roonbug commited on
Commit
76b2d40
·
verified ·
1 Parent(s): 061289c

Upload folder using huggingface_hub

Browse files
.gitattributes CHANGED
@@ -34,3 +34,4 @@ saved_model/**/* 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
  checkpoint-500/tokenizer.json 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
  checkpoint-500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
37
+ checkpoint-1000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
checkpoint-1000/added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "<image_soft_token>": 262144
3
+ }
checkpoint-1000/chat_template.jinja ADDED
@@ -0,0 +1,47 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {{ bos_token }}
2
+ {%- if messages[0]['role'] == 'system' -%}
3
+ {%- if messages[0]['content'] is string -%}
4
+ {%- set first_user_prefix = messages[0]['content'] + '
5
+
6
+ ' -%}
7
+ {%- else -%}
8
+ {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
9
+
10
+ ' -%}
11
+ {%- endif -%}
12
+ {%- set loop_messages = messages[1:] -%}
13
+ {%- else -%}
14
+ {%- set first_user_prefix = "" -%}
15
+ {%- set loop_messages = messages -%}
16
+ {%- endif -%}
17
+ {%- for message in loop_messages -%}
18
+ {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
19
+ {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
20
+ {%- endif -%}
21
+ {%- if (message['role'] == 'assistant') -%}
22
+ {%- set role = "model" -%}
23
+ {%- else -%}
24
+ {%- set role = message['role'] -%}
25
+ {%- endif -%}
26
+ {{ '<start_of_turn>' + role + '
27
+ ' + (first_user_prefix if loop.first else "") }}
28
+ {%- if message['content'] is string -%}
29
+ {{ message['content'] | trim }}
30
+ {%- elif message['content'] is iterable -%}
31
+ {%- for item in message['content'] -%}
32
+ {%- if item['type'] == 'image' -%}
33
+ {{ '<start_of_image>' }}
34
+ {%- elif item['type'] == 'text' -%}
35
+ {{ item['text'] | trim }}
36
+ {%- endif -%}
37
+ {%- endfor -%}
38
+ {%- else -%}
39
+ {{ raise_exception("Invalid content type") }}
40
+ {%- endif -%}
41
+ {{ '<end_of_turn>
42
+ ' }}
43
+ {%- endfor -%}
44
+ {%- if add_generation_prompt -%}
45
+ {{'<start_of_turn>model
46
+ '}}
47
+ {%- endif -%}
checkpoint-1000/config.json ADDED
@@ -0,0 +1,110 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Gemma3ForConditionalGeneration"
4
+ ],
5
+ "boi_token_index": 255999,
6
+ "bos_token_id": 2,
7
+ "dtype": "bfloat16",
8
+ "eoi_token_index": 256000,
9
+ "eos_token_id": 1,
10
+ "image_token_index": 262144,
11
+ "initializer_range": 0.02,
12
+ "mm_tokens_per_image": 256,
13
+ "model_type": "gemma3",
14
+ "pad_token_id": 0,
15
+ "text_config": {
16
+ "_sliding_window_pattern": 6,
17
+ "attention_bias": false,
18
+ "attention_dropout": 0.0,
19
+ "attn_logit_softcapping": null,
20
+ "dtype": "bfloat16",
21
+ "final_logit_softcapping": null,
22
+ "head_dim": 256,
23
+ "hidden_activation": "gelu_pytorch_tanh",
24
+ "hidden_size": 3840,
25
+ "initializer_range": 0.02,
26
+ "intermediate_size": 15360,
27
+ "layer_types": [
28
+ "sliding_attention",
29
+ "sliding_attention",
30
+ "sliding_attention",
31
+ "sliding_attention",
32
+ "sliding_attention",
33
+ "full_attention",
34
+ "sliding_attention",
35
+ "sliding_attention",
36
+ "sliding_attention",
37
+ "sliding_attention",
38
+ "sliding_attention",
39
+ "full_attention",
40
+ "sliding_attention",
41
+ "sliding_attention",
42
+ "sliding_attention",
43
+ "sliding_attention",
44
+ "sliding_attention",
45
+ "full_attention",
46
+ "sliding_attention",
47
+ "sliding_attention",
48
+ "sliding_attention",
49
+ "sliding_attention",
50
+ "sliding_attention",
51
+ "full_attention",
52
+ "sliding_attention",
53
+ "sliding_attention",
54
+ "sliding_attention",
55
+ "sliding_attention",
56
+ "sliding_attention",
57
+ "full_attention",
58
+ "sliding_attention",
59
+ "sliding_attention",
60
+ "sliding_attention",
61
+ "sliding_attention",
62
+ "sliding_attention",
63
+ "full_attention",
64
+ "sliding_attention",
65
+ "sliding_attention",
66
+ "sliding_attention",
67
+ "sliding_attention",
68
+ "sliding_attention",
69
+ "full_attention",
70
+ "sliding_attention",
71
+ "sliding_attention",
72
+ "sliding_attention",
73
+ "sliding_attention",
74
+ "sliding_attention",
75
+ "full_attention"
76
+ ],
77
+ "max_position_embeddings": 131072,
78
+ "model_type": "gemma3_text",
79
+ "num_attention_heads": 16,
80
+ "num_hidden_layers": 48,
81
+ "num_key_value_heads": 8,
82
+ "query_pre_attn_scalar": 256,
83
+ "rms_norm_eps": 1e-06,
84
+ "rope_local_base_freq": 10000.0,
85
+ "rope_scaling": {
86
+ "factor": 8.0,
87
+ "rope_type": "linear"
88
+ },
89
+ "rope_theta": 1000000.0,
90
+ "sliding_window": 1024,
91
+ "use_cache": true,
92
+ "vocab_size": 262208
93
+ },
94
+ "transformers_version": "4.56.1",
95
+ "vision_config": {
96
+ "attention_dropout": 0.0,
97
+ "dtype": "bfloat16",
98
+ "hidden_act": "gelu_pytorch_tanh",
99
+ "hidden_size": 1152,
100
+ "image_size": 896,
101
+ "intermediate_size": 4304,
102
+ "layer_norm_eps": 1e-06,
103
+ "model_type": "siglip_vision_model",
104
+ "num_attention_heads": 16,
105
+ "num_channels": 3,
106
+ "num_hidden_layers": 27,
107
+ "patch_size": 14,
108
+ "vision_use_head": false
109
+ }
110
+ }
checkpoint-1000/generation_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 2,
3
+ "do_sample": true,
4
+ "eos_token_id": [
5
+ 1,
6
+ 1,
7
+ 106
8
+ ],
9
+ "pad_token_id": 0,
10
+ "top_k": 64,
11
+ "top_p": 0.95,
12
+ "transformers_version": "4.56.1"
13
+ }
checkpoint-1000/model-00001-of-00005.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4847447e92599833e8dbaa3067cd201c3bb5c052efa91f11ba891e43234f7832
3
+ size 4979902192
checkpoint-1000/model-00002-of-00005.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:891bd54eed03cba9ee1e705533a02a8217fcc29f356e4a1f53e5fd0d178883ad
3
+ size 4931296592
checkpoint-1000/model-00003-of-00005.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7cee411d9d57324e50ce064a192cc5a858276d508611b12fc599e0c9767112e0
3
+ size 4931296656
checkpoint-1000/model-00004-of-00005.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ced74b8af92d9e5e67820172ef3c3406d5189c8f96d044d36daf7589a4226110
3
+ size 4931296656
checkpoint-1000/model-00005-of-00005.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4a91795167ac06e6778091ecc09ed7dfd6c5b1a67b8c09f377c1863c080e4478
3
+ size 4601000928
checkpoint-1000/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-1000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f2a1a00cb06e4e82d353ad89e55ce1052554c8158dd4d8c39e7600c8b0839a3f
3
+ size 7286558019
checkpoint-1000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:acd74ad34e3c5060506b53a8a3fafbfb35d44065ee496b2abf3a0b9a3b93ba09
3
+ size 14645
checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6078e52773a379f026b8fbbbc5546ab7ed9418d05b713f8aff2a4f3c7e12f108
3
+ size 1465
checkpoint-1000/special_tokens_map.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "boi_token": "<start_of_image>",
3
+ "bos_token": {
4
+ "content": "<bos>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false
9
+ },
10
+ "eoi_token": "<end_of_image>",
11
+ "eos_token": {
12
+ "content": "<eos>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false
17
+ },
18
+ "image_token": "<image_soft_token>",
19
+ "pad_token": {
20
+ "content": "<pad>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false
25
+ },
26
+ "unk_token": {
27
+ "content": "<unk>",
28
+ "lstrip": false,
29
+ "normalized": false,
30
+ "rstrip": false,
31
+ "single_word": false
32
+ }
33
+ }
checkpoint-1000/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4667f2089529e8e7657cfb6d1c19910ae71ff5f28aa7ab2ff2763330affad795
3
+ size 33384568
checkpoint-1000/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1299c11d7cf632ef3b4e11937501358ada021bbdf7c47638d13c0ee982f2e79c
3
+ size 4689074
checkpoint-1000/tokenizer_config.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-1000/trainer_state.json ADDED
@@ -0,0 +1,1254 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.4451121126133645,
6
+ "eval_steps": 100,
7
+ "global_step": 1000,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "entropy": 0.8066153490915895,
14
+ "epoch": 0.004451121126133645,
15
+ "grad_norm": 107.5,
16
+ "learning_rate": 1.8e-07,
17
+ "loss": 16.3591,
18
+ "mean_token_accuracy": 0.7447933696210385,
19
+ "num_tokens": 63133.0,
20
+ "step": 10
21
+ },
22
+ {
23
+ "entropy": 0.8401238698512316,
24
+ "epoch": 0.00890224225226729,
25
+ "grad_norm": 103.5,
26
+ "learning_rate": 3.8e-07,
27
+ "loss": 17.2936,
28
+ "mean_token_accuracy": 0.7330440735444427,
29
+ "num_tokens": 128968.0,
30
+ "step": 20
31
+ },
32
+ {
33
+ "entropy": 0.828661117143929,
34
+ "epoch": 0.013353363378400934,
35
+ "grad_norm": 96.5,
36
+ "learning_rate": 5.800000000000001e-07,
37
+ "loss": 16.5809,
38
+ "mean_token_accuracy": 0.7381731692701579,
39
+ "num_tokens": 193314.0,
40
+ "step": 30
41
+ },
42
+ {
43
+ "entropy": 0.8600894263014197,
44
+ "epoch": 0.01780448450453458,
45
+ "grad_norm": 96.0,
46
+ "learning_rate": 7.8e-07,
47
+ "loss": 17.7444,
48
+ "mean_token_accuracy": 0.7324823886156082,
49
+ "num_tokens": 257189.0,
50
+ "step": 40
51
+ },
52
+ {
53
+ "entropy": 0.8017022017389536,
54
+ "epoch": 0.022255605630668224,
55
+ "grad_norm": 84.5,
56
+ "learning_rate": 9.800000000000001e-07,
57
+ "loss": 15.9063,
58
+ "mean_token_accuracy": 0.7455223135650157,
59
+ "num_tokens": 324224.0,
60
+ "step": 50
61
+ },
62
+ {
63
+ "entropy": 0.846268966794014,
64
+ "epoch": 0.026706726756801868,
65
+ "grad_norm": 78.5,
66
+ "learning_rate": 1.1800000000000001e-06,
67
+ "loss": 16.6006,
68
+ "mean_token_accuracy": 0.7411292420700193,
69
+ "num_tokens": 394654.0,
70
+ "step": 60
71
+ },
72
+ {
73
+ "entropy": 0.8181243563070894,
74
+ "epoch": 0.031157847882935515,
75
+ "grad_norm": 76.5,
76
+ "learning_rate": 1.3800000000000001e-06,
77
+ "loss": 15.9846,
78
+ "mean_token_accuracy": 0.746401545777917,
79
+ "num_tokens": 461658.0,
80
+ "step": 70
81
+ },
82
+ {
83
+ "entropy": 0.8314600124955177,
84
+ "epoch": 0.03560896900906916,
85
+ "grad_norm": 75.0,
86
+ "learning_rate": 1.5800000000000001e-06,
87
+ "loss": 15.8227,
88
+ "mean_token_accuracy": 0.7472479769960045,
89
+ "num_tokens": 528285.0,
90
+ "step": 80
91
+ },
92
+ {
93
+ "entropy": 0.8807666478678584,
94
+ "epoch": 0.04006009013520281,
95
+ "grad_norm": 72.0,
96
+ "learning_rate": 1.7800000000000001e-06,
97
+ "loss": 16.6834,
98
+ "mean_token_accuracy": 0.7396679904311896,
99
+ "num_tokens": 595336.0,
100
+ "step": 90
101
+ },
102
+ {
103
+ "entropy": 0.8991396678611636,
104
+ "epoch": 0.04451121126133645,
105
+ "grad_norm": 79.5,
106
+ "learning_rate": 1.98e-06,
107
+ "loss": 16.3968,
108
+ "mean_token_accuracy": 0.7396075185388327,
109
+ "num_tokens": 663126.0,
110
+ "step": 100
111
+ },
112
+ {
113
+ "epoch": 0.04451121126133645,
114
+ "eval_biology_entropy": 1.1188154411315918,
115
+ "eval_biology_loss": 1.2691835165023804,
116
+ "eval_biology_mean_token_accuracy": 0.6881385813951493,
117
+ "eval_biology_num_tokens": 663126.0,
118
+ "eval_biology_runtime": 40.4565,
119
+ "eval_biology_samples_per_second": 12.359,
120
+ "eval_biology_steps_per_second": 12.359,
121
+ "step": 100
122
+ },
123
+ {
124
+ "epoch": 0.04451121126133645,
125
+ "eval_chemistry_entropy": 0.8804921235442161,
126
+ "eval_chemistry_loss": 1.0042469501495361,
127
+ "eval_chemistry_mean_token_accuracy": 0.7444319971203804,
128
+ "eval_chemistry_num_tokens": 663126.0,
129
+ "eval_chemistry_runtime": 46.3585,
130
+ "eval_chemistry_samples_per_second": 10.786,
131
+ "eval_chemistry_steps_per_second": 10.786,
132
+ "step": 100
133
+ },
134
+ {
135
+ "entropy": 0.8988691195845604,
136
+ "epoch": 0.048962332387470095,
137
+ "grad_norm": 100.5,
138
+ "learning_rate": 2.1800000000000003e-06,
139
+ "loss": 16.3063,
140
+ "mean_token_accuracy": 0.7414613764733076,
141
+ "num_tokens": 731060.0,
142
+ "step": 110
143
+ },
144
+ {
145
+ "entropy": 0.8918870648369193,
146
+ "epoch": 0.053413453513603736,
147
+ "grad_norm": 77.5,
148
+ "learning_rate": 2.38e-06,
149
+ "loss": 15.8311,
150
+ "mean_token_accuracy": 0.7483173958957196,
151
+ "num_tokens": 801635.0,
152
+ "step": 120
153
+ },
154
+ {
155
+ "entropy": 0.9162682231515646,
156
+ "epoch": 0.05786457463973738,
157
+ "grad_norm": 66.0,
158
+ "learning_rate": 2.5800000000000003e-06,
159
+ "loss": 16.1464,
160
+ "mean_token_accuracy": 0.7448406910523773,
161
+ "num_tokens": 867260.0,
162
+ "step": 130
163
+ },
164
+ {
165
+ "entropy": 0.9749668512493372,
166
+ "epoch": 0.06231569576587103,
167
+ "grad_norm": 79.5,
168
+ "learning_rate": 2.7800000000000005e-06,
169
+ "loss": 16.9562,
170
+ "mean_token_accuracy": 0.7369577366858721,
171
+ "num_tokens": 931344.0,
172
+ "step": 140
173
+ },
174
+ {
175
+ "entropy": 0.939477625861764,
176
+ "epoch": 0.06676681689200467,
177
+ "grad_norm": 56.75,
178
+ "learning_rate": 2.9800000000000003e-06,
179
+ "loss": 15.7327,
180
+ "mean_token_accuracy": 0.7471803797408938,
181
+ "num_tokens": 993586.0,
182
+ "step": 150
183
+ },
184
+ {
185
+ "entropy": 0.9718907386064529,
186
+ "epoch": 0.07121793801813832,
187
+ "grad_norm": 53.75,
188
+ "learning_rate": 3.1800000000000005e-06,
189
+ "loss": 16.2036,
190
+ "mean_token_accuracy": 0.7392314806580543,
191
+ "num_tokens": 1060812.0,
192
+ "step": 160
193
+ },
194
+ {
195
+ "entropy": 0.9165311623364687,
196
+ "epoch": 0.07566905914427197,
197
+ "grad_norm": 63.25,
198
+ "learning_rate": 3.3800000000000007e-06,
199
+ "loss": 15.2657,
200
+ "mean_token_accuracy": 0.7518215283751488,
201
+ "num_tokens": 1131832.0,
202
+ "step": 170
203
+ },
204
+ {
205
+ "entropy": 0.9465073021128774,
206
+ "epoch": 0.08012018027040561,
207
+ "grad_norm": 50.25,
208
+ "learning_rate": 3.58e-06,
209
+ "loss": 15.5603,
210
+ "mean_token_accuracy": 0.7473610159009695,
211
+ "num_tokens": 1200650.0,
212
+ "step": 180
213
+ },
214
+ {
215
+ "entropy": 0.9450380651280283,
216
+ "epoch": 0.08457130139653925,
217
+ "grad_norm": 54.0,
218
+ "learning_rate": 3.7800000000000002e-06,
219
+ "loss": 15.56,
220
+ "mean_token_accuracy": 0.748077143356204,
221
+ "num_tokens": 1265107.0,
222
+ "step": 190
223
+ },
224
+ {
225
+ "entropy": 0.9630168141797185,
226
+ "epoch": 0.0890224225226729,
227
+ "grad_norm": 58.75,
228
+ "learning_rate": 3.980000000000001e-06,
229
+ "loss": 15.5029,
230
+ "mean_token_accuracy": 0.7486971555277705,
231
+ "num_tokens": 1327380.0,
232
+ "step": 200
233
+ },
234
+ {
235
+ "epoch": 0.0890224225226729,
236
+ "eval_biology_entropy": 1.19531831908226,
237
+ "eval_biology_loss": 1.2584387063980103,
238
+ "eval_biology_mean_token_accuracy": 0.6882349443435669,
239
+ "eval_biology_num_tokens": 1327380.0,
240
+ "eval_biology_runtime": 40.8783,
241
+ "eval_biology_samples_per_second": 12.231,
242
+ "eval_biology_steps_per_second": 12.231,
243
+ "step": 200
244
+ },
245
+ {
246
+ "epoch": 0.0890224225226729,
247
+ "eval_chemistry_entropy": 0.9437598274946213,
248
+ "eval_chemistry_loss": 0.9655953645706177,
249
+ "eval_chemistry_mean_token_accuracy": 0.7501006088852883,
250
+ "eval_chemistry_num_tokens": 1327380.0,
251
+ "eval_chemistry_runtime": 46.6023,
252
+ "eval_chemistry_samples_per_second": 10.729,
253
+ "eval_chemistry_steps_per_second": 10.729,
254
+ "step": 200
255
+ },
256
+ {
257
+ "entropy": 0.9782480053603649,
258
+ "epoch": 0.09347354364880654,
259
+ "grad_norm": 64.0,
260
+ "learning_rate": 4.18e-06,
261
+ "loss": 15.9821,
262
+ "mean_token_accuracy": 0.7436690799891948,
263
+ "num_tokens": 1393379.0,
264
+ "step": 210
265
+ },
266
+ {
267
+ "entropy": 0.9402019061148167,
268
+ "epoch": 0.09792466477494019,
269
+ "grad_norm": 54.0,
270
+ "learning_rate": 4.38e-06,
271
+ "loss": 15.3729,
272
+ "mean_token_accuracy": 0.7516820874065161,
273
+ "num_tokens": 1460130.0,
274
+ "step": 220
275
+ },
276
+ {
277
+ "entropy": 0.9247835712507367,
278
+ "epoch": 0.10237578590107384,
279
+ "grad_norm": 54.5,
280
+ "learning_rate": 4.58e-06,
281
+ "loss": 15.0731,
282
+ "mean_token_accuracy": 0.7559274602681398,
283
+ "num_tokens": 1529183.0,
284
+ "step": 230
285
+ },
286
+ {
287
+ "entropy": 0.9673028320074082,
288
+ "epoch": 0.10682690702720747,
289
+ "grad_norm": 71.5,
290
+ "learning_rate": 4.78e-06,
291
+ "loss": 15.731,
292
+ "mean_token_accuracy": 0.7473661951720715,
293
+ "num_tokens": 1597405.0,
294
+ "step": 240
295
+ },
296
+ {
297
+ "entropy": 0.9974556604400278,
298
+ "epoch": 0.11127802815334112,
299
+ "grad_norm": 55.0,
300
+ "learning_rate": 4.980000000000001e-06,
301
+ "loss": 15.8962,
302
+ "mean_token_accuracy": 0.7434635870158672,
303
+ "num_tokens": 1661767.0,
304
+ "step": 250
305
+ },
306
+ {
307
+ "entropy": 0.985609365440905,
308
+ "epoch": 0.11572914927947477,
309
+ "grad_norm": 50.75,
310
+ "learning_rate": 5.18e-06,
311
+ "loss": 16.0267,
312
+ "mean_token_accuracy": 0.7446854375302792,
313
+ "num_tokens": 1728207.0,
314
+ "step": 260
315
+ },
316
+ {
317
+ "entropy": 0.9456103699281811,
318
+ "epoch": 0.12018027040560841,
319
+ "grad_norm": 58.0,
320
+ "learning_rate": 5.380000000000001e-06,
321
+ "loss": 15.3377,
322
+ "mean_token_accuracy": 0.750850186496973,
323
+ "num_tokens": 1796055.0,
324
+ "step": 270
325
+ },
326
+ {
327
+ "entropy": 0.9541573049500585,
328
+ "epoch": 0.12463139153174206,
329
+ "grad_norm": 56.5,
330
+ "learning_rate": 5.580000000000001e-06,
331
+ "loss": 15.2532,
332
+ "mean_token_accuracy": 0.7530262626707553,
333
+ "num_tokens": 1859684.0,
334
+ "step": 280
335
+ },
336
+ {
337
+ "entropy": 0.9941559780389071,
338
+ "epoch": 0.1290825126578757,
339
+ "grad_norm": 59.25,
340
+ "learning_rate": 5.78e-06,
341
+ "loss": 15.94,
342
+ "mean_token_accuracy": 0.745009395852685,
343
+ "num_tokens": 1921704.0,
344
+ "step": 290
345
+ },
346
+ {
347
+ "entropy": 0.981252990104258,
348
+ "epoch": 0.13353363378400934,
349
+ "grad_norm": 56.75,
350
+ "learning_rate": 5.98e-06,
351
+ "loss": 15.8943,
352
+ "mean_token_accuracy": 0.7449573867022992,
353
+ "num_tokens": 1985766.0,
354
+ "step": 300
355
+ },
356
+ {
357
+ "epoch": 0.13353363378400934,
358
+ "eval_biology_entropy": 1.2063790675401687,
359
+ "eval_biology_loss": 1.2592713832855225,
360
+ "eval_biology_mean_token_accuracy": 0.6874204781055451,
361
+ "eval_biology_num_tokens": 1985766.0,
362
+ "eval_biology_runtime": 40.5061,
363
+ "eval_biology_samples_per_second": 12.344,
364
+ "eval_biology_steps_per_second": 12.344,
365
+ "step": 300
366
+ },
367
+ {
368
+ "epoch": 0.13353363378400934,
369
+ "eval_chemistry_entropy": 0.9332761432528496,
370
+ "eval_chemistry_loss": 0.9383891820907593,
371
+ "eval_chemistry_mean_token_accuracy": 0.7543213546276093,
372
+ "eval_chemistry_num_tokens": 1985766.0,
373
+ "eval_chemistry_runtime": 46.2485,
374
+ "eval_chemistry_samples_per_second": 10.811,
375
+ "eval_chemistry_steps_per_second": 10.811,
376
+ "step": 300
377
+ },
378
+ {
379
+ "entropy": 0.9657653540372848,
380
+ "epoch": 0.137984754910143,
381
+ "grad_norm": 66.5,
382
+ "learning_rate": 6.18e-06,
383
+ "loss": 15.5291,
384
+ "mean_token_accuracy": 0.7452911786735058,
385
+ "num_tokens": 2049041.0,
386
+ "step": 310
387
+ },
388
+ {
389
+ "entropy": 0.922235100530088,
390
+ "epoch": 0.14243587603627664,
391
+ "grad_norm": 49.75,
392
+ "learning_rate": 6.380000000000001e-06,
393
+ "loss": 14.7109,
394
+ "mean_token_accuracy": 0.75825478695333,
395
+ "num_tokens": 2118068.0,
396
+ "step": 320
397
+ },
398
+ {
399
+ "entropy": 0.9044711474329233,
400
+ "epoch": 0.14688699716241027,
401
+ "grad_norm": 60.25,
402
+ "learning_rate": 6.5800000000000005e-06,
403
+ "loss": 14.5687,
404
+ "mean_token_accuracy": 0.7618395145982504,
405
+ "num_tokens": 2186387.0,
406
+ "step": 330
407
+ },
408
+ {
409
+ "entropy": 0.946225673891604,
410
+ "epoch": 0.15133811828854393,
411
+ "grad_norm": 52.5,
412
+ "learning_rate": 6.780000000000001e-06,
413
+ "loss": 15.1959,
414
+ "mean_token_accuracy": 0.7535316452383996,
415
+ "num_tokens": 2252650.0,
416
+ "step": 340
417
+ },
418
+ {
419
+ "entropy": 0.9036338411271572,
420
+ "epoch": 0.15578923941467757,
421
+ "grad_norm": 57.75,
422
+ "learning_rate": 6.98e-06,
423
+ "loss": 14.5854,
424
+ "mean_token_accuracy": 0.7611672822386026,
425
+ "num_tokens": 2320358.0,
426
+ "step": 350
427
+ },
428
+ {
429
+ "entropy": 0.9015818448737264,
430
+ "epoch": 0.16024036054081123,
431
+ "grad_norm": 49.5,
432
+ "learning_rate": 7.180000000000001e-06,
433
+ "loss": 14.5381,
434
+ "mean_token_accuracy": 0.7606555309146643,
435
+ "num_tokens": 2388824.0,
436
+ "step": 360
437
+ },
438
+ {
439
+ "entropy": 0.8864203749224544,
440
+ "epoch": 0.16469148166694486,
441
+ "grad_norm": 49.25,
442
+ "learning_rate": 7.3800000000000005e-06,
443
+ "loss": 14.1936,
444
+ "mean_token_accuracy": 0.7665066320449114,
445
+ "num_tokens": 2456144.0,
446
+ "step": 370
447
+ },
448
+ {
449
+ "entropy": 0.9866490814834833,
450
+ "epoch": 0.1691426027930785,
451
+ "grad_norm": 49.5,
452
+ "learning_rate": 7.58e-06,
453
+ "loss": 15.8412,
454
+ "mean_token_accuracy": 0.7478409979492426,
455
+ "num_tokens": 2515325.0,
456
+ "step": 380
457
+ },
458
+ {
459
+ "entropy": 0.9080646676942706,
460
+ "epoch": 0.17359372391921216,
461
+ "grad_norm": 48.25,
462
+ "learning_rate": 7.78e-06,
463
+ "loss": 14.5343,
464
+ "mean_token_accuracy": 0.7594566397368908,
465
+ "num_tokens": 2580490.0,
466
+ "step": 390
467
+ },
468
+ {
469
+ "entropy": 0.9095059128478169,
470
+ "epoch": 0.1780448450453458,
471
+ "grad_norm": 43.5,
472
+ "learning_rate": 7.980000000000002e-06,
473
+ "loss": 14.6647,
474
+ "mean_token_accuracy": 0.7607249341905117,
475
+ "num_tokens": 2644330.0,
476
+ "step": 400
477
+ },
478
+ {
479
+ "epoch": 0.1780448450453458,
480
+ "eval_biology_entropy": 1.1971934199333192,
481
+ "eval_biology_loss": 1.2638347148895264,
482
+ "eval_biology_mean_token_accuracy": 0.6877701328396797,
483
+ "eval_biology_num_tokens": 2644330.0,
484
+ "eval_biology_runtime": 39.8022,
485
+ "eval_biology_samples_per_second": 12.562,
486
+ "eval_biology_steps_per_second": 12.562,
487
+ "step": 400
488
+ },
489
+ {
490
+ "epoch": 0.1780448450453458,
491
+ "eval_chemistry_entropy": 0.8957328157424926,
492
+ "eval_chemistry_loss": 0.9148933291435242,
493
+ "eval_chemistry_mean_token_accuracy": 0.7592848987579346,
494
+ "eval_chemistry_num_tokens": 2644330.0,
495
+ "eval_chemistry_runtime": 46.2334,
496
+ "eval_chemistry_samples_per_second": 10.815,
497
+ "eval_chemistry_steps_per_second": 10.815,
498
+ "step": 400
499
+ },
500
+ {
501
+ "entropy": 0.8526191784068942,
502
+ "epoch": 0.18249596617147945,
503
+ "grad_norm": 50.0,
504
+ "learning_rate": 8.18e-06,
505
+ "loss": 13.7182,
506
+ "mean_token_accuracy": 0.7745671790093184,
507
+ "num_tokens": 2713234.0,
508
+ "step": 410
509
+ },
510
+ {
511
+ "entropy": 0.932603782787919,
512
+ "epoch": 0.18694708729761308,
513
+ "grad_norm": 47.5,
514
+ "learning_rate": 8.380000000000001e-06,
515
+ "loss": 15.028,
516
+ "mean_token_accuracy": 0.7569812458008528,
517
+ "num_tokens": 2783261.0,
518
+ "step": 420
519
+ },
520
+ {
521
+ "entropy": 0.9143912255764007,
522
+ "epoch": 0.19139820842374672,
523
+ "grad_norm": 43.0,
524
+ "learning_rate": 8.580000000000001e-06,
525
+ "loss": 14.7724,
526
+ "mean_token_accuracy": 0.7596961252391339,
527
+ "num_tokens": 2850170.0,
528
+ "step": 430
529
+ },
530
+ {
531
+ "entropy": 0.9113649705424904,
532
+ "epoch": 0.19584932954988038,
533
+ "grad_norm": 58.5,
534
+ "learning_rate": 8.78e-06,
535
+ "loss": 14.6432,
536
+ "mean_token_accuracy": 0.760087676718831,
537
+ "num_tokens": 2913700.0,
538
+ "step": 440
539
+ },
540
+ {
541
+ "entropy": 0.9072460785508156,
542
+ "epoch": 0.200300450676014,
543
+ "grad_norm": 46.5,
544
+ "learning_rate": 8.98e-06,
545
+ "loss": 14.609,
546
+ "mean_token_accuracy": 0.7596194025129079,
547
+ "num_tokens": 2981212.0,
548
+ "step": 450
549
+ },
550
+ {
551
+ "entropy": 0.8693920068442822,
552
+ "epoch": 0.20475157180214767,
553
+ "grad_norm": 53.25,
554
+ "learning_rate": 9.180000000000002e-06,
555
+ "loss": 13.9505,
556
+ "mean_token_accuracy": 0.771946213953197,
557
+ "num_tokens": 3048973.0,
558
+ "step": 460
559
+ },
560
+ {
561
+ "entropy": 0.9109398307278752,
562
+ "epoch": 0.2092026929282813,
563
+ "grad_norm": 47.25,
564
+ "learning_rate": 9.38e-06,
565
+ "loss": 14.614,
566
+ "mean_token_accuracy": 0.7599313069134951,
567
+ "num_tokens": 3117033.0,
568
+ "step": 470
569
+ },
570
+ {
571
+ "entropy": 0.8936059167608619,
572
+ "epoch": 0.21365381405441494,
573
+ "grad_norm": 50.25,
574
+ "learning_rate": 9.58e-06,
575
+ "loss": 14.4662,
576
+ "mean_token_accuracy": 0.7608913701027632,
577
+ "num_tokens": 3185255.0,
578
+ "step": 480
579
+ },
580
+ {
581
+ "entropy": 0.9031545946374535,
582
+ "epoch": 0.2181049351805486,
583
+ "grad_norm": 51.5,
584
+ "learning_rate": 9.780000000000001e-06,
585
+ "loss": 14.5197,
586
+ "mean_token_accuracy": 0.7608289115130902,
587
+ "num_tokens": 3253121.0,
588
+ "step": 490
589
+ },
590
+ {
591
+ "entropy": 0.8450184227898717,
592
+ "epoch": 0.22255605630668224,
593
+ "grad_norm": 49.5,
594
+ "learning_rate": 9.980000000000001e-06,
595
+ "loss": 13.4565,
596
+ "mean_token_accuracy": 0.7740382503718137,
597
+ "num_tokens": 3322823.0,
598
+ "step": 500
599
+ },
600
+ {
601
+ "epoch": 0.22255605630668224,
602
+ "eval_biology_entropy": 1.2025449865460396,
603
+ "eval_biology_loss": 1.267388939857483,
604
+ "eval_biology_mean_token_accuracy": 0.6874357106685638,
605
+ "eval_biology_num_tokens": 3322823.0,
606
+ "eval_biology_runtime": 40.0827,
607
+ "eval_biology_samples_per_second": 12.474,
608
+ "eval_biology_steps_per_second": 12.474,
609
+ "step": 500
610
+ },
611
+ {
612
+ "epoch": 0.22255605630668224,
613
+ "eval_chemistry_entropy": 0.8795891938209534,
614
+ "eval_chemistry_loss": 0.8957814574241638,
615
+ "eval_chemistry_mean_token_accuracy": 0.7629851229190826,
616
+ "eval_chemistry_num_tokens": 3322823.0,
617
+ "eval_chemistry_runtime": 46.3478,
618
+ "eval_chemistry_samples_per_second": 10.788,
619
+ "eval_chemistry_steps_per_second": 10.788,
620
+ "step": 500
621
+ },
622
+ {
623
+ "entropy": 0.8574318964034319,
624
+ "epoch": 0.2270071774328159,
625
+ "grad_norm": 45.25,
626
+ "learning_rate": 1.018e-05,
627
+ "loss": 13.9042,
628
+ "mean_token_accuracy": 0.7678481444716454,
629
+ "num_tokens": 3393393.0,
630
+ "step": 510
631
+ },
632
+ {
633
+ "entropy": 0.9288356432691216,
634
+ "epoch": 0.23145829855894953,
635
+ "grad_norm": 57.25,
636
+ "learning_rate": 1.038e-05,
637
+ "loss": 15.193,
638
+ "mean_token_accuracy": 0.7506348451599478,
639
+ "num_tokens": 3454750.0,
640
+ "step": 520
641
+ },
642
+ {
643
+ "entropy": 0.8564269673079252,
644
+ "epoch": 0.2359094196850832,
645
+ "grad_norm": 50.75,
646
+ "learning_rate": 1.0580000000000002e-05,
647
+ "loss": 13.7644,
648
+ "mean_token_accuracy": 0.7751363463699817,
649
+ "num_tokens": 3526914.0,
650
+ "step": 530
651
+ },
652
+ {
653
+ "entropy": 0.93802858479321,
654
+ "epoch": 0.24036054081121683,
655
+ "grad_norm": 45.0,
656
+ "learning_rate": 1.0780000000000002e-05,
657
+ "loss": 15.0494,
658
+ "mean_token_accuracy": 0.7559149663895368,
659
+ "num_tokens": 3589449.0,
660
+ "step": 540
661
+ },
662
+ {
663
+ "entropy": 0.9012869004160166,
664
+ "epoch": 0.24481166193735046,
665
+ "grad_norm": 50.25,
666
+ "learning_rate": 1.0980000000000002e-05,
667
+ "loss": 14.2344,
668
+ "mean_token_accuracy": 0.7642816316336394,
669
+ "num_tokens": 3655092.0,
670
+ "step": 550
671
+ },
672
+ {
673
+ "entropy": 0.8946880368515849,
674
+ "epoch": 0.24926278306348412,
675
+ "grad_norm": 57.25,
676
+ "learning_rate": 1.1180000000000001e-05,
677
+ "loss": 14.5832,
678
+ "mean_token_accuracy": 0.7632339514791966,
679
+ "num_tokens": 3719113.0,
680
+ "step": 560
681
+ },
682
+ {
683
+ "entropy": 0.8879899585619568,
684
+ "epoch": 0.25371390418961776,
685
+ "grad_norm": 44.5,
686
+ "learning_rate": 1.138e-05,
687
+ "loss": 14.3692,
688
+ "mean_token_accuracy": 0.7596106130629778,
689
+ "num_tokens": 3785282.0,
690
+ "step": 570
691
+ },
692
+ {
693
+ "entropy": 0.9068781601265072,
694
+ "epoch": 0.2581650253157514,
695
+ "grad_norm": 42.25,
696
+ "learning_rate": 1.1580000000000001e-05,
697
+ "loss": 14.6096,
698
+ "mean_token_accuracy": 0.761234056390822,
699
+ "num_tokens": 3852276.0,
700
+ "step": 580
701
+ },
702
+ {
703
+ "entropy": 0.8276193620637059,
704
+ "epoch": 0.262616146441885,
705
+ "grad_norm": 43.25,
706
+ "learning_rate": 1.178e-05,
707
+ "loss": 13.3892,
708
+ "mean_token_accuracy": 0.7755540499463678,
709
+ "num_tokens": 3925649.0,
710
+ "step": 590
711
+ },
712
+ {
713
+ "entropy": 0.8920110030099749,
714
+ "epoch": 0.2670672675680187,
715
+ "grad_norm": 52.25,
716
+ "learning_rate": 1.198e-05,
717
+ "loss": 14.0197,
718
+ "mean_token_accuracy": 0.766492671892047,
719
+ "num_tokens": 3990505.0,
720
+ "step": 600
721
+ },
722
+ {
723
+ "epoch": 0.2670672675680187,
724
+ "eval_biology_entropy": 1.2123423200249672,
725
+ "eval_biology_loss": 1.2729928493499756,
726
+ "eval_biology_mean_token_accuracy": 0.6870606996417046,
727
+ "eval_biology_num_tokens": 3990505.0,
728
+ "eval_biology_runtime": 39.7983,
729
+ "eval_biology_samples_per_second": 12.563,
730
+ "eval_biology_steps_per_second": 12.563,
731
+ "step": 600
732
+ },
733
+ {
734
+ "epoch": 0.2670672675680187,
735
+ "eval_chemistry_entropy": 0.8659857953190804,
736
+ "eval_chemistry_loss": 0.8781383037567139,
737
+ "eval_chemistry_mean_token_accuracy": 0.7664026654362679,
738
+ "eval_chemistry_num_tokens": 3990505.0,
739
+ "eval_chemistry_runtime": 46.1775,
740
+ "eval_chemistry_samples_per_second": 10.828,
741
+ "eval_chemistry_steps_per_second": 10.828,
742
+ "step": 600
743
+ },
744
+ {
745
+ "entropy": 0.8550863016396761,
746
+ "epoch": 0.27151838869415235,
747
+ "grad_norm": 51.25,
748
+ "learning_rate": 1.218e-05,
749
+ "loss": 13.8181,
750
+ "mean_token_accuracy": 0.7682087656110526,
751
+ "num_tokens": 4060657.0,
752
+ "step": 610
753
+ },
754
+ {
755
+ "entropy": 0.8537623688578606,
756
+ "epoch": 0.275969509820286,
757
+ "grad_norm": 42.25,
758
+ "learning_rate": 1.2380000000000002e-05,
759
+ "loss": 13.6601,
760
+ "mean_token_accuracy": 0.772542554140091,
761
+ "num_tokens": 4133119.0,
762
+ "step": 620
763
+ },
764
+ {
765
+ "entropy": 0.8621461872011423,
766
+ "epoch": 0.2804206309464196,
767
+ "grad_norm": 49.25,
768
+ "learning_rate": 1.2580000000000002e-05,
769
+ "loss": 13.9865,
770
+ "mean_token_accuracy": 0.7680005200207234,
771
+ "num_tokens": 4200051.0,
772
+ "step": 630
773
+ },
774
+ {
775
+ "entropy": 0.9106066713109613,
776
+ "epoch": 0.2848717520725533,
777
+ "grad_norm": 49.25,
778
+ "learning_rate": 1.2780000000000001e-05,
779
+ "loss": 14.717,
780
+ "mean_token_accuracy": 0.7637291874736547,
781
+ "num_tokens": 4267403.0,
782
+ "step": 640
783
+ },
784
+ {
785
+ "entropy": 0.868153141438961,
786
+ "epoch": 0.28932287319868694,
787
+ "grad_norm": 52.25,
788
+ "learning_rate": 1.2980000000000001e-05,
789
+ "loss": 13.8988,
790
+ "mean_token_accuracy": 0.7680647127330303,
791
+ "num_tokens": 4333268.0,
792
+ "step": 650
793
+ },
794
+ {
795
+ "entropy": 0.8451825473457575,
796
+ "epoch": 0.29377399432482054,
797
+ "grad_norm": 53.5,
798
+ "learning_rate": 1.3180000000000001e-05,
799
+ "loss": 13.5394,
800
+ "mean_token_accuracy": 0.7747031616047024,
801
+ "num_tokens": 4400242.0,
802
+ "step": 660
803
+ },
804
+ {
805
+ "entropy": 0.8826779069378972,
806
+ "epoch": 0.2982251154509542,
807
+ "grad_norm": 49.75,
808
+ "learning_rate": 1.3380000000000002e-05,
809
+ "loss": 14.0787,
810
+ "mean_token_accuracy": 0.7660945057868958,
811
+ "num_tokens": 4464714.0,
812
+ "step": 670
813
+ },
814
+ {
815
+ "entropy": 0.8183048281818628,
816
+ "epoch": 0.30267623657708786,
817
+ "grad_norm": 47.75,
818
+ "learning_rate": 1.3580000000000002e-05,
819
+ "loss": 13.4564,
820
+ "mean_token_accuracy": 0.7773742496967315,
821
+ "num_tokens": 4535458.0,
822
+ "step": 680
823
+ },
824
+ {
825
+ "entropy": 0.8536316430196166,
826
+ "epoch": 0.3071273577032215,
827
+ "grad_norm": 48.25,
828
+ "learning_rate": 1.378e-05,
829
+ "loss": 13.552,
830
+ "mean_token_accuracy": 0.7754087567329406,
831
+ "num_tokens": 4599099.0,
832
+ "step": 690
833
+ },
834
+ {
835
+ "entropy": 0.816833440028131,
836
+ "epoch": 0.31157847882935513,
837
+ "grad_norm": 39.0,
838
+ "learning_rate": 1.398e-05,
839
+ "loss": 13.1791,
840
+ "mean_token_accuracy": 0.777547013387084,
841
+ "num_tokens": 4662903.0,
842
+ "step": 700
843
+ },
844
+ {
845
+ "epoch": 0.31157847882935513,
846
+ "eval_biology_entropy": 1.2325178788900375,
847
+ "eval_biology_loss": 1.278289556503296,
848
+ "eval_biology_mean_token_accuracy": 0.686549211382866,
849
+ "eval_biology_num_tokens": 4662903.0,
850
+ "eval_biology_runtime": 39.6837,
851
+ "eval_biology_samples_per_second": 12.6,
852
+ "eval_biology_steps_per_second": 12.6,
853
+ "step": 700
854
+ },
855
+ {
856
+ "epoch": 0.31157847882935513,
857
+ "eval_chemistry_entropy": 0.870070047557354,
858
+ "eval_chemistry_loss": 0.8674882650375366,
859
+ "eval_chemistry_mean_token_accuracy": 0.7685190732479096,
860
+ "eval_chemistry_num_tokens": 4662903.0,
861
+ "eval_chemistry_runtime": 46.0896,
862
+ "eval_chemistry_samples_per_second": 10.848,
863
+ "eval_chemistry_steps_per_second": 10.848,
864
+ "step": 700
865
+ },
866
+ {
867
+ "entropy": 0.8818355791270733,
868
+ "epoch": 0.3160295999554888,
869
+ "grad_norm": 54.0,
870
+ "learning_rate": 1.418e-05,
871
+ "loss": 14.2266,
872
+ "mean_token_accuracy": 0.7657015427947045,
873
+ "num_tokens": 4729166.0,
874
+ "step": 710
875
+ },
876
+ {
877
+ "entropy": 0.910587764903903,
878
+ "epoch": 0.32048072108162245,
879
+ "grad_norm": 54.5,
880
+ "learning_rate": 1.4380000000000001e-05,
881
+ "loss": 14.7617,
882
+ "mean_token_accuracy": 0.7591140177100897,
883
+ "num_tokens": 4796815.0,
884
+ "step": 720
885
+ },
886
+ {
887
+ "entropy": 0.8099086729809641,
888
+ "epoch": 0.32493184220775606,
889
+ "grad_norm": 51.25,
890
+ "learning_rate": 1.4580000000000001e-05,
891
+ "loss": 12.9674,
892
+ "mean_token_accuracy": 0.7837777521461249,
893
+ "num_tokens": 4865172.0,
894
+ "step": 730
895
+ },
896
+ {
897
+ "entropy": 0.8765029039233923,
898
+ "epoch": 0.3293829633338897,
899
+ "grad_norm": 44.25,
900
+ "learning_rate": 1.478e-05,
901
+ "loss": 14.0705,
902
+ "mean_token_accuracy": 0.7659166298806668,
903
+ "num_tokens": 4932671.0,
904
+ "step": 740
905
+ },
906
+ {
907
+ "entropy": 0.8784225210547447,
908
+ "epoch": 0.3338340844600234,
909
+ "grad_norm": 44.0,
910
+ "learning_rate": 1.498e-05,
911
+ "loss": 14.1774,
912
+ "mean_token_accuracy": 0.7669123791158199,
913
+ "num_tokens": 4998710.0,
914
+ "step": 750
915
+ },
916
+ {
917
+ "entropy": 0.8316720003262162,
918
+ "epoch": 0.338285205586157,
919
+ "grad_norm": 42.75,
920
+ "learning_rate": 1.5180000000000002e-05,
921
+ "loss": 13.3239,
922
+ "mean_token_accuracy": 0.7748636573553085,
923
+ "num_tokens": 5066948.0,
924
+ "step": 760
925
+ },
926
+ {
927
+ "entropy": 0.8594924572855234,
928
+ "epoch": 0.34273632671229065,
929
+ "grad_norm": 49.5,
930
+ "learning_rate": 1.5380000000000002e-05,
931
+ "loss": 13.8153,
932
+ "mean_token_accuracy": 0.7711022242903709,
933
+ "num_tokens": 5129950.0,
934
+ "step": 770
935
+ },
936
+ {
937
+ "entropy": 0.895938608981669,
938
+ "epoch": 0.3471874478384243,
939
+ "grad_norm": 48.25,
940
+ "learning_rate": 1.5580000000000003e-05,
941
+ "loss": 14.4493,
942
+ "mean_token_accuracy": 0.7591051306575537,
943
+ "num_tokens": 5193519.0,
944
+ "step": 780
945
+ },
946
+ {
947
+ "entropy": 0.8509046232327819,
948
+ "epoch": 0.351638568964558,
949
+ "grad_norm": 56.5,
950
+ "learning_rate": 1.578e-05,
951
+ "loss": 13.6439,
952
+ "mean_token_accuracy": 0.7696408761665225,
953
+ "num_tokens": 5256503.0,
954
+ "step": 790
955
+ },
956
+ {
957
+ "entropy": 0.9008656185120344,
958
+ "epoch": 0.3560896900906916,
959
+ "grad_norm": 48.75,
960
+ "learning_rate": 1.5980000000000003e-05,
961
+ "loss": 14.7305,
962
+ "mean_token_accuracy": 0.7615581404417753,
963
+ "num_tokens": 5324751.0,
964
+ "step": 800
965
+ },
966
+ {
967
+ "epoch": 0.3560896900906916,
968
+ "eval_biology_entropy": 1.267573108136654,
969
+ "eval_biology_loss": 1.2815055847167969,
970
+ "eval_biology_mean_token_accuracy": 0.6854563910365105,
971
+ "eval_biology_num_tokens": 5324751.0,
972
+ "eval_biology_runtime": 39.7889,
973
+ "eval_biology_samples_per_second": 12.566,
974
+ "eval_biology_steps_per_second": 12.566,
975
+ "step": 800
976
+ },
977
+ {
978
+ "epoch": 0.3560896900906916,
979
+ "eval_chemistry_entropy": 0.8897559930682182,
980
+ "eval_chemistry_loss": 0.8585976362228394,
981
+ "eval_chemistry_mean_token_accuracy": 0.7700544927716255,
982
+ "eval_chemistry_num_tokens": 5324751.0,
983
+ "eval_chemistry_runtime": 45.9968,
984
+ "eval_chemistry_samples_per_second": 10.87,
985
+ "eval_chemistry_steps_per_second": 10.87,
986
+ "step": 800
987
+ },
988
+ {
989
+ "entropy": 0.7946027474477887,
990
+ "epoch": 0.36054081121682524,
991
+ "grad_norm": 50.0,
992
+ "learning_rate": 1.618e-05,
993
+ "loss": 12.8437,
994
+ "mean_token_accuracy": 0.7842005740851163,
995
+ "num_tokens": 5391664.0,
996
+ "step": 810
997
+ },
998
+ {
999
+ "entropy": 0.8090571435168386,
1000
+ "epoch": 0.3649919323429589,
1001
+ "grad_norm": 45.0,
1002
+ "learning_rate": 1.638e-05,
1003
+ "loss": 12.7173,
1004
+ "mean_token_accuracy": 0.7849415507167578,
1005
+ "num_tokens": 5458519.0,
1006
+ "step": 820
1007
+ },
1008
+ {
1009
+ "entropy": 0.8408348582684994,
1010
+ "epoch": 0.3694430534690925,
1011
+ "grad_norm": 46.0,
1012
+ "learning_rate": 1.658e-05,
1013
+ "loss": 13.9973,
1014
+ "mean_token_accuracy": 0.7693964328616858,
1015
+ "num_tokens": 5523391.0,
1016
+ "step": 830
1017
+ },
1018
+ {
1019
+ "entropy": 0.8412857724353671,
1020
+ "epoch": 0.37389417459522617,
1021
+ "grad_norm": 44.0,
1022
+ "learning_rate": 1.6780000000000002e-05,
1023
+ "loss": 13.3081,
1024
+ "mean_token_accuracy": 0.7760034879669547,
1025
+ "num_tokens": 5585508.0,
1026
+ "step": 840
1027
+ },
1028
+ {
1029
+ "entropy": 0.836153868213296,
1030
+ "epoch": 0.37834529572135983,
1031
+ "grad_norm": 48.5,
1032
+ "learning_rate": 1.698e-05,
1033
+ "loss": 13.6747,
1034
+ "mean_token_accuracy": 0.775155283510685,
1035
+ "num_tokens": 5650755.0,
1036
+ "step": 850
1037
+ },
1038
+ {
1039
+ "entropy": 0.8424218002706766,
1040
+ "epoch": 0.38279641684749344,
1041
+ "grad_norm": 45.0,
1042
+ "learning_rate": 1.718e-05,
1043
+ "loss": 13.4803,
1044
+ "mean_token_accuracy": 0.7742194497957826,
1045
+ "num_tokens": 5716321.0,
1046
+ "step": 860
1047
+ },
1048
+ {
1049
+ "entropy": 0.883666661940515,
1050
+ "epoch": 0.3872475379736271,
1051
+ "grad_norm": 50.75,
1052
+ "learning_rate": 1.7380000000000003e-05,
1053
+ "loss": 14.1247,
1054
+ "mean_token_accuracy": 0.7642681807279587,
1055
+ "num_tokens": 5782698.0,
1056
+ "step": 870
1057
+ },
1058
+ {
1059
+ "entropy": 0.8131563207134604,
1060
+ "epoch": 0.39169865909976076,
1061
+ "grad_norm": 39.0,
1062
+ "learning_rate": 1.758e-05,
1063
+ "loss": 13.2211,
1064
+ "mean_token_accuracy": 0.7782290887087584,
1065
+ "num_tokens": 5848889.0,
1066
+ "step": 880
1067
+ },
1068
+ {
1069
+ "entropy": 0.8608601313084364,
1070
+ "epoch": 0.3961497802258944,
1071
+ "grad_norm": 44.0,
1072
+ "learning_rate": 1.7780000000000003e-05,
1073
+ "loss": 13.9079,
1074
+ "mean_token_accuracy": 0.7717852048575878,
1075
+ "num_tokens": 5912919.0,
1076
+ "step": 890
1077
+ },
1078
+ {
1079
+ "entropy": 0.8315491866320371,
1080
+ "epoch": 0.400600901352028,
1081
+ "grad_norm": 42.25,
1082
+ "learning_rate": 1.798e-05,
1083
+ "loss": 13.4828,
1084
+ "mean_token_accuracy": 0.7756550934165716,
1085
+ "num_tokens": 5980229.0,
1086
+ "step": 900
1087
+ },
1088
+ {
1089
+ "epoch": 0.400600901352028,
1090
+ "eval_biology_entropy": 1.2335028433203696,
1091
+ "eval_biology_loss": 1.2846676111221313,
1092
+ "eval_biology_mean_token_accuracy": 0.6853172712922097,
1093
+ "eval_biology_num_tokens": 5980229.0,
1094
+ "eval_biology_runtime": 39.7136,
1095
+ "eval_biology_samples_per_second": 12.59,
1096
+ "eval_biology_steps_per_second": 12.59,
1097
+ "step": 900
1098
+ },
1099
+ {
1100
+ "epoch": 0.400600901352028,
1101
+ "eval_chemistry_entropy": 0.8548561576008796,
1102
+ "eval_chemistry_loss": 0.8478842973709106,
1103
+ "eval_chemistry_mean_token_accuracy": 0.773112800002098,
1104
+ "eval_chemistry_num_tokens": 5980229.0,
1105
+ "eval_chemistry_runtime": 45.9488,
1106
+ "eval_chemistry_samples_per_second": 10.882,
1107
+ "eval_chemistry_steps_per_second": 10.882,
1108
+ "step": 900
1109
+ },
1110
+ {
1111
+ "entropy": 0.7898269753903151,
1112
+ "epoch": 0.4050520224781617,
1113
+ "grad_norm": 40.25,
1114
+ "learning_rate": 1.8180000000000002e-05,
1115
+ "loss": 12.7209,
1116
+ "mean_token_accuracy": 0.784660654142499,
1117
+ "num_tokens": 6048508.0,
1118
+ "step": 910
1119
+ },
1120
+ {
1121
+ "entropy": 0.8143722828477621,
1122
+ "epoch": 0.40950314360429535,
1123
+ "grad_norm": 44.5,
1124
+ "learning_rate": 1.8380000000000004e-05,
1125
+ "loss": 13.2591,
1126
+ "mean_token_accuracy": 0.7791141759604215,
1127
+ "num_tokens": 6114855.0,
1128
+ "step": 920
1129
+ },
1130
+ {
1131
+ "entropy": 0.8494924793019891,
1132
+ "epoch": 0.41395426473042896,
1133
+ "grad_norm": 40.0,
1134
+ "learning_rate": 1.858e-05,
1135
+ "loss": 13.506,
1136
+ "mean_token_accuracy": 0.776913444697857,
1137
+ "num_tokens": 6179050.0,
1138
+ "step": 930
1139
+ },
1140
+ {
1141
+ "entropy": 0.8335931519046426,
1142
+ "epoch": 0.4184053858565626,
1143
+ "grad_norm": 45.25,
1144
+ "learning_rate": 1.878e-05,
1145
+ "loss": 13.6217,
1146
+ "mean_token_accuracy": 0.7724651444703341,
1147
+ "num_tokens": 6244987.0,
1148
+ "step": 940
1149
+ },
1150
+ {
1151
+ "entropy": 0.8118610519915819,
1152
+ "epoch": 0.4228565069826963,
1153
+ "grad_norm": 49.75,
1154
+ "learning_rate": 1.898e-05,
1155
+ "loss": 13.0423,
1156
+ "mean_token_accuracy": 0.7783929593861103,
1157
+ "num_tokens": 6310799.0,
1158
+ "step": 950
1159
+ },
1160
+ {
1161
+ "entropy": 0.8028015844523907,
1162
+ "epoch": 0.4273076281088299,
1163
+ "grad_norm": 48.25,
1164
+ "learning_rate": 1.918e-05,
1165
+ "loss": 13.0412,
1166
+ "mean_token_accuracy": 0.778607621230185,
1167
+ "num_tokens": 6378152.0,
1168
+ "step": 960
1169
+ },
1170
+ {
1171
+ "entropy": 0.8370830919593573,
1172
+ "epoch": 0.43175874923496355,
1173
+ "grad_norm": 44.75,
1174
+ "learning_rate": 1.938e-05,
1175
+ "loss": 13.2798,
1176
+ "mean_token_accuracy": 0.7766136281192303,
1177
+ "num_tokens": 6442491.0,
1178
+ "step": 970
1179
+ },
1180
+ {
1181
+ "entropy": 0.862270618416369,
1182
+ "epoch": 0.4362098703610972,
1183
+ "grad_norm": 39.5,
1184
+ "learning_rate": 1.9580000000000002e-05,
1185
+ "loss": 14.0144,
1186
+ "mean_token_accuracy": 0.7675615277141332,
1187
+ "num_tokens": 6510231.0,
1188
+ "step": 980
1189
+ },
1190
+ {
1191
+ "entropy": 0.8360511595383286,
1192
+ "epoch": 0.44066099148723087,
1193
+ "grad_norm": 48.0,
1194
+ "learning_rate": 1.978e-05,
1195
+ "loss": 13.4815,
1196
+ "mean_token_accuracy": 0.7760949255898595,
1197
+ "num_tokens": 6572858.0,
1198
+ "step": 990
1199
+ },
1200
+ {
1201
+ "entropy": 0.8388594528660178,
1202
+ "epoch": 0.4451121126133645,
1203
+ "grad_norm": 43.25,
1204
+ "learning_rate": 1.9980000000000002e-05,
1205
+ "loss": 13.6609,
1206
+ "mean_token_accuracy": 0.7700441874563694,
1207
+ "num_tokens": 6637273.0,
1208
+ "step": 1000
1209
+ },
1210
+ {
1211
+ "epoch": 0.4451121126133645,
1212
+ "eval_biology_entropy": 1.187324990749359,
1213
+ "eval_biology_loss": 1.2962696552276611,
1214
+ "eval_biology_mean_token_accuracy": 0.6844774860739707,
1215
+ "eval_biology_num_tokens": 6637273.0,
1216
+ "eval_biology_runtime": 39.733,
1217
+ "eval_biology_samples_per_second": 12.584,
1218
+ "eval_biology_steps_per_second": 12.584,
1219
+ "step": 1000
1220
+ },
1221
+ {
1222
+ "epoch": 0.4451121126133645,
1223
+ "eval_chemistry_entropy": 0.8146472455263137,
1224
+ "eval_chemistry_loss": 0.8452854156494141,
1225
+ "eval_chemistry_mean_token_accuracy": 0.7733621709942817,
1226
+ "eval_chemistry_num_tokens": 6637273.0,
1227
+ "eval_chemistry_runtime": 46.1309,
1228
+ "eval_chemistry_samples_per_second": 10.839,
1229
+ "eval_chemistry_steps_per_second": 10.839,
1230
+ "step": 1000
1231
+ }
1232
+ ],
1233
+ "logging_steps": 10,
1234
+ "max_steps": 10000,
1235
+ "num_input_tokens_seen": 0,
1236
+ "num_train_epochs": 5,
1237
+ "save_steps": 500,
1238
+ "stateful_callbacks": {
1239
+ "TrainerControl": {
1240
+ "args": {
1241
+ "should_epoch_stop": false,
1242
+ "should_evaluate": false,
1243
+ "should_log": false,
1244
+ "should_save": true,
1245
+ "should_training_stop": false
1246
+ },
1247
+ "attributes": {}
1248
+ }
1249
+ },
1250
+ "total_flos": 4.450581354264347e+17,
1251
+ "train_batch_size": 1,
1252
+ "trial_name": null,
1253
+ "trial_params": null
1254
+ }
checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6fab7fd8d13d8cb7edd1499a2ea354d8c2eb72f507a5456e0592668296e6fd8d
3
+ size 6481