213830671725,
70
+ "learning_rate": 9.230769230769232e-06,
71
+ "loss": 0.8505,
72
+ "step": 9
73
+ },
74
+ {
75
+ "epoch": 0.12738853503184713,
76
+ "grad_norm": 1.924658811896663,
77
+ "learning_rate": 1.0256410256410256e-05,
78
+ "loss": 0.7536,
79
+ "step": 10
80
+ },
81
+ {
82
+ "epoch": 0.14012738853503184,
83
+ "grad_norm": 1.7407165339891915,
84
+ "learning_rate": 1.1282051282051283e-05,
85
+ "loss": 0.7524,
86
+ "step": 11
87
+ },
88
+ {
89
+ "epoch": 0.15286624203821655,
90
+ "grad_norm": 1.3558527366110018,
91
+ "learning_rate": 1.230769230769231e-05,
92
+ "loss": 0.7382,
93
+ "step": 12
94
+ },
95
+ {
96
+ "epoch": 0.16560509554140126,
97
+ "grad_norm": 1.0929459173306966,
98
+ "learning_rate": 1.3333333333333333e-05,
99
+ "loss": 0.695,
100
+ "step": 13
101
+ },
102
+ {
103
+ "epoch": 0.17834394904458598,
104
+ "grad_norm": 1.2749157546624197,
105
+ "learning_rate": 1.435897435897436e-05,
106
+ "loss": 0.7111,
107
+ "step": 14
108
+ },
109
+ {
110
+ "epoch": 0.1910828025477707,
111
+ "grad_norm": 1.0998533610546595,
112
+ "learning_rate": 1.5384615384615387e-05,
113
+ "loss": 0.7091,
114
+ "step": 15
115
+ },
116
+ {
117
+ "epoch": 0.20382165605095542,
118
+ "grad_norm": 0.913903420689088,
119
+ "learning_rate": 1.641025641025641e-05,
120
+ "loss": 0.6703,
121
+ "step": 16
122
+ },
123
+ {
124
+ "epoch": 0.21656050955414013,
125
+ "grad_norm": 0.7661216217896927,
126
+ "learning_rate": 1.7435897435897438e-05,
127
+ "loss": 0.6948,
128
+ "step": 17
129
+ },
130
+ {
131
+ "epoch": 0.22929936305732485,
132
+ "grad_norm": 0.7437430749983936,
133
+ "learning_rate": 1.8461538461538465e-05,
134
+ "loss": 0.6176,
135
+ "step": 18
136
+ },
137
+ {
138
+ "epoch": 0.24203821656050956,
139
+ "grad_norm": 0.7446006831365574,
140
+ "learning_rate": 1.9487179487179488e-05,
141
+ "loss": 0.6591,
142
+ "step": 19
143
+ },
144
+ {
145
+ "epoch": 0.25477707006369427,
146
+ "grad_norm": 0.6792039966401283,
147
+ "learning_rate": 2.0512820512820512e-05,
148
+ "loss": 0.6359,
149
+ "step": 20
150
+ },
151
+ {
152
+ "epoch": 0.267515923566879,
153
+ "grad_norm": 0.5849626006077434,
154
+ "learning_rate": 2.153846153846154e-05,
155
+ "loss": 0.6271,
156
+ "step": 21
157
+ },
158
+ {
159
+ "epoch": 0.2802547770700637,
160
+ "grad_norm": 0.6919741782724808,
161
+ "learning_rate": 2.2564102564102566e-05,
162
+ "loss": 0.665,
163
+ "step": 22
164
+ },
165
+ {
166
+ "epoch": 0.2929936305732484,
167
+ "grad_norm": 0.7570356242077189,
168
+ "learning_rate": 2.3589743589743593e-05,
169
+ "loss": 0.6547,
170
+ "step": 23
171
+ },
172
+ {
173
+ "epoch": 0.3057324840764331,
174
+ "grad_norm": 0.6189824489932229,
175
+ "learning_rate": 2.461538461538462e-05,
176
+ "loss": 0.6146,
177
+ "step": 24
178
+ },
179
+ {
180
+ "epoch": 0.3184713375796178,
181
+ "grad_norm": 0.5416486550934623,
182
+ "learning_rate": 2.5641025641025646e-05,
183
+ "loss": 0.6467,
184
+ "step": 25
185
+ },
186
+ {
187
+ "epoch": 0.33121019108280253,
188
+ "grad_norm": 0.6011316533516919,
189
+ "learning_rate": 2.6666666666666667e-05,
190
+ "loss": 0.6231,
191
+ "step": 26
192
+ },
193
+ {
194
+ "epoch": 0.34394904458598724,
195
+ "grad_norm": 0.5744339851945378,
196
+ "learning_rate": 2.7692307692307694e-05,
197
+ "loss": 0.5921,
198
+ "step": 27
199
+ },
200
+ {
201
+ "epoch": 0.35668789808917195,
202
+ "grad_norm": 0.5427273841273952,
203
+ "learning_rate": 2.871794871794872e-05,
204
+ "loss": 0.5984,
205
+ "step": 28
206
+ },
207
+ {
208
+ "epoch": 0.36942675159235666,
209
+ "grad_norm": 0.6587706620684388,
210
+ "learning_rate": 2.9743589743589747e-05,
211
+ "loss": 0.6258,
212
+ "step": 29
213
+ },
214
+ {
215
+ "epoch": 0.3821656050955414,
216
+ "grad_norm": 0.5241407477072185,
217
+ "learning_rate": 3.0769230769230774e-05,
218
+ "loss": 0.6057,
219
+ "step": 30
220
+ },
221
+ {
222
+ "epoch": 0.39490445859872614,
223
+ "grad_norm": 0.6221622861877016,
224
+ "learning_rate": 3.1794871794871795e-05,
225
+ "loss": 0.6028,
226
+ "step": 31
227
+ },
228
+ {
229
+ "epoch": 0.40764331210191085,
230
+ "grad_norm": 0.5873239727341781,
231
+ "learning_rate": 3.282051282051282e-05,
232
+ "loss": 0.5984,
233
+ "step": 32
234
+ },
235
+ {
236
+ "epoch": 0.42038216560509556,
237
+ "grad_norm": 0.5306206208481558,
238
+ "learning_rate": 3.384615384615385e-05,
239
+ "loss": 0.559,
240
+ "step": 33
241
+ },
242
+ {
243
+ "epoch": 0.43312101910828027,
244
+ "grad_norm": 0.5582669312616446,
245
+ "learning_rate": 3.4871794871794875e-05,
246
+ "loss": 0.6149,
247
+ "step": 34
248
+ },
249
+ {
250
+ "epoch": 0.445859872611465,
251
+ "grad_norm": 0.5000854344725365,
252
+ "learning_rate": 3.58974358974359e-05,
253
+ "loss": 0.6189,
254
+ "step": 35
255
+ },
256
+ {
257
+ "epoch": 0.4585987261146497,
258
+ "grad_norm": 0.527484918197263,
259
+ "learning_rate": 3.692307692307693e-05,
260
+ "loss": 0.6085,
261
+ "step": 36
262
+ },
263
+ {
264
+ "epoch": 0.4713375796178344,
265
+ "grad_norm": 0.48032262139418297,
266
+ "learning_rate": 3.794871794871795e-05,
267
+ "loss": 0.5931,
268
+ "step": 37
269
+ },
270
+ {
271
+ "epoch": 0.4840764331210191,
272
+ "grad_norm": 0.4279530947525605,
273
+ "learning_rate": 3.8974358974358976e-05,
274
+ "loss": 0.5781,
275
+ "step": 38
276
+ },
277
+ {
278
+ "epoch": 0.4968152866242038,
279
+ "grad_norm": 0.4900464691732963,
280
+ "learning_rate": 4e-05,
281
+ "loss": 0.6013,
282
+ "step": 39
283
+ },
284
+ {
285
+ "epoch": 0.5095541401273885,
286
+ "grad_norm": 0.4882711712476194,
287
+ "learning_rate": 3.9999198907597046e-05,
288
+ "loss": 0.5778,
289
+ "step": 40
290
+ },
291
+ {
292
+ "epoch": 0.5222929936305732,
293
+ "grad_norm": 0.4368372090967015,
294
+ "learning_rate": 3.9996795694563096e-05,
295
+ "loss": 0.567,
296
+ "step": 41
297
+ },
298
+ {
299
+ "epoch": 0.535031847133758,
300
+ "grad_norm": 0.46003529609029364,
301
+ "learning_rate": 3.999279055341771e-05,
302
+ "loss": 0.5919,
303
+ "step": 42
304
+ },
305
+ {
306
+ "epoch": 0.5477707006369427,
307
+ "grad_norm": 0.433443291072034,
308
+ "learning_rate": 3.998718380500971e-05,
309
+ "loss": 0.5608,
310
+ "step": 43
311
+ },
312
+ {
313
+ "epoch": 0.5605095541401274,
314
+ "grad_norm": 0.5579094629535033,
315
+ "learning_rate": 3.997997589849145e-05,
316
+ "loss": 0.5919,
317
+ "step": 44
318
+ },
319
+ {
320
+ "epoch": 0.5732484076433121,
321
+ "grad_norm": 0.5002127081158247,
322
+ "learning_rate": 3.9971167411282835e-05,
323
+ "loss": 0.6202,
324
+ "step": 45
325
+ },
326
+ {
327
+ "epoch": 0.5859872611464968,
328
+ "grad_norm": 0.45513028054084476,
329
+ "learning_rate": 3.99607590490251e-05,
330
+ "loss": 0.583,
331
+ "step": 46
332
+ },
333
+ {
334
+ "epoch": 0.5987261146496815,
335
+ "grad_norm": 0.49480268233907687,
336
+ "learning_rate": 3.9948751645524235e-05,
337
+ "loss": 0.6031,
338
+ "step": 47
339
+ },
340
+ {
341
+ "epoch": 0.6114649681528662,
342
+ "grad_norm": 0.4676370814108011,
343
+ "learning_rate": 3.9935146162684206e-05,
344
+ "loss": 0.5774,
345
+ "step": 48
346
+ },
347
+ {
348
+ "epoch": 0.6242038216560509,
349
+ "grad_norm": 0.42161416955688974,
350
+ "learning_rate": 3.9919943690429906e-05,
351
+ "loss": 0.5836,
352
+ "step": 49
353
+ },
354
+ {
355
+ "epoch": 0.6369426751592356,
356
+ "grad_norm": 0.4758290404818582,
357
+ "learning_rate": 3.9903145446619837e-05,
358
+ "loss": 0.5961,
359
+ "step": 50
360
+ },
361
+ {
362
+ "epoch": 0.6496815286624203,
363
+ "grad_norm": 0.4465799608039247,
364
+ "learning_rate": 3.9884752776948564e-05,
365
+ "loss": 0.5712,
366
+ "step": 51
367
+ },
368
+ {
369
+ "epoch": 0.6624203821656051,
370
+ "grad_norm": 0.4783306849945249,
371
+ "learning_rate": 3.9864767154838864e-05,
372
+ "loss": 0.5781,
373
+ "step": 52
374
+ },
375
+ {
376
+ "epoch": 0.6751592356687898,
377
+ "grad_norm": 0.47408070441336125,
378
+ "learning_rate": 3.9843190181323744e-05,
379
+ "loss": 0.5631,
380
+ "step": 53
381
+ },
382
+ {
383
+ "epoch": 0.6878980891719745,
384
+ "grad_norm": 0.44160447990297685,
385
+ "learning_rate": 3.982002358491817e-05,
386
+ "loss": 0.5793,
387
+ "step": 54
388
+ },
389
+ {
390
+ "epoch": 0.7006369426751592,
391
+ "grad_norm": 0.5440809599078149,
392
+ "learning_rate": 3.979526922148058e-05,
393
+ "loss": 0.5739,
394
+ "step": 55
395
+ },
396
+ {
397
+ "epoch": 0.7133757961783439,
398
+ "grad_norm": 0.4030353591762494,
399
+ "learning_rate": 3.9768929074064206e-05,
400
+ "loss": 0.5731,
401
+ "step": 56
402
+ },
403
+ {
404
+ "epoch": 0.7261146496815286,
405
+ "grad_norm": 0.47204667115023985,
406
+ "learning_rate": 3.9741005252758255e-05,
407
+ "loss": 0.5752,
408
+ "step": 57
409
+ },
410
+ {
411
+ "epoch": 0.7388535031847133,
412
+ "grad_norm": 0.3924967377535464,
413
+ "learning_rate": 3.971149999451886e-05,
414
+ "loss": 0.5723,
415
+ "step": 58
416
+ },
417
+ {
418
+ "epoch": 0.7515923566878981,
419
+ "grad_norm": 0.5036651859656847,
420
+ "learning_rate": 3.9680415662989806e-05,
421
+ "loss": 0.5816,
422
+ "step": 59
423
+ },
424
+ {
425
+ "epoch": 0.7643312101910829,
426
+ "grad_norm": 0.44083451558702347,
427
+ "learning_rate": 3.9647754748313294e-05,
428
+ "loss": 0.5878,
429
+ "step": 60
430
+ },
431
+ {
432
+ "epoch": 0.7770700636942676,
433
+ "grad_norm": 0.4848776951212216,
434
+ "learning_rate": 3.96135198669304e-05,
435
+ "loss": 0.5733,
436
+ "step": 61
437
+ },
438
+ {
439
+ "epoch": 0.7898089171974523,
440
+ "grad_norm": 0.4525502560691572,
441
+ "learning_rate": 3.957771376137144e-05,
442
+ "loss": 0.5558,
443
+ "step": 62
444
+ },
445
+ {
446
+ "epoch": 0.802547770700637,
447
+ "grad_norm": 0.4368149673012582,
448
+ "learning_rate": 3.954033930003634e-05,
449
+ "loss": 0.5766,
450
+ "step": 63
451
+ },
452
+ {
453
+ "epoch": 0.8152866242038217,
454
+ "grad_norm": 0.4530850844628041,
455
+ "learning_rate": 3.9501399476964806e-05,
456
+ "loss": 0.567,
457
+ "step": 64
458
+ },
459
+ {
460
+ "epoch": 0.8280254777070064,
461
+ "grad_norm": 0.4520029723511962,
462
+ "learning_rate": 3.946089741159648e-05,
463
+ "loss": 0.5747,
464
+ "step": 65
465
+ },
466
+ {
467
+ "epoch": 0.8407643312101911,
468
+ "grad_norm": 0.4458792601241768,
469
+ "learning_rate": 3.9418836348521045e-05,
470
+ "loss": 0.5567,
471
+ "step": 66
472
+ },
473
+ {
474
+ "epoch": 0.8535031847133758,
475
+ "grad_norm": 0.49824351969919184,
476
+ "learning_rate": 3.937521965721831e-05,
477
+ "loss": 0.5707,
478
+ "step": 67
479
+ },
480
+ {
481
+ "epoch": 0.8662420382165605,
482
+ "grad_norm": 0.5483292524184726,
483
+ "learning_rate": 3.933005083178828e-05,
484
+ "loss": 0.5947,
485
+ "step": 68
486
+ },
487
+ {
488
+ "epoch": 0.8789808917197452,
489
+ "grad_norm": 0.48089214203643743,
490
+ "learning_rate": 3.928333349067125e-05,
491
+ "loss": 0.5389,
492
+ "step": 69
493
+ },
494
+ {
495
+ "epoch": 0.89171974522293,
496
+ "grad_norm": 0.49454234464079616,
497
+ "learning_rate": 3.923507137635792e-05,
498
+ "loss": 0.5396,
499
+ "step": 70
500
+ },
501
+ {
502
+ "epoch": 0.9044585987261147,
503
+ "grad_norm": 0.43160503319375604,
504
+ "learning_rate": 3.9185268355089606e-05,
505
+ "loss": 0.5714,
506
+ "step": 71
507
+ },
508
+ {
509
+ "epoch": 0.9171974522292994,
510
+ "grad_norm": 0.4419560432757935,
511
+ "learning_rate": 3.913392841654851e-05,
512
+ "loss": 0.5623,
513
+ "step": 72
514
+ },
515
+ {
516
+ "epoch": 0.9299363057324841,
517
+ "grad_norm": 0.44484601437802485,
518
+ "learning_rate": 3.9081055673538093e-05,
519
+ "loss": 0.5606,
520
+ "step": 73
521
+ },
522
+ {
523
+ "epoch": 0.9426751592356688,
524
+ "grad_norm": 0.4085811628730557,
525
+ "learning_rate": 3.902665436165364e-05,
526
+ "loss": 0.5457,
527
+ "step": 74
528
+ },
529
+ {
530
+ "epoch": 0.9554140127388535,
531
+ "grad_norm": 0.4101366746857519,
532
+ "learning_rate": 3.897072883894291e-05,
533
+ "loss": 0.6062,
534
+ "step": 75
535
+ },
536
+ {
537
+ "epoch": 0.9681528662420382,
538
+ "grad_norm": 0.4532752791503924,
539
+ "learning_rate": 3.8913283585557054e-05,
540
+ "loss": 0.5861,
541
+ "step": 76
542
+ },
543
+ {
544
+ "epoch": 0.9808917197452229,
545
+ "grad_norm": 0.3783522323934715,
546
+ "learning_rate": 3.885432320339167e-05,
547
+ "loss": 0.5515,
548
+ "step": 77
549
+ },
550
+ {
551
+ "epoch": 0.9936305732484076,
552
+ "grad_norm": 0.4477233989258556,
553
+ "learning_rate": 3.879385241571817e-05,
554
+ "loss": 0.5707,
555
+ "step": 78
556
+ },
557
+ {
558
+ "epoch": 1.0063694267515924,
559
+ "grad_norm": 0.4629761435051509,
560
+ "learning_rate": 3.873187606680543e-05,
561
+ "loss": 0.4782,
562
+ "step": 79
563
+ },
564
+ {
565
+ "epoch": 1.019108280254777,
566
+ "grad_norm": 0.4840535828792894,
567
+ "learning_rate": 3.866839912153168e-05,
568
+ "loss": 0.418,
569
+ "step": 80
570
+ },
571
+ {
572
+ "epoch": 1.0318471337579618,
573
+ "grad_norm": 0.5036326924866047,
574
+ "learning_rate": 3.860342666498677e-05,
575
+ "loss": 0.3971,
576
+ "step": 81
577
+ },
578
+ {
579
+ "epoch": 1.0445859872611465,
580
+ "grad_norm": 0.6228426093815846,
581
+ "learning_rate": 3.853696390206484e-05,
582
+ "loss": 0.3766,
583
+ "step": 82
584
+ },
585
+ {
586
+ "epoch": 1.0573248407643312,
587
+ "grad_norm": 0.5561176865138412,
588
+ "learning_rate": 3.846901615704734e-05,
589
+ "loss": 0.3907,
590
+ "step": 83
591
+ },
592
+ {
593
+ "epoch": 1.070063694267516,
594
+ "grad_norm": 0.5545736824358007,
595
+ "learning_rate": 3.839958887317649e-05,
596
+ "loss": 0.4015,
597
+ "step": 84
598
+ },
599
+ {
600
+ "epoch": 1.0828025477707006,
601
+ "grad_norm": 0.5837935204920232,
602
+ "learning_rate": 3.832868761221926e-05,
603
+ "loss": 0.4234,
604
+ "step": 85
605
+ },
606
+ {
607
+ "epoch": 1.0955414012738853,
608
+ "grad_norm": 0.492326405924025,
609
+ "learning_rate": 3.825631805402182e-05,
610
+ "loss": 0.3696,
611
+ "step": 86
612
+ },
613
+ {
614
+ "epoch": 1.10828025477707,
615
+ "grad_norm": 0.5209834439614793,
616
+ "learning_rate": 3.818248599605448e-05,
617
+ "loss": 0.4019,
618
+ "step": 87
619
+ },
620
+ {
621
+ "epoch": 1.1210191082802548,
622
+ "grad_norm": 0.4885566025553211,
623
+ "learning_rate": 3.810719735294731e-05,
624
+ "loss": 0.3943,
625
+ "step": 88
626
+ },
627
+ {
628
+ "epoch": 1.1337579617834395,
629
+ "grad_norm": 0.47450364241586784,
630
+ "learning_rate": 3.8030458156016326e-05,
631
+ "loss": 0.3907,
632
+ "step": 89
633
+ },
634
+ {
635
+ "epoch": 1.1464968152866242,
636
+ "grad_norm": 0.4445266649179879,
637
+ "learning_rate": 3.795227455278029e-05,
638
+ "loss": 0.411,
639
+ "step": 90
640
+ },
641
+ {
642
+ "epoch": 1.1592356687898089,
643
+ "grad_norm": 0.46006224511432137,
644
+ "learning_rate": 3.787265280646825e-05,
645
+ "loss": 0.3926,
646
+ "step": 91
647
+ },
648
+ {
649
+ "epoch": 1.1719745222929936,
650
+ "grad_norm": 0.4270153987174869,
651
+ "learning_rate": 3.7791599295517825e-05,
652
+ "loss": 0.3655,
653
+ "step": 92
654
+ },
655
+ {
656
+ "epoch": 1.1847133757961783,
657
+ "grad_norm": 0.4878817140154042,
658
+ "learning_rate": 3.7709120513064196e-05,
659
+ "loss": 0.3798,
660
+ "step": 93
661
+ },
662
+ {
663
+ "epoch": 1.197452229299363,
664
+ "grad_norm": 0.46867955444134074,
665
+ "learning_rate": 3.762522306641998e-05,
666
+ "loss": 0.3631,
667
+ "step": 94
668
+ },
669
+ {
670
+ "epoch": 1.2101910828025477,
671
+ "grad_norm": 0.4419027945938377,
672
+ "learning_rate": 3.7539913676545874e-05,
673
+ "loss": 0.3884,
674
+ "step": 95
675
+ },
676
+ {
677
+ "epoch": 1.2229299363057324,
678
+ "grad_norm": 0.40849750356307674,
679
+ "learning_rate": 3.745319917751229e-05,
680
+ "loss": 0.3933,
681
+ "step": 96
682
+ },
683
+ {
684
+ "epoch": 1.2356687898089171,
685
+ "grad_norm": 0.5655828636490249,
686
+ "learning_rate": 3.736508651595188e-05,
687
+ "loss": 0.3992,
688
+ "step": 97
689
+ },
690
+ {
691
+ "epoch": 1.2484076433121019,
692
+ "grad_norm": 0.4781160371870933,
693
+ "learning_rate": 3.727558275050301e-05,
694
+ "loss": 0.4138,
695
+ "step": 98
696
+ },
697
+ {
698
+ "epoch": 1.2611464968152866,
699
+ "grad_norm": 0.4213212240808343,
700
+ "learning_rate": 3.718469505124434e-05,
701
+ "loss": 0.3848,
702
+ "step": 99
703
+ },
704
+ {
705
+ "epoch": 1.2738853503184713,
706
+ "grad_norm": 0.4274617938587094,
707
+ "learning_rate": 3.709243069912041e-05,
708
+ "loss": 0.3763,
709
+ "step": 100
710
+ },
711
+ {
712
+ "epoch": 1.286624203821656,
713
+ "grad_norm": 0.4220383272407581,
714
+ "learning_rate": 3.699879708535838e-05,
715
+ "loss": 0.378,
716
+ "step": 101
717
+ },
718
+ {
719
+ "epoch": 1.2993630573248407,
720
+ "grad_norm": 0.4171375820693096,
721
+ "learning_rate": 3.69038017108759e-05,
722
+ "loss": 0.41,
723
+ "step": 102
724
+ },
725
+ {
726
+ "epoch": 1.3121019108280254,
727
+ "grad_norm": 0.4382305912659545,
728
+ "learning_rate": 3.680745218568026e-05,
729
+ "loss": 0.4059,
730
+ "step": 103
731
+ },
732
+ {
733
+ "epoch": 1.3248407643312101,
734
+ "grad_norm": 0.4165670382998542,
735
+ "learning_rate": 3.6709756228258735e-05,
736
+ "loss": 0.3844,
737
+ "step": 104
738
+ },
739
+ {
740
+ "epoch": 1.3375796178343948,
741
+ "grad_norm": 0.4406466854897885,
742
+ "learning_rate": 3.6610721664960236e-05,
743
+ "loss": 0.4006,
744
+ "step": 105
745
+ },
746
+ {
747
+ "epoch": 1.3503184713375795,
748
+ "grad_norm": 0.4481397864298152,
749
+ "learning_rate": 3.65103564293684e-05,
750
+ "loss": 0.384,
751
+ "step": 106
752
+ },
753
+ {
754
+ "epoch": 1.3630573248407643,
755
+ "grad_norm": 0.43576569305511864,
756
+ "learning_rate": 3.640866856166601e-05,
757
+ "loss": 0.3816,
758
+ "step": 107
759
+ },
760
+ {
761
+ "epoch": 1.3757961783439492,
762
+ "grad_norm": 0.41729038843610433,
763
+ "learning_rate": 3.6305666207990886e-05,
764
+ "loss": 0.417,
765
+ "step": 108
766
+ },
767
+ {
768
+ "epoch": 1.388535031847134,
769
+ "grad_norm": 0.4880218086268087,
770
+ "learning_rate": 3.6201357619783336e-05,
771
+ "loss": 0.4033,
772
+ "step": 109
773
+ },
774
+ {
775
+ "epoch": 1.4012738853503186,
776
+ "grad_norm": 0.41047644449018167,
777
+ "learning_rate": 3.609575115312511e-05,
778
+ "loss": 0.3797,
779
+ "step": 110
780
+ },
781
+ {
782
+ "epoch": 1.4140127388535033,
783
+ "grad_norm": 0.4822480986437211,
784
+ "learning_rate": 3.598885526807003e-05,
785
+ "loss": 0.4076,
786
+ "step": 111
787
+ },
788
+ {
789
+ "epoch": 1.426751592356688,
790
+ "grad_norm": 0.4459969963783435,
791
+ "learning_rate": 3.5880678527966224e-05,
792
+ "loss": 0.384,
793
+ "step": 112
794
+ },
795
+ {
796
+ "epoch": 1.4394904458598727,
797
+ "grad_norm": 0.4108885261851228,
798
+ "learning_rate": 3.577122959877017e-05,
799
+ "loss": 0.3775,
800
+ "step": 113
801
+ },
802
+ {
803
+ "epoch": 1.4522292993630574,
804
+ "grad_norm": 0.4264177407871251,
805
+ "learning_rate": 3.566051724835245e-05,
806
+ "loss": 0.3723,
807
+ "step": 114
808
+ },
809
+ {
810
+ "epoch": 1.4649681528662422,
811
+ "grad_norm": 0.46254534003867953,
812
+ "learning_rate": 3.554855034579532e-05,
813
+ "loss": 0.3905,
814
+ "step": 115
815
+ },
816
+ {
817
+ "epoch": 1.4777070063694269,
818
+ "grad_norm": 0.43159939406328157,
819
+ "learning_rate": 3.5435337860682304e-05,
820
+ "loss": 0.3968,
821
+ "step": 116
822
+ },
823
+ {
824
+ "epoch": 1.4904458598726116,
825
+ "grad_norm": 0.40378318420157067,
826
+ "learning_rate": 3.532088886237956e-05,
827
+ "loss": 0.3986,
828
+ "step": 117
829
+ },
830
+ {
831
+ "epoch": 1.5031847133757963,
832
+ "grad_norm": 0.4010856819942095,
833
+ "learning_rate": 3.520521251930941e-05,
834
+ "loss": 0.4018,
835
+ "step": 118
836
+ },
837
+ {
838
+ "epoch": 1.515923566878981,
839
+ "grad_norm": 0.3953489181151893,
840
+ "learning_rate": 3.5088318098215805e-05,
841
+ "loss": 0.3877,
842
+ "step": 119
843
+ },
844
+ {
845
+ "epoch": 1.5286624203821657,
846
+ "grad_norm": 0.4360231716627973,
847
+ "learning_rate": 3.497021496342203e-05,
848
+ "loss": 0.3619,
849
+ "step": 120
850
+ },
851
+ {
852
+ "epoch": 1.5414012738853504,
853
+ "grad_norm": 0.42883040333822336,
854
+ "learning_rate": 3.485091257608047e-05,
855
+ "loss": 0.3917,
856
+ "step": 121
857
+ },
858
+ {
859
+ "epoch": 1.5541401273885351,
860
+ "grad_norm": 0.3949540394409347,
861
+ "learning_rate": 3.473042049341474e-05,
862
+ "loss": 0.3795,
863
+ "step": 122
864
+ },
865
+ {
866
+ "epoch": 1.5668789808917198,
867
+ "grad_norm": 0.4613117397583663,
868
+ "learning_rate": 3.4608748367954064e-05,
869
+ "loss": 0.3996,
870
+ "step": 123
871
+ },
872
+ {
873
+ "epoch": 1.5796178343949046,
874
+ "grad_norm": 0.3602038452810683,
875
+ "learning_rate": 3.4485905946759965e-05,
876
+ "loss": 0.3716,
877
+ "step": 124
878
+ },
879
+ {
880
+ "epoch": 1.5923566878980893,
881
+ "grad_norm": 0.3931259570104483,
882
+ "learning_rate": 3.4361903070645484e-05,
883
+ "loss": 0.3847,
884
+ "step": 125
885
+ },
886
+ {
887
+ "epoch": 1.605095541401274,
888
+ "grad_norm": 0.4433362238715447,
889
+ "learning_rate": 3.423674967338681e-05,
890
+ "loss": 0.3824,
891
+ "step": 126
892
+ },
893
+ {
894
+ "epoch": 1.6178343949044587,
895
+ "grad_norm": 0.3897182981017782,
896
+ "learning_rate": 3.411045578092754e-05,
897
+ "loss": 0.382,
898
+ "step": 127
899
+ },
900
+ {
901
+ "epoch": 1.6305732484076434,
902
+ "grad_norm": 0.39580662710002257,
903
+ "learning_rate": 3.398303151057543e-05,
904
+ "loss": 0.3872,
905
+ "step": 128
906
+ },
907
+ {
908
+ "epoch": 1.643312101910828,
909
+ "grad_norm": 0.36389859578115363,
910
+ "learning_rate": 3.385448707019199e-05,
911
+ "loss": 0.393,
912
+ "step": 129
913
+ },
914
+ {
915
+ "epoch": 1.6560509554140128,
916
+ "grad_norm": 0.3760822336320154,
917
+ "learning_rate": 3.372483275737468e-05,
918
+ "loss": 0.4038,
919
+ "step": 130
920
+ },
921
+ {
922
+ "epoch": 1.6687898089171975,
923
+ "grad_norm": 0.40124915655150517,
924
+ "learning_rate": 3.359407895863199e-05,
925
+ "loss": 0.3828,
926
+ "step": 131
927
+ },
928
+ {
929
+ "epoch": 1.6815286624203822,
930
+ "grad_norm": 0.3492681667679969,
931
+ "learning_rate": 3.34622361485514e-05,
932
+ "loss": 0.3988,
933
+ "step": 132
934
+ },
935
+ {
936
+ "epoch": 1.694267515923567,
937
+ "grad_norm": 0.4359639458909479,
938
+ "learning_rate": 3.332931488896029e-05,
939
+ "loss": 0.4123,
940
+ "step": 133
941
+ },
942
+ {
943
+ "epoch": 1.7070063694267517,
944
+ "grad_norm": 0.35765716673817444,
945
+ "learning_rate": 3.319532582807977e-05,
946
+ "loss": 0.4025,
947
+ "step": 134
948
+ },
949
+ {
950
+ "epoch": 1.7197452229299364,
951
+ "grad_norm": 0.4181343414020294,
952
+ "learning_rate": 3.30602796996717e-05,
953
+ "loss": 0.3988,
954
+ "step": 135
955
+ },
956
+ {
957
+ "epoch": 1.732484076433121,
958
+ "grad_norm": 0.3671754645675639,
959
+ "learning_rate": 3.2924187322178865e-05,
960
+ "loss": 0.3866,
961
+ "step": 136
962
+ },
963
+ {
964
+ "epoch": 1.7452229299363058,
965
+ "grad_norm": 0.40706707942452913,
966
+ "learning_rate": 3.278705959785821e-05,
967
+ "loss": 0.4071,
968
+ "step": 137
969
+ },
970
+ {
971
+ "epoch": 1.7579617834394905,
972
+ "grad_norm": 0.40132966074561627,
973
+ "learning_rate": 3.2648907511907544e-05,
974
+ "loss": 0.3915,
975
+ "step": 138
976
+ },
977
+ {
978
+ "epoch": 1.7707006369426752,
979
+ "grad_norm": 0.356092810738192,
980
+ "learning_rate": 3.250974213158555e-05,
981
+ "loss": 0.3881,
982
+ "step": 139
983
+ },
984
+ {
985
+ "epoch": 1.78343949044586,
986
+ "grad_norm": 0.4088495881297229,
987
+ "learning_rate": 3.23695746053251e-05,
988
+ "loss": 0.3956,
989
+ "step": 140
990
+ },
991
+ {
992
+ "epoch": 1.7961783439490446,
993
+ "grad_norm": 0.3568174986894459,
994
+ "learning_rate": 3.222841616184025e-05,
995
+ "loss": 0.4264,
996
+ "step": 141
997
+ },
998
+ {
999
+ "epoch": 1.8089171974522293,
1000
+ "grad_norm": 0.3828743980988108,
1001
+ "learning_rate": 3.208627810922665e-05,
1002
+ "loss": 0.3934,
1003
+ "step": 142
1004
+ },
1005
+ {
1006
+ "epoch": 1.821656050955414,
1007
+ "grad_norm": 0.40695432826523725,
1008
+ "learning_rate": 3.194317183405573e-05,
1009
+ "loss": 0.3538,
1010
+ "step": 143
1011
+ },
1012
+ {
1013
+ "epoch": 1.8343949044585988,
1014
+ "grad_norm": 0.34980415470555404,
1015
+ "learning_rate": 3.1799108800462466e-05,
1016
+ "loss": 0.3751,
1017
+ "step": 144
1018
+ },
1019
+ {
1020
+ "epoch": 1.8471337579617835,
1021
+ "grad_norm": 0.4171552280293834,
1022
+ "learning_rate": 3.1654100549227024e-05,
1023
+ "loss": 0.3823,
1024
+ "step": 145
1025
+ },
1026
+ {
1027
+ "epoch": 1.8598726114649682,
1028
+ "grad_norm": 0.3856839605835017,
1029
+ "learning_rate": 3.1508158696850275e-05,
1030
+ "loss": 0.3732,
1031
+ "step": 146
1032
+ },
1033
+ {
1034
+ "epoch": 1.872611464968153,
1035
+ "grad_norm": 0.3534737799953714,
1036
+ "learning_rate": 3.136129493462312e-05,
1037
+ "loss": 0.4172,
1038
+ "step": 147
1039
+ },
1040
+ {
1041
+ "epoch": 1.8853503184713376,
1042
+ "grad_norm": 0.40251325811058225,
1043
+ "learning_rate": 3.121352102768998e-05,
1044
+ "loss": 0.3775,
1045
+ "step": 148
1046
+ },
1047
+ {
1048
+ "epoch": 1.8980891719745223,
1049
+ "grad_norm": 0.39323479639785724,
1050
+ "learning_rate": 3.106484881410628e-05,
1051
+ "loss": 0.3929,
1052
+ "step": 149
1053
+ },
1054
+ {
1055
+ "epoch": 1.910828025477707,
1056
+ "grad_norm": 0.3948424894576429,
1057
+ "learning_rate": 3.091529020389009e-05,
1058
+ "loss": 0.4098,
1059
+ "step": 150
1060
+ },
1061
+ {
1062
+ "epoch": 1.9235668789808917,
1063
+ "grad_norm": 0.3798104391658232,
1064
+ "learning_rate": 3.076485717806808e-05,
1065
+ "loss": 0.3865,
1066
+ "step": 151
1067
+ },
1068
+ {
1069
+ "epoch": 1.9363057324840764,
1070
+ "grad_norm": 0.42664388139031734,
1071
+ "learning_rate": 3.061356178771564e-05,
1072
+ "loss": 0.4049,
1073
+ "step": 152
1074
+ },
1075
+ {
1076
+ "epoch": 1.9490445859872612,
1077
+ "grad_norm": 0.38754913155372867,
1078
+ "learning_rate": 3.0461416152991555e-05,
1079
+ "loss": 0.3863,
1080
+ "step": 153
1081
+ },
1082
+ {
1083
+ "epoch": 1.9617834394904459,
1084
+ "grad_norm": 0.41477873550604283,
1085
+ "learning_rate": 3.0308432462167045e-05,
1086
+ "loss": 0.3887,
1087
+ "step": 154
1088
+ },
1089
+ {
1090
+ "epoch": 1.9745222929936306,
1091
+ "grad_norm": 0.4140640968780079,
1092
+ "learning_rate": 3.015462297064936e-05,
1093
+ "loss": 0.3936,
1094
+ "step": 155
1095
+ },
1096
+ {
1097
+ "epoch": 1.9872611464968153,
1098
+ "grad_norm": 0.38731104462679833,
1099
+ "learning_rate": 3.0000000000000004e-05,
1100
+ "loss": 0.383,
1101
+ "step": 156
1102
+ },
1103
+ {
1104
+ "epoch": 2.0,
1105
+ "grad_norm": 0.3728294872032267,
1106
+ "learning_rate": 2.98445759369477e-05,
1107
+ "loss": 0.3533,
1108
+ "step": 157
1109
+ },
1110
+ {
1111
+ "epoch": 2.0127388535031847,
1112
+ "grad_norm": 0.5326083329193402,
1113
+ "learning_rate": 2.9688363232396056e-05,
1114
+ "loss": 0.2746,
1115
+ "step": 158
1116
+ },
1117
+ {
1118
+ "epoch": 2.0254777070063694,
1119
+ "grad_norm": 0.4241037663468422,
1120
+ "learning_rate": 2.9531374400426158e-05,
1121
+ "loss": 0.2451,
1122
+ "step": 159
1123
+ },
1124
+ {
1125
+ "epoch": 2.038216560509554,
1126
+ "grad_norm": 0.5238076547115884,
1127
+ "learning_rate": 2.9373622017294075e-05,
1128
+ "loss": 0.2426,
1129
+ "step": 160
1130
+ },
1131
+ {
1132
+ "epoch": 2.050955414012739,
1133
+ "grad_norm": 0.43750634458426324,
1134
+ "learning_rate": 2.9215118720423375e-05,
1135
+ "loss": 0.2461,
1136
+ "step": 161
1137
+ },
1138
+ {
1139
+ "epoch": 2.0636942675159236,
1140
+ "grad_norm": 0.5426569564960375,
1141
+ "learning_rate": 2.9055877207392752e-05,
1142
+ "loss": 0.2396,
1143
+ "step": 162
1144
+ },
1145
+ {
1146
+ "epoch": 2.0764331210191083,
1147
+ "grad_norm": 0.4603442806084417,
1148
+ "learning_rate": 2.8895910234918828e-05,
1149
+ "loss": 0.2423,
1150
+ "step": 163
1151
+ },
1152
+ {
1153
+ "epoch": 2.089171974522293,
1154
+ "grad_norm": 0.4783694185949688,
1155
+ "learning_rate": 2.873523061783426e-05,
1156
+ "loss": 0.2522,
1157
+ "step": 164
1158
+ },
1159
+ {
1160
+ "epoch": 2.1019108280254777,
1161
+ "grad_norm": 0.4155346213843112,
1162
+ "learning_rate": 2.8573851228061084e-05,
1163
+ "loss": 0.2384,
1164
+ "step": 165
1165
+ },
1166
+ {
1167
+ "epoch": 2.1146496815286624,
1168
+ "grad_norm": 0.4032696957971972,
1169
+ "learning_rate": 2.8411784993579633e-05,
1170
+ "loss": 0.2505,
1171
+ "step": 166
1172
+ },
1173
+ {
1174
+ "epoch": 2.127388535031847,
1175
+ "grad_norm": 0.3898038039651499,
1176
+ "learning_rate": 2.8249044897392814e-05,
1177
+ "loss": 0.2701,
1178
+ "step": 167
1179
+ },
1180
+ {
1181
+ "epoch": 2.140127388535032,
1182
+ "grad_norm": 0.40575399978730575,
1183
+ "learning_rate": 2.80856439764861e-05,
1184
+ "loss": 0.2305,
1185
+ "step": 168
1186
+ },
1187
+ {
1188
+ "epoch": 2.1528662420382165,
1189
+ "grad_norm": 0.37004856559051524,
1190
+ "learning_rate": 2.792159532078314e-05,
1191
+ "loss": 0.2451,
1192
+ "step": 169
1193
+ },
1194
+ {
1195
+ "epoch": 2.1656050955414012,
1196
+ "grad_norm": 0.39808666444917457,
1197
+ "learning_rate": 2.77569120720971e-05,
1198
+ "loss": 0.258,
1199
+ "step": 170
1200
+ },
1201
+ {
1202
+ "epoch": 2.178343949044586,
1203
+ "grad_norm": 0.36955071620464847,
1204
+ "learning_rate": 2.7591607423077932e-05,
1205
+ "loss": 0.2292,
1206
+ "step": 171
1207
+ },
1208
+ {
1209
+ "epoch": 2.1910828025477707,
1210
+ "grad_norm": 0.336518551638344,
1211
+ "learning_rate": 2.7425694616155474e-05,
1212
+ "loss": 0.2477,
1213
+ "step": 172
1214
+ },
1215
+ {
1216
+ "epoch": 2.2038216560509554,
1217
+ "grad_norm": 0.4140957079330811,
1218
+ "learning_rate": 2.7259186942478656e-05,
1219
+ "loss": 0.2255,
1220
+ "step": 173
1221
+ },
1222
+ {
1223
+ "epoch": 2.21656050955414,
1224
+ "grad_norm": 0.32308874630904505,
1225
+ "learning_rate": 2.7092097740850712e-05,
1226
+ "loss": 0.237,
1227
+ "step": 174
1228
+ },
1229
+ {
1230
+ "epoch": 2.229299363057325,
1231
+ "grad_norm": 0.38917894801947195,
1232
+ "learning_rate": 2.692444039666066e-05,
1233
+ "loss": 0.2514,
1234
+ "step": 175
1235
+ },
1236
+ {
1237
+ "epoch": 2.2420382165605095,
1238
+ "grad_norm": 0.32626408176829913,
1239
+ "learning_rate": 2.6756228340810946e-05,
1240
+ "loss": 0.2317,
1241
+ "step": 176
1242
+ },
1243
+ {
1244
+ "epoch": 2.254777070063694,
1245
+ "grad_norm": 0.34013561380822094,
1246
+ "learning_rate": 2.6587475048641596e-05,
1247
+ "loss": 0.2553,
1248
+ "step": 177
1249
+ },
1250
+ {
1251
+ "epoch": 2.267515923566879,
1252
+ "grad_norm": 0.37726290374312765,
1253
+ "learning_rate": 2.6418194038850634e-05,
1254
+ "loss": 0.2438,
1255
+ "step": 178
1256
+ },
1257
+ {
1258
+ "epoch": 2.2802547770700636,
1259
+ "grad_norm": 0.3524165115081538,
1260
+ "learning_rate": 2.624839887241115e-05,
1261
+ "loss": 0.231,
1262
+ "step": 179
1263
+ },
1264
+ {
1265
+ "epoch": 2.2929936305732483,
1266
+ "grad_norm": 0.32629627151791846,
1267
+ "learning_rate": 2.607810315148494e-05,
1268
+ "loss": 0.2278,
1269
+ "step": 180
1270
+ },
1271
+ {
1272
+ "epoch": 2.305732484076433,
1273
+ "grad_norm": 0.41583661942751143,
1274
+ "learning_rate": 2.5907320518332827e-05,
1275
+ "loss": 0.242,
1276
+ "step": 181
1277
+ },
1278
+ {
1279
+ "epoch": 2.3184713375796178,
1280
+ "grad_norm": 0.30326537675872595,
1281
+ "learning_rate": 2.5736064654221808e-05,
1282
+ "loss": 0.2488,
1283
+ "step": 182
1284
+ },
1285
+ {
1286
+ "epoch": 2.3312101910828025,
1287
+ "grad_norm": 0.3469983421489293,
1288
+ "learning_rate": 2.5564349278329056e-05,
1289
+ "loss": 0.2512,
1290
+ "step": 183
1291
+ },
1292
+ {
1293
+ "epoch": 2.343949044585987,
1294
+ "grad_norm": 0.34347379957192603,
1295
+ "learning_rate": 2.539218814664288e-05,
1296
+ "loss": 0.2477,
1297
+ "step": 184
1298
+ },
1299
+ {
1300
+ "epoch": 2.356687898089172,
1301
+ "grad_norm": 0.28641537582695287,
1302
+ "learning_rate": 2.521959505086075e-05,
1303
+ "loss": 0.2362,
1304
+ "step": 185
1305
+ },
1306
+ {
1307
+ "epoch": 2.3694267515923566,
1308
+ "grad_norm": 0.3690167657528722,
1309
+ "learning_rate": 2.5046583817284437e-05,
1310
+ "loss": 0.2587,
1311
+ "step": 186
1312
+ },
1313
+ {
1314
+ "epoch": 2.3821656050955413,
1315
+ "grad_norm": 0.2961653899116515,
1316
+ "learning_rate": 2.487316830571244e-05,
1317
+ "loss": 0.2311,
1318
+ "step": 187
1319
+ },
1320
+ {
1321
+ "epoch": 2.394904458598726,
1322
+ "grad_norm": 0.29521846638552274,
1323
+ "learning_rate": 2.4699362408329646e-05,
1324
+ "loss": 0.2327,
1325
+ "step": 188
1326
+ },
1327
+ {
1328
+ "epoch": 2.4076433121019107,
1329
+ "grad_norm": 0.29770180340093805,
1330
+ "learning_rate": 2.4525180048594452e-05,
1331
+ "loss": 0.2544,
1332
+ "step": 189
1333
+ },
1334
+ {
1335
+ "epoch": 2.4203821656050954,
1336
+ "grad_norm": 0.2913325444490102,
1337
+ "learning_rate": 2.435063518012335e-05,
1338
+ "loss": 0.2247,
1339
+ "step": 190
1340
+ },
1341
+ {
1342
+ "epoch": 2.43312101910828,
1343
+ "grad_norm": 0.3384396420749506,
1344
+ "learning_rate": 2.4175741785573177e-05,
1345
+ "loss": 0.2379,
1346
+ "step": 191
1347
+ },
1348
+ {
1349
+ "epoch": 2.445859872611465,
1350
+ "grad_norm": 0.30159462621181515,
1351
+ "learning_rate": 2.4000513875520892e-05,
1352
+ "loss": 0.2152,
1353
+ "step": 192
1354
+ },
1355
+ {
1356
+ "epoch": 2.4585987261146496,
1357
+ "grad_norm": 0.29567483557148394,
1358
+ "learning_rate": 2.3824965487341247e-05,
1359
+ "loss": 0.2402,
1360
+ "step": 193
1361
+ },
1362
+ {
1363
+ "epoch": 2.4713375796178343,
1364
+ "grad_norm": 0.2769989858820714,
1365
+ "learning_rate": 2.3649110684082258e-05,
1366
+ "loss": 0.2597,
1367
+ "step": 194
1368
+ },
1369
+ {
1370
+ "epoch": 2.484076433121019,
1371
+ "grad_norm": 0.29336394029366775,
1372
+ "learning_rate": 2.3472963553338614e-05,
1373
+ "loss": 0.2273,
1374
+ "step": 195
1375
+ },
1376
+ {
1377
+ "epoch": 2.4968152866242037,
1378
+ "grad_norm": 0.2838756230989672,
1379
+ "learning_rate": 2.3296538206123134e-05,
1380
+ "loss": 0.2389,
1381
+ "step": 196
1382
+ },
1383
+ {
1384
+ "epoch": 2.5095541401273884,
1385
+ "grad_norm": 0.2626418361521103,
1386
+ "learning_rate": 2.311984877573636e-05,
1387
+ "loss": 0.2352,
1388
+ "step": 197
1389
+ },
1390
+ {
1391
+ "epoch": 2.522292993630573,
1392
+ "grad_norm": 0.26718958075760085,
1393
+ "learning_rate": 2.2942909416634326e-05,
1394
+ "loss": 0.2357,
1395
+ "step": 198
1396
+ },
1397
+ {
1398
+ "epoch": 2.535031847133758,
1399
+ "grad_norm": 0.2845956555682719,
1400
+ "learning_rate": 2.2765734303294666e-05,
1401
+ "loss": 0.2371,
1402
+ "step": 199
1403
+ },
1404
+ {
1405
+ "epoch": 2.5477707006369426,
1406
+ "grad_norm": 0.2640776304490595,
1407
+ "learning_rate": 2.2588337629081107e-05,
1408
+ "loss": 0.2384,
1409
+ "step": 200
1410
+ },
1411
+ {
1412
+ "epoch": 2.5605095541401273,
1413
+ "grad_norm": 0.2848898116027439,
1414
+ "learning_rate": 2.2410733605106462e-05,
1415
+ "loss": 0.2217,
1416
+ "step": 201
1417
+ },
1418
+ {
1419
+ "epoch": 2.573248407643312,
1420
+ "grad_norm": 0.2812237619598167,
1421
+ "learning_rate": 2.2232936459094158e-05,
1422
+ "loss": 0.2403,
1423
+ "step": 202
1424
+ },
1425
+ {
1426
+ "epoch": 2.5859872611464967,
1427
+ "grad_norm": 0.29409588083070315,
1428
+ "learning_rate": 2.205496043423849e-05,
1429
+ "loss": 0.2266,
1430
+ "step": 203
1431
+ },
1432
+ {
1433
+ "epoch": 2.5987261146496814,
1434
+ "grad_norm": 0.32256548006968067,
1435
+ "learning_rate": 2.1876819788063586e-05,
1436
+ "loss": 0.2547,
1437
+ "step": 204
1438
+ },
1439
+ {
1440
+ "epoch": 2.611464968152866,
1441
+ "grad_norm": 0.32874439082160606,
1442
+ "learning_rate": 2.16985287912813e-05,
1443
+ "loss": 0.2261,
1444
+ "step": 205
1445
+ },
1446
+ {
1447
+ "epoch": 2.624203821656051,
1448
+ "grad_norm": 0.3266191143621023,
1449
+ "learning_rate": 2.1520101726647922e-05,
1450
+ "loss": 0.2321,
1451
+ "step": 206
1452
+ },
1453
+ {
1454
+ "epoch": 2.6369426751592355,
1455
+ "grad_norm": 0.30563025473541944,
1456
+ "learning_rate": 2.1341552887820048e-05,
1457
+ "loss": 0.2394,
1458
+ "step": 207
1459
+ },
1460
+ {
1461
+ "epoch": 2.6496815286624202,
1462
+ "grad_norm": 0.2869377241233248,
1463
+ "learning_rate": 2.1162896578209517e-05,
1464
+ "loss": 0.2409,
1465
+ "step": 208
1466
+ },
1467
+ {
1468
+ "epoch": 2.662420382165605,
1469
+ "grad_norm": 0.30715126599292863,
1470
+ "learning_rate": 2.0984147109837564e-05,
1471
+ "loss": 0.2423,
1472
+ "step": 209
1473
+ },
1474
+ {
1475
+ "epoch": 2.6751592356687897,
1476
+ "grad_norm": 0.2894859890668604,
1477
+ "learning_rate": 2.0805318802188307e-05,
1478
+ "loss": 0.264,
1479
+ "step": 210
1480
+ },
1481
+ {
1482
+ "epoch": 2.6878980891719744,
1483
+ "grad_norm": 0.28673853671855953,
1484
+ "learning_rate": 2.0626425981061608e-05,
1485
+ "loss": 0.2632,
1486
+ "step": 211
1487
+ },
1488
+ {
1489
+ "epoch": 2.700636942675159,
1490
+ "grad_norm": 0.26547959114709113,
1491
+ "learning_rate": 2.0447482977425465e-05,
1492
+ "loss": 0.2165,
1493
+ "step": 212
1494
+ },
1495
+ {
1496
+ "epoch": 2.713375796178344,
1497
+ "grad_norm": 0.29689259712489163,
1498
+ "learning_rate": 2.0268504126267952e-05,
1499
+ "loss": 0.254,
1500
+ "step": 213
1501
+ },
1502
+ {
1503
+ "epoch": 2.7261146496815285,
1504
+ "grad_norm": 0.30794580549232115,
1505
+ "learning_rate": 2.008950376544887e-05,
1506
+ "loss": 0.234,
1507
+ "step": 214
1508
+ },
1509
+ {
1510
+ "epoch": 2.738853503184713,
1511
+ "grad_norm": 0.29029164891637815,
1512
+ "learning_rate": 1.9910496234551132e-05,
1513
+ "loss": 0.2423,
1514
+ "step": 215
1515
+ },
1516
+ {
1517
+ "epoch": 2.7515923566878984,
1518
+ "grad_norm": 0.28445043266611614,
1519
+ "learning_rate": 1.9731495873732055e-05,
1520
+ "loss": 0.2399,
1521
+ "step": 216
1522
+ },
1523
+ {
1524
+ "epoch": 2.7643312101910826,
1525
+ "grad_norm": 0.28463685628918173,
1526
+ "learning_rate": 1.9552517022574542e-05,
1527
+ "loss": 0.2246,
1528
+ "step": 217
1529
+ },
1530
+ {
1531
+ "epoch": 2.777070063694268,
1532
+ "grad_norm": 0.29242221813409025,
1533
+ "learning_rate": 1.93735740189384e-05,
1534
+ "loss": 0.2276,
1535
+ "step": 218
1536
+ },
1537
+ {
1538
+ "epoch": 2.789808917197452,
1539
+ "grad_norm": 0.309031276466736,
1540
+ "learning_rate": 1.9194681197811703e-05,
1541
+ "loss": 0.2522,
1542
+ "step": 219
1543
+ },
1544
+ {
1545
+ "epoch": 2.802547770700637,
1546
+ "grad_norm": 0.2769431867934569,
1547
+ "learning_rate": 1.901585289016244e-05,
1548
+ "loss": 0.2247,
1549
+ "step": 220
1550
+ },
1551
+ {
1552
+ "epoch": 2.8152866242038215,
1553
+ "grad_norm": 0.27968185341237434,
1554
+ "learning_rate": 1.8837103421790486e-05,
1555
+ "loss": 0.2227,
1556
+ "step": 221
1557
+ },
1558
+ {
1559
+ "epoch": 2.8280254777070066,
1560
+ "grad_norm": 0.2888083036658392,
1561
+ "learning_rate": 1.8658447112179952e-05,
1562
+ "loss": 0.2374,
1563
+ "step": 222
1564
+ },
1565
+ {
1566
+ "epoch": 2.840764331210191,
1567
+ "grad_norm": 0.28582364336230653,
1568
+ "learning_rate": 1.8479898273352084e-05,
1569
+ "loss": 0.2536,
1570
+ "step": 223
1571
+ },
1572
+ {
1573
+ "epoch": 2.853503184713376,
1574
+ "grad_norm": 0.2935452815809231,
1575
+ "learning_rate": 1.83014712087187e-05,
1576
+ "loss": 0.2181,
1577
+ "step": 224
1578
+ },
1579
+ {
1580
+ "epoch": 2.8662420382165603,
1581
+ "grad_norm": 0.30167338921175085,
1582
+ "learning_rate": 1.8123180211936417e-05,
1583
+ "loss": 0.2279,
1584
+ "step": 225
1585
+ },
1586
+ {
1587
+ "epoch": 2.8789808917197455,
1588
+ "grad_norm": 0.28313988031191023,
1589
+ "learning_rate": 1.794503956576152e-05,
1590
+ "loss": 0.2136,
1591
+ "step": 226
1592
+ },
1593
+ {
1594
+ "epoch": 2.8917197452229297,
1595
+ "grad_norm": 0.3036257204617132,
1596
+ "learning_rate": 1.776706354090585e-05,
1597
+ "loss": 0.2431,
1598
+ "step": 227
1599
+ },
1600
+ {
1601
+ "epoch": 2.904458598726115,
1602
+ "grad_norm": 0.28848979346617437,
1603
+ "learning_rate": 1.758926639489354e-05,
1604
+ "loss": 0.2423,
1605
+ "step": 228
1606
+ },
1607
+ {
1608
+ "epoch": 2.917197452229299,
1609
+ "grad_norm": 0.272764590445302,
1610
+ "learning_rate": 1.7411662370918893e-05,
1611
+ "loss": 0.2301,
1612
+ "step": 229
1613
+ },
1614
+ {
1615
+ "epoch": 2.9299363057324843,
1616
+ "grad_norm": 0.3047366111218703,
1617
+ "learning_rate": 1.7234265696705344e-05,
1618
+ "loss": 0.2548,
1619
+ "step": 230
1620
+ },
1621
+ {
1622
+ "epoch": 2.9426751592356686,
1623
+ "grad_norm": 0.29141657758684564,
1624
+ "learning_rate": 1.7057090583365678e-05,
1625
+ "loss": 0.2363,
1626
+ "step": 231
1627
+ },
1628
+ {
1629
+ "epoch": 2.9554140127388537,
1630
+ "grad_norm": 0.2905701392836334,
1631
+ "learning_rate": 1.6880151224263646e-05,
1632
+ "loss": 0.2429,
1633
+ "step": 232
1634
+ },
1635
+ {
1636
+ "epoch": 2.968152866242038,
1637
+ "grad_norm": 0.3055726799006245,
1638
+ "learning_rate": 1.6703461793876876e-05,
1639
+ "loss": 0.241,
1640
+ "step": 233
1641
+ },
1642
+ {
1643
+ "epoch": 2.980891719745223,
1644
+ "grad_norm": 0.27702453651096914,
1645
+ "learning_rate": 1.6527036446661396e-05,
1646
+ "loss": 0.2785,
1647
+ "step": 234
1648
+ },
1649
+ {
1650
+ "epoch": 2.9936305732484074,
1651
+ "grad_norm": 0.2958171896040813,
1652
+ "learning_rate": 1.635088931591775e-05,
1653
+ "loss": 0.2494,
1654
+ "step": 235
1655
+ },
1656
+ {
1657
+ "epoch": 3.0063694267515926,
1658
+ "grad_norm": 0.3339252283259285,
1659
+ "learning_rate": 1.6175034512658753e-05,
1660
+ "loss": 0.1814,
1661
+ "step": 236
1662
+ },
1663
+ {
1664
+ "epoch": 3.0191082802547773,
1665
+ "grad_norm": 0.33323829415209344,
1666
+ "learning_rate": 1.5999486124479115e-05,
1667
+ "loss": 0.1519,
1668
+ "step": 237
1669
+ },
1670
+ {
1671
+ "epoch": 3.031847133757962,
1672
+ "grad_norm": 0.263725808295815,
1673
+ "learning_rate": 1.5824258214426833e-05,
1674
+ "loss": 0.1756,
1675
+ "step": 238
1676
+ },
1677
+ {
1678
+ "epoch": 3.0445859872611467,
1679
+ "grad_norm": 0.3263063861580799,
1680
+ "learning_rate": 1.5649364819876655e-05,
1681
+ "loss": 0.1681,
1682
+ "step": 239
1683
+ },
1684
+ {
1685
+ "epoch": 3.0573248407643314,
1686
+ "grad_norm": 0.3332704699662475,
1687
+ "learning_rate": 1.547481995140556e-05,
1688
+ "loss": 0.1533,
1689
+ "step": 240
1690
+ },
1691
+ {
1692
+ "epoch": 3.070063694267516,
1693
+ "grad_norm": 0.3290607170386027,
1694
+ "learning_rate": 1.5300637591670357e-05,
1695
+ "loss": 0.1491,
1696
+ "step": 241
1697
+ },
1698
+ {
1699
+ "epoch": 3.082802547770701,
1700
+ "grad_norm": 0.27626642274366475,
1701
+ "learning_rate": 1.5126831694287564e-05,
1702
+ "loss": 0.1386,
1703
+ "step": 242
1704
+ },
1705
+ {
1706
+ "epoch": 3.0955414012738856,
1707
+ "grad_norm": 0.24383948766661087,
1708
+ "learning_rate": 1.4953416182715566e-05,
1709
+ "loss": 0.1454,
1710
+ "step": 243
1711
+ },
1712
+ {
1713
+ "epoch": 3.1082802547770703,
1714
+ "grad_norm": 0.2565144051732309,
1715
+ "learning_rate": 1.478040494913926e-05,
1716
+ "loss": 0.1297,
1717
+ "step": 244
1718
+ },
1719
+ {
1720
+ "epoch": 3.121019108280255,
1721
+ "grad_norm": 0.2854715023231579,
1722
+ "learning_rate": 1.460781185335713e-05,
1723
+ "loss": 0.1611,
1724
+ "step": 245
1725
+ },
1726
+ {
1727
+ "epoch": 3.1337579617834397,
1728
+ "grad_norm": 0.23688489896659726,
1729
+ "learning_rate": 1.443565072167095e-05,
1730
+ "loss": 0.1537,
1731
+ "step": 246
1732
+ },
1733
+ {
1734
+ "epoch": 3.1464968152866244,
1735
+ "grad_norm": 0.23498017729946113,
1736
+ "learning_rate": 1.4263935345778202e-05,
1737
+ "loss": 0.128,
1738
+ "step": 247
1739
+ },
1740
+ {
1741
+ "epoch": 3.159235668789809,
1742
+ "grad_norm": 0.25210838232322463,
1743
+ "learning_rate": 1.409267948166718e-05,
1744
+ "loss": 0.1529,
1745
+ "step": 248
1746
+ },
1747
+ {
1748
+ "epoch": 3.171974522292994,
1749
+ "grad_norm": 0.23817238492456902,
1750
+ "learning_rate": 1.3921896848515064e-05,
1751
+ "loss": 0.1431,
1752
+ "step": 249
1753
+ },
1754
+ {
1755
+ "epoch": 3.1847133757961785,
1756
+ "grad_norm": 0.23878818674795252,
1757
+ "learning_rate": 1.3751601127588849e-05,
1758
+ "loss": 0.1511,
1759
+ "step": 250
1760
+ },
1761
+ {
1762
+ "epoch": 3.1974522292993632,
1763
+ "grad_norm": 0.2253169671801177,
1764
+ "learning_rate": 1.3581805961149371e-05,
1765
+ "loss": 0.171,
1766
+ "step": 251
1767
+ },
1768
+ {
1769
+ "epoch": 3.210191082802548,
1770
+ "grad_norm": 0.2399433284877313,
1771
+ "learning_rate": 1.341252495135841e-05,
1772
+ "loss": 0.125,
1773
+ "step": 252
1774
+ },
1775
+ {
1776
+ "epoch": 3.2229299363057327,
1777
+ "grad_norm": 0.2482037509566062,
1778
+ "learning_rate": 1.324377165918906e-05,
1779
+ "loss": 0.1332,
1780
+ "step": 253
1781
+ },
1782
+ {
1783
+ "epoch": 3.2356687898089174,
1784
+ "grad_norm": 0.21581244708702438,
1785
+ "learning_rate": 1.3075559603339354e-05,
1786
+ "loss": 0.1479,
1787
+ "step": 254
1788
+ },
1789
+ {
1790
+ "epoch": 3.248407643312102,
1791
+ "grad_norm": 0.2075414978318476,
1792
+ "learning_rate": 1.2907902259149287e-05,
1793
+ "loss": 0.1289,
1794
+ "step": 255
1795
+ },
1796
+ {
1797
+ "epoch": 3.261146496815287,
1798
+ "grad_norm": 0.24464034758512393,
1799
+ "learning_rate": 1.274081305752135e-05,
1800
+ "loss": 0.1505,
1801
+ "step": 256
1802
+ },
1803
+ {
1804
+ "epoch": 3.2738853503184715,
1805
+ "grad_norm": 0.23054729756063358,
1806
+ "learning_rate": 1.2574305383844528e-05,
1807
+ "loss": 0.1534,
1808
+ "step": 257
1809
+ },
1810
+ {
1811
+ "epoch": 3.286624203821656,
1812
+ "grad_norm": 0.2020043798081996,
1813
+ "learning_rate": 1.2408392576922075e-05,
1814
+ "loss": 0.1381,
1815
+ "step": 258
1816
+ },
1817
+ {
1818
+ "epoch": 3.299363057324841,
1819
+ "grad_norm": 0.22799906894946498,
1820
+ "learning_rate": 1.2243087927902905e-05,
1821
+ "loss": 0.1719,
1822
+ "step": 259
1823
+ },
1824
+ {
1825
+ "epoch": 3.3121019108280256,
1826
+ "grad_norm": 0.24533082757636485,
1827
+ "learning_rate": 1.2078404679216864e-05,
1828
+ "loss": 0.1609,
1829
+ "step": 260
1830
+ },
1831
+ {
1832
+ "epoch": 3.3248407643312103,
1833
+ "grad_norm": 0.20714305709401984,
1834
+ "learning_rate": 1.1914356023513904e-05,
1835
+ "loss": 0.1179,
1836
+ "step": 261
1837
+ },
1838
+ {
1839
+ "epoch": 3.337579617834395,
1840
+ "grad_norm": 0.21668069968026532,
1841
+ "learning_rate": 1.1750955102607193e-05,
1842
+ "loss": 0.1647,
1843
+ "step": 262
1844
+ },
1845
+ {
1846
+ "epoch": 3.3503184713375798,
1847
+ "grad_norm": 0.2184496195596042,
1848
+ "learning_rate": 1.1588215006420374e-05,
1849
+ "loss": 0.1553,
1850
+ "step": 263
1851
+ },
1852
+ {
1853
+ "epoch": 3.3630573248407645,
1854
+ "grad_norm": 0.20337547245152646,
1855
+ "learning_rate": 1.1426148771938915e-05,
1856
+ "loss": 0.1315,
1857
+ "step": 264
1858
+ },
1859
+ {
1860
+ "epoch": 3.375796178343949,
1861
+ "grad_norm": 0.2037566957772752,
1862
+ "learning_rate": 1.1264769382165748e-05,
1863
+ "loss": 0.1539,
1864
+ "step": 265
1865
+ },
1866
+ {
1867
+ "epoch": 3.388535031847134,
1868
+ "grad_norm": 0.20032474440683493,
1869
+ "learning_rate": 1.110408976508118e-05,
1870
+ "loss": 0.1215,
1871
+ "step": 266
1872
+ },
1873
+ {
1874
+ "epoch": 3.4012738853503186,
1875
+ "grad_norm": 0.2136164552846967,
1876
+ "learning_rate": 1.094412279260726e-05,
1877
+ "loss": 0.1428,
1878
+ "step": 267
1879
+ },
1880
+ {
1881
+ "epoch": 3.4140127388535033,
1882
+ "grad_norm": 0.21177053858403921,
1883
+ "learning_rate": 1.0784881279576635e-05,
1884
+ "loss": 0.1595,
1885
+ "step": 268
1886
+ },
1887
+ {
1888
+ "epoch": 3.426751592356688,
1889
+ "grad_norm": 0.2029892437448442,
1890
+ "learning_rate": 1.0626377982705929e-05,
1891
+ "loss": 0.1383,
1892
+ "step": 269
1893
+ },
1894
+ {
1895
+ "epoch": 3.4394904458598727,
1896
+ "grad_norm": 0.20919896892050535,
1897
+ "learning_rate": 1.0468625599573842e-05,
1898
+ "loss": 0.1424,
1899
+ "step": 270
1900
+ },
1901
+ {
1902
+ "epoch": 3.4522292993630574,
1903
+ "grad_norm": 0.2070233782186885,
1904
+ "learning_rate": 1.0311636767603952e-05,
1905
+ "loss": 0.1467,
1906
+ "step": 271
1907
+ },
1908
+ {
1909
+ "epoch": 3.464968152866242,
1910
+ "grad_norm": 0.19702950998811397,
1911
+ "learning_rate": 1.0155424063052306e-05,
1912
+ "loss": 0.1341,
1913
+ "step": 272
1914
+ },
1915
+ {
1916
+ "epoch": 3.477707006369427,
1917
+ "grad_norm": 0.2063307034689398,
1918
+ "learning_rate": 1.0000000000000006e-05,
1919
+ "loss": 0.1333,
1920
+ "step": 273
1921
+ },
1922
+ {
1923
+ "epoch": 3.4904458598726116,
1924
+ "grad_norm": 0.20719963685656465,
1925
+ "learning_rate": 9.84537702935065e-06,
1926
+ "loss": 0.1388,
1927
+ "step": 274
1928
+ },
1929
+ {
1930
+ "epoch": 3.5031847133757963,
1931
+ "grad_norm": 0.215076053264252,
1932
+ "learning_rate": 9.691567537832964e-06,
1933
+ "loss": 0.1382,
1934
+ "step": 275
1935
+ },
1936
+ {
1937
+ "epoch": 3.515923566878981,
1938
+ "grad_norm": 0.2088643606927074,
1939
+ "learning_rate": 9.538583847008452e-06,
1940
+ "loss": 0.138,
1941
+ "step": 276
1942
+ },
1943
+ {
1944
+ "epoch": 3.5286624203821657,
1945
+ "grad_norm": 0.19464590100491783,
1946
+ "learning_rate": 9.386438212284372e-06,
1947
+ "loss": 0.1602,
1948
+ "step": 277
1949
+ },
1950
+ {
1951
+ "epoch": 3.5414012738853504,
1952
+ "grad_norm": 0.19748474073166788,
1953
+ "learning_rate": 9.235142821931928e-06,
1954
+ "loss": 0.1205,
1955
+ "step": 278
1956
+ },
1957
+ {
1958
+ "epoch": 3.554140127388535,
1959
+ "grad_norm": 0.20462175920146233,
1960
+ "learning_rate": 9.084709796109907e-06,
1961
+ "loss": 0.1243,
1962
+ "step": 279
1963
+ },
1964
+ {
1965
+ "epoch": 3.56687898089172,
1966
+ "grad_norm": 0.19781750645119578,
1967
+ "learning_rate": 8.93515118589373e-06,
1968
+ "loss": 0.1184,
1969
+ "step": 280
1970
+ },
1971
+ {
1972
+ "epoch": 3.5796178343949046,
1973
+ "grad_norm": 0.1932570190163277,
1974
+ "learning_rate": 8.786478972310023e-06,
1975
+ "loss": 0.1402,
1976
+ "step": 281
1977
+ },
1978
+ {
1979
+ "epoch": 3.5923566878980893,
1980
+ "grad_norm": 0.2130457950220486,
1981
+ "learning_rate": 8.638705065376887e-06,
1982
+ "loss": 0.1572,
1983
+ "step": 282
1984
+ },
1985
+ {
1986
+ "epoch": 3.605095541401274,
1987
+ "grad_norm": 0.21977855143398284,
1988
+ "learning_rate": 8.491841303149728e-06,
1989
+ "loss": 0.1666,
1990
+ "step": 283
1991
+ },
1992
+ {
1993
+ "epoch": 3.6178343949044587,
1994
+ "grad_norm": 0.19809617527946968,
1995
+ "learning_rate": 8.345899450772975e-06,
1996
+ "loss": 0.1423,
1997
+ "step": 284
1998
+ },
1999
+ {
2000
+ "epoch": 3.6305732484076434,
2001
+ "grad_norm": 0.18690090576844234,
2002
+ "learning_rate": 8.200891199537549e-06,
2003
+ "loss": 0.1364,
2004
+ "step": 285
2005
+ },
2006
+ {
2007
+ "epoch": 3.643312101910828,
2008
+ "grad_norm": 0.18446815748564052,
2009
+ "learning_rate": 8.056828165944282e-06,
2010
+ "loss": 0.1276,
2011
+ "step": 286
2012
+ },
2013
+ {
2014
+ "epoch": 3.656050955414013,
2015
+ "grad_norm": 0.19914110063591683,
2016
+ "learning_rate": 7.913721890773354e-06,
2017
+ "loss": 0.1452,
2018
+ "step": 287
2019
+ },
2020
+ {
2021
+ "epoch": 3.6687898089171975,
2022
+ "grad_norm": 0.19784197594445632,
2023
+ "learning_rate": 7.771583838159756e-06,
2024
+ "loss": 0.1393,
2025
+ "step": 288
2026
+ },
2027
+ {
2028
+ "epoch": 3.6815286624203822,
2029
+ "grad_norm": 0.18779371172725068,
2030
+ "learning_rate": 7.630425394674903e-06,
2031
+ "loss": 0.1486,
2032
+ "step": 289
2033
+ },
2034
+ {
2035
+ "epoch": 3.694267515923567,
2036
+ "grad_norm": 0.19091202103101299,
2037
+ "learning_rate": 7.49025786841445e-06,
2038
+ "loss": 0.1532,
2039
+ "step": 290
2040
+ },
2041
+ {
2042
+ "epoch": 3.7070063694267517,
2043
+ "grad_norm": 0.19572576915772363,
2044
+ "learning_rate": 7.3510924880924575e-06,
2045
+ "loss": 0.1498,
2046
+ "step": 291
2047
+ },
2048
+ {
2049
+ "epoch": 3.7197452229299364,
2050
+ "grad_norm": 0.19383886127171826,
2051
+ "learning_rate": 7.212940402141808e-06,
2052
+ "loss": 0.1243,
2053
+ "step": 292
2054
+ },
2055
+ {
2056
+ "epoch": 3.732484076433121,
2057
+ "grad_norm": 0.18951075996073852,
2058
+ "learning_rate": 7.075812677821145e-06,
2059
+ "loss": 0.1408,
2060
+ "step": 293
2061
+ },
2062
+ {
2063
+ "epoch": 3.745222929936306,
2064
+ "grad_norm": 0.18725185654705204,
2065
+ "learning_rate": 6.939720300328303e-06,
2066
+ "loss": 0.1445,
2067
+ "step": 294
2068
+ },
2069
+ {
2070
+ "epoch": 3.7579617834394905,
2071
+ "grad_norm": 0.1923854581483771,
2072
+ "learning_rate": 6.8046741719202385e-06,
2073
+ "loss": 0.1322,
2074
+ "step": 295
2075
+ },
2076
+ {
2077
+ "epoch": 3.770700636942675,
2078
+ "grad_norm": 0.19857273274849133,
2079
+ "learning_rate": 6.67068511103971e-06,
2080
+ "loss": 0.1472,
2081
+ "step": 296
2082
+ },
2083
+ {
2084
+ "epoch": 3.78343949044586,
2085
+ "grad_norm": 0.18958112392058588,
2086
+ "learning_rate": 6.537763851448593e-06,
2087
+ "loss": 0.1188,
2088
+ "step": 297
2089
+ },
2090
+ {
2091
+ "epoch": 3.7961783439490446,
2092
+ "grad_norm": 0.1754562981627312,
2093
+ "learning_rate": 6.4059210413680175e-06,
2094
+ "loss": 0.1352,
2095
+ "step": 298
2096
+ },
2097
+ {
2098
+ "epoch": 3.8089171974522293,
2099
+ "grad_norm": 0.1827948964704896,
2100
+ "learning_rate": 6.275167242625331e-06,
2101
+ "loss": 0.1343,
2102
+ "step": 299
2103
+ },
2104
+ {
2105
+ "epoch": 3.821656050955414,
2106
+ "grad_norm": 0.17640473509943577,
2107
+ "learning_rate": 6.145512929808013e-06,
2108
+ "loss": 0.1463,
2109
+ "step": 300
2110
+ },
2111
+ {
2112
+ "epoch": 3.8343949044585988,
2113
+ "grad_norm": 0.19352956970296584,
2114
+ "learning_rate": 6.016968489424572e-06,
2115
+ "loss": 0.136,
2116
+ "step": 301
2117
+ },
2118
+ {
2119
+ "epoch": 3.8471337579617835,
2120
+ "grad_norm": 0.17765227219483207,
2121
+ "learning_rate": 5.889544219072465e-06,
2122
+ "loss": 0.1371,
2123
+ "step": 302
2124
+ },
2125
+ {
2126
+ "epoch": 3.859872611464968,
2127
+ "grad_norm": 0.17378510781197484,
2128
+ "learning_rate": 5.7632503266131925e-06,
2129
+ "loss": 0.1508,
2130
+ "step": 303
2131
+ },
2132
+ {
2133
+ "epoch": 3.872611464968153,
2134
+ "grad_norm": 0.1754866346893611,
2135
+ "learning_rate": 5.638096929354522e-06,
2136
+ "loss": 0.1371,
2137
+ "step": 304
2138
+ },
2139
+ {
2140
+ "epoch": 3.8853503184713376,
2141
+ "grad_norm": 0.17990543007311735,
2142
+ "learning_rate": 5.514094053240035e-06,
2143
+ "loss": 0.1504,
2144
+ "step": 305
2145
+ },
2146
+ {
2147
+ "epoch": 3.8980891719745223,
2148
+ "grad_norm": 0.1767314352665911,
2149
+ "learning_rate": 5.39125163204594e-06,
2150
+ "loss": 0.1384,
2151
+ "step": 306
2152
+ },
2153
+ {
2154
+ "epoch": 3.910828025477707,
2155
+ "grad_norm": 0.18388289790416432,
2156
+ "learning_rate": 5.269579506585259e-06,
2157
+ "loss": 0.125,
2158
+ "step": 307
2159
+ },
2160
+ {
2161
+ "epoch": 3.9235668789808917,
2162
+ "grad_norm": 0.18709650841448822,
2163
+ "learning_rate": 5.149087423919541e-06,
2164
+ "loss": 0.1719,
2165
+ "step": 308
2166
+ },
2167
+ {
2168
+ "epoch": 3.9363057324840764,
2169
+ "grad_norm": 0.1765027020500491,
2170
+ "learning_rate": 5.029785036577976e-06,
2171
+ "loss": 0.1365,
2172
+ "step": 309
2173
+ },
2174
+ {
2175
+ "epoch": 3.949044585987261,
2176
+ "grad_norm": 0.1740407642246799,
2177
+ "learning_rate": 4.911681901784198e-06,
2178
+ "loss": 0.1455,
2179
+ "step": 310
2180
+ },
2181
+ {
2182
+ "epoch": 3.961783439490446,
2183
+ "grad_norm": 0.17955395869877427,
2184
+ "learning_rate": 4.794787480690597e-06,
2185
+ "loss": 0.135,
2186
+ "step": 311
2187
+ },
2188
+ {
2189
+ "epoch": 3.9745222929936306,
2190
+ "grad_norm": 0.18507428221903918,
2191
+ "learning_rate": 4.679111137620442e-06,
2192
+ "loss": 0.1575,
2193
+ "step": 312
2194
+ },
2195
+ {
2196
+ "epoch": 3.9872611464968153,
2197
+ "grad_norm": 0.1805033770468359,
2198
+ "learning_rate": 4.5646621393177e-06,
2199
+ "loss": 0.1388,
2200
+ "step": 313
2201
+ },
2202
+ {
2203
+ "epoch": 4.0,
2204
+ "grad_norm": 0.16048437681749295,
2205
+ "learning_rate": 4.451449654204685e-06,
2206
+ "loss": 0.1153,
2207
+ "step": 314
2208
+ },
2209
+ {
2210
+ "epoch": 4.012738853503185,
2211
+ "grad_norm": 0.24959599645461245,
2212
+ "learning_rate": 4.339482751647557e-06,
2213
+ "loss": 0.1103,
2214
+ "step": 315
2215
+ },
2216
+ {
2217
+ "epoch": 4.025477707006369,
2218
+ "grad_norm": 0.23271125680775137,
2219
+ "learning_rate": 4.228770401229824e-06,
2220
+ "loss": 0.1186,
2221
+ "step": 316
2222
+ },
2223
+ {
2224
+ "epoch": 4.038216560509555,
2225
+ "grad_norm": 0.2122069247359907,
2226
+ "learning_rate": 4.119321472033779e-06,
2227
+ "loss": 0.1008,
2228
+ "step": 317
2229
+ },
2230
+ {
2231
+ "epoch": 4.050955414012739,
2232
+ "grad_norm": 0.15556832545642946,
2233
+ "learning_rate": 4.011144731929981e-06,
2234
+ "loss": 0.0879,
2235
+ "step": 318
2236
+ },
2237
+ {
2238
+ "epoch": 4.063694267515924,
2239
+ "grad_norm": 0.13068949666043236,
2240
+ "learning_rate": 3.904248846874894e-06,
2241
+ "loss": 0.101,
2242
+ "step": 319
2243
+ },
2244
+ {
2245
+ "epoch": 4.076433121019108,
2246
+ "grad_norm": 0.15988306601505795,
2247
+ "learning_rate": 3.7986423802166705e-06,
2248
+ "loss": 0.1042,
2249
+ "step": 320
2250
+ },
2251
+ {
2252
+ "epoch": 4.089171974522293,
2253
+ "grad_norm": 0.2008005962776904,
2254
+ "learning_rate": 3.694333792009115e-06,
2255
+ "loss": 0.1186,
2256
+ "step": 321
2257
+ },
2258
+ {
2259
+ "epoch": 4.101910828025478,
2260
+ "grad_norm": 0.22309601017298253,
2261
+ "learning_rate": 3.5913314383339937e-06,
2262
+ "loss": 0.1079,
2263
+ "step": 322
2264
+ },
2265
+ {
2266
+ "epoch": 4.114649681528663,
2267
+ "grad_norm": 0.2171458668604706,
2268
+ "learning_rate": 3.4896435706316e-06,
2269
+ "loss": 0.0992,
2270
+ "step": 323
2271
+ },
2272
+ {
2273
+ "epoch": 4.127388535031847,
2274
+ "grad_norm": 0.1754223239205233,
2275
+ "learning_rate": 3.3892783350397675e-06,
2276
+ "loss": 0.0966,
2277
+ "step": 324
2278
+ },
2279
+ {
2280
+ "epoch": 4.140127388535032,
2281
+ "grad_norm": 0.17536902204340468,
2282
+ "learning_rate": 3.290243771741275e-06,
2283
+ "loss": 0.1002,
2284
+ "step": 325
2285
+ },
2286
+ {
2287
+ "epoch": 4.1528662420382165,
2288
+ "grad_norm": 0.1349302126131413,
2289
+ "learning_rate": 3.1925478143197418e-06,
2290
+ "loss": 0.0864,
2291
+ "step": 326
2292
+ },
2293
+ {
2294
+ "epoch": 4.165605095541402,
2295
+ "grad_norm": 0.14115081640236687,
2296
+ "learning_rate": 3.0961982891241083e-06,
2297
+ "loss": 0.1042,
2298
+ "step": 327
2299
+ },
2300
+ {
2301
+ "epoch": 4.178343949044586,
2302
+ "grad_norm": 0.14152556061179694,
2303
+ "learning_rate": 3.001202914641628e-06,
2304
+ "loss": 0.0832,
2305
+ "step": 328
2306
+ },
2307
+ {
2308
+ "epoch": 4.191082802547771,
2309
+ "grad_norm": 0.14396584823928577,
2310
+ "learning_rate": 2.907569300879596e-06,
2311
+ "loss": 0.0863,
2312
+ "step": 329
2313
+ },
2314
+ {
2315
+ "epoch": 4.203821656050955,
2316
+ "grad_norm": 0.13656039310896373,
2317
+ "learning_rate": 2.815304948755664e-06,
2318
+ "loss": 0.105,
2319
+ "step": 330
2320
+ },
2321
+ {
2322
+ "epoch": 4.2165605095541405,
2323
+ "grad_norm": 0.14707618001361536,
2324
+ "learning_rate": 2.7244172494969978e-06,
2325
+ "loss": 0.1121,
2326
+ "step": 331
2327
+ },
2328
+ {
2329
+ "epoch": 4.229299363057325,
2330
+ "grad_norm": 0.14952406400214932,
2331
+ "learning_rate": 2.6349134840481294e-06,
2332
+ "loss": 0.1039,
2333
+ "step": 332
2334
+ },
2335
+ {
2336
+ "epoch": 4.24203821656051,
2337
+ "grad_norm": 0.45773988775828156,
2338
+ "learning_rate": 2.546800822487714e-06,
2339
+ "loss": 0.1006,
2340
+ "step": 333
2341
+ },
2342
+ {
2343
+ "epoch": 4.254777070063694,
2344
+ "grad_norm": 0.14086482905242464,
2345
+ "learning_rate": 2.4600863234541338e-06,
2346
+ "loss": 0.103,
2347
+ "step": 334
2348
+ },
2349
+ {
2350
+ "epoch": 4.267515923566879,
2351
+ "grad_norm": 0.13406472253081728,
2352
+ "learning_rate": 2.374776933580025e-06,
2353
+ "loss": 0.1102,
2354
+ "step": 335
2355
+ },
2356
+ {
2357
+ "epoch": 4.280254777070064,
2358
+ "grad_norm": 0.14550219594207334,
2359
+ "learning_rate": 2.2908794869358044e-06,
2360
+ "loss": 0.1125,
2361
+ "step": 336
2362
+ },
2363
+ {
2364
+ "epoch": 4.292993630573249,
2365
+ "grad_norm": 0.1347484077586462,
2366
+ "learning_rate": 2.2084007044821764e-06,
2367
+ "loss": 0.0835,
2368
+ "step": 337
2369
+ },
2370
+ {
2371
+ "epoch": 4.305732484076433,
2372
+ "grad_norm": 0.1290805497722182,
2373
+ "learning_rate": 2.127347193531757e-06,
2374
+ "loss": 0.1041,
2375
+ "step": 338
2376
+ },
2377
+ {
2378
+ "epoch": 4.318471337579618,
2379
+ "grad_norm": 0.13309082164534528,
2380
+ "learning_rate": 2.0477254472197237e-06,
2381
+ "loss": 0.0967,
2382
+ "step": 339
2383
+ },
2384
+ {
2385
+ "epoch": 4.3312101910828025,
2386
+ "grad_norm": 0.13761067925193232,
2387
+ "learning_rate": 1.96954184398368e-06,
2388
+ "loss": 0.1113,
2389
+ "step": 340
2390
+ },
2391
+ {
2392
+ "epoch": 4.343949044585988,
2393
+ "grad_norm": 0.13572266975580224,
2394
+ "learning_rate": 1.8928026470526917e-06,
2395
+ "loss": 0.1053,
2396
+ "step": 341
2397
+ },
2398
+ {
2399
+ "epoch": 4.356687898089172,
2400
+ "grad_norm": 0.12800656352242193,
2401
+ "learning_rate": 1.817514003945524e-06,
2402
+ "loss": 0.0867,
2403
+ "step": 342
2404
+ },
2405
+ {
2406
+ "epoch": 4.369426751592357,
2407
+ "grad_norm": 0.13496217012761974,
2408
+ "learning_rate": 1.743681945978184e-06,
2409
+ "loss": 0.11,
2410
+ "step": 343
2411
+ },
2412
+ {
2413
+ "epoch": 4.382165605095541,
2414
+ "grad_norm": 0.13765332230964036,
2415
+ "learning_rate": 1.6713123877807413e-06,
2416
+ "loss": 0.1133,
2417
+ "step": 344
2418
+ },
2419
+ {
2420
+ "epoch": 4.3949044585987265,
2421
+ "grad_norm": 0.13352458324587824,
2422
+ "learning_rate": 1.6004111268235156e-06,
2423
+ "loss": 0.1058,
2424
+ "step": 345
2425
+ },
2426
+ {
2427
+ "epoch": 4.407643312101911,
2428
+ "grad_norm": 0.13191561231330995,
2429
+ "learning_rate": 1.5309838429526714e-06,
2430
+ "loss": 0.1051,
2431
+ "step": 346
2432
+ },
2433
+ {
2434
+ "epoch": 4.420382165605096,
2435
+ "grad_norm": 0.12490688145220193,
2436
+ "learning_rate": 1.4630360979351644e-06,
2437
+ "loss": 0.09,
2438
+ "step": 347
2439
+ },
2440
+ {
2441
+ "epoch": 4.43312101910828,
2442
+ "grad_norm": 0.15947738459761718,
2443
+ "learning_rate": 1.396573335013236e-06,
2444
+ "loss": 0.0964,
2445
+ "step": 348
2446
+ },
2447
+ {
2448
+ "epoch": 4.445859872611465,
2449
+ "grad_norm": 0.12096509710405506,
2450
+ "learning_rate": 1.3316008784683265e-06,
2451
+ "loss": 0.0907,
2452
+ "step": 349
2453
+ },
2454
+ {
2455
+ "epoch": 4.45859872611465,
2456
+ "grad_norm": 0.12188980792759932,
2457
+ "learning_rate": 1.2681239331945695e-06,
2458
+ "loss": 0.1016,
2459
+ "step": 350
2460
+ },
2461
+ {
2462
+ "epoch": 4.471337579617835,
2463
+ "grad_norm": 0.13559724068658183,
2464
+ "learning_rate": 1.2061475842818337e-06,
2465
+ "loss": 0.1072,
2466
+ "step": 351
2467
+ },
2468
+ {
2469
+ "epoch": 4.484076433121019,
2470
+ "grad_norm": 0.12292648232545614,
2471
+ "learning_rate": 1.1456767966083393e-06,
2472
+ "loss": 0.1087,
2473
+ "step": 352
2474
+ },
2475
+ {
2476
+ "epoch": 4.496815286624204,
2477
+ "grad_norm": 0.1334907116310344,
2478
+ "learning_rate": 1.086716414442952e-06,
2479
+ "loss": 0.1161,
2480
+ "step": 353
2481
+ },
2482
+ {
2483
+ "epoch": 4.509554140127388,
2484
+ "grad_norm": 0.13271118963324863,
2485
+ "learning_rate": 1.0292711610570904e-06,
2486
+ "loss": 0.0958,
2487
+ "step": 354
2488
+ },
2489
+ {
2490
+ "epoch": 4.522292993630574,
2491
+ "grad_norm": 0.13680838872833267,
2492
+ "learning_rate": 9.733456383463658e-07,
2493
+ "loss": 0.0943,
2494
+ "step": 355
2495
+ },
2496
+ {
2497
+ "epoch": 4.535031847133758,
2498
+ "grad_norm": 0.13801489368931033,
2499
+ "learning_rate": 9.189443264619102e-07,
2500
+ "loss": 0.123,
2501
+ "step": 356
2502
+ },
2503
+ {
2504
+ "epoch": 4.547770700636943,
2505
+ "grad_norm": 0.12529844924959202,
2506
+ "learning_rate": 8.660715834514977e-07,
2507
+ "loss": 0.0975,
2508
+ "step": 357
2509
+ },
2510
+ {
2511
+ "epoch": 4.560509554140127,
2512
+ "grad_norm": 0.12938877893813736,
2513
+ "learning_rate": 8.147316449103959e-07,
2514
+ "loss": 0.0918,
2515
+ "step": 358
2516
+ },
2517
+ {
2518
+ "epoch": 4.573248407643312,
2519
+ "grad_norm": 0.12115317481092167,
2520
+ "learning_rate": 7.649286236420806e-07,
2521
+ "loss": 0.1135,
2522
+ "step": 359
2523
+ },
2524
+ {
2525
+ "epoch": 4.585987261146497,
2526
+ "grad_norm": 0.11833791739810552,
2527
+ "learning_rate": 7.166665093287539e-07,
2528
+ "loss": 0.0854,
2529
+ "step": 360
2530
+ },
2531
+ {
2532
+ "epoch": 4.598726114649682,
2533
+ "grad_norm": 0.12705837653951255,
2534
+ "learning_rate": 6.69949168211721e-07,
2535
+ "loss": 0.1109,
2536
+ "step": 361
2537
+ },
2538
+ {
2539
+ "epoch": 4.611464968152866,
2540
+ "grad_norm": 0.12508614962340267,
2541
+ "learning_rate": 6.247803427816945e-07,
2542
+ "loss": 0.0974,
2543
+ "step": 362
2544
+ },
2545
+ {
2546
+ "epoch": 4.624203821656051,
2547
+ "grad_norm": 0.1452236579744733,
2548
+ "learning_rate": 5.811636514789598e-07,
2549
+ "loss": 0.1157,
2550
+ "step": 363
2551
+ },
2552
+ {
2553
+ "epoch": 4.6369426751592355,
2554
+ "grad_norm": 0.1300660477914448,
2555
+ "learning_rate": 5.391025884035239e-07,
2556
+ "loss": 0.0915,
2557
+ "step": 364
2558
+ },
2559
+ {
2560
+ "epoch": 4.649681528662421,
2561
+ "grad_norm": 0.1200540678678256,
2562
+ "learning_rate": 4.986005230351954e-07,
2563
+ "loss": 0.0872,
2564
+ "step": 365
2565
+ },
2566
+ {
2567
+ "epoch": 4.662420382165605,
2568
+ "grad_norm": 0.11487997298035525,
2569
+ "learning_rate": 4.5966069996365993e-07,
2570
+ "loss": 0.1165,
2571
+ "step": 366
2572
+ },
2573
+ {
2574
+ "epoch": 4.67515923566879,
2575
+ "grad_norm": 0.139543063544079,
2576
+ "learning_rate": 4.22286238628562e-07,
2577
+ "loss": 0.1053,
2578
+ "step": 367
2579
+ },
2580
+ {
2581
+ "epoch": 4.687898089171974,
2582
+ "grad_norm": 0.12478098586523038,
2583
+ "learning_rate": 3.8648013306960664e-07,
2584
+ "loss": 0.0858,
2585
+ "step": 368
2586
+ },
2587
+ {
2588
+ "epoch": 4.7006369426751595,
2589
+ "grad_norm": 0.1314853226678359,
2590
+ "learning_rate": 3.522452516867048e-07,
2591
+ "loss": 0.1179,
2592
+ "step": 369
2593
+ },
2594
+ {
2595
+ "epoch": 4.713375796178344,
2596
+ "grad_norm": 0.1561600576660889,
2597
+ "learning_rate": 3.1958433701019697e-07,
2598
+ "loss": 0.0977,
2599
+ "step": 370
2600
+ },
2601
+ {
2602
+ "epoch": 4.726114649681529,
2603
+ "grad_norm": 0.1275347994190016,
2604
+ "learning_rate": 2.8850000548115155e-07,
2605
+ "loss": 0.112,
2606
+ "step": 371
2607
+ },
2608
+ {
2609
+ "epoch": 4.738853503184713,
2610
+ "grad_norm": 0.12188252873510248,
2611
+ "learning_rate": 2.5899474724174313e-07,
2612
+ "loss": 0.0862,
2613
+ "step": 372
2614
+ },
2615
+ {
2616
+ "epoch": 4.751592356687898,
2617
+ "grad_norm": 0.1318771763835626,
2618
+ "learning_rate": 2.3107092593579905e-07,
2619
+ "loss": 0.1142,
2620
+ "step": 373
2621
+ },
2622
+ {
2623
+ "epoch": 4.764331210191083,
2624
+ "grad_norm": 0.14220929835983984,
2625
+ "learning_rate": 2.0473077851942858e-07,
2626
+ "loss": 0.1118,
2627
+ "step": 374
2628
+ },
2629
+ {
2630
+ "epoch": 4.777070063694268,
2631
+ "grad_norm": 0.1219858566908112,
2632
+ "learning_rate": 1.799764150818306e-07,
2633
+ "loss": 0.1084,
2634
+ "step": 375
2635
+ },
2636
+ {
2637
+ "epoch": 4.789808917197452,
2638
+ "grad_norm": 0.12566174956929266,
2639
+ "learning_rate": 1.5680981867625566e-07,
2640
+ "loss": 0.1077,
2641
+ "step": 376
2642
+ },
2643
+ {
2644
+ "epoch": 4.802547770700637,
2645
+ "grad_norm": 0.12253664048339573,
2646
+ "learning_rate": 1.3523284516113955e-07,
2647
+ "loss": 0.0899,
2648
+ "step": 377
2649
+ },
2650
+ {
2651
+ "epoch": 4.8152866242038215,
2652
+ "grad_norm": 0.12142275297990943,
2653
+ "learning_rate": 1.1524722305144231e-07,
2654
+ "loss": 0.0956,
2655
+ "step": 378
2656
+ },
2657
+ {
2658
+ "epoch": 4.828025477707007,
2659
+ "grad_norm": 0.1235160540744103,
2660
+ "learning_rate": 9.685455338016347e-08,
2661
+ "loss": 0.1077,
2662
+ "step": 379
2663
+ },
2664
+ {
2665
+ "epoch": 4.840764331210191,
2666
+ "grad_norm": 0.12467941962401048,
2667
+ "learning_rate": 8.005630957010014e-08,
2668
+ "loss": 0.0793,
2669
+ "step": 380
2670
+ },
2671
+ {
2672
+ "epoch": 4.853503184713376,
2673
+ "grad_norm": 0.20795298981735666,
2674
+ "learning_rate": 6.485383731580142e-08,
2675
+ "loss": 0.1045,
2676
+ "step": 381
2677
+ },
2678
+ {
2679
+ "epoch": 4.86624203821656,
2680
+ "grad_norm": 0.12799721551039098,
2681
+ "learning_rate": 5.1248354475768034e-08,
2682
+ "loss": 0.101,
2683
+ "step": 382
2684
+ },
2685
+ {
2686
+ "epoch": 4.8789808917197455,
2687
+ "grad_norm": 0.1258631802102174,
2688
+ "learning_rate": 3.924095097489922e-08,
2689
+ "loss": 0.0944,
2690
+ "step": 383
2691
+ },
2692
+ {
2693
+ "epoch": 4.89171974522293,
2694
+ "grad_norm": 0.12812579528294077,
2695
+ "learning_rate": 2.8832588717164766e-08,
2696
+ "loss": 0.1053,
2697
+ "step": 384
2698
+ },
2699
+ {
2700
+ "epoch": 4.904458598726115,
2701
+ "grad_norm": 0.12751773731121074,
2702
+ "learning_rate": 2.0024101508555604e-08,
2703
+ "loss": 0.0982,
2704
+ "step": 385
2705
+ },
2706
+ {
2707
+ "epoch": 4.917197452229299,
2708
+ "grad_norm": 0.13129186435975,
2709
+ "learning_rate": 1.281619499029274e-08,
2710
+ "loss": 0.1091,
2711
+ "step": 386
2712
+ },
2713
+ {
2714
+ "epoch": 4.929936305732484,
2715
+ "grad_norm": 0.12192936592060051,
2716
+ "learning_rate": 7.209446582292501e-09,
2717
+ "loss": 0.1065,
2718
+ "step": 387
2719
+ },
2720
+ {
2721
+ "epoch": 4.942675159235669,
2722
+ "grad_norm": 0.13025711875394239,
2723
+ "learning_rate": 3.2043054369057523e-09,
2724
+ "loss": 0.1076,
2725
+ "step": 388
2726
+ },
2727
+ {
2728
+ "epoch": 4.955414012738854,
2729
+ "grad_norm": 0.11678098849828318,
2730
+ "learning_rate": 8.010924029533406e-10,
2731
+ "loss": 0.1073,
2732
+ "step": 389
2733
+ },
2734
+ {
2735
+ "epoch": 4.968152866242038,
2736
+ "grad_norm": 0.12263643856967697,
2737
+ "learning_rate": 0.0,
2738
+ "loss": 0.1206,
2739
+ "step": 390
2740
+ },
2741
+ {
2742
+ "epoch": 4.968152866242038,
2743
+ "step": 390,
2744
+ "total_flos": 4.6317288847048704e+17,
2745
+ "train_loss": 0.30433785139750213,
2746
+ "train_runtime": 26201.4476,
2747
+ "train_samples_per_second": 1.908,
2748
+ "train_steps_per_second": 0.015
2749
+ }
2750
+ ],
2751
+ "logging_steps": 1,
2752
+ "max_steps": 390,
2753
+ "num_input_tokens_seen": 0,
2754
+ "num_train_epochs": 5,
2755
+ "save_steps": 500,
2756
+ "stateful_callbacks": {
2757
+ "TrainerControl": {
2758
+ "args": {
2759
+ "should_epoch_stop": false,
2760
+ "should_evaluate": false,
2761
+ "should_log": false,
2762
+ "should_save": true,
2763
+ "should_training_stop": true
2764
+ },
2765
+ "attributes": {}
2766
+ }
2767
+ },
2768
+ "total_flos": 4.6317288847048704e+17,
2769
+ "train_batch_size": 1,
2770
+ "trial_name": null,
2771
+ "trial_params": null
2772
+ }
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c54959564ce89463eee89264102be47c5db47e35f4fdf3151a0e247e22037b63
3
- size 7224
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:94da09efce32ded52b07167a713c7e2d5d21f294e8b090ddb07077d42205eac3
3
+ size 7352
training_loss.png ADDED