DGurgurov commited on
Commit
a71463a
·
verified ·
1 Parent(s): 9e9b528

Uploading checkpoint-98000 for xlm-r - lvs-latn

Browse files
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,67 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers
3
+ license: mit
4
+ base_model: FacebookAI/xlm-roberta-base
5
+ tags:
6
+ - generated_from_trainer
7
+ metrics:
8
+ - accuracy
9
+ model-index:
10
+ - name: lvs-Latn
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
+ # lvs-Latn
18
+
19
+ This model is a fine-tuned version of [FacebookAI/xlm-roberta-base](https://huggingface.co/FacebookAI/xlm-roberta-base) on an unknown dataset.
20
+ It achieves the following results on the evaluation set:
21
+ - Loss: nan
22
+ - Accuracy: 0.8365
23
+
24
+ ## Model description
25
+
26
+ More information needed
27
+
28
+ ## Intended uses & limitations
29
+
30
+ More information needed
31
+
32
+ ## Training and evaluation data
33
+
34
+ More information needed
35
+
36
+ ## Training procedure
37
+
38
+ ### Training hyperparameters
39
+
40
+ The following hyperparameters were used during training:
41
+ - learning_rate: 0.0001
42
+ - train_batch_size: 16
43
+ - eval_batch_size: 16
44
+ - seed: 42
45
+ - distributed_type: multi-GPU
46
+ - num_devices: 2
47
+ - total_train_batch_size: 32
48
+ - total_eval_batch_size: 32
49
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
50
+ - lr_scheduler_type: linear
51
+ - training_steps: 100000
52
+
53
+
54
+ ### Citation Information
55
+
56
+ If you use this model in your work, please cite the following paper. Additionally, if you require more details on training and performance, refer to the paper:
57
+
58
+ @misc{gurgurov2025smallmodelsbigimpact,
59
+ title={Small Models, Big Impact: Efficient Corpus and Graph-Based Adaptation of Small Multilingual Language Models for Low-Resource Languages},
60
+ author={Daniil Gurgurov and Ivan Vykopal and Josef van Genabith and Simon Ostermann},
61
+ year={2025},
62
+ eprint={2502.10140},
63
+ archivePrefix={arXiv},
64
+ primaryClass={cs.CL},
65
+ url={https://arxiv.org/abs/2502.10140},
66
+ }
67
+
config.json ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "FacebookAI/xlm-roberta-base",
3
+ "adapters": {
4
+ "adapters": {},
5
+ "config_map": {},
6
+ "fusion_config_map": {},
7
+ "fusions": {}
8
+ },
9
+ "architectures": [
10
+ "XLMRobertaForMaskedLM"
11
+ ],
12
+ "attention_probs_dropout_prob": 0.1,
13
+ "bos_token_id": 0,
14
+ "classifier_dropout": null,
15
+ "eos_token_id": 2,
16
+ "hidden_act": "gelu",
17
+ "hidden_dropout_prob": 0.1,
18
+ "hidden_size": 768,
19
+ "initializer_range": 0.02,
20
+ "intermediate_size": 3072,
21
+ "layer_norm_eps": 1e-05,
22
+ "max_position_embeddings": 514,
23
+ "model_type": "xlm-roberta",
24
+ "num_attention_heads": 12,
25
+ "num_hidden_layers": 12,
26
+ "output_past": true,
27
+ "pad_token_id": 1,
28
+ "position_embedding_type": "absolute",
29
+ "torch_dtype": "float32",
30
+ "transformers_version": "4.45.2",
31
+ "type_vocab_size": 1,
32
+ "use_cache": true,
33
+ "vocab_size": 250002
34
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cd831dbe95ed32fb4c07e28c0e341b519fe6e64e07a905a787d3876077baa09e
3
+ size 1113205088
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:451e726db12f6e6fbae11c11f5c0933486ad7cddc4e9625bd348d3f707545247
3
+ size 2226530490
rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:58b2470cf50b1d4e53de3757dc3879ecadef3eebeb7523188f3f284bf0db7a03
3
+ size 14512
rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fed968cd1ae77c0cf8f9dac366cb3e79d890622f59b24e3988eea5e64165962c
3
+ size 14512
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a3fd067b8f5b6dee2ca565b8ea5c9f00e6d80f59b5809618fff53eb072beb787
3
+ size 1064
sentencepiece.bpe.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cfc8146abe2a0488e9e2a0c56de7952f7c11ab059eca145a0a727afce0db2865
3
+ size 5069051
special_tokens_map.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "cls_token": "<s>",
4
+ "eos_token": "</s>",
5
+ "mask_token": {
6
+ "content": "<mask>",
7
+ "lstrip": true,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false
11
+ },
12
+ "pad_token": "<pad>",
13
+ "sep_token": "</s>",
14
+ "unk_token": "<unk>"
15
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3a56def25aa40facc030ea8b0b87f3688e4b3c39eb8b45d5702b3a1300fe2a20
3
+ size 17082734
tokenizer_config.json ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<s>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<pad>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "</s>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "<unk>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ },
35
+ "250001": {
36
+ "content": "<mask>",
37
+ "lstrip": true,
38
+ "normalized": false,
39
+ "rstrip": false,
40
+ "single_word": false,
41
+ "special": true
42
+ }
43
+ },
44
+ "bos_token": "<s>",
45
+ "clean_up_tokenization_spaces": false,
46
+ "cls_token": "<s>",
47
+ "eos_token": "</s>",
48
+ "mask_token": "<mask>",
49
+ "model_max_length": 512,
50
+ "pad_token": "<pad>",
51
+ "sep_token": "</s>",
52
+ "tokenizer_class": "XLMRobertaTokenizer",
53
+ "unk_token": "<unk>"
54
+ }
trainer_state.json ADDED
@@ -0,0 +1,3169 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.8651660084724426,
3
+ "best_model_checkpoint": "./model_fine-tune/glot/xlm-r/lvs-Latn/checkpoint-98000",
4
+ "epoch": 9.5535192045233,
5
+ "eval_steps": 500,
6
+ "global_step": 98000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.04874244492103724,
13
+ "grad_norm": 5.49888801574707,
14
+ "learning_rate": 9.95e-05,
15
+ "loss": 1.542,
16
+ "step": 500
17
+ },
18
+ {
19
+ "epoch": 0.04874244492103724,
20
+ "eval_accuracy": 0.7314449847229756,
21
+ "eval_loss": 1.6759803295135498,
22
+ "eval_runtime": 412.1515,
23
+ "eval_samples_per_second": 150.714,
24
+ "eval_steps_per_second": 4.712,
25
+ "step": 500
26
+ },
27
+ {
28
+ "epoch": 0.09748488984207448,
29
+ "grad_norm": 6.478686809539795,
30
+ "learning_rate": 9.900000000000001e-05,
31
+ "loss": 1.4892,
32
+ "step": 1000
33
+ },
34
+ {
35
+ "epoch": 0.09748488984207448,
36
+ "eval_accuracy": 0.7361168583960936,
37
+ "eval_loss": 1.5948923826217651,
38
+ "eval_runtime": 404.9613,
39
+ "eval_samples_per_second": 153.39,
40
+ "eval_steps_per_second": 4.796,
41
+ "step": 1000
42
+ },
43
+ {
44
+ "epoch": 0.14622733476311173,
45
+ "grad_norm": 4.809515476226807,
46
+ "learning_rate": 9.850000000000001e-05,
47
+ "loss": 1.4572,
48
+ "step": 1500
49
+ },
50
+ {
51
+ "epoch": 0.14622733476311173,
52
+ "eval_accuracy": 0.742282605428477,
53
+ "eval_loss": 1.5350223779678345,
54
+ "eval_runtime": 416.0878,
55
+ "eval_samples_per_second": 149.288,
56
+ "eval_steps_per_second": 4.667,
57
+ "step": 1500
58
+ },
59
+ {
60
+ "epoch": 0.19496977968414897,
61
+ "grad_norm": 4.836926460266113,
62
+ "learning_rate": 9.8e-05,
63
+ "loss": 1.4184,
64
+ "step": 2000
65
+ },
66
+ {
67
+ "epoch": 0.19496977968414897,
68
+ "eval_accuracy": 0.7447425901720343,
69
+ "eval_loss": 1.4902225732803345,
70
+ "eval_runtime": 416.3726,
71
+ "eval_samples_per_second": 149.186,
72
+ "eval_steps_per_second": 4.664,
73
+ "step": 2000
74
+ },
75
+ {
76
+ "epoch": 0.24371222460518618,
77
+ "grad_norm": 5.625064373016357,
78
+ "learning_rate": 9.75e-05,
79
+ "loss": 1.3971,
80
+ "step": 2500
81
+ },
82
+ {
83
+ "epoch": 0.24371222460518618,
84
+ "eval_accuracy": 0.7491213937548641,
85
+ "eval_loss": 1.470357060432434,
86
+ "eval_runtime": 415.6377,
87
+ "eval_samples_per_second": 149.45,
88
+ "eval_steps_per_second": 4.672,
89
+ "step": 2500
90
+ },
91
+ {
92
+ "epoch": 0.29245466952622345,
93
+ "grad_norm": 4.774477958679199,
94
+ "learning_rate": 9.7e-05,
95
+ "loss": 1.377,
96
+ "step": 3000
97
+ },
98
+ {
99
+ "epoch": 0.29245466952622345,
100
+ "eval_accuracy": 0.751029615792176,
101
+ "eval_loss": 1.4384983777999878,
102
+ "eval_runtime": 415.9349,
103
+ "eval_samples_per_second": 149.343,
104
+ "eval_steps_per_second": 4.669,
105
+ "step": 3000
106
+ },
107
+ {
108
+ "epoch": 0.3411971144472607,
109
+ "grad_norm": 5.25926399230957,
110
+ "learning_rate": 9.65e-05,
111
+ "loss": 1.3586,
112
+ "step": 3500
113
+ },
114
+ {
115
+ "epoch": 0.3411971144472607,
116
+ "eval_accuracy": 0.7538155299545674,
117
+ "eval_loss": 1.4148088693618774,
118
+ "eval_runtime": 406.8693,
119
+ "eval_samples_per_second": 152.671,
120
+ "eval_steps_per_second": 4.773,
121
+ "step": 3500
122
+ },
123
+ {
124
+ "epoch": 0.38993955936829794,
125
+ "grad_norm": 6.997039794921875,
126
+ "learning_rate": 9.6e-05,
127
+ "loss": 1.3364,
128
+ "step": 4000
129
+ },
130
+ {
131
+ "epoch": 0.38993955936829794,
132
+ "eval_accuracy": 0.7563830000327777,
133
+ "eval_loss": 1.37319016456604,
134
+ "eval_runtime": 416.5936,
135
+ "eval_samples_per_second": 149.107,
136
+ "eval_steps_per_second": 4.662,
137
+ "step": 4000
138
+ },
139
+ {
140
+ "epoch": 0.4386820042893351,
141
+ "grad_norm": 5.220494747161865,
142
+ "learning_rate": 9.55e-05,
143
+ "loss": 1.3298,
144
+ "step": 4500
145
+ },
146
+ {
147
+ "epoch": 0.4386820042893351,
148
+ "eval_accuracy": 0.7592725706718406,
149
+ "eval_loss": 1.3646610975265503,
150
+ "eval_runtime": 406.1249,
151
+ "eval_samples_per_second": 152.951,
152
+ "eval_steps_per_second": 4.782,
153
+ "step": 4500
154
+ },
155
+ {
156
+ "epoch": 0.48742444921037237,
157
+ "grad_norm": 4.78454065322876,
158
+ "learning_rate": 9.5e-05,
159
+ "loss": 1.3013,
160
+ "step": 5000
161
+ },
162
+ {
163
+ "epoch": 0.48742444921037237,
164
+ "eval_accuracy": 0.7610639838712026,
165
+ "eval_loss": 1.347570776939392,
166
+ "eval_runtime": 405.9899,
167
+ "eval_samples_per_second": 153.001,
168
+ "eval_steps_per_second": 4.783,
169
+ "step": 5000
170
+ },
171
+ {
172
+ "epoch": 0.5361668941314096,
173
+ "grad_norm": 5.818552017211914,
174
+ "learning_rate": 9.449999999999999e-05,
175
+ "loss": 1.3015,
176
+ "step": 5500
177
+ },
178
+ {
179
+ "epoch": 0.5361668941314096,
180
+ "eval_accuracy": 0.7620598261338267,
181
+ "eval_loss": 1.3511470556259155,
182
+ "eval_runtime": 404.932,
183
+ "eval_samples_per_second": 153.401,
184
+ "eval_steps_per_second": 4.796,
185
+ "step": 5500
186
+ },
187
+ {
188
+ "epoch": 0.5849093390524469,
189
+ "grad_norm": 4.825132369995117,
190
+ "learning_rate": 9.4e-05,
191
+ "loss": 1.275,
192
+ "step": 6000
193
+ },
194
+ {
195
+ "epoch": 0.5849093390524469,
196
+ "eval_accuracy": 0.7646006262510697,
197
+ "eval_loss": 1.3294358253479004,
198
+ "eval_runtime": 416.996,
199
+ "eval_samples_per_second": 148.963,
200
+ "eval_steps_per_second": 4.657,
201
+ "step": 6000
202
+ },
203
+ {
204
+ "epoch": 0.6336517839734841,
205
+ "grad_norm": 4.281984806060791,
206
+ "learning_rate": 9.350000000000001e-05,
207
+ "loss": 1.2767,
208
+ "step": 6500
209
+ },
210
+ {
211
+ "epoch": 0.6336517839734841,
212
+ "eval_accuracy": 0.7639123003074906,
213
+ "eval_loss": 1.3210957050323486,
214
+ "eval_runtime": 416.2294,
215
+ "eval_samples_per_second": 149.237,
216
+ "eval_steps_per_second": 4.666,
217
+ "step": 6500
218
+ },
219
+ {
220
+ "epoch": 0.6823942288945214,
221
+ "grad_norm": 4.695993900299072,
222
+ "learning_rate": 9.300000000000001e-05,
223
+ "loss": 1.2587,
224
+ "step": 7000
225
+ },
226
+ {
227
+ "epoch": 0.6823942288945214,
228
+ "eval_accuracy": 0.7661378977535546,
229
+ "eval_loss": 1.3145350217819214,
230
+ "eval_runtime": 411.5116,
231
+ "eval_samples_per_second": 150.948,
232
+ "eval_steps_per_second": 4.719,
233
+ "step": 7000
234
+ },
235
+ {
236
+ "epoch": 0.7311366738155586,
237
+ "grad_norm": 4.380468368530273,
238
+ "learning_rate": 9.250000000000001e-05,
239
+ "loss": 1.2413,
240
+ "step": 7500
241
+ },
242
+ {
243
+ "epoch": 0.7311366738155586,
244
+ "eval_accuracy": 0.7678338508433707,
245
+ "eval_loss": 1.3006008863449097,
246
+ "eval_runtime": 405.5396,
247
+ "eval_samples_per_second": 153.171,
248
+ "eval_steps_per_second": 4.789,
249
+ "step": 7500
250
+ },
251
+ {
252
+ "epoch": 0.7798791187365959,
253
+ "grad_norm": 4.586111545562744,
254
+ "learning_rate": 9.200000000000001e-05,
255
+ "loss": 1.2439,
256
+ "step": 8000
257
+ },
258
+ {
259
+ "epoch": 0.7798791187365959,
260
+ "eval_accuracy": 0.7698155524324797,
261
+ "eval_loss": 1.3006173372268677,
262
+ "eval_runtime": 416.5011,
263
+ "eval_samples_per_second": 149.14,
264
+ "eval_steps_per_second": 4.663,
265
+ "step": 8000
266
+ },
267
+ {
268
+ "epoch": 0.8286215636576331,
269
+ "grad_norm": 4.449491024017334,
270
+ "learning_rate": 9.15e-05,
271
+ "loss": 1.2394,
272
+ "step": 8500
273
+ },
274
+ {
275
+ "epoch": 0.8286215636576331,
276
+ "eval_accuracy": 0.7712381365247952,
277
+ "eval_loss": 1.2807519435882568,
278
+ "eval_runtime": 406.4615,
279
+ "eval_samples_per_second": 152.824,
280
+ "eval_steps_per_second": 4.778,
281
+ "step": 8500
282
+ },
283
+ {
284
+ "epoch": 0.8773640085786703,
285
+ "grad_norm": 4.266624450683594,
286
+ "learning_rate": 9.1e-05,
287
+ "loss": 1.2314,
288
+ "step": 9000
289
+ },
290
+ {
291
+ "epoch": 0.8773640085786703,
292
+ "eval_accuracy": 0.7721473946396241,
293
+ "eval_loss": 1.2844055891036987,
294
+ "eval_runtime": 405.1662,
295
+ "eval_samples_per_second": 153.312,
296
+ "eval_steps_per_second": 4.793,
297
+ "step": 9000
298
+ },
299
+ {
300
+ "epoch": 0.9261064534997075,
301
+ "grad_norm": 3.9454383850097656,
302
+ "learning_rate": 9.05e-05,
303
+ "loss": 1.2235,
304
+ "step": 9500
305
+ },
306
+ {
307
+ "epoch": 0.9261064534997075,
308
+ "eval_accuracy": 0.7743032006713836,
309
+ "eval_loss": 1.2606672048568726,
310
+ "eval_runtime": 416.4441,
311
+ "eval_samples_per_second": 149.16,
312
+ "eval_steps_per_second": 4.663,
313
+ "step": 9500
314
+ },
315
+ {
316
+ "epoch": 0.9748488984207447,
317
+ "grad_norm": 4.444428443908691,
318
+ "learning_rate": 9e-05,
319
+ "loss": 1.2177,
320
+ "step": 10000
321
+ },
322
+ {
323
+ "epoch": 0.9748488984207447,
324
+ "eval_accuracy": 0.7739040185569158,
325
+ "eval_loss": 1.2562477588653564,
326
+ "eval_runtime": 417.8832,
327
+ "eval_samples_per_second": 148.647,
328
+ "eval_steps_per_second": 4.647,
329
+ "step": 10000
330
+ },
331
+ {
332
+ "epoch": 1.023591343341782,
333
+ "grad_norm": 3.7575981616973877,
334
+ "learning_rate": 8.950000000000001e-05,
335
+ "loss": 1.2069,
336
+ "step": 10500
337
+ },
338
+ {
339
+ "epoch": 1.023591343341782,
340
+ "eval_accuracy": 0.7760550010984028,
341
+ "eval_loss": 1.2498961687088013,
342
+ "eval_runtime": 405.9854,
343
+ "eval_samples_per_second": 153.003,
344
+ "eval_steps_per_second": 4.783,
345
+ "step": 10500
346
+ },
347
+ {
348
+ "epoch": 1.0723337882628192,
349
+ "grad_norm": 4.82220983505249,
350
+ "learning_rate": 8.900000000000001e-05,
351
+ "loss": 1.1901,
352
+ "step": 11000
353
+ },
354
+ {
355
+ "epoch": 1.0723337882628192,
356
+ "eval_accuracy": 0.7766217461325415,
357
+ "eval_loss": 1.2388906478881836,
358
+ "eval_runtime": 405.8717,
359
+ "eval_samples_per_second": 153.046,
360
+ "eval_steps_per_second": 4.785,
361
+ "step": 11000
362
+ },
363
+ {
364
+ "epoch": 1.1210762331838564,
365
+ "grad_norm": 4.685916900634766,
366
+ "learning_rate": 8.850000000000001e-05,
367
+ "loss": 1.1861,
368
+ "step": 11500
369
+ },
370
+ {
371
+ "epoch": 1.1210762331838564,
372
+ "eval_accuracy": 0.7772882734600404,
373
+ "eval_loss": 1.2301160097122192,
374
+ "eval_runtime": 416.4452,
375
+ "eval_samples_per_second": 149.16,
376
+ "eval_steps_per_second": 4.663,
377
+ "step": 11500
378
+ },
379
+ {
380
+ "epoch": 1.1698186781048938,
381
+ "grad_norm": 5.34528923034668,
382
+ "learning_rate": 8.800000000000001e-05,
383
+ "loss": 1.1817,
384
+ "step": 12000
385
+ },
386
+ {
387
+ "epoch": 1.1698186781048938,
388
+ "eval_accuracy": 0.7785483554561496,
389
+ "eval_loss": 1.235237717628479,
390
+ "eval_runtime": 406.0432,
391
+ "eval_samples_per_second": 152.981,
392
+ "eval_steps_per_second": 4.783,
393
+ "step": 12000
394
+ },
395
+ {
396
+ "epoch": 1.218561123025931,
397
+ "grad_norm": 5.657125473022461,
398
+ "learning_rate": 8.75e-05,
399
+ "loss": 1.171,
400
+ "step": 12500
401
+ },
402
+ {
403
+ "epoch": 1.218561123025931,
404
+ "eval_accuracy": 0.7794601307367932,
405
+ "eval_loss": 1.2307161092758179,
406
+ "eval_runtime": 404.5582,
407
+ "eval_samples_per_second": 153.543,
408
+ "eval_steps_per_second": 4.8,
409
+ "step": 12500
410
+ },
411
+ {
412
+ "epoch": 1.2673035679469682,
413
+ "grad_norm": 4.351173400878906,
414
+ "learning_rate": 8.7e-05,
415
+ "loss": 1.1627,
416
+ "step": 13000
417
+ },
418
+ {
419
+ "epoch": 1.2673035679469682,
420
+ "eval_accuracy": 0.7807187204409836,
421
+ "eval_loss": 1.2187020778656006,
422
+ "eval_runtime": 415.7588,
423
+ "eval_samples_per_second": 149.406,
424
+ "eval_steps_per_second": 4.671,
425
+ "step": 13000
426
+ },
427
+ {
428
+ "epoch": 1.3160460128680054,
429
+ "grad_norm": 3.8281147480010986,
430
+ "learning_rate": 8.65e-05,
431
+ "loss": 1.1649,
432
+ "step": 13500
433
+ },
434
+ {
435
+ "epoch": 1.3160460128680054,
436
+ "eval_accuracy": 0.7814787354583909,
437
+ "eval_loss": 1.2125664949417114,
438
+ "eval_runtime": 416.765,
439
+ "eval_samples_per_second": 149.046,
440
+ "eval_steps_per_second": 4.66,
441
+ "step": 13500
442
+ },
443
+ {
444
+ "epoch": 1.3647884577890426,
445
+ "grad_norm": 3.9933149814605713,
446
+ "learning_rate": 8.6e-05,
447
+ "loss": 1.1627,
448
+ "step": 14000
449
+ },
450
+ {
451
+ "epoch": 1.3647884577890426,
452
+ "eval_accuracy": 0.7817388124597489,
453
+ "eval_loss": 1.2068222761154175,
454
+ "eval_runtime": 406.9099,
455
+ "eval_samples_per_second": 152.655,
456
+ "eval_steps_per_second": 4.773,
457
+ "step": 14000
458
+ },
459
+ {
460
+ "epoch": 1.41353090271008,
461
+ "grad_norm": 4.470147609710693,
462
+ "learning_rate": 8.55e-05,
463
+ "loss": 1.1378,
464
+ "step": 14500
465
+ },
466
+ {
467
+ "epoch": 1.41353090271008,
468
+ "eval_accuracy": 0.7828488018378367,
469
+ "eval_loss": 1.1938220262527466,
470
+ "eval_runtime": 406.0802,
471
+ "eval_samples_per_second": 152.967,
472
+ "eval_steps_per_second": 4.782,
473
+ "step": 14500
474
+ },
475
+ {
476
+ "epoch": 1.4622733476311172,
477
+ "grad_norm": 4.2107977867126465,
478
+ "learning_rate": 8.5e-05,
479
+ "loss": 1.1476,
480
+ "step": 15000
481
+ },
482
+ {
483
+ "epoch": 1.4622733476311172,
484
+ "eval_accuracy": 0.7845405728249568,
485
+ "eval_loss": 1.1901649236679077,
486
+ "eval_runtime": 406.9716,
487
+ "eval_samples_per_second": 152.632,
488
+ "eval_steps_per_second": 4.772,
489
+ "step": 15000
490
+ },
491
+ {
492
+ "epoch": 1.5110157925521546,
493
+ "grad_norm": 3.608473300933838,
494
+ "learning_rate": 8.450000000000001e-05,
495
+ "loss": 1.1363,
496
+ "step": 15500
497
+ },
498
+ {
499
+ "epoch": 1.5110157925521546,
500
+ "eval_accuracy": 0.7837757811554797,
501
+ "eval_loss": 1.1860324144363403,
502
+ "eval_runtime": 407.0867,
503
+ "eval_samples_per_second": 152.589,
504
+ "eval_steps_per_second": 4.77,
505
+ "step": 15500
506
+ },
507
+ {
508
+ "epoch": 1.5597582374731918,
509
+ "grad_norm": 4.442079067230225,
510
+ "learning_rate": 8.4e-05,
511
+ "loss": 1.1429,
512
+ "step": 16000
513
+ },
514
+ {
515
+ "epoch": 1.5597582374731918,
516
+ "eval_accuracy": 0.7845283762127719,
517
+ "eval_loss": 1.182504653930664,
518
+ "eval_runtime": 404.9789,
519
+ "eval_samples_per_second": 153.383,
520
+ "eval_steps_per_second": 4.795,
521
+ "step": 16000
522
+ },
523
+ {
524
+ "epoch": 1.608500682394229,
525
+ "grad_norm": 5.191744804382324,
526
+ "learning_rate": 8.35e-05,
527
+ "loss": 1.1351,
528
+ "step": 16500
529
+ },
530
+ {
531
+ "epoch": 1.608500682394229,
532
+ "eval_accuracy": 0.7856157559268981,
533
+ "eval_loss": 1.1625466346740723,
534
+ "eval_runtime": 416.36,
535
+ "eval_samples_per_second": 149.191,
536
+ "eval_steps_per_second": 4.664,
537
+ "step": 16500
538
+ },
539
+ {
540
+ "epoch": 1.6572431273152661,
541
+ "grad_norm": 3.691304922103882,
542
+ "learning_rate": 8.3e-05,
543
+ "loss": 1.1215,
544
+ "step": 17000
545
+ },
546
+ {
547
+ "epoch": 1.6572431273152661,
548
+ "eval_accuracy": 0.7858090227903097,
549
+ "eval_loss": 1.1740021705627441,
550
+ "eval_runtime": 405.1783,
551
+ "eval_samples_per_second": 153.308,
552
+ "eval_steps_per_second": 4.793,
553
+ "step": 17000
554
+ },
555
+ {
556
+ "epoch": 1.7059855722363033,
557
+ "grad_norm": 3.727795362472534,
558
+ "learning_rate": 8.25e-05,
559
+ "loss": 1.1254,
560
+ "step": 17500
561
+ },
562
+ {
563
+ "epoch": 1.7059855722363033,
564
+ "eval_accuracy": 0.7880051408492428,
565
+ "eval_loss": 1.1576285362243652,
566
+ "eval_runtime": 404.8472,
567
+ "eval_samples_per_second": 153.433,
568
+ "eval_steps_per_second": 4.797,
569
+ "step": 17500
570
+ },
571
+ {
572
+ "epoch": 1.7547280171573405,
573
+ "grad_norm": 4.758922100067139,
574
+ "learning_rate": 8.2e-05,
575
+ "loss": 1.1302,
576
+ "step": 18000
577
+ },
578
+ {
579
+ "epoch": 1.7547280171573405,
580
+ "eval_accuracy": 0.7874455488255201,
581
+ "eval_loss": 1.1678783893585205,
582
+ "eval_runtime": 404.8165,
583
+ "eval_samples_per_second": 153.445,
584
+ "eval_steps_per_second": 4.797,
585
+ "step": 18000
586
+ },
587
+ {
588
+ "epoch": 1.8034704620783777,
589
+ "grad_norm": 3.915757417678833,
590
+ "learning_rate": 8.15e-05,
591
+ "loss": 1.1137,
592
+ "step": 18500
593
+ },
594
+ {
595
+ "epoch": 1.8034704620783777,
596
+ "eval_accuracy": 0.7881269778283002,
597
+ "eval_loss": 1.1509817838668823,
598
+ "eval_runtime": 405.6793,
599
+ "eval_samples_per_second": 153.118,
600
+ "eval_steps_per_second": 4.787,
601
+ "step": 18500
602
+ },
603
+ {
604
+ "epoch": 1.852212906999415,
605
+ "grad_norm": 4.390922546386719,
606
+ "learning_rate": 8.1e-05,
607
+ "loss": 1.1161,
608
+ "step": 19000
609
+ },
610
+ {
611
+ "epoch": 1.852212906999415,
612
+ "eval_accuracy": 0.790121277336312,
613
+ "eval_loss": 1.1375824213027954,
614
+ "eval_runtime": 405.7037,
615
+ "eval_samples_per_second": 153.109,
616
+ "eval_steps_per_second": 4.787,
617
+ "step": 19000
618
+ },
619
+ {
620
+ "epoch": 1.9009553519204523,
621
+ "grad_norm": 3.9364380836486816,
622
+ "learning_rate": 8.05e-05,
623
+ "loss": 1.1013,
624
+ "step": 19500
625
+ },
626
+ {
627
+ "epoch": 1.9009553519204523,
628
+ "eval_accuracy": 0.7908922207777267,
629
+ "eval_loss": 1.1409724950790405,
630
+ "eval_runtime": 412.4408,
631
+ "eval_samples_per_second": 150.608,
632
+ "eval_steps_per_second": 4.709,
633
+ "step": 19500
634
+ },
635
+ {
636
+ "epoch": 1.9496977968414897,
637
+ "grad_norm": 4.358945846557617,
638
+ "learning_rate": 8e-05,
639
+ "loss": 1.1176,
640
+ "step": 20000
641
+ },
642
+ {
643
+ "epoch": 1.9496977968414897,
644
+ "eval_accuracy": 0.7898279761421843,
645
+ "eval_loss": 1.1488791704177856,
646
+ "eval_runtime": 404.5288,
647
+ "eval_samples_per_second": 153.554,
648
+ "eval_steps_per_second": 4.801,
649
+ "step": 20000
650
+ },
651
+ {
652
+ "epoch": 1.9984402417625269,
653
+ "grad_norm": 3.865518569946289,
654
+ "learning_rate": 7.950000000000001e-05,
655
+ "loss": 1.1046,
656
+ "step": 20500
657
+ },
658
+ {
659
+ "epoch": 1.9984402417625269,
660
+ "eval_accuracy": 0.7907846225745591,
661
+ "eval_loss": 1.1414114236831665,
662
+ "eval_runtime": 417.3398,
663
+ "eval_samples_per_second": 148.84,
664
+ "eval_steps_per_second": 4.653,
665
+ "step": 20500
666
+ },
667
+ {
668
+ "epoch": 2.047182686683564,
669
+ "grad_norm": 4.270514488220215,
670
+ "learning_rate": 7.900000000000001e-05,
671
+ "loss": 1.0834,
672
+ "step": 21000
673
+ },
674
+ {
675
+ "epoch": 2.047182686683564,
676
+ "eval_accuracy": 0.7908569503746256,
677
+ "eval_loss": NaN,
678
+ "eval_runtime": 417.0997,
679
+ "eval_samples_per_second": 148.926,
680
+ "eval_steps_per_second": 4.656,
681
+ "step": 21000
682
+ },
683
+ {
684
+ "epoch": 2.0959251316046013,
685
+ "grad_norm": 4.67491340637207,
686
+ "learning_rate": 7.850000000000001e-05,
687
+ "loss": 1.0893,
688
+ "step": 21500
689
+ },
690
+ {
691
+ "epoch": 2.0959251316046013,
692
+ "eval_accuracy": 0.7910673718246737,
693
+ "eval_loss": 1.136673927307129,
694
+ "eval_runtime": 404.5912,
695
+ "eval_samples_per_second": 153.53,
696
+ "eval_steps_per_second": 4.8,
697
+ "step": 21500
698
+ },
699
+ {
700
+ "epoch": 2.1446675765256384,
701
+ "grad_norm": 15.46749496459961,
702
+ "learning_rate": 7.800000000000001e-05,
703
+ "loss": 1.0936,
704
+ "step": 22000
705
+ },
706
+ {
707
+ "epoch": 2.1446675765256384,
708
+ "eval_accuracy": 0.7924174903562226,
709
+ "eval_loss": 1.1254655122756958,
710
+ "eval_runtime": 413.6747,
711
+ "eval_samples_per_second": 150.159,
712
+ "eval_steps_per_second": 4.695,
713
+ "step": 22000
714
+ },
715
+ {
716
+ "epoch": 2.1934100214466756,
717
+ "grad_norm": 3.755019187927246,
718
+ "learning_rate": 7.75e-05,
719
+ "loss": 1.0861,
720
+ "step": 22500
721
+ },
722
+ {
723
+ "epoch": 2.1934100214466756,
724
+ "eval_accuracy": 0.7921287985056727,
725
+ "eval_loss": 1.1365504264831543,
726
+ "eval_runtime": 417.2166,
727
+ "eval_samples_per_second": 148.884,
728
+ "eval_steps_per_second": 4.655,
729
+ "step": 22500
730
+ },
731
+ {
732
+ "epoch": 2.242152466367713,
733
+ "grad_norm": 5.085185527801514,
734
+ "learning_rate": 7.7e-05,
735
+ "loss": 1.0846,
736
+ "step": 23000
737
+ },
738
+ {
739
+ "epoch": 2.242152466367713,
740
+ "eval_accuracy": 0.7926971124998237,
741
+ "eval_loss": 1.1271023750305176,
742
+ "eval_runtime": 413.8039,
743
+ "eval_samples_per_second": 150.112,
744
+ "eval_steps_per_second": 4.693,
745
+ "step": 23000
746
+ },
747
+ {
748
+ "epoch": 2.29089491128875,
749
+ "grad_norm": 3.165522575378418,
750
+ "learning_rate": 7.65e-05,
751
+ "loss": 1.0823,
752
+ "step": 23500
753
+ },
754
+ {
755
+ "epoch": 2.29089491128875,
756
+ "eval_accuracy": 0.7926174090551544,
757
+ "eval_loss": 1.1266652345657349,
758
+ "eval_runtime": 404.7022,
759
+ "eval_samples_per_second": 153.488,
760
+ "eval_steps_per_second": 4.799,
761
+ "step": 23500
762
+ },
763
+ {
764
+ "epoch": 2.3396373562097876,
765
+ "grad_norm": 3.573643684387207,
766
+ "learning_rate": 7.6e-05,
767
+ "loss": 1.0797,
768
+ "step": 24000
769
+ },
770
+ {
771
+ "epoch": 2.3396373562097876,
772
+ "eval_accuracy": 0.7946712997916913,
773
+ "eval_loss": 1.1183183193206787,
774
+ "eval_runtime": 405.4486,
775
+ "eval_samples_per_second": 153.206,
776
+ "eval_steps_per_second": 4.79,
777
+ "step": 24000
778
+ },
779
+ {
780
+ "epoch": 2.388379801130825,
781
+ "grad_norm": 3.7613110542297363,
782
+ "learning_rate": 7.55e-05,
783
+ "loss": 1.0628,
784
+ "step": 24500
785
+ },
786
+ {
787
+ "epoch": 2.388379801130825,
788
+ "eval_accuracy": 0.7942687964007779,
789
+ "eval_loss": 1.1270796060562134,
790
+ "eval_runtime": 416.8987,
791
+ "eval_samples_per_second": 148.998,
792
+ "eval_steps_per_second": 4.658,
793
+ "step": 24500
794
+ },
795
+ {
796
+ "epoch": 2.437122246051862,
797
+ "grad_norm": 3.7550013065338135,
798
+ "learning_rate": 7.500000000000001e-05,
799
+ "loss": 1.0699,
800
+ "step": 25000
801
+ },
802
+ {
803
+ "epoch": 2.437122246051862,
804
+ "eval_accuracy": 0.7951970388990339,
805
+ "eval_loss": 1.1055256128311157,
806
+ "eval_runtime": 406.0229,
807
+ "eval_samples_per_second": 152.989,
808
+ "eval_steps_per_second": 4.783,
809
+ "step": 25000
810
+ },
811
+ {
812
+ "epoch": 2.485864690972899,
813
+ "grad_norm": 4.42018461227417,
814
+ "learning_rate": 7.450000000000001e-05,
815
+ "loss": 1.0693,
816
+ "step": 25500
817
+ },
818
+ {
819
+ "epoch": 2.485864690972899,
820
+ "eval_accuracy": 0.796027678802998,
821
+ "eval_loss": 1.1022557020187378,
822
+ "eval_runtime": 405.75,
823
+ "eval_samples_per_second": 153.092,
824
+ "eval_steps_per_second": 4.786,
825
+ "step": 25500
826
+ },
827
+ {
828
+ "epoch": 2.5346071358939364,
829
+ "grad_norm": 3.7746379375457764,
830
+ "learning_rate": 7.4e-05,
831
+ "loss": 1.0632,
832
+ "step": 26000
833
+ },
834
+ {
835
+ "epoch": 2.5346071358939364,
836
+ "eval_accuracy": 0.7964411431203794,
837
+ "eval_loss": 1.0998525619506836,
838
+ "eval_runtime": 408.282,
839
+ "eval_samples_per_second": 152.142,
840
+ "eval_steps_per_second": 4.757,
841
+ "step": 26000
842
+ },
843
+ {
844
+ "epoch": 2.5833495808149736,
845
+ "grad_norm": 3.6922600269317627,
846
+ "learning_rate": 7.35e-05,
847
+ "loss": 1.065,
848
+ "step": 26500
849
+ },
850
+ {
851
+ "epoch": 2.5833495808149736,
852
+ "eval_accuracy": 0.7965003715883248,
853
+ "eval_loss": 1.0927073955535889,
854
+ "eval_runtime": 418.0036,
855
+ "eval_samples_per_second": 148.604,
856
+ "eval_steps_per_second": 4.646,
857
+ "step": 26500
858
+ },
859
+ {
860
+ "epoch": 2.6320920257360108,
861
+ "grad_norm": 3.770766019821167,
862
+ "learning_rate": 7.3e-05,
863
+ "loss": 1.051,
864
+ "step": 27000
865
+ },
866
+ {
867
+ "epoch": 2.6320920257360108,
868
+ "eval_accuracy": 0.7983237835847491,
869
+ "eval_loss": 1.0924187898635864,
870
+ "eval_runtime": 417.8408,
871
+ "eval_samples_per_second": 148.662,
872
+ "eval_steps_per_second": 4.648,
873
+ "step": 27000
874
+ },
875
+ {
876
+ "epoch": 2.6808344706570484,
877
+ "grad_norm": 3.686671733856201,
878
+ "learning_rate": 7.25e-05,
879
+ "loss": 1.051,
880
+ "step": 27500
881
+ },
882
+ {
883
+ "epoch": 2.6808344706570484,
884
+ "eval_accuracy": 0.7981858316674941,
885
+ "eval_loss": 1.092360019683838,
886
+ "eval_runtime": 406.1188,
887
+ "eval_samples_per_second": 152.953,
888
+ "eval_steps_per_second": 4.782,
889
+ "step": 27500
890
+ },
891
+ {
892
+ "epoch": 2.729576915578085,
893
+ "grad_norm": 3.7387359142303467,
894
+ "learning_rate": 7.2e-05,
895
+ "loss": 1.0472,
896
+ "step": 28000
897
+ },
898
+ {
899
+ "epoch": 2.729576915578085,
900
+ "eval_accuracy": 0.7984156631850994,
901
+ "eval_loss": 1.0957449674606323,
902
+ "eval_runtime": 417.9276,
903
+ "eval_samples_per_second": 148.631,
904
+ "eval_steps_per_second": 4.647,
905
+ "step": 28000
906
+ },
907
+ {
908
+ "epoch": 2.7783193604991228,
909
+ "grad_norm": 4.267466068267822,
910
+ "learning_rate": 7.15e-05,
911
+ "loss": 1.0467,
912
+ "step": 28500
913
+ },
914
+ {
915
+ "epoch": 2.7783193604991228,
916
+ "eval_accuracy": 0.7987033025065455,
917
+ "eval_loss": 1.1051925420761108,
918
+ "eval_runtime": 417.6089,
919
+ "eval_samples_per_second": 148.744,
920
+ "eval_steps_per_second": 4.65,
921
+ "step": 28500
922
+ },
923
+ {
924
+ "epoch": 2.82706180542016,
925
+ "grad_norm": 4.151378631591797,
926
+ "learning_rate": 7.1e-05,
927
+ "loss": 1.0383,
928
+ "step": 29000
929
+ },
930
+ {
931
+ "epoch": 2.82706180542016,
932
+ "eval_accuracy": 0.7991770419924266,
933
+ "eval_loss": 1.0816470384597778,
934
+ "eval_runtime": 417.8895,
935
+ "eval_samples_per_second": 148.645,
936
+ "eval_steps_per_second": 4.647,
937
+ "step": 29000
938
+ },
939
+ {
940
+ "epoch": 2.875804250341197,
941
+ "grad_norm": 3.69924259185791,
942
+ "learning_rate": 7.05e-05,
943
+ "loss": 1.0304,
944
+ "step": 29500
945
+ },
946
+ {
947
+ "epoch": 2.875804250341197,
948
+ "eval_accuracy": 0.8006994660557517,
949
+ "eval_loss": 1.0710912942886353,
950
+ "eval_runtime": 417.7606,
951
+ "eval_samples_per_second": 148.69,
952
+ "eval_steps_per_second": 4.649,
953
+ "step": 29500
954
+ },
955
+ {
956
+ "epoch": 2.9245466952622343,
957
+ "grad_norm": 3.548632860183716,
958
+ "learning_rate": 7e-05,
959
+ "loss": 1.0422,
960
+ "step": 30000
961
+ },
962
+ {
963
+ "epoch": 2.9245466952622343,
964
+ "eval_accuracy": 0.8007965409547303,
965
+ "eval_loss": 1.0759388208389282,
966
+ "eval_runtime": 418.895,
967
+ "eval_samples_per_second": 148.288,
968
+ "eval_steps_per_second": 4.636,
969
+ "step": 30000
970
+ },
971
+ {
972
+ "epoch": 2.9732891401832715,
973
+ "grad_norm": 3.8006861209869385,
974
+ "learning_rate": 6.95e-05,
975
+ "loss": 1.0338,
976
+ "step": 30500
977
+ },
978
+ {
979
+ "epoch": 2.9732891401832715,
980
+ "eval_accuracy": 0.8004465507568449,
981
+ "eval_loss": 1.0845800638198853,
982
+ "eval_runtime": 418.1301,
983
+ "eval_samples_per_second": 148.559,
984
+ "eval_steps_per_second": 4.644,
985
+ "step": 30500
986
+ },
987
+ {
988
+ "epoch": 3.0220315851043087,
989
+ "grad_norm": 3.686068058013916,
990
+ "learning_rate": 6.9e-05,
991
+ "loss": 1.0241,
992
+ "step": 31000
993
+ },
994
+ {
995
+ "epoch": 3.0220315851043087,
996
+ "eval_accuracy": 0.8011609180647845,
997
+ "eval_loss": 1.06769597530365,
998
+ "eval_runtime": 406.9954,
999
+ "eval_samples_per_second": 152.623,
1000
+ "eval_steps_per_second": 4.772,
1001
+ "step": 31000
1002
+ },
1003
+ {
1004
+ "epoch": 3.070774030025346,
1005
+ "grad_norm": 3.8143186569213867,
1006
+ "learning_rate": 6.850000000000001e-05,
1007
+ "loss": 1.0159,
1008
+ "step": 31500
1009
+ },
1010
+ {
1011
+ "epoch": 3.070774030025346,
1012
+ "eval_accuracy": 0.8013632123849934,
1013
+ "eval_loss": 1.0762627124786377,
1014
+ "eval_runtime": 419.6348,
1015
+ "eval_samples_per_second": 148.026,
1016
+ "eval_steps_per_second": 4.628,
1017
+ "step": 31500
1018
+ },
1019
+ {
1020
+ "epoch": 3.1195164749463835,
1021
+ "grad_norm": 4.718136787414551,
1022
+ "learning_rate": 6.800000000000001e-05,
1023
+ "loss": 1.0156,
1024
+ "step": 32000
1025
+ },
1026
+ {
1027
+ "epoch": 3.1195164749463835,
1028
+ "eval_accuracy": 0.8020213422523931,
1029
+ "eval_loss": 1.0555154085159302,
1030
+ "eval_runtime": 416.7929,
1031
+ "eval_samples_per_second": 149.036,
1032
+ "eval_steps_per_second": 4.659,
1033
+ "step": 32000
1034
+ },
1035
+ {
1036
+ "epoch": 3.1682589198674207,
1037
+ "grad_norm": 4.645066738128662,
1038
+ "learning_rate": 6.750000000000001e-05,
1039
+ "loss": 1.0125,
1040
+ "step": 32500
1041
+ },
1042
+ {
1043
+ "epoch": 3.1682589198674207,
1044
+ "eval_accuracy": 0.8024784559116106,
1045
+ "eval_loss": 1.0634628534317017,
1046
+ "eval_runtime": 406.8171,
1047
+ "eval_samples_per_second": 152.69,
1048
+ "eval_steps_per_second": 4.774,
1049
+ "step": 32500
1050
+ },
1051
+ {
1052
+ "epoch": 3.217001364788458,
1053
+ "grad_norm": 4.164463996887207,
1054
+ "learning_rate": 6.7e-05,
1055
+ "loss": 1.0139,
1056
+ "step": 33000
1057
+ },
1058
+ {
1059
+ "epoch": 3.217001364788458,
1060
+ "eval_accuracy": 0.8032846270208084,
1061
+ "eval_loss": 1.0512491464614868,
1062
+ "eval_runtime": 416.8591,
1063
+ "eval_samples_per_second": 149.012,
1064
+ "eval_steps_per_second": 4.659,
1065
+ "step": 33000
1066
+ },
1067
+ {
1068
+ "epoch": 3.265743809709495,
1069
+ "grad_norm": 3.9483578205108643,
1070
+ "learning_rate": 6.65e-05,
1071
+ "loss": 1.0162,
1072
+ "step": 33500
1073
+ },
1074
+ {
1075
+ "epoch": 3.265743809709495,
1076
+ "eval_accuracy": 0.8031067271435598,
1077
+ "eval_loss": 1.05771005153656,
1078
+ "eval_runtime": 416.679,
1079
+ "eval_samples_per_second": 149.076,
1080
+ "eval_steps_per_second": 4.661,
1081
+ "step": 33500
1082
+ },
1083
+ {
1084
+ "epoch": 3.3144862546305323,
1085
+ "grad_norm": 3.5133304595947266,
1086
+ "learning_rate": 6.6e-05,
1087
+ "loss": 1.004,
1088
+ "step": 34000
1089
+ },
1090
+ {
1091
+ "epoch": 3.3144862546305323,
1092
+ "eval_accuracy": 0.8038770049824707,
1093
+ "eval_loss": 1.0553202629089355,
1094
+ "eval_runtime": 405.7221,
1095
+ "eval_samples_per_second": 153.102,
1096
+ "eval_steps_per_second": 4.787,
1097
+ "step": 34000
1098
+ },
1099
+ {
1100
+ "epoch": 3.3632286995515694,
1101
+ "grad_norm": 3.9474008083343506,
1102
+ "learning_rate": 6.55e-05,
1103
+ "loss": 0.9951,
1104
+ "step": 34500
1105
+ },
1106
+ {
1107
+ "epoch": 3.3632286995515694,
1108
+ "eval_accuracy": 0.8041849568994928,
1109
+ "eval_loss": 1.0525976419448853,
1110
+ "eval_runtime": 416.6873,
1111
+ "eval_samples_per_second": 149.073,
1112
+ "eval_steps_per_second": 4.661,
1113
+ "step": 34500
1114
+ },
1115
+ {
1116
+ "epoch": 3.4119711444726066,
1117
+ "grad_norm": 3.9235286712646484,
1118
+ "learning_rate": 6.500000000000001e-05,
1119
+ "loss": 0.9993,
1120
+ "step": 35000
1121
+ },
1122
+ {
1123
+ "epoch": 3.4119711444726066,
1124
+ "eval_accuracy": 0.804642827866986,
1125
+ "eval_loss": NaN,
1126
+ "eval_runtime": 417.3671,
1127
+ "eval_samples_per_second": 148.831,
1128
+ "eval_steps_per_second": 4.653,
1129
+ "step": 35000
1130
+ },
1131
+ {
1132
+ "epoch": 3.460713589393644,
1133
+ "grad_norm": 5.006605625152588,
1134
+ "learning_rate": 6.450000000000001e-05,
1135
+ "loss": 1.0039,
1136
+ "step": 35500
1137
+ },
1138
+ {
1139
+ "epoch": 3.460713589393644,
1140
+ "eval_accuracy": 0.8052872089101958,
1141
+ "eval_loss": 1.0506031513214111,
1142
+ "eval_runtime": 417.1878,
1143
+ "eval_samples_per_second": 148.895,
1144
+ "eval_steps_per_second": 4.655,
1145
+ "step": 35500
1146
+ },
1147
+ {
1148
+ "epoch": 3.509456034314681,
1149
+ "grad_norm": 3.453827142715454,
1150
+ "learning_rate": 6.400000000000001e-05,
1151
+ "loss": 1.0052,
1152
+ "step": 36000
1153
+ },
1154
+ {
1155
+ "epoch": 3.509456034314681,
1156
+ "eval_accuracy": 0.8053784625961633,
1157
+ "eval_loss": 1.0422734022140503,
1158
+ "eval_runtime": 405.9932,
1159
+ "eval_samples_per_second": 153.0,
1160
+ "eval_steps_per_second": 4.783,
1161
+ "step": 36000
1162
+ },
1163
+ {
1164
+ "epoch": 3.5581984792357186,
1165
+ "grad_norm": 4.76883602142334,
1166
+ "learning_rate": 6.35e-05,
1167
+ "loss": 0.9931,
1168
+ "step": 36500
1169
+ },
1170
+ {
1171
+ "epoch": 3.5581984792357186,
1172
+ "eval_accuracy": 0.8062917178441975,
1173
+ "eval_loss": 1.038949728012085,
1174
+ "eval_runtime": 417.023,
1175
+ "eval_samples_per_second": 148.953,
1176
+ "eval_steps_per_second": 4.657,
1177
+ "step": 36500
1178
+ },
1179
+ {
1180
+ "epoch": 3.606940924156756,
1181
+ "grad_norm": 4.503074645996094,
1182
+ "learning_rate": 6.3e-05,
1183
+ "loss": 0.9913,
1184
+ "step": 37000
1185
+ },
1186
+ {
1187
+ "epoch": 3.606940924156756,
1188
+ "eval_accuracy": 0.8066731304811298,
1189
+ "eval_loss": 1.0464720726013184,
1190
+ "eval_runtime": 417.9854,
1191
+ "eval_samples_per_second": 148.61,
1192
+ "eval_steps_per_second": 4.646,
1193
+ "step": 37000
1194
+ },
1195
+ {
1196
+ "epoch": 3.655683369077793,
1197
+ "grad_norm": 3.608944892883301,
1198
+ "learning_rate": 6.25e-05,
1199
+ "loss": 0.9954,
1200
+ "step": 37500
1201
+ },
1202
+ {
1203
+ "epoch": 3.655683369077793,
1204
+ "eval_accuracy": 0.8066646897067399,
1205
+ "eval_loss": 1.0526313781738281,
1206
+ "eval_runtime": 417.3272,
1207
+ "eval_samples_per_second": 148.845,
1208
+ "eval_steps_per_second": 4.653,
1209
+ "step": 37500
1210
+ },
1211
+ {
1212
+ "epoch": 3.70442581399883,
1213
+ "grad_norm": 4.124717712402344,
1214
+ "learning_rate": 6.2e-05,
1215
+ "loss": 0.9934,
1216
+ "step": 38000
1217
+ },
1218
+ {
1219
+ "epoch": 3.70442581399883,
1220
+ "eval_accuracy": 0.8077767724973488,
1221
+ "eval_loss": 1.0216541290283203,
1222
+ "eval_runtime": 407.2038,
1223
+ "eval_samples_per_second": 152.545,
1224
+ "eval_steps_per_second": 4.769,
1225
+ "step": 38000
1226
+ },
1227
+ {
1228
+ "epoch": 3.7531682589198674,
1229
+ "grad_norm": 4.475372314453125,
1230
+ "learning_rate": 6.15e-05,
1231
+ "loss": 0.9801,
1232
+ "step": 38500
1233
+ },
1234
+ {
1235
+ "epoch": 3.7531682589198674,
1236
+ "eval_accuracy": 0.8071986044608647,
1237
+ "eval_loss": 1.0320008993148804,
1238
+ "eval_runtime": 417.0159,
1239
+ "eval_samples_per_second": 148.956,
1240
+ "eval_steps_per_second": 4.657,
1241
+ "step": 38500
1242
+ },
1243
+ {
1244
+ "epoch": 3.8019107038409046,
1245
+ "grad_norm": 3.5460593700408936,
1246
+ "learning_rate": 6.1e-05,
1247
+ "loss": 0.9793,
1248
+ "step": 39000
1249
+ },
1250
+ {
1251
+ "epoch": 3.8019107038409046,
1252
+ "eval_accuracy": 0.8078482967630676,
1253
+ "eval_loss": 1.0231602191925049,
1254
+ "eval_runtime": 417.531,
1255
+ "eval_samples_per_second": 148.772,
1256
+ "eval_steps_per_second": 4.651,
1257
+ "step": 39000
1258
+ },
1259
+ {
1260
+ "epoch": 3.8506531487619418,
1261
+ "grad_norm": 3.9331490993499756,
1262
+ "learning_rate": 6.05e-05,
1263
+ "loss": 0.9839,
1264
+ "step": 39500
1265
+ },
1266
+ {
1267
+ "epoch": 3.8506531487619418,
1268
+ "eval_accuracy": 0.8081606791887387,
1269
+ "eval_loss": 1.023592472076416,
1270
+ "eval_runtime": 418.8767,
1271
+ "eval_samples_per_second": 148.294,
1272
+ "eval_steps_per_second": 4.636,
1273
+ "step": 39500
1274
+ },
1275
+ {
1276
+ "epoch": 3.8993955936829794,
1277
+ "grad_norm": 3.0806970596313477,
1278
+ "learning_rate": 6e-05,
1279
+ "loss": 0.9679,
1280
+ "step": 40000
1281
+ },
1282
+ {
1283
+ "epoch": 3.8993955936829794,
1284
+ "eval_accuracy": 0.8086802173852721,
1285
+ "eval_loss": 1.02633535861969,
1286
+ "eval_runtime": 406.3522,
1287
+ "eval_samples_per_second": 152.865,
1288
+ "eval_steps_per_second": 4.779,
1289
+ "step": 40000
1290
+ },
1291
+ {
1292
+ "epoch": 3.948138038604016,
1293
+ "grad_norm": 3.3825278282165527,
1294
+ "learning_rate": 5.95e-05,
1295
+ "loss": 0.9725,
1296
+ "step": 40500
1297
+ },
1298
+ {
1299
+ "epoch": 3.948138038604016,
1300
+ "eval_accuracy": 0.80833269637443,
1301
+ "eval_loss": 1.04069983959198,
1302
+ "eval_runtime": 405.9788,
1303
+ "eval_samples_per_second": 153.006,
1304
+ "eval_steps_per_second": 4.784,
1305
+ "step": 40500
1306
+ },
1307
+ {
1308
+ "epoch": 3.9968804835250538,
1309
+ "grad_norm": 3.728228807449341,
1310
+ "learning_rate": 5.9e-05,
1311
+ "loss": 0.9734,
1312
+ "step": 41000
1313
+ },
1314
+ {
1315
+ "epoch": 3.9968804835250538,
1316
+ "eval_accuracy": 0.8088856850276098,
1317
+ "eval_loss": 1.0243360996246338,
1318
+ "eval_runtime": 418.2067,
1319
+ "eval_samples_per_second": 148.532,
1320
+ "eval_steps_per_second": 4.644,
1321
+ "step": 41000
1322
+ },
1323
+ {
1324
+ "epoch": 4.0456229284460905,
1325
+ "grad_norm": 3.1366865634918213,
1326
+ "learning_rate": 5.85e-05,
1327
+ "loss": 0.9615,
1328
+ "step": 41500
1329
+ },
1330
+ {
1331
+ "epoch": 4.0456229284460905,
1332
+ "eval_accuracy": 0.8099273789501316,
1333
+ "eval_loss": 1.0140135288238525,
1334
+ "eval_runtime": 417.4989,
1335
+ "eval_samples_per_second": 148.784,
1336
+ "eval_steps_per_second": 4.652,
1337
+ "step": 41500
1338
+ },
1339
+ {
1340
+ "epoch": 4.094365373367128,
1341
+ "grad_norm": 4.486032009124756,
1342
+ "learning_rate": 5.8e-05,
1343
+ "loss": 0.9606,
1344
+ "step": 42000
1345
+ },
1346
+ {
1347
+ "epoch": 4.094365373367128,
1348
+ "eval_accuracy": 0.8099747385653199,
1349
+ "eval_loss": 1.0296921730041504,
1350
+ "eval_runtime": 420.3851,
1351
+ "eval_samples_per_second": 147.762,
1352
+ "eval_steps_per_second": 4.62,
1353
+ "step": 42000
1354
+ },
1355
+ {
1356
+ "epoch": 4.143107818288165,
1357
+ "grad_norm": 3.989729881286621,
1358
+ "learning_rate": 5.7499999999999995e-05,
1359
+ "loss": 0.9634,
1360
+ "step": 42500
1361
+ },
1362
+ {
1363
+ "epoch": 4.143107818288165,
1364
+ "eval_accuracy": 0.8108008770195908,
1365
+ "eval_loss": 1.0081605911254883,
1366
+ "eval_runtime": 406.5978,
1367
+ "eval_samples_per_second": 152.773,
1368
+ "eval_steps_per_second": 4.776,
1369
+ "step": 42500
1370
+ },
1371
+ {
1372
+ "epoch": 4.1918502632092025,
1373
+ "grad_norm": 3.4221575260162354,
1374
+ "learning_rate": 5.6999999999999996e-05,
1375
+ "loss": 0.9628,
1376
+ "step": 43000
1377
+ },
1378
+ {
1379
+ "epoch": 4.1918502632092025,
1380
+ "eval_accuracy": 0.8107686235924147,
1381
+ "eval_loss": 1.018919825553894,
1382
+ "eval_runtime": 419.8469,
1383
+ "eval_samples_per_second": 147.952,
1384
+ "eval_steps_per_second": 4.625,
1385
+ "step": 43000
1386
+ },
1387
+ {
1388
+ "epoch": 4.24059270813024,
1389
+ "grad_norm": 3.121842861175537,
1390
+ "learning_rate": 5.65e-05,
1391
+ "loss": 0.9544,
1392
+ "step": 43500
1393
+ },
1394
+ {
1395
+ "epoch": 4.24059270813024,
1396
+ "eval_accuracy": 0.8112811526798606,
1397
+ "eval_loss": 1.005543828010559,
1398
+ "eval_runtime": 418.6747,
1399
+ "eval_samples_per_second": 148.366,
1400
+ "eval_steps_per_second": 4.638,
1401
+ "step": 43500
1402
+ },
1403
+ {
1404
+ "epoch": 4.289335153051277,
1405
+ "grad_norm": 3.5650501251220703,
1406
+ "learning_rate": 5.6000000000000006e-05,
1407
+ "loss": 0.949,
1408
+ "step": 44000
1409
+ },
1410
+ {
1411
+ "epoch": 4.289335153051277,
1412
+ "eval_accuracy": 0.8118240945446494,
1413
+ "eval_loss": 1.0131685733795166,
1414
+ "eval_runtime": 407.6808,
1415
+ "eval_samples_per_second": 152.367,
1416
+ "eval_steps_per_second": 4.764,
1417
+ "step": 44000
1418
+ },
1419
+ {
1420
+ "epoch": 4.3380775979723145,
1421
+ "grad_norm": 3.2165634632110596,
1422
+ "learning_rate": 5.550000000000001e-05,
1423
+ "loss": 0.9522,
1424
+ "step": 44500
1425
+ },
1426
+ {
1427
+ "epoch": 4.3380775979723145,
1428
+ "eval_accuracy": 0.8115780246613691,
1429
+ "eval_loss": 1.0220428705215454,
1430
+ "eval_runtime": 405.9231,
1431
+ "eval_samples_per_second": 153.027,
1432
+ "eval_steps_per_second": 4.784,
1433
+ "step": 44500
1434
+ },
1435
+ {
1436
+ "epoch": 4.386820042893351,
1437
+ "grad_norm": 2.9322192668914795,
1438
+ "learning_rate": 5.500000000000001e-05,
1439
+ "loss": 0.9431,
1440
+ "step": 45000
1441
+ },
1442
+ {
1443
+ "epoch": 4.386820042893351,
1444
+ "eval_accuracy": 0.8120671097872612,
1445
+ "eval_loss": 1.0120651721954346,
1446
+ "eval_runtime": 418.1628,
1447
+ "eval_samples_per_second": 148.547,
1448
+ "eval_steps_per_second": 4.644,
1449
+ "step": 45000
1450
+ },
1451
+ {
1452
+ "epoch": 4.435562487814389,
1453
+ "grad_norm": 3.452725887298584,
1454
+ "learning_rate": 5.45e-05,
1455
+ "loss": 0.9448,
1456
+ "step": 45500
1457
+ },
1458
+ {
1459
+ "epoch": 4.435562487814389,
1460
+ "eval_accuracy": 0.8127470780385738,
1461
+ "eval_loss": 0.9997569918632507,
1462
+ "eval_runtime": 418.1188,
1463
+ "eval_samples_per_second": 148.563,
1464
+ "eval_steps_per_second": 4.645,
1465
+ "step": 45500
1466
+ },
1467
+ {
1468
+ "epoch": 4.484304932735426,
1469
+ "grad_norm": 3.1269946098327637,
1470
+ "learning_rate": 5.4000000000000005e-05,
1471
+ "loss": 0.9424,
1472
+ "step": 46000
1473
+ },
1474
+ {
1475
+ "epoch": 4.484304932735426,
1476
+ "eval_accuracy": 0.812740042747058,
1477
+ "eval_loss": 1.001397967338562,
1478
+ "eval_runtime": 407.1549,
1479
+ "eval_samples_per_second": 152.564,
1480
+ "eval_steps_per_second": 4.77,
1481
+ "step": 46000
1482
+ },
1483
+ {
1484
+ "epoch": 4.533047377656463,
1485
+ "grad_norm": 2.9431774616241455,
1486
+ "learning_rate": 5.3500000000000006e-05,
1487
+ "loss": 0.9404,
1488
+ "step": 46500
1489
+ },
1490
+ {
1491
+ "epoch": 4.533047377656463,
1492
+ "eval_accuracy": 0.8133600475471777,
1493
+ "eval_loss": 0.9982110261917114,
1494
+ "eval_runtime": 419.2597,
1495
+ "eval_samples_per_second": 148.159,
1496
+ "eval_steps_per_second": 4.632,
1497
+ "step": 46500
1498
+ },
1499
+ {
1500
+ "epoch": 4.5817898225775,
1501
+ "grad_norm": 3.8127102851867676,
1502
+ "learning_rate": 5.300000000000001e-05,
1503
+ "loss": 0.9474,
1504
+ "step": 47000
1505
+ },
1506
+ {
1507
+ "epoch": 4.5817898225775,
1508
+ "eval_accuracy": 0.8133919109774917,
1509
+ "eval_loss": 0.9953769445419312,
1510
+ "eval_runtime": 407.8044,
1511
+ "eval_samples_per_second": 152.321,
1512
+ "eval_steps_per_second": 4.762,
1513
+ "step": 47000
1514
+ },
1515
+ {
1516
+ "epoch": 4.630532267498538,
1517
+ "grad_norm": 3.686004161834717,
1518
+ "learning_rate": 5.25e-05,
1519
+ "loss": 0.94,
1520
+ "step": 47500
1521
+ },
1522
+ {
1523
+ "epoch": 4.630532267498538,
1524
+ "eval_accuracy": 0.8134523016816173,
1525
+ "eval_loss": 1.0122311115264893,
1526
+ "eval_runtime": 410.1647,
1527
+ "eval_samples_per_second": 151.444,
1528
+ "eval_steps_per_second": 4.735,
1529
+ "step": 47500
1530
+ },
1531
+ {
1532
+ "epoch": 4.679274712419575,
1533
+ "grad_norm": 4.049985885620117,
1534
+ "learning_rate": 5.2000000000000004e-05,
1535
+ "loss": 0.9386,
1536
+ "step": 48000
1537
+ },
1538
+ {
1539
+ "epoch": 4.679274712419575,
1540
+ "eval_accuracy": 0.8137329160966164,
1541
+ "eval_loss": 0.9996346235275269,
1542
+ "eval_runtime": 407.6298,
1543
+ "eval_samples_per_second": 152.386,
1544
+ "eval_steps_per_second": 4.764,
1545
+ "step": 48000
1546
+ },
1547
+ {
1548
+ "epoch": 4.728017157340612,
1549
+ "grad_norm": 3.2507503032684326,
1550
+ "learning_rate": 5.1500000000000005e-05,
1551
+ "loss": 0.9384,
1552
+ "step": 48500
1553
+ },
1554
+ {
1555
+ "epoch": 4.728017157340612,
1556
+ "eval_accuracy": 0.8149296571374698,
1557
+ "eval_loss": 0.9956032633781433,
1558
+ "eval_runtime": 406.976,
1559
+ "eval_samples_per_second": 152.631,
1560
+ "eval_steps_per_second": 4.772,
1561
+ "step": 48500
1562
+ },
1563
+ {
1564
+ "epoch": 4.77675960226165,
1565
+ "grad_norm": 3.3173000812530518,
1566
+ "learning_rate": 5.1000000000000006e-05,
1567
+ "loss": 0.9364,
1568
+ "step": 49000
1569
+ },
1570
+ {
1571
+ "epoch": 4.77675960226165,
1572
+ "eval_accuracy": 0.8157672114344285,
1573
+ "eval_loss": 0.9892277121543884,
1574
+ "eval_runtime": 408.661,
1575
+ "eval_samples_per_second": 152.001,
1576
+ "eval_steps_per_second": 4.752,
1577
+ "step": 49000
1578
+ },
1579
+ {
1580
+ "epoch": 4.825502047182686,
1581
+ "grad_norm": 3.604827642440796,
1582
+ "learning_rate": 5.05e-05,
1583
+ "loss": 0.9378,
1584
+ "step": 49500
1585
+ },
1586
+ {
1587
+ "epoch": 4.825502047182686,
1588
+ "eval_accuracy": 0.8150183266816289,
1589
+ "eval_loss": 0.9937326908111572,
1590
+ "eval_runtime": 419.6753,
1591
+ "eval_samples_per_second": 148.012,
1592
+ "eval_steps_per_second": 4.627,
1593
+ "step": 49500
1594
+ },
1595
+ {
1596
+ "epoch": 4.874244492103724,
1597
+ "grad_norm": 3.5434885025024414,
1598
+ "learning_rate": 5e-05,
1599
+ "loss": 0.9281,
1600
+ "step": 50000
1601
+ },
1602
+ {
1603
+ "epoch": 4.874244492103724,
1604
+ "eval_accuracy": 0.8152856845597338,
1605
+ "eval_loss": 0.9860332012176514,
1606
+ "eval_runtime": 418.3134,
1607
+ "eval_samples_per_second": 148.494,
1608
+ "eval_steps_per_second": 4.642,
1609
+ "step": 50000
1610
+ },
1611
+ {
1612
+ "epoch": 4.922986937024762,
1613
+ "grad_norm": 3.5801565647125244,
1614
+ "learning_rate": 4.9500000000000004e-05,
1615
+ "loss": 0.9259,
1616
+ "step": 50500
1617
+ },
1618
+ {
1619
+ "epoch": 4.922986937024762,
1620
+ "eval_accuracy": 0.8157059324837188,
1621
+ "eval_loss": 0.9821184873580933,
1622
+ "eval_runtime": 418.3936,
1623
+ "eval_samples_per_second": 148.465,
1624
+ "eval_steps_per_second": 4.642,
1625
+ "step": 50500
1626
+ },
1627
+ {
1628
+ "epoch": 4.971729381945798,
1629
+ "grad_norm": 3.7391278743743896,
1630
+ "learning_rate": 4.9e-05,
1631
+ "loss": 0.9245,
1632
+ "step": 51000
1633
+ },
1634
+ {
1635
+ "epoch": 4.971729381945798,
1636
+ "eval_accuracy": 0.8167135547079718,
1637
+ "eval_loss": 0.9834054112434387,
1638
+ "eval_runtime": 418.1441,
1639
+ "eval_samples_per_second": 148.554,
1640
+ "eval_steps_per_second": 4.644,
1641
+ "step": 51000
1642
+ },
1643
+ {
1644
+ "epoch": 5.020471826866836,
1645
+ "grad_norm": 3.5294556617736816,
1646
+ "learning_rate": 4.85e-05,
1647
+ "loss": 0.93,
1648
+ "step": 51500
1649
+ },
1650
+ {
1651
+ "epoch": 5.020471826866836,
1652
+ "eval_accuracy": 0.816996770373217,
1653
+ "eval_loss": 0.9669159054756165,
1654
+ "eval_runtime": 416.8904,
1655
+ "eval_samples_per_second": 149.001,
1656
+ "eval_steps_per_second": 4.658,
1657
+ "step": 51500
1658
+ },
1659
+ {
1660
+ "epoch": 5.069214271787873,
1661
+ "grad_norm": 3.1137163639068604,
1662
+ "learning_rate": 4.8e-05,
1663
+ "loss": 0.9247,
1664
+ "step": 52000
1665
+ },
1666
+ {
1667
+ "epoch": 5.069214271787873,
1668
+ "eval_accuracy": 0.8173001472872028,
1669
+ "eval_loss": 0.969613254070282,
1670
+ "eval_runtime": 405.5038,
1671
+ "eval_samples_per_second": 153.185,
1672
+ "eval_steps_per_second": 4.789,
1673
+ "step": 52000
1674
+ },
1675
+ {
1676
+ "epoch": 5.11795671670891,
1677
+ "grad_norm": 2.89302659034729,
1678
+ "learning_rate": 4.75e-05,
1679
+ "loss": 0.9154,
1680
+ "step": 52500
1681
+ },
1682
+ {
1683
+ "epoch": 5.11795671670891,
1684
+ "eval_accuracy": 0.8169749903064084,
1685
+ "eval_loss": 0.9733508229255676,
1686
+ "eval_runtime": 408.6617,
1687
+ "eval_samples_per_second": 152.001,
1688
+ "eval_steps_per_second": 4.752,
1689
+ "step": 52500
1690
+ },
1691
+ {
1692
+ "epoch": 5.166699161629947,
1693
+ "grad_norm": 3.131472587585449,
1694
+ "learning_rate": 4.7e-05,
1695
+ "loss": 0.9153,
1696
+ "step": 53000
1697
+ },
1698
+ {
1699
+ "epoch": 5.166699161629947,
1700
+ "eval_accuracy": 0.8171414906960495,
1701
+ "eval_loss": 0.9911856651306152,
1702
+ "eval_runtime": 406.327,
1703
+ "eval_samples_per_second": 152.874,
1704
+ "eval_steps_per_second": 4.779,
1705
+ "step": 53000
1706
+ },
1707
+ {
1708
+ "epoch": 5.215441606550985,
1709
+ "grad_norm": 3.7570555210113525,
1710
+ "learning_rate": 4.6500000000000005e-05,
1711
+ "loss": 0.9104,
1712
+ "step": 53500
1713
+ },
1714
+ {
1715
+ "epoch": 5.215441606550985,
1716
+ "eval_accuracy": 0.8174119137884183,
1717
+ "eval_loss": 0.9812771677970886,
1718
+ "eval_runtime": 406.9743,
1719
+ "eval_samples_per_second": 152.631,
1720
+ "eval_steps_per_second": 4.772,
1721
+ "step": 53500
1722
+ },
1723
+ {
1724
+ "epoch": 5.2641840514720215,
1725
+ "grad_norm": 3.2733726501464844,
1726
+ "learning_rate": 4.600000000000001e-05,
1727
+ "loss": 0.912,
1728
+ "step": 54000
1729
+ },
1730
+ {
1731
+ "epoch": 5.2641840514720215,
1732
+ "eval_accuracy": 0.8181780925962017,
1733
+ "eval_loss": 0.9795467257499695,
1734
+ "eval_runtime": 417.1257,
1735
+ "eval_samples_per_second": 148.917,
1736
+ "eval_steps_per_second": 4.656,
1737
+ "step": 54000
1738
+ },
1739
+ {
1740
+ "epoch": 5.312926496393059,
1741
+ "grad_norm": 3.646557569503784,
1742
+ "learning_rate": 4.55e-05,
1743
+ "loss": 0.9099,
1744
+ "step": 54500
1745
+ },
1746
+ {
1747
+ "epoch": 5.312926496393059,
1748
+ "eval_accuracy": 0.8182088182333788,
1749
+ "eval_loss": 0.9727326035499573,
1750
+ "eval_runtime": 417.132,
1751
+ "eval_samples_per_second": 148.914,
1752
+ "eval_steps_per_second": 4.656,
1753
+ "step": 54500
1754
+ },
1755
+ {
1756
+ "epoch": 5.361668941314097,
1757
+ "grad_norm": 3.503202199935913,
1758
+ "learning_rate": 4.5e-05,
1759
+ "loss": 0.9075,
1760
+ "step": 55000
1761
+ },
1762
+ {
1763
+ "epoch": 5.361668941314097,
1764
+ "eval_accuracy": 0.8184407949709821,
1765
+ "eval_loss": 0.9728838205337524,
1766
+ "eval_runtime": 417.243,
1767
+ "eval_samples_per_second": 148.875,
1768
+ "eval_steps_per_second": 4.654,
1769
+ "step": 55000
1770
+ },
1771
+ {
1772
+ "epoch": 5.4104113862351335,
1773
+ "grad_norm": 3.7343647480010986,
1774
+ "learning_rate": 4.4500000000000004e-05,
1775
+ "loss": 0.908,
1776
+ "step": 55500
1777
+ },
1778
+ {
1779
+ "epoch": 5.4104113862351335,
1780
+ "eval_accuracy": 0.8189042158006526,
1781
+ "eval_loss": 0.9732007384300232,
1782
+ "eval_runtime": 417.2141,
1783
+ "eval_samples_per_second": 148.885,
1784
+ "eval_steps_per_second": 4.655,
1785
+ "step": 55500
1786
+ },
1787
+ {
1788
+ "epoch": 5.459153831156171,
1789
+ "grad_norm": 3.133802652359009,
1790
+ "learning_rate": 4.4000000000000006e-05,
1791
+ "loss": 0.9026,
1792
+ "step": 56000
1793
+ },
1794
+ {
1795
+ "epoch": 5.459153831156171,
1796
+ "eval_accuracy": 0.8201767667683212,
1797
+ "eval_loss": 0.9600487947463989,
1798
+ "eval_runtime": 417.4136,
1799
+ "eval_samples_per_second": 148.814,
1800
+ "eval_steps_per_second": 4.652,
1801
+ "step": 56000
1802
+ },
1803
+ {
1804
+ "epoch": 5.507896276077208,
1805
+ "grad_norm": 3.379943609237671,
1806
+ "learning_rate": 4.35e-05,
1807
+ "loss": 0.9079,
1808
+ "step": 56500
1809
+ },
1810
+ {
1811
+ "epoch": 5.507896276077208,
1812
+ "eval_accuracy": 0.8192980524609018,
1813
+ "eval_loss": 0.9664751887321472,
1814
+ "eval_runtime": 417.6211,
1815
+ "eval_samples_per_second": 148.74,
1816
+ "eval_steps_per_second": 4.65,
1817
+ "step": 56500
1818
+ },
1819
+ {
1820
+ "epoch": 5.5566387209982455,
1821
+ "grad_norm": 3.35117244720459,
1822
+ "learning_rate": 4.3e-05,
1823
+ "loss": 0.8984,
1824
+ "step": 57000
1825
+ },
1826
+ {
1827
+ "epoch": 5.5566387209982455,
1828
+ "eval_accuracy": 0.8200202500664683,
1829
+ "eval_loss": 0.9683465957641602,
1830
+ "eval_runtime": 416.9849,
1831
+ "eval_samples_per_second": 148.967,
1832
+ "eval_steps_per_second": 4.657,
1833
+ "step": 57000
1834
+ },
1835
+ {
1836
+ "epoch": 5.605381165919282,
1837
+ "grad_norm": 3.655966281890869,
1838
+ "learning_rate": 4.25e-05,
1839
+ "loss": 0.9024,
1840
+ "step": 57500
1841
+ },
1842
+ {
1843
+ "epoch": 5.605381165919282,
1844
+ "eval_accuracy": 0.8204076766761219,
1845
+ "eval_loss": 0.9583476781845093,
1846
+ "eval_runtime": 407.0914,
1847
+ "eval_samples_per_second": 152.587,
1848
+ "eval_steps_per_second": 4.77,
1849
+ "step": 57500
1850
+ },
1851
+ {
1852
+ "epoch": 5.65412361084032,
1853
+ "grad_norm": 3.7306268215179443,
1854
+ "learning_rate": 4.2e-05,
1855
+ "loss": 0.8921,
1856
+ "step": 58000
1857
+ },
1858
+ {
1859
+ "epoch": 5.65412361084032,
1860
+ "eval_accuracy": 0.8197779799818017,
1861
+ "eval_loss": 0.9624991416931152,
1862
+ "eval_runtime": 417.1391,
1863
+ "eval_samples_per_second": 148.912,
1864
+ "eval_steps_per_second": 4.656,
1865
+ "step": 58000
1866
+ },
1867
+ {
1868
+ "epoch": 5.702866055761357,
1869
+ "grad_norm": 3.412425994873047,
1870
+ "learning_rate": 4.15e-05,
1871
+ "loss": 0.8981,
1872
+ "step": 58500
1873
+ },
1874
+ {
1875
+ "epoch": 5.702866055761357,
1876
+ "eval_accuracy": 0.8202077126346001,
1877
+ "eval_loss": 0.959416389465332,
1878
+ "eval_runtime": 407.0106,
1879
+ "eval_samples_per_second": 152.618,
1880
+ "eval_steps_per_second": 4.771,
1881
+ "step": 58500
1882
+ },
1883
+ {
1884
+ "epoch": 5.751608500682394,
1885
+ "grad_norm": 3.5848515033721924,
1886
+ "learning_rate": 4.1e-05,
1887
+ "loss": 0.8963,
1888
+ "step": 59000
1889
+ },
1890
+ {
1891
+ "epoch": 5.751608500682394,
1892
+ "eval_accuracy": 0.8210654104614679,
1893
+ "eval_loss": 0.9516875743865967,
1894
+ "eval_runtime": 417.8109,
1895
+ "eval_samples_per_second": 148.673,
1896
+ "eval_steps_per_second": 4.648,
1897
+ "step": 59000
1898
+ },
1899
+ {
1900
+ "epoch": 5.800350945603432,
1901
+ "grad_norm": 3.3994343280792236,
1902
+ "learning_rate": 4.05e-05,
1903
+ "loss": 0.8882,
1904
+ "step": 59500
1905
+ },
1906
+ {
1907
+ "epoch": 5.800350945603432,
1908
+ "eval_accuracy": 0.8211162443184222,
1909
+ "eval_loss": 0.9653207063674927,
1910
+ "eval_runtime": 406.7456,
1911
+ "eval_samples_per_second": 152.717,
1912
+ "eval_steps_per_second": 4.774,
1913
+ "step": 59500
1914
+ },
1915
+ {
1916
+ "epoch": 5.849093390524469,
1917
+ "grad_norm": 3.743419885635376,
1918
+ "learning_rate": 4e-05,
1919
+ "loss": 0.8981,
1920
+ "step": 60000
1921
+ },
1922
+ {
1923
+ "epoch": 5.849093390524469,
1924
+ "eval_accuracy": 0.8213700909383002,
1925
+ "eval_loss": 0.950171947479248,
1926
+ "eval_runtime": 404.3291,
1927
+ "eval_samples_per_second": 153.63,
1928
+ "eval_steps_per_second": 4.803,
1929
+ "step": 60000
1930
+ },
1931
+ {
1932
+ "epoch": 5.897835835445506,
1933
+ "grad_norm": 3.55446720123291,
1934
+ "learning_rate": 3.9500000000000005e-05,
1935
+ "loss": 0.8783,
1936
+ "step": 60500
1937
+ },
1938
+ {
1939
+ "epoch": 5.897835835445506,
1940
+ "eval_accuracy": 0.8216780241795132,
1941
+ "eval_loss": 0.9581466913223267,
1942
+ "eval_runtime": 406.5142,
1943
+ "eval_samples_per_second": 152.804,
1944
+ "eval_steps_per_second": 4.777,
1945
+ "step": 60500
1946
+ },
1947
+ {
1948
+ "epoch": 5.946578280366543,
1949
+ "grad_norm": 3.7575368881225586,
1950
+ "learning_rate": 3.9000000000000006e-05,
1951
+ "loss": 0.8877,
1952
+ "step": 61000
1953
+ },
1954
+ {
1955
+ "epoch": 5.946578280366543,
1956
+ "eval_accuracy": 0.8223021733371326,
1957
+ "eval_loss": 0.9538587331771851,
1958
+ "eval_runtime": 418.445,
1959
+ "eval_samples_per_second": 148.447,
1960
+ "eval_steps_per_second": 4.641,
1961
+ "step": 61000
1962
+ },
1963
+ {
1964
+ "epoch": 5.995320725287581,
1965
+ "grad_norm": 3.153002977371216,
1966
+ "learning_rate": 3.85e-05,
1967
+ "loss": 0.8868,
1968
+ "step": 61500
1969
+ },
1970
+ {
1971
+ "epoch": 5.995320725287581,
1972
+ "eval_accuracy": 0.8223680440139836,
1973
+ "eval_loss": 0.9499430656433105,
1974
+ "eval_runtime": 405.4074,
1975
+ "eval_samples_per_second": 153.221,
1976
+ "eval_steps_per_second": 4.79,
1977
+ "step": 61500
1978
+ },
1979
+ {
1980
+ "epoch": 6.044063170208617,
1981
+ "grad_norm": 4.0667500495910645,
1982
+ "learning_rate": 3.8e-05,
1983
+ "loss": 0.8899,
1984
+ "step": 62000
1985
+ },
1986
+ {
1987
+ "epoch": 6.044063170208617,
1988
+ "eval_accuracy": 0.8224810332579563,
1989
+ "eval_loss": 0.9389235377311707,
1990
+ "eval_runtime": 407.9926,
1991
+ "eval_samples_per_second": 152.25,
1992
+ "eval_steps_per_second": 4.76,
1993
+ "step": 62000
1994
+ },
1995
+ {
1996
+ "epoch": 6.092805615129655,
1997
+ "grad_norm": 3.6726677417755127,
1998
+ "learning_rate": 3.7500000000000003e-05,
1999
+ "loss": 0.8682,
2000
+ "step": 62500
2001
+ },
2002
+ {
2003
+ "epoch": 6.092805615129655,
2004
+ "eval_accuracy": 0.8231253184052942,
2005
+ "eval_loss": 0.9457637071609497,
2006
+ "eval_runtime": 409.6859,
2007
+ "eval_samples_per_second": 151.621,
2008
+ "eval_steps_per_second": 4.74,
2009
+ "step": 62500
2010
+ },
2011
+ {
2012
+ "epoch": 6.141548060050692,
2013
+ "grad_norm": 3.4400100708007812,
2014
+ "learning_rate": 3.7e-05,
2015
+ "loss": 0.885,
2016
+ "step": 63000
2017
+ },
2018
+ {
2019
+ "epoch": 6.141548060050692,
2020
+ "eval_accuracy": 0.8236364297929629,
2021
+ "eval_loss": 0.9425154328346252,
2022
+ "eval_runtime": 406.4124,
2023
+ "eval_samples_per_second": 152.842,
2024
+ "eval_steps_per_second": 4.778,
2025
+ "step": 63000
2026
+ },
2027
+ {
2028
+ "epoch": 6.190290504971729,
2029
+ "grad_norm": 4.634922027587891,
2030
+ "learning_rate": 3.65e-05,
2031
+ "loss": 0.8745,
2032
+ "step": 63500
2033
+ },
2034
+ {
2035
+ "epoch": 6.190290504971729,
2036
+ "eval_accuracy": 0.8235738463597942,
2037
+ "eval_loss": 0.945899486541748,
2038
+ "eval_runtime": 405.9732,
2039
+ "eval_samples_per_second": 153.008,
2040
+ "eval_steps_per_second": 4.784,
2041
+ "step": 63500
2042
+ },
2043
+ {
2044
+ "epoch": 6.239032949892767,
2045
+ "grad_norm": 4.0379509925842285,
2046
+ "learning_rate": 3.6e-05,
2047
+ "loss": 0.8665,
2048
+ "step": 64000
2049
+ },
2050
+ {
2051
+ "epoch": 6.239032949892767,
2052
+ "eval_accuracy": 0.8238192088542137,
2053
+ "eval_loss": 0.9333313703536987,
2054
+ "eval_runtime": 406.2352,
2055
+ "eval_samples_per_second": 152.909,
2056
+ "eval_steps_per_second": 4.78,
2057
+ "step": 64000
2058
+ },
2059
+ {
2060
+ "epoch": 6.287775394813804,
2061
+ "grad_norm": 4.645595550537109,
2062
+ "learning_rate": 3.55e-05,
2063
+ "loss": 0.8775,
2064
+ "step": 64500
2065
+ },
2066
+ {
2067
+ "epoch": 6.287775394813804,
2068
+ "eval_accuracy": 0.8239884570656095,
2069
+ "eval_loss": 0.9405332207679749,
2070
+ "eval_runtime": 418.2514,
2071
+ "eval_samples_per_second": 148.516,
2072
+ "eval_steps_per_second": 4.643,
2073
+ "step": 64500
2074
+ },
2075
+ {
2076
+ "epoch": 6.336517839734841,
2077
+ "grad_norm": 4.875068187713623,
2078
+ "learning_rate": 3.5e-05,
2079
+ "loss": 0.8642,
2080
+ "step": 65000
2081
+ },
2082
+ {
2083
+ "epoch": 6.336517839734841,
2084
+ "eval_accuracy": 0.8242385226771245,
2085
+ "eval_loss": 0.9486621618270874,
2086
+ "eval_runtime": 417.1053,
2087
+ "eval_samples_per_second": 148.924,
2088
+ "eval_steps_per_second": 4.656,
2089
+ "step": 65000
2090
+ },
2091
+ {
2092
+ "epoch": 6.385260284655878,
2093
+ "grad_norm": 3.5531485080718994,
2094
+ "learning_rate": 3.45e-05,
2095
+ "loss": 0.8646,
2096
+ "step": 65500
2097
+ },
2098
+ {
2099
+ "epoch": 6.385260284655878,
2100
+ "eval_accuracy": 0.8246846366774455,
2101
+ "eval_loss": 0.9260883331298828,
2102
+ "eval_runtime": 419.2646,
2103
+ "eval_samples_per_second": 148.157,
2104
+ "eval_steps_per_second": 4.632,
2105
+ "step": 65500
2106
+ },
2107
+ {
2108
+ "epoch": 6.434002729576916,
2109
+ "grad_norm": 4.116947174072266,
2110
+ "learning_rate": 3.4000000000000007e-05,
2111
+ "loss": 0.8681,
2112
+ "step": 66000
2113
+ },
2114
+ {
2115
+ "epoch": 6.434002729576916,
2116
+ "eval_accuracy": 0.8250557813416932,
2117
+ "eval_loss": 0.9300101399421692,
2118
+ "eval_runtime": 417.6044,
2119
+ "eval_samples_per_second": 148.746,
2120
+ "eval_steps_per_second": 4.65,
2121
+ "step": 66000
2122
+ },
2123
+ {
2124
+ "epoch": 6.4827451744979525,
2125
+ "grad_norm": 3.6558291912078857,
2126
+ "learning_rate": 3.35e-05,
2127
+ "loss": 0.8694,
2128
+ "step": 66500
2129
+ },
2130
+ {
2131
+ "epoch": 6.4827451744979525,
2132
+ "eval_accuracy": 0.8257142012386084,
2133
+ "eval_loss": 0.9325526356697083,
2134
+ "eval_runtime": 407.9768,
2135
+ "eval_samples_per_second": 152.256,
2136
+ "eval_steps_per_second": 4.76,
2137
+ "step": 66500
2138
+ },
2139
+ {
2140
+ "epoch": 6.53148761941899,
2141
+ "grad_norm": 3.5170669555664062,
2142
+ "learning_rate": 3.3e-05,
2143
+ "loss": 0.8605,
2144
+ "step": 67000
2145
+ },
2146
+ {
2147
+ "epoch": 6.53148761941899,
2148
+ "eval_accuracy": 0.8253691543918565,
2149
+ "eval_loss": 0.9221523404121399,
2150
+ "eval_runtime": 417.7475,
2151
+ "eval_samples_per_second": 148.695,
2152
+ "eval_steps_per_second": 4.649,
2153
+ "step": 67000
2154
+ },
2155
+ {
2156
+ "epoch": 6.580230064340027,
2157
+ "grad_norm": 3.9361133575439453,
2158
+ "learning_rate": 3.2500000000000004e-05,
2159
+ "loss": 0.8636,
2160
+ "step": 67500
2161
+ },
2162
+ {
2163
+ "epoch": 6.580230064340027,
2164
+ "eval_accuracy": 0.8256220792916974,
2165
+ "eval_loss": 0.9299176335334778,
2166
+ "eval_runtime": 405.8973,
2167
+ "eval_samples_per_second": 153.036,
2168
+ "eval_steps_per_second": 4.784,
2169
+ "step": 67500
2170
+ },
2171
+ {
2172
+ "epoch": 6.6289725092610645,
2173
+ "grad_norm": 2.9539661407470703,
2174
+ "learning_rate": 3.2000000000000005e-05,
2175
+ "loss": 0.855,
2176
+ "step": 68000
2177
+ },
2178
+ {
2179
+ "epoch": 6.6289725092610645,
2180
+ "eval_accuracy": 0.8260808646917535,
2181
+ "eval_loss": 0.9166622161865234,
2182
+ "eval_runtime": 418.4165,
2183
+ "eval_samples_per_second": 148.457,
2184
+ "eval_steps_per_second": 4.641,
2185
+ "step": 68000
2186
+ },
2187
+ {
2188
+ "epoch": 6.677714954182102,
2189
+ "grad_norm": 3.324882745742798,
2190
+ "learning_rate": 3.15e-05,
2191
+ "loss": 0.8565,
2192
+ "step": 68500
2193
+ },
2194
+ {
2195
+ "epoch": 6.677714954182102,
2196
+ "eval_accuracy": 0.8265228789366582,
2197
+ "eval_loss": 0.9246287941932678,
2198
+ "eval_runtime": 404.3212,
2199
+ "eval_samples_per_second": 153.633,
2200
+ "eval_steps_per_second": 4.803,
2201
+ "step": 68500
2202
+ },
2203
+ {
2204
+ "epoch": 6.726457399103139,
2205
+ "grad_norm": 4.108856201171875,
2206
+ "learning_rate": 3.1e-05,
2207
+ "loss": 0.8531,
2208
+ "step": 69000
2209
+ },
2210
+ {
2211
+ "epoch": 6.726457399103139,
2212
+ "eval_accuracy": 0.826138264736776,
2213
+ "eval_loss": 0.9267178773880005,
2214
+ "eval_runtime": 417.5496,
2215
+ "eval_samples_per_second": 148.766,
2216
+ "eval_steps_per_second": 4.651,
2217
+ "step": 69000
2218
+ },
2219
+ {
2220
+ "epoch": 6.7751998440241765,
2221
+ "grad_norm": 2.7647440433502197,
2222
+ "learning_rate": 3.05e-05,
2223
+ "loss": 0.8598,
2224
+ "step": 69500
2225
+ },
2226
+ {
2227
+ "epoch": 6.7751998440241765,
2228
+ "eval_accuracy": 0.8265823639247751,
2229
+ "eval_loss": 0.9271634221076965,
2230
+ "eval_runtime": 406.295,
2231
+ "eval_samples_per_second": 152.886,
2232
+ "eval_steps_per_second": 4.78,
2233
+ "step": 69500
2234
+ },
2235
+ {
2236
+ "epoch": 6.823942288945213,
2237
+ "grad_norm": 3.1595242023468018,
2238
+ "learning_rate": 3e-05,
2239
+ "loss": 0.8476,
2240
+ "step": 70000
2241
+ },
2242
+ {
2243
+ "epoch": 6.823942288945213,
2244
+ "eval_accuracy": 0.8272099679007218,
2245
+ "eval_loss": 0.9176299571990967,
2246
+ "eval_runtime": 407.4234,
2247
+ "eval_samples_per_second": 152.463,
2248
+ "eval_steps_per_second": 4.767,
2249
+ "step": 70000
2250
+ },
2251
+ {
2252
+ "epoch": 6.872684733866251,
2253
+ "grad_norm": 3.26577091217041,
2254
+ "learning_rate": 2.95e-05,
2255
+ "loss": 0.8546,
2256
+ "step": 70500
2257
+ },
2258
+ {
2259
+ "epoch": 6.872684733866251,
2260
+ "eval_accuracy": 0.8276587581698454,
2261
+ "eval_loss": 0.9238181114196777,
2262
+ "eval_runtime": 417.6745,
2263
+ "eval_samples_per_second": 148.721,
2264
+ "eval_steps_per_second": 4.65,
2265
+ "step": 70500
2266
+ },
2267
+ {
2268
+ "epoch": 6.921427178787288,
2269
+ "grad_norm": 4.188171863555908,
2270
+ "learning_rate": 2.9e-05,
2271
+ "loss": 0.857,
2272
+ "step": 71000
2273
+ },
2274
+ {
2275
+ "epoch": 6.921427178787288,
2276
+ "eval_accuracy": 0.8276207824632725,
2277
+ "eval_loss": 0.9164689779281616,
2278
+ "eval_runtime": 410.2588,
2279
+ "eval_samples_per_second": 151.409,
2280
+ "eval_steps_per_second": 4.734,
2281
+ "step": 71000
2282
+ },
2283
+ {
2284
+ "epoch": 6.970169623708325,
2285
+ "grad_norm": 2.763139247894287,
2286
+ "learning_rate": 2.8499999999999998e-05,
2287
+ "loss": 0.8451,
2288
+ "step": 71500
2289
+ },
2290
+ {
2291
+ "epoch": 6.970169623708325,
2292
+ "eval_accuracy": 0.8271887327119185,
2293
+ "eval_loss": 0.9177712798118591,
2294
+ "eval_runtime": 407.3065,
2295
+ "eval_samples_per_second": 152.507,
2296
+ "eval_steps_per_second": 4.768,
2297
+ "step": 71500
2298
+ },
2299
+ {
2300
+ "epoch": 7.018912068629362,
2301
+ "grad_norm": 4.059235095977783,
2302
+ "learning_rate": 2.8000000000000003e-05,
2303
+ "loss": 0.8555,
2304
+ "step": 72000
2305
+ },
2306
+ {
2307
+ "epoch": 7.018912068629362,
2308
+ "eval_accuracy": 0.8276455598958156,
2309
+ "eval_loss": 0.9244613647460938,
2310
+ "eval_runtime": 407.3569,
2311
+ "eval_samples_per_second": 152.488,
2312
+ "eval_steps_per_second": 4.767,
2313
+ "step": 72000
2314
+ },
2315
+ {
2316
+ "epoch": 7.0676545135504,
2317
+ "grad_norm": 3.997545003890991,
2318
+ "learning_rate": 2.7500000000000004e-05,
2319
+ "loss": 0.8462,
2320
+ "step": 72500
2321
+ },
2322
+ {
2323
+ "epoch": 7.0676545135504,
2324
+ "eval_accuracy": 0.8275193754333664,
2325
+ "eval_loss": 0.9270544052124023,
2326
+ "eval_runtime": 410.4712,
2327
+ "eval_samples_per_second": 151.331,
2328
+ "eval_steps_per_second": 4.731,
2329
+ "step": 72500
2330
+ },
2331
+ {
2332
+ "epoch": 7.116396958471437,
2333
+ "grad_norm": 3.3070528507232666,
2334
+ "learning_rate": 2.7000000000000002e-05,
2335
+ "loss": 0.8406,
2336
+ "step": 73000
2337
+ },
2338
+ {
2339
+ "epoch": 7.116396958471437,
2340
+ "eval_accuracy": 0.827724122124491,
2341
+ "eval_loss": 0.907126247882843,
2342
+ "eval_runtime": 420.0303,
2343
+ "eval_samples_per_second": 147.887,
2344
+ "eval_steps_per_second": 4.623,
2345
+ "step": 73000
2346
+ },
2347
+ {
2348
+ "epoch": 7.165139403392474,
2349
+ "grad_norm": 3.3446497917175293,
2350
+ "learning_rate": 2.6500000000000004e-05,
2351
+ "loss": 0.8361,
2352
+ "step": 73500
2353
+ },
2354
+ {
2355
+ "epoch": 7.165139403392474,
2356
+ "eval_accuracy": 0.8287420232113283,
2357
+ "eval_loss": 0.90281742811203,
2358
+ "eval_runtime": 415.0764,
2359
+ "eval_samples_per_second": 149.652,
2360
+ "eval_steps_per_second": 4.679,
2361
+ "step": 73500
2362
+ },
2363
+ {
2364
+ "epoch": 7.213881848313512,
2365
+ "grad_norm": 3.962263345718384,
2366
+ "learning_rate": 2.6000000000000002e-05,
2367
+ "loss": 0.8407,
2368
+ "step": 74000
2369
+ },
2370
+ {
2371
+ "epoch": 7.213881848313512,
2372
+ "eval_accuracy": 0.8290002931584822,
2373
+ "eval_loss": 0.908825695514679,
2374
+ "eval_runtime": 421.665,
2375
+ "eval_samples_per_second": 147.314,
2376
+ "eval_steps_per_second": 4.606,
2377
+ "step": 74000
2378
+ },
2379
+ {
2380
+ "epoch": 7.262624293234548,
2381
+ "grad_norm": 3.412740468978882,
2382
+ "learning_rate": 2.5500000000000003e-05,
2383
+ "loss": 0.831,
2384
+ "step": 74500
2385
+ },
2386
+ {
2387
+ "epoch": 7.262624293234548,
2388
+ "eval_accuracy": 0.8298599086687342,
2389
+ "eval_loss": 0.9019548892974854,
2390
+ "eval_runtime": 411.246,
2391
+ "eval_samples_per_second": 151.046,
2392
+ "eval_steps_per_second": 4.722,
2393
+ "step": 74500
2394
+ },
2395
+ {
2396
+ "epoch": 7.311366738155586,
2397
+ "grad_norm": 3.3244595527648926,
2398
+ "learning_rate": 2.5e-05,
2399
+ "loss": 0.8394,
2400
+ "step": 75000
2401
+ },
2402
+ {
2403
+ "epoch": 7.311366738155586,
2404
+ "eval_accuracy": 0.829160436990889,
2405
+ "eval_loss": 0.9075309038162231,
2406
+ "eval_runtime": 411.1707,
2407
+ "eval_samples_per_second": 151.074,
2408
+ "eval_steps_per_second": 4.723,
2409
+ "step": 75000
2410
+ },
2411
+ {
2412
+ "epoch": 7.360109183076623,
2413
+ "grad_norm": 3.855215549468994,
2414
+ "learning_rate": 2.45e-05,
2415
+ "loss": 0.8319,
2416
+ "step": 75500
2417
+ },
2418
+ {
2419
+ "epoch": 7.360109183076623,
2420
+ "eval_accuracy": 0.8290864150792293,
2421
+ "eval_loss": 0.914759635925293,
2422
+ "eval_runtime": 410.8744,
2423
+ "eval_samples_per_second": 151.182,
2424
+ "eval_steps_per_second": 4.727,
2425
+ "step": 75500
2426
+ },
2427
+ {
2428
+ "epoch": 7.40885162799766,
2429
+ "grad_norm": 3.3140082359313965,
2430
+ "learning_rate": 2.4e-05,
2431
+ "loss": 0.8382,
2432
+ "step": 76000
2433
+ },
2434
+ {
2435
+ "epoch": 7.40885162799766,
2436
+ "eval_accuracy": 0.8291830831243966,
2437
+ "eval_loss": 0.9062051773071289,
2438
+ "eval_runtime": 415.0784,
2439
+ "eval_samples_per_second": 149.651,
2440
+ "eval_steps_per_second": 4.679,
2441
+ "step": 76000
2442
+ },
2443
+ {
2444
+ "epoch": 7.457594072918697,
2445
+ "grad_norm": 3.379228115081787,
2446
+ "learning_rate": 2.35e-05,
2447
+ "loss": 0.8393,
2448
+ "step": 76500
2449
+ },
2450
+ {
2451
+ "epoch": 7.457594072918697,
2452
+ "eval_accuracy": 0.8302764580660623,
2453
+ "eval_loss": 0.904826283454895,
2454
+ "eval_runtime": 420.5555,
2455
+ "eval_samples_per_second": 147.702,
2456
+ "eval_steps_per_second": 4.618,
2457
+ "step": 76500
2458
+ },
2459
+ {
2460
+ "epoch": 7.506336517839735,
2461
+ "grad_norm": 4.405816555023193,
2462
+ "learning_rate": 2.3000000000000003e-05,
2463
+ "loss": 0.8329,
2464
+ "step": 77000
2465
+ },
2466
+ {
2467
+ "epoch": 7.506336517839735,
2468
+ "eval_accuracy": 0.8302014924762132,
2469
+ "eval_loss": 0.901425302028656,
2470
+ "eval_runtime": 411.0172,
2471
+ "eval_samples_per_second": 151.13,
2472
+ "eval_steps_per_second": 4.725,
2473
+ "step": 77000
2474
+ },
2475
+ {
2476
+ "epoch": 7.555078962760772,
2477
+ "grad_norm": 3.3257713317871094,
2478
+ "learning_rate": 2.25e-05,
2479
+ "loss": 0.821,
2480
+ "step": 77500
2481
+ },
2482
+ {
2483
+ "epoch": 7.555078962760772,
2484
+ "eval_accuracy": 0.8307996486461996,
2485
+ "eval_loss": 0.9093529582023621,
2486
+ "eval_runtime": 421.3837,
2487
+ "eval_samples_per_second": 147.412,
2488
+ "eval_steps_per_second": 4.609,
2489
+ "step": 77500
2490
+ },
2491
+ {
2492
+ "epoch": 7.603821407681809,
2493
+ "grad_norm": 3.748304843902588,
2494
+ "learning_rate": 2.2000000000000003e-05,
2495
+ "loss": 0.8182,
2496
+ "step": 78000
2497
+ },
2498
+ {
2499
+ "epoch": 7.603821407681809,
2500
+ "eval_accuracy": 0.8306815632390625,
2501
+ "eval_loss": 0.9139160513877869,
2502
+ "eval_runtime": 421.7982,
2503
+ "eval_samples_per_second": 147.267,
2504
+ "eval_steps_per_second": 4.604,
2505
+ "step": 78000
2506
+ },
2507
+ {
2508
+ "epoch": 7.652563852602847,
2509
+ "grad_norm": 3.9193367958068848,
2510
+ "learning_rate": 2.15e-05,
2511
+ "loss": 0.8235,
2512
+ "step": 78500
2513
+ },
2514
+ {
2515
+ "epoch": 7.652563852602847,
2516
+ "eval_accuracy": 0.8304462857947059,
2517
+ "eval_loss": 0.9012885093688965,
2518
+ "eval_runtime": 410.9874,
2519
+ "eval_samples_per_second": 151.141,
2520
+ "eval_steps_per_second": 4.725,
2521
+ "step": 78500
2522
+ },
2523
+ {
2524
+ "epoch": 7.7013062975238835,
2525
+ "grad_norm": 3.0447959899902344,
2526
+ "learning_rate": 2.1e-05,
2527
+ "loss": 0.8231,
2528
+ "step": 79000
2529
+ },
2530
+ {
2531
+ "epoch": 7.7013062975238835,
2532
+ "eval_accuracy": 0.8311366484422618,
2533
+ "eval_loss": 0.9142009615898132,
2534
+ "eval_runtime": 421.0783,
2535
+ "eval_samples_per_second": 147.519,
2536
+ "eval_steps_per_second": 4.612,
2537
+ "step": 79000
2538
+ },
2539
+ {
2540
+ "epoch": 7.750048742444921,
2541
+ "grad_norm": 3.178105354309082,
2542
+ "learning_rate": 2.05e-05,
2543
+ "loss": 0.8295,
2544
+ "step": 79500
2545
+ },
2546
+ {
2547
+ "epoch": 7.750048742444921,
2548
+ "eval_accuracy": 0.8311365986906916,
2549
+ "eval_loss": 0.8990747332572937,
2550
+ "eval_runtime": 411.2163,
2551
+ "eval_samples_per_second": 151.057,
2552
+ "eval_steps_per_second": 4.723,
2553
+ "step": 79500
2554
+ },
2555
+ {
2556
+ "epoch": 7.798791187365958,
2557
+ "grad_norm": 3.5765883922576904,
2558
+ "learning_rate": 2e-05,
2559
+ "loss": 0.8226,
2560
+ "step": 80000
2561
+ },
2562
+ {
2563
+ "epoch": 7.798791187365958,
2564
+ "eval_accuracy": 0.831212165348238,
2565
+ "eval_loss": 0.9084219336509705,
2566
+ "eval_runtime": 421.425,
2567
+ "eval_samples_per_second": 147.398,
2568
+ "eval_steps_per_second": 4.608,
2569
+ "step": 80000
2570
+ },
2571
+ {
2572
+ "epoch": 7.8475336322869955,
2573
+ "grad_norm": 3.9313628673553467,
2574
+ "learning_rate": 1.9500000000000003e-05,
2575
+ "loss": 0.8259,
2576
+ "step": 80500
2577
+ },
2578
+ {
2579
+ "epoch": 7.8475336322869955,
2580
+ "eval_accuracy": 0.8318195921167911,
2581
+ "eval_loss": 0.9043404459953308,
2582
+ "eval_runtime": 421.3189,
2583
+ "eval_samples_per_second": 147.435,
2584
+ "eval_steps_per_second": 4.609,
2585
+ "step": 80500
2586
+ },
2587
+ {
2588
+ "epoch": 7.896276077208032,
2589
+ "grad_norm": 3.5046310424804688,
2590
+ "learning_rate": 1.9e-05,
2591
+ "loss": 0.8214,
2592
+ "step": 81000
2593
+ },
2594
+ {
2595
+ "epoch": 7.896276077208032,
2596
+ "eval_accuracy": 0.8321367445173339,
2597
+ "eval_loss": 0.8911643028259277,
2598
+ "eval_runtime": 421.1955,
2599
+ "eval_samples_per_second": 147.478,
2600
+ "eval_steps_per_second": 4.611,
2601
+ "step": 81000
2602
+ },
2603
+ {
2604
+ "epoch": 7.94501852212907,
2605
+ "grad_norm": 3.291712999343872,
2606
+ "learning_rate": 1.85e-05,
2607
+ "loss": 0.8245,
2608
+ "step": 81500
2609
+ },
2610
+ {
2611
+ "epoch": 7.94501852212907,
2612
+ "eval_accuracy": 0.8320552879695452,
2613
+ "eval_loss": 0.8937601447105408,
2614
+ "eval_runtime": 411.1067,
2615
+ "eval_samples_per_second": 151.097,
2616
+ "eval_steps_per_second": 4.724,
2617
+ "step": 81500
2618
+ },
2619
+ {
2620
+ "epoch": 7.9937609670501075,
2621
+ "grad_norm": 3.106718063354492,
2622
+ "learning_rate": 1.8e-05,
2623
+ "loss": 0.8231,
2624
+ "step": 82000
2625
+ },
2626
+ {
2627
+ "epoch": 7.9937609670501075,
2628
+ "eval_accuracy": 0.8324106123010148,
2629
+ "eval_loss": 0.892302393913269,
2630
+ "eval_runtime": 421.5985,
2631
+ "eval_samples_per_second": 147.337,
2632
+ "eval_steps_per_second": 4.606,
2633
+ "step": 82000
2634
+ },
2635
+ {
2636
+ "epoch": 8.042503411971145,
2637
+ "grad_norm": 3.7286880016326904,
2638
+ "learning_rate": 1.75e-05,
2639
+ "loss": 0.8103,
2640
+ "step": 82500
2641
+ },
2642
+ {
2643
+ "epoch": 8.042503411971145,
2644
+ "eval_accuracy": 0.8321463976856007,
2645
+ "eval_loss": 0.8813316822052002,
2646
+ "eval_runtime": 410.8912,
2647
+ "eval_samples_per_second": 151.176,
2648
+ "eval_steps_per_second": 4.726,
2649
+ "step": 82500
2650
+ },
2651
+ {
2652
+ "epoch": 8.091245856892181,
2653
+ "grad_norm": 4.125603199005127,
2654
+ "learning_rate": 1.7000000000000003e-05,
2655
+ "loss": 0.816,
2656
+ "step": 83000
2657
+ },
2658
+ {
2659
+ "epoch": 8.091245856892181,
2660
+ "eval_accuracy": 0.8330290185255811,
2661
+ "eval_loss": 0.90044766664505,
2662
+ "eval_runtime": 410.932,
2663
+ "eval_samples_per_second": 151.161,
2664
+ "eval_steps_per_second": 4.726,
2665
+ "step": 83000
2666
+ },
2667
+ {
2668
+ "epoch": 8.139988301813219,
2669
+ "grad_norm": 3.6977617740631104,
2670
+ "learning_rate": 1.65e-05,
2671
+ "loss": 0.8178,
2672
+ "step": 83500
2673
+ },
2674
+ {
2675
+ "epoch": 8.139988301813219,
2676
+ "eval_accuracy": 0.8330648677426203,
2677
+ "eval_loss": 0.8834232687950134,
2678
+ "eval_runtime": 410.273,
2679
+ "eval_samples_per_second": 151.404,
2680
+ "eval_steps_per_second": 4.733,
2681
+ "step": 83500
2682
+ },
2683
+ {
2684
+ "epoch": 8.188730746734256,
2685
+ "grad_norm": 3.645653247833252,
2686
+ "learning_rate": 1.6000000000000003e-05,
2687
+ "loss": 0.8004,
2688
+ "step": 84000
2689
+ },
2690
+ {
2691
+ "epoch": 8.188730746734256,
2692
+ "eval_accuracy": 0.8328894631127913,
2693
+ "eval_loss": 0.8885020017623901,
2694
+ "eval_runtime": 421.6692,
2695
+ "eval_samples_per_second": 147.312,
2696
+ "eval_steps_per_second": 4.606,
2697
+ "step": 84000
2698
+ },
2699
+ {
2700
+ "epoch": 8.237473191655294,
2701
+ "grad_norm": 4.153274059295654,
2702
+ "learning_rate": 1.55e-05,
2703
+ "loss": 0.8062,
2704
+ "step": 84500
2705
+ },
2706
+ {
2707
+ "epoch": 8.237473191655294,
2708
+ "eval_accuracy": 0.8329369957505026,
2709
+ "eval_loss": 0.8888145089149475,
2710
+ "eval_runtime": 409.367,
2711
+ "eval_samples_per_second": 151.739,
2712
+ "eval_steps_per_second": 4.744,
2713
+ "step": 84500
2714
+ },
2715
+ {
2716
+ "epoch": 8.28621563657633,
2717
+ "grad_norm": 3.958219528198242,
2718
+ "learning_rate": 1.5e-05,
2719
+ "loss": 0.8139,
2720
+ "step": 85000
2721
+ },
2722
+ {
2723
+ "epoch": 8.28621563657633,
2724
+ "eval_accuracy": 0.8327970192226268,
2725
+ "eval_loss": 0.8985949754714966,
2726
+ "eval_runtime": 420.7578,
2727
+ "eval_samples_per_second": 147.631,
2728
+ "eval_steps_per_second": 4.615,
2729
+ "step": 85000
2730
+ },
2731
+ {
2732
+ "epoch": 8.334958081497367,
2733
+ "grad_norm": 2.9045639038085938,
2734
+ "learning_rate": 1.45e-05,
2735
+ "loss": 0.8062,
2736
+ "step": 85500
2737
+ },
2738
+ {
2739
+ "epoch": 8.334958081497367,
2740
+ "eval_accuracy": 0.8340764025345829,
2741
+ "eval_loss": 0.8828948140144348,
2742
+ "eval_runtime": 421.0352,
2743
+ "eval_samples_per_second": 147.534,
2744
+ "eval_steps_per_second": 4.612,
2745
+ "step": 85500
2746
+ },
2747
+ {
2748
+ "epoch": 8.383700526418405,
2749
+ "grad_norm": 3.1342079639434814,
2750
+ "learning_rate": 1.4000000000000001e-05,
2751
+ "loss": 0.7989,
2752
+ "step": 86000
2753
+ },
2754
+ {
2755
+ "epoch": 8.383700526418405,
2756
+ "eval_accuracy": 0.8341633931538641,
2757
+ "eval_loss": 0.886264979839325,
2758
+ "eval_runtime": 421.0291,
2759
+ "eval_samples_per_second": 147.536,
2760
+ "eval_steps_per_second": 4.613,
2761
+ "step": 86000
2762
+ },
2763
+ {
2764
+ "epoch": 8.432442971339443,
2765
+ "grad_norm": 3.8171074390411377,
2766
+ "learning_rate": 1.3500000000000001e-05,
2767
+ "loss": 0.7996,
2768
+ "step": 86500
2769
+ },
2770
+ {
2771
+ "epoch": 8.432442971339443,
2772
+ "eval_accuracy": 0.8343804133323081,
2773
+ "eval_loss": 0.8822252154350281,
2774
+ "eval_runtime": 410.5394,
2775
+ "eval_samples_per_second": 151.306,
2776
+ "eval_steps_per_second": 4.73,
2777
+ "step": 86500
2778
+ },
2779
+ {
2780
+ "epoch": 8.48118541626048,
2781
+ "grad_norm": 3.815415620803833,
2782
+ "learning_rate": 1.3000000000000001e-05,
2783
+ "loss": 0.8077,
2784
+ "step": 87000
2785
+ },
2786
+ {
2787
+ "epoch": 8.48118541626048,
2788
+ "eval_accuracy": 0.8346840748418042,
2789
+ "eval_loss": 0.8820429444313049,
2790
+ "eval_runtime": 409.395,
2791
+ "eval_samples_per_second": 151.729,
2792
+ "eval_steps_per_second": 4.744,
2793
+ "step": 87000
2794
+ },
2795
+ {
2796
+ "epoch": 8.529927861181516,
2797
+ "grad_norm": 3.472749710083008,
2798
+ "learning_rate": 1.25e-05,
2799
+ "loss": 0.8,
2800
+ "step": 87500
2801
+ },
2802
+ {
2803
+ "epoch": 8.529927861181516,
2804
+ "eval_accuracy": 0.8344507719507152,
2805
+ "eval_loss": NaN,
2806
+ "eval_runtime": 410.8123,
2807
+ "eval_samples_per_second": 151.205,
2808
+ "eval_steps_per_second": 4.727,
2809
+ "step": 87500
2810
+ },
2811
+ {
2812
+ "epoch": 8.578670306102554,
2813
+ "grad_norm": 3.4210853576660156,
2814
+ "learning_rate": 1.2e-05,
2815
+ "loss": 0.7954,
2816
+ "step": 88000
2817
+ },
2818
+ {
2819
+ "epoch": 8.578670306102554,
2820
+ "eval_accuracy": 0.8346871655612746,
2821
+ "eval_loss": 0.889224112033844,
2822
+ "eval_runtime": 421.2172,
2823
+ "eval_samples_per_second": 147.47,
2824
+ "eval_steps_per_second": 4.61,
2825
+ "step": 88000
2826
+ },
2827
+ {
2828
+ "epoch": 8.627412751023591,
2829
+ "grad_norm": 4.314051628112793,
2830
+ "learning_rate": 1.1500000000000002e-05,
2831
+ "loss": 0.7948,
2832
+ "step": 88500
2833
+ },
2834
+ {
2835
+ "epoch": 8.627412751023591,
2836
+ "eval_accuracy": 0.8345174375290413,
2837
+ "eval_loss": 0.8909358382225037,
2838
+ "eval_runtime": 421.2829,
2839
+ "eval_samples_per_second": 147.447,
2840
+ "eval_steps_per_second": 4.61,
2841
+ "step": 88500
2842
+ },
2843
+ {
2844
+ "epoch": 8.676155195944629,
2845
+ "grad_norm": 4.1729841232299805,
2846
+ "learning_rate": 1.1000000000000001e-05,
2847
+ "loss": 0.8054,
2848
+ "step": 89000
2849
+ },
2850
+ {
2851
+ "epoch": 8.676155195944629,
2852
+ "eval_accuracy": 0.8349649013896184,
2853
+ "eval_loss": 0.8881335854530334,
2854
+ "eval_runtime": 417.162,
2855
+ "eval_samples_per_second": 148.904,
2856
+ "eval_steps_per_second": 4.655,
2857
+ "step": 89000
2858
+ },
2859
+ {
2860
+ "epoch": 8.724897640865667,
2861
+ "grad_norm": 4.046889305114746,
2862
+ "learning_rate": 1.05e-05,
2863
+ "loss": 0.8025,
2864
+ "step": 89500
2865
+ },
2866
+ {
2867
+ "epoch": 8.724897640865667,
2868
+ "eval_accuracy": 0.8351938054358102,
2869
+ "eval_loss": 0.8830729126930237,
2870
+ "eval_runtime": 410.8378,
2871
+ "eval_samples_per_second": 151.196,
2872
+ "eval_steps_per_second": 4.727,
2873
+ "step": 89500
2874
+ },
2875
+ {
2876
+ "epoch": 8.773640085786703,
2877
+ "grad_norm": 3.559530258178711,
2878
+ "learning_rate": 1e-05,
2879
+ "loss": 0.7931,
2880
+ "step": 90000
2881
+ },
2882
+ {
2883
+ "epoch": 8.773640085786703,
2884
+ "eval_accuracy": 0.8350349267242381,
2885
+ "eval_loss": 0.8810989260673523,
2886
+ "eval_runtime": 410.8806,
2887
+ "eval_samples_per_second": 151.18,
2888
+ "eval_steps_per_second": 4.726,
2889
+ "step": 90000
2890
+ },
2891
+ {
2892
+ "epoch": 8.82238253070774,
2893
+ "grad_norm": 2.676598072052002,
2894
+ "learning_rate": 9.5e-06,
2895
+ "loss": 0.7991,
2896
+ "step": 90500
2897
+ },
2898
+ {
2899
+ "epoch": 8.82238253070774,
2900
+ "eval_accuracy": 0.835213564425697,
2901
+ "eval_loss": 0.8790693879127502,
2902
+ "eval_runtime": 411.3497,
2903
+ "eval_samples_per_second": 151.008,
2904
+ "eval_steps_per_second": 4.721,
2905
+ "step": 90500
2906
+ },
2907
+ {
2908
+ "epoch": 8.871124975628778,
2909
+ "grad_norm": 4.853274345397949,
2910
+ "learning_rate": 9e-06,
2911
+ "loss": 0.7932,
2912
+ "step": 91000
2913
+ },
2914
+ {
2915
+ "epoch": 8.871124975628778,
2916
+ "eval_accuracy": 0.8354392726720316,
2917
+ "eval_loss": 0.884040892124176,
2918
+ "eval_runtime": 410.218,
2919
+ "eval_samples_per_second": 151.424,
2920
+ "eval_steps_per_second": 4.734,
2921
+ "step": 91000
2922
+ },
2923
+ {
2924
+ "epoch": 8.919867420549815,
2925
+ "grad_norm": 3.395076274871826,
2926
+ "learning_rate": 8.500000000000002e-06,
2927
+ "loss": 0.8038,
2928
+ "step": 91500
2929
+ },
2930
+ {
2931
+ "epoch": 8.919867420549815,
2932
+ "eval_accuracy": 0.8354977290802013,
2933
+ "eval_loss": 0.8750784397125244,
2934
+ "eval_runtime": 420.6483,
2935
+ "eval_samples_per_second": 147.67,
2936
+ "eval_steps_per_second": 4.617,
2937
+ "step": 91500
2938
+ },
2939
+ {
2940
+ "epoch": 8.968609865470851,
2941
+ "grad_norm": 3.1790411472320557,
2942
+ "learning_rate": 8.000000000000001e-06,
2943
+ "loss": 0.797,
2944
+ "step": 92000
2945
+ },
2946
+ {
2947
+ "epoch": 8.968609865470851,
2948
+ "eval_accuracy": 0.8356597168660417,
2949
+ "eval_loss": 0.8808930516242981,
2950
+ "eval_runtime": 421.2453,
2951
+ "eval_samples_per_second": 147.46,
2952
+ "eval_steps_per_second": 4.61,
2953
+ "step": 92000
2954
+ },
2955
+ {
2956
+ "epoch": 9.017352310391889,
2957
+ "grad_norm": 3.0785293579101562,
2958
+ "learning_rate": 7.5e-06,
2959
+ "loss": 0.7886,
2960
+ "step": 92500
2961
+ },
2962
+ {
2963
+ "epoch": 9.017352310391889,
2964
+ "eval_accuracy": 0.8362075555143024,
2965
+ "eval_loss": 0.8746069669723511,
2966
+ "eval_runtime": 411.8984,
2967
+ "eval_samples_per_second": 150.807,
2968
+ "eval_steps_per_second": 4.715,
2969
+ "step": 92500
2970
+ },
2971
+ {
2972
+ "epoch": 9.066094755312927,
2973
+ "grad_norm": 4.20958137512207,
2974
+ "learning_rate": 7.000000000000001e-06,
2975
+ "loss": 0.7857,
2976
+ "step": 93000
2977
+ },
2978
+ {
2979
+ "epoch": 9.066094755312927,
2980
+ "eval_accuracy": 0.8365400512194333,
2981
+ "eval_loss": 0.8794592022895813,
2982
+ "eval_runtime": 411.5406,
2983
+ "eval_samples_per_second": 150.938,
2984
+ "eval_steps_per_second": 4.719,
2985
+ "step": 93000
2986
+ },
2987
+ {
2988
+ "epoch": 9.114837200233964,
2989
+ "grad_norm": 3.4888412952423096,
2990
+ "learning_rate": 6.5000000000000004e-06,
2991
+ "loss": 0.7905,
2992
+ "step": 93500
2993
+ },
2994
+ {
2995
+ "epoch": 9.114837200233964,
2996
+ "eval_accuracy": 0.8357585983151461,
2997
+ "eval_loss": NaN,
2998
+ "eval_runtime": 411.342,
2999
+ "eval_samples_per_second": 151.011,
3000
+ "eval_steps_per_second": 4.721,
3001
+ "step": 93500
3002
+ },
3003
+ {
3004
+ "epoch": 9.163579645155002,
3005
+ "grad_norm": 3.387164831161499,
3006
+ "learning_rate": 6e-06,
3007
+ "loss": 0.7871,
3008
+ "step": 94000
3009
+ },
3010
+ {
3011
+ "epoch": 9.163579645155002,
3012
+ "eval_accuracy": 0.8367420900181296,
3013
+ "eval_loss": 0.8770105242729187,
3014
+ "eval_runtime": 421.3937,
3015
+ "eval_samples_per_second": 147.408,
3016
+ "eval_steps_per_second": 4.609,
3017
+ "step": 94000
3018
+ },
3019
+ {
3020
+ "epoch": 9.212322090076038,
3021
+ "grad_norm": 2.8375401496887207,
3022
+ "learning_rate": 5.500000000000001e-06,
3023
+ "loss": 0.7943,
3024
+ "step": 94500
3025
+ },
3026
+ {
3027
+ "epoch": 9.212322090076038,
3028
+ "eval_accuracy": 0.8363407742570033,
3029
+ "eval_loss": 0.8801371455192566,
3030
+ "eval_runtime": 412.0432,
3031
+ "eval_samples_per_second": 150.754,
3032
+ "eval_steps_per_second": 4.713,
3033
+ "step": 94500
3034
+ },
3035
+ {
3036
+ "epoch": 9.261064534997075,
3037
+ "grad_norm": 3.558210849761963,
3038
+ "learning_rate": 5e-06,
3039
+ "loss": 0.7827,
3040
+ "step": 95000
3041
+ },
3042
+ {
3043
+ "epoch": 9.261064534997075,
3044
+ "eval_accuracy": 0.8367376661868028,
3045
+ "eval_loss": 0.8815721273422241,
3046
+ "eval_runtime": 411.572,
3047
+ "eval_samples_per_second": 150.926,
3048
+ "eval_steps_per_second": 4.718,
3049
+ "step": 95000
3050
+ },
3051
+ {
3052
+ "epoch": 9.309806979918113,
3053
+ "grad_norm": 3.5054867267608643,
3054
+ "learning_rate": 4.5e-06,
3055
+ "loss": 0.7817,
3056
+ "step": 95500
3057
+ },
3058
+ {
3059
+ "epoch": 9.309806979918113,
3060
+ "eval_accuracy": 0.8367934091134708,
3061
+ "eval_loss": 0.8681079149246216,
3062
+ "eval_runtime": 422.2461,
3063
+ "eval_samples_per_second": 147.111,
3064
+ "eval_steps_per_second": 4.599,
3065
+ "step": 95500
3066
+ },
3067
+ {
3068
+ "epoch": 9.35854942483915,
3069
+ "grad_norm": 3.3173916339874268,
3070
+ "learning_rate": 4.000000000000001e-06,
3071
+ "loss": 0.7869,
3072
+ "step": 96000
3073
+ },
3074
+ {
3075
+ "epoch": 9.35854942483915,
3076
+ "eval_accuracy": 0.8368133488425433,
3077
+ "eval_loss": 0.8694866895675659,
3078
+ "eval_runtime": 416.5981,
3079
+ "eval_samples_per_second": 149.105,
3080
+ "eval_steps_per_second": 4.662,
3081
+ "step": 96000
3082
+ },
3083
+ {
3084
+ "epoch": 9.407291869760186,
3085
+ "grad_norm": 3.7099990844726562,
3086
+ "learning_rate": 3.5000000000000004e-06,
3087
+ "loss": 0.7824,
3088
+ "step": 96500
3089
+ },
3090
+ {
3091
+ "epoch": 9.407291869760186,
3092
+ "eval_accuracy": 0.8365284158336549,
3093
+ "eval_loss": 0.8713687062263489,
3094
+ "eval_runtime": 421.857,
3095
+ "eval_samples_per_second": 147.247,
3096
+ "eval_steps_per_second": 4.603,
3097
+ "step": 96500
3098
+ },
3099
+ {
3100
+ "epoch": 9.456034314681224,
3101
+ "grad_norm": 3.0227816104888916,
3102
+ "learning_rate": 3e-06,
3103
+ "loss": 0.7798,
3104
+ "step": 97000
3105
+ },
3106
+ {
3107
+ "epoch": 9.456034314681224,
3108
+ "eval_accuracy": 0.8368876518310658,
3109
+ "eval_loss": 0.8661150336265564,
3110
+ "eval_runtime": 421.5746,
3111
+ "eval_samples_per_second": 147.345,
3112
+ "eval_steps_per_second": 4.607,
3113
+ "step": 97000
3114
+ },
3115
+ {
3116
+ "epoch": 9.504776759602262,
3117
+ "grad_norm": 3.0522849559783936,
3118
+ "learning_rate": 2.5e-06,
3119
+ "loss": 0.7781,
3120
+ "step": 97500
3121
+ },
3122
+ {
3123
+ "epoch": 9.504776759602262,
3124
+ "eval_accuracy": 0.8375290357797354,
3125
+ "eval_loss": 0.8749486804008484,
3126
+ "eval_runtime": 412.2604,
3127
+ "eval_samples_per_second": 150.674,
3128
+ "eval_steps_per_second": 4.711,
3129
+ "step": 97500
3130
+ },
3131
+ {
3132
+ "epoch": 9.5535192045233,
3133
+ "grad_norm": 3.0036725997924805,
3134
+ "learning_rate": 2.0000000000000003e-06,
3135
+ "loss": 0.7817,
3136
+ "step": 98000
3137
+ },
3138
+ {
3139
+ "epoch": 9.5535192045233,
3140
+ "eval_accuracy": 0.8371932481486926,
3141
+ "eval_loss": 0.8651660084724426,
3142
+ "eval_runtime": 421.7398,
3143
+ "eval_samples_per_second": 147.287,
3144
+ "eval_steps_per_second": 4.605,
3145
+ "step": 98000
3146
+ }
3147
+ ],
3148
+ "logging_steps": 500,
3149
+ "max_steps": 100000,
3150
+ "num_input_tokens_seen": 0,
3151
+ "num_train_epochs": 10,
3152
+ "save_steps": 500,
3153
+ "stateful_callbacks": {
3154
+ "TrainerControl": {
3155
+ "args": {
3156
+ "should_epoch_stop": false,
3157
+ "should_evaluate": false,
3158
+ "should_log": false,
3159
+ "should_save": true,
3160
+ "should_training_stop": false
3161
+ },
3162
+ "attributes": {}
3163
+ }
3164
+ },
3165
+ "total_flos": 8.27282260465025e+17,
3166
+ "train_batch_size": 16,
3167
+ "trial_name": null,
3168
+ "trial_params": null
3169
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ee3f623a57bd8bde67e813c7aabc9017c36ee0d916c976805df4142ef7ec8f1b
3
+ size 5176