roonbug commited on
Commit
afa7ef9
·
verified ·
1 Parent(s): 69a6f08

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 sha256:8bc75a29a730c9e743cad013feda3b0991a913fafe787c58a1c6e20afad97723
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 sha256:49153ec5495fd0788ee87b3f87aa8d607e56b6e7b978aaf17aff59d0985a6ffb
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:f43e4abd7a982ad8ce3750e536a27dad1b5df423d9303413f3231d7adc7f9bf3
3
+ size 3643278603
checkpoint-1000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6839153725e0ef373aee5916d51b3599fb6e0b69dfc1019957bfa474c6ebcaff
3
+ size 14645
checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1047a716c066f9f4e78c42060834e4ca319f06029bfedda77528f4fcc6970b7e
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 +1,1474 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.4,
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": 0.8839326679706574,
14
+ "epoch": 0.004,
15
+ "grad_norm": 1312.0,
16
+ "learning_rate": 4.5000000000000003e-07,
17
+ "loss": 62.0743,
18
+ "mean_token_accuracy": 0.501955559849739,
19
+ "num_tokens": 51831.0,
20
+ "step": 10
21
+ },
22
+ {
23
+ "entropy": 0.9008006965741515,
24
+ "epoch": 0.008,
25
+ "grad_norm": 1304.0,
26
+ "learning_rate": 9.500000000000001e-07,
27
+ "loss": 62.2143,
28
+ "mean_token_accuracy": 0.4949474561959505,
29
+ "num_tokens": 103172.0,
30
+ "step": 20
31
+ },
32
+ {
33
+ "entropy": 1.0253256265074016,
34
+ "epoch": 0.012,
35
+ "grad_norm": 1024.0,
36
+ "learning_rate": 1.45e-06,
37
+ "loss": 59.54,
38
+ "mean_token_accuracy": 0.4916029818356037,
39
+ "num_tokens": 147915.0,
40
+ "step": 30
41
+ },
42
+ {
43
+ "entropy": 1.1394393384456634,
44
+ "epoch": 0.016,
45
+ "grad_norm": 480.0,
46
+ "learning_rate": 1.9500000000000004e-06,
47
+ "loss": 53.5,
48
+ "mean_token_accuracy": 0.5029474480077625,
49
+ "num_tokens": 195524.0,
50
+ "step": 40
51
+ },
52
+ {
53
+ "entropy": 1.4098424911499023,
54
+ "epoch": 0.02,
55
+ "grad_norm": 210.0,
56
+ "learning_rate": 2.4500000000000003e-06,
57
+ "loss": 45.9974,
58
+ "mean_token_accuracy": 0.5211129426956177,
59
+ "num_tokens": 243747.0,
60
+ "step": 50
61
+ },
62
+ {
63
+ "entropy": 1.9254061240702867,
64
+ "epoch": 0.024,
65
+ "grad_norm": 115.0,
66
+ "learning_rate": 2.95e-06,
67
+ "loss": 42.274,
68
+ "mean_token_accuracy": 0.5261368529871107,
69
+ "num_tokens": 296999.0,
70
+ "step": 60
71
+ },
72
+ {
73
+ "entropy": 2.4110619977116583,
74
+ "epoch": 0.028,
75
+ "grad_norm": 98.0,
76
+ "learning_rate": 3.45e-06,
77
+ "loss": 40.0546,
78
+ "mean_token_accuracy": 0.5326165046542883,
79
+ "num_tokens": 342114.0,
80
+ "step": 70
81
+ },
82
+ {
83
+ "entropy": 2.179915499314666,
84
+ "epoch": 0.032,
85
+ "grad_norm": 84.0,
86
+ "learning_rate": 3.95e-06,
87
+ "loss": 35.2015,
88
+ "mean_token_accuracy": 0.5648044316098094,
89
+ "num_tokens": 390903.0,
90
+ "step": 80
91
+ },
92
+ {
93
+ "entropy": 2.016700956225395,
94
+ "epoch": 0.036,
95
+ "grad_norm": 76.0,
96
+ "learning_rate": 4.450000000000001e-06,
97
+ "loss": 33.5383,
98
+ "mean_token_accuracy": 0.5738874508067966,
99
+ "num_tokens": 440359.0,
100
+ "step": 90
101
+ },
102
+ {
103
+ "entropy": 2.017691622674465,
104
+ "epoch": 0.04,
105
+ "grad_norm": 70.5,
106
+ "learning_rate": 4.95e-06,
107
+ "loss": 32.5671,
108
+ "mean_token_accuracy": 0.5853316407650709,
109
+ "num_tokens": 488481.0,
110
+ "step": 100
111
+ },
112
+ {
113
+ "epoch": 0.04,
114
+ "eval_biology_entropy": 1.9829271322488784,
115
+ "eval_biology_loss": 1.9226267337799072,
116
+ "eval_biology_mean_token_accuracy": 0.5983601321578026,
117
+ "eval_biology_num_tokens": 488481.0,
118
+ "eval_biology_runtime": 48.7684,
119
+ "eval_biology_samples_per_second": 10.253,
120
+ "eval_biology_steps_per_second": 10.253,
121
+ "step": 100
122
+ },
123
+ {
124
+ "epoch": 0.04,
125
+ "eval_chemistry_entropy": 1.4066607907414437,
126
+ "eval_chemistry_loss": 1.3990291357040405,
127
+ "eval_chemistry_mean_token_accuracy": 0.6932575578689575,
128
+ "eval_chemistry_num_tokens": 488481.0,
129
+ "eval_chemistry_runtime": 56.2604,
130
+ "eval_chemistry_samples_per_second": 8.887,
131
+ "eval_chemistry_steps_per_second": 8.887,
132
+ "step": 100
133
+ },
134
+ {
135
+ "epoch": 0.04,
136
+ "eval_math_entropy": 1.0215888382792473,
137
+ "eval_math_loss": 1.2960768938064575,
138
+ "eval_math_mean_token_accuracy": 0.7188282798528671,
139
+ "eval_math_num_tokens": 488481.0,
140
+ "eval_math_runtime": 60.4104,
141
+ "eval_math_samples_per_second": 8.277,
142
+ "eval_math_steps_per_second": 8.277,
143
+ "step": 100
144
+ },
145
+ {
146
+ "epoch": 0.04,
147
+ "eval_physics_entropy": 1.2605198118686676,
148
+ "eval_physics_loss": 1.3288249969482422,
149
+ "eval_physics_mean_token_accuracy": 0.7086242160797119,
150
+ "eval_physics_num_tokens": 488481.0,
151
+ "eval_physics_runtime": 64.6678,
152
+ "eval_physics_samples_per_second": 7.732,
153
+ "eval_physics_steps_per_second": 7.732,
154
+ "step": 100
155
+ },
156
+ {
157
+ "entropy": 1.9778225660324096,
158
+ "epoch": 0.044,
159
+ "grad_norm": 63.0,
160
+ "learning_rate": 5.450000000000001e-06,
161
+ "loss": 30.8512,
162
+ "mean_token_accuracy": 0.5971941664814949,
163
+ "num_tokens": 537471.0,
164
+ "step": 110
165
+ },
166
+ {
167
+ "entropy": 1.8523553270846604,
168
+ "epoch": 0.048,
169
+ "grad_norm": 61.0,
170
+ "learning_rate": 5.950000000000001e-06,
171
+ "loss": 29.5146,
172
+ "mean_token_accuracy": 0.6044289888814092,
173
+ "num_tokens": 589868.0,
174
+ "step": 120
175
+ },
176
+ {
177
+ "entropy": 1.9029788821935654,
178
+ "epoch": 0.052,
179
+ "grad_norm": 60.75,
180
+ "learning_rate": 6.450000000000001e-06,
181
+ "loss": 29.5234,
182
+ "mean_token_accuracy": 0.6029617737978696,
183
+ "num_tokens": 639512.0,
184
+ "step": 130
185
+ },
186
+ {
187
+ "entropy": 1.8046179365366697,
188
+ "epoch": 0.056,
189
+ "grad_norm": 58.0,
190
+ "learning_rate": 6.95e-06,
191
+ "loss": 27.9151,
192
+ "mean_token_accuracy": 0.6184237884357572,
193
+ "num_tokens": 690754.0,
194
+ "step": 140
195
+ },
196
+ {
197
+ "entropy": 1.757765456289053,
198
+ "epoch": 0.06,
199
+ "grad_norm": 60.0,
200
+ "learning_rate": 7.450000000000001e-06,
201
+ "loss": 27.9707,
202
+ "mean_token_accuracy": 0.6147179694846272,
203
+ "num_tokens": 737075.0,
204
+ "step": 150
205
+ },
206
+ {
207
+ "entropy": 1.6650845151394607,
208
+ "epoch": 0.064,
209
+ "grad_norm": 56.5,
210
+ "learning_rate": 7.950000000000002e-06,
211
+ "loss": 26.4519,
212
+ "mean_token_accuracy": 0.6242706684395671,
213
+ "num_tokens": 791190.0,
214
+ "step": 160
215
+ },
216
+ {
217
+ "entropy": 1.6960168480873108,
218
+ "epoch": 0.068,
219
+ "grad_norm": 58.0,
220
+ "learning_rate": 8.45e-06,
221
+ "loss": 27.3757,
222
+ "mean_token_accuracy": 0.6216507468372583,
223
+ "num_tokens": 841344.0,
224
+ "step": 170
225
+ },
226
+ {
227
+ "entropy": 1.6937061734497547,
228
+ "epoch": 0.072,
229
+ "grad_norm": 56.75,
230
+ "learning_rate": 8.95e-06,
231
+ "loss": 27.2176,
232
+ "mean_token_accuracy": 0.6197895903140307,
233
+ "num_tokens": 890075.0,
234
+ "step": 180
235
+ },
236
+ {
237
+ "entropy": 1.6111037533730268,
238
+ "epoch": 0.076,
239
+ "grad_norm": 54.0,
240
+ "learning_rate": 9.450000000000001e-06,
241
+ "loss": 25.6739,
242
+ "mean_token_accuracy": 0.6365846041589975,
243
+ "num_tokens": 940599.0,
244
+ "step": 190
245
+ },
246
+ {
247
+ "entropy": 1.5951560869812966,
248
+ "epoch": 0.08,
249
+ "grad_norm": 59.5,
250
+ "learning_rate": 9.950000000000001e-06,
251
+ "loss": 25.9919,
252
+ "mean_token_accuracy": 0.6377695228904485,
253
+ "num_tokens": 989860.0,
254
+ "step": 200
255
+ },
256
+ {
257
+ "epoch": 0.08,
258
+ "eval_biology_entropy": 1.5966026850938797,
259
+ "eval_biology_loss": 1.6000874042510986,
260
+ "eval_biology_mean_token_accuracy": 0.6389070873260498,
261
+ "eval_biology_num_tokens": 989860.0,
262
+ "eval_biology_runtime": 49.0141,
263
+ "eval_biology_samples_per_second": 10.201,
264
+ "eval_biology_steps_per_second": 10.201,
265
+ "step": 200
266
+ },
267
+ {
268
+ "epoch": 0.08,
269
+ "eval_chemistry_entropy": 1.1679962225556373,
270
+ "eval_chemistry_loss": 1.1516532897949219,
271
+ "eval_chemistry_mean_token_accuracy": 0.7243751650452613,
272
+ "eval_chemistry_num_tokens": 989860.0,
273
+ "eval_chemistry_runtime": 56.2166,
274
+ "eval_chemistry_samples_per_second": 8.894,
275
+ "eval_chemistry_steps_per_second": 8.894,
276
+ "step": 200
277
+ },
278
+ {
279
+ "epoch": 0.08,
280
+ "eval_math_entropy": 0.8919650472998619,
281
+ "eval_math_loss": 1.1023179292678833,
282
+ "eval_math_mean_token_accuracy": 0.742928339600563,
283
+ "eval_math_num_tokens": 989860.0,
284
+ "eval_math_runtime": 57.8672,
285
+ "eval_math_samples_per_second": 8.64,
286
+ "eval_math_steps_per_second": 8.64,
287
+ "step": 200
288
+ },
289
+ {
290
+ "epoch": 0.08,
291
+ "eval_physics_entropy": 1.0691025341749192,
292
+ "eval_physics_loss": 1.101883888244629,
293
+ "eval_physics_mean_token_accuracy": 0.740790363073349,
294
+ "eval_physics_num_tokens": 989860.0,
295
+ "eval_physics_runtime": 64.8886,
296
+ "eval_physics_samples_per_second": 7.706,
297
+ "eval_physics_steps_per_second": 7.706,
298
+ "step": 200
299
+ },
300
+ {
301
+ "entropy": 1.5743534132838248,
302
+ "epoch": 0.084,
303
+ "grad_norm": 58.0,
304
+ "learning_rate": 1.045e-05,
305
+ "loss": 25.1989,
306
+ "mean_token_accuracy": 0.639858372323215,
307
+ "num_tokens": 1041199.0,
308
+ "step": 210
309
+ },
310
+ {
311
+ "entropy": 1.7016947157680988,
312
+ "epoch": 0.088,
313
+ "grad_norm": 61.0,
314
+ "learning_rate": 1.095e-05,
315
+ "loss": 27.1763,
316
+ "mean_token_accuracy": 0.6213466323912143,
317
+ "num_tokens": 1087099.0,
318
+ "step": 220
319
+ },
320
+ {
321
+ "entropy": 1.5750356163829564,
322
+ "epoch": 0.092,
323
+ "grad_norm": 54.25,
324
+ "learning_rate": 1.145e-05,
325
+ "loss": 25.6523,
326
+ "mean_token_accuracy": 0.6347338335588575,
327
+ "num_tokens": 1135090.0,
328
+ "step": 230
329
+ },
330
+ {
331
+ "entropy": 1.616741643473506,
332
+ "epoch": 0.096,
333
+ "grad_norm": 61.75,
334
+ "learning_rate": 1.195e-05,
335
+ "loss": 26.0902,
336
+ "mean_token_accuracy": 0.6296613601967692,
337
+ "num_tokens": 1181777.0,
338
+ "step": 240
339
+ },
340
+ {
341
+ "entropy": 1.5834291957318782,
342
+ "epoch": 0.1,
343
+ "grad_norm": 57.75,
344
+ "learning_rate": 1.2450000000000003e-05,
345
+ "loss": 25.5346,
346
+ "mean_token_accuracy": 0.636902061291039,
347
+ "num_tokens": 1229242.0,
348
+ "step": 250
349
+ },
350
+ {
351
+ "entropy": 1.5806551307439805,
352
+ "epoch": 0.104,
353
+ "grad_norm": 51.0,
354
+ "learning_rate": 1.295e-05,
355
+ "loss": 25.4876,
356
+ "mean_token_accuracy": 0.6371098335832357,
357
+ "num_tokens": 1283881.0,
358
+ "step": 260
359
+ },
360
+ {
361
+ "entropy": 1.5482000581920148,
362
+ "epoch": 0.108,
363
+ "grad_norm": 54.75,
364
+ "learning_rate": 1.3450000000000002e-05,
365
+ "loss": 24.9879,
366
+ "mean_token_accuracy": 0.6427959805354476,
367
+ "num_tokens": 1332507.0,
368
+ "step": 270
369
+ },
370
+ {
371
+ "entropy": 1.4955736376345157,
372
+ "epoch": 0.112,
373
+ "grad_norm": 47.25,
374
+ "learning_rate": 1.3950000000000002e-05,
375
+ "loss": 24.3535,
376
+ "mean_token_accuracy": 0.6494839109480381,
377
+ "num_tokens": 1385513.0,
378
+ "step": 280
379
+ },
380
+ {
381
+ "entropy": 1.585788345709443,
382
+ "epoch": 0.116,
383
+ "grad_norm": 55.0,
384
+ "learning_rate": 1.4450000000000002e-05,
385
+ "loss": 25.6047,
386
+ "mean_token_accuracy": 0.6335254821926355,
387
+ "num_tokens": 1432251.0,
388
+ "step": 290
389
+ },
390
+ {
391
+ "entropy": 1.4942802522331475,
392
+ "epoch": 0.12,
393
+ "grad_norm": 53.25,
394
+ "learning_rate": 1.4950000000000003e-05,
395
+ "loss": 23.8796,
396
+ "mean_token_accuracy": 0.6505829736590385,
397
+ "num_tokens": 1483667.0,
398
+ "step": 300
399
+ },
400
+ {
401
+ "epoch": 0.12,
402
+ "eval_biology_entropy": 1.5278286596536637,
403
+ "eval_biology_loss": 1.5206036567687988,
404
+ "eval_biology_mean_token_accuracy": 0.6507429304122925,
405
+ "eval_biology_num_tokens": 1483667.0,
406
+ "eval_biology_runtime": 48.7887,
407
+ "eval_biology_samples_per_second": 10.248,
408
+ "eval_biology_steps_per_second": 10.248,
409
+ "step": 300
410
+ },
411
+ {
412
+ "epoch": 0.12,
413
+ "eval_chemistry_entropy": 1.1177591310739516,
414
+ "eval_chemistry_loss": 1.092423677444458,
415
+ "eval_chemistry_mean_token_accuracy": 0.7334695355892181,
416
+ "eval_chemistry_num_tokens": 1483667.0,
417
+ "eval_chemistry_runtime": 56.2344,
418
+ "eval_chemistry_samples_per_second": 8.891,
419
+ "eval_chemistry_steps_per_second": 8.891,
420
+ "step": 300
421
+ },
422
+ {
423
+ "epoch": 0.12,
424
+ "eval_math_entropy": 0.8748624550104142,
425
+ "eval_math_loss": 1.0690035820007324,
426
+ "eval_math_mean_token_accuracy": 0.7478210436105728,
427
+ "eval_math_num_tokens": 1483667.0,
428
+ "eval_math_runtime": 57.9846,
429
+ "eval_math_samples_per_second": 8.623,
430
+ "eval_math_steps_per_second": 8.623,
431
+ "step": 300
432
+ },
433
+ {
434
+ "epoch": 0.12,
435
+ "eval_physics_entropy": 1.0364689626097678,
436
+ "eval_physics_loss": 1.0411994457244873,
437
+ "eval_physics_mean_token_accuracy": 0.7504333697557449,
438
+ "eval_physics_num_tokens": 1483667.0,
439
+ "eval_physics_runtime": 64.8677,
440
+ "eval_physics_samples_per_second": 7.708,
441
+ "eval_physics_steps_per_second": 7.708,
442
+ "step": 300
443
+ },
444
+ {
445
+ "entropy": 1.5230422887951136,
446
+ "epoch": 0.124,
447
+ "grad_norm": 46.0,
448
+ "learning_rate": 1.545e-05,
449
+ "loss": 24.9963,
450
+ "mean_token_accuracy": 0.6401191424578428,
451
+ "num_tokens": 1534286.0,
452
+ "step": 310
453
+ },
454
+ {
455
+ "entropy": 1.5620095640420915,
456
+ "epoch": 0.128,
457
+ "grad_norm": 48.75,
458
+ "learning_rate": 1.595e-05,
459
+ "loss": 24.9707,
460
+ "mean_token_accuracy": 0.6434399029240012,
461
+ "num_tokens": 1582368.0,
462
+ "step": 320
463
+ },
464
+ {
465
+ "entropy": 1.5268741115927695,
466
+ "epoch": 0.132,
467
+ "grad_norm": 53.75,
468
+ "learning_rate": 1.645e-05,
469
+ "loss": 24.6543,
470
+ "mean_token_accuracy": 0.6455080669373274,
471
+ "num_tokens": 1630878.0,
472
+ "step": 330
473
+ },
474
+ {
475
+ "entropy": 1.5756058074533938,
476
+ "epoch": 0.136,
477
+ "grad_norm": 55.25,
478
+ "learning_rate": 1.6950000000000002e-05,
479
+ "loss": 25.1865,
480
+ "mean_token_accuracy": 0.6419553246349097,
481
+ "num_tokens": 1677060.0,
482
+ "step": 340
483
+ },
484
+ {
485
+ "entropy": 1.4673658799380065,
486
+ "epoch": 0.14,
487
+ "grad_norm": 45.25,
488
+ "learning_rate": 1.7450000000000004e-05,
489
+ "loss": 24.2051,
490
+ "mean_token_accuracy": 0.650670082308352,
491
+ "num_tokens": 1725186.0,
492
+ "step": 350
493
+ },
494
+ {
495
+ "entropy": 1.5301035284996032,
496
+ "epoch": 0.144,
497
+ "grad_norm": 63.5,
498
+ "learning_rate": 1.795e-05,
499
+ "loss": 24.6432,
500
+ "mean_token_accuracy": 0.6437291771173477,
501
+ "num_tokens": 1773764.0,
502
+ "step": 360
503
+ },
504
+ {
505
+ "entropy": 1.4943597927689551,
506
+ "epoch": 0.148,
507
+ "grad_norm": 49.75,
508
+ "learning_rate": 1.845e-05,
509
+ "loss": 24.1058,
510
+ "mean_token_accuracy": 0.6495298765599727,
511
+ "num_tokens": 1823706.0,
512
+ "step": 370
513
+ },
514
+ {
515
+ "entropy": 1.534142181277275,
516
+ "epoch": 0.152,
517
+ "grad_norm": 61.75,
518
+ "learning_rate": 1.8950000000000003e-05,
519
+ "loss": 24.7723,
520
+ "mean_token_accuracy": 0.6464893363416195,
521
+ "num_tokens": 1873167.0,
522
+ "step": 380
523
+ },
524
+ {
525
+ "entropy": 1.524210822209716,
526
+ "epoch": 0.156,
527
+ "grad_norm": 52.0,
528
+ "learning_rate": 1.9450000000000002e-05,
529
+ "loss": 24.4617,
530
+ "mean_token_accuracy": 0.6497795518487692,
531
+ "num_tokens": 1921484.0,
532
+ "step": 390
533
+ },
534
+ {
535
+ "entropy": 1.479876821860671,
536
+ "epoch": 0.16,
537
+ "grad_norm": 56.75,
538
+ "learning_rate": 1.9950000000000004e-05,
539
+ "loss": 24.2586,
540
+ "mean_token_accuracy": 0.6476251525804401,
541
+ "num_tokens": 1970077.0,
542
+ "step": 400
543
+ },
544
+ {
545
+ "epoch": 0.16,
546
+ "eval_biology_entropy": 1.4946268691420554,
547
+ "eval_biology_loss": 1.4757590293884277,
548
+ "eval_biology_mean_token_accuracy": 0.6582760804891586,
549
+ "eval_biology_num_tokens": 1970077.0,
550
+ "eval_biology_runtime": 48.864,
551
+ "eval_biology_samples_per_second": 10.232,
552
+ "eval_biology_steps_per_second": 10.232,
553
+ "step": 400
554
+ },
555
+ {
556
+ "epoch": 0.16,
557
+ "eval_chemistry_entropy": 1.0978138650655747,
558
+ "eval_chemistry_loss": 1.0629022121429443,
559
+ "eval_chemistry_mean_token_accuracy": 0.7388500568270683,
560
+ "eval_chemistry_num_tokens": 1970077.0,
561
+ "eval_chemistry_runtime": 56.2753,
562
+ "eval_chemistry_samples_per_second": 8.885,
563
+ "eval_chemistry_steps_per_second": 8.885,
564
+ "step": 400
565
+ },
566
+ {
567
+ "epoch": 0.16,
568
+ "eval_math_entropy": 0.8591428139805793,
569
+ "eval_math_loss": 1.05092191696167,
570
+ "eval_math_mean_token_accuracy": 0.7511963348388672,
571
+ "eval_math_num_tokens": 1970077.0,
572
+ "eval_math_runtime": 58.021,
573
+ "eval_math_samples_per_second": 8.618,
574
+ "eval_math_steps_per_second": 8.618,
575
+ "step": 400
576
+ },
577
+ {
578
+ "epoch": 0.16,
579
+ "eval_physics_entropy": 1.0165277478098869,
580
+ "eval_physics_loss": 1.0104260444641113,
581
+ "eval_physics_mean_token_accuracy": 0.7560842056274414,
582
+ "eval_physics_num_tokens": 1970077.0,
583
+ "eval_physics_runtime": 64.8626,
584
+ "eval_physics_samples_per_second": 7.709,
585
+ "eval_physics_steps_per_second": 7.709,
586
+ "step": 400
587
+ },
588
+ {
589
+ "entropy": 1.5483303785324096,
590
+ "epoch": 0.164,
591
+ "grad_norm": 48.25,
592
+ "learning_rate": 1.9950000000000004e-05,
593
+ "loss": 25.1551,
594
+ "mean_token_accuracy": 0.6388038458302617,
595
+ "num_tokens": 2019053.0,
596
+ "step": 410
597
+ },
598
+ {
599
+ "entropy": 1.454981879517436,
600
+ "epoch": 0.168,
601
+ "grad_norm": 39.25,
602
+ "learning_rate": 1.9894444444444446e-05,
603
+ "loss": 23.3291,
604
+ "mean_token_accuracy": 0.6591469643637538,
605
+ "num_tokens": 2066815.0,
606
+ "step": 420
607
+ },
608
+ {
609
+ "entropy": 1.5025325208902358,
610
+ "epoch": 0.172,
611
+ "grad_norm": 42.75,
612
+ "learning_rate": 1.983888888888889e-05,
613
+ "loss": 24.4368,
614
+ "mean_token_accuracy": 0.6499477269127965,
615
+ "num_tokens": 2116402.0,
616
+ "step": 430
617
+ },
618
+ {
619
+ "entropy": 1.4382389456033706,
620
+ "epoch": 0.176,
621
+ "grad_norm": 43.75,
622
+ "learning_rate": 1.9783333333333334e-05,
623
+ "loss": 23.1478,
624
+ "mean_token_accuracy": 0.6576124299317598,
625
+ "num_tokens": 2168354.0,
626
+ "step": 440
627
+ },
628
+ {
629
+ "entropy": 1.4807231716811657,
630
+ "epoch": 0.18,
631
+ "grad_norm": 44.0,
632
+ "learning_rate": 1.972777777777778e-05,
633
+ "loss": 23.9247,
634
+ "mean_token_accuracy": 0.6508983993902803,
635
+ "num_tokens": 2219158.0,
636
+ "step": 450
637
+ },
638
+ {
639
+ "entropy": 1.4961057174950838,
640
+ "epoch": 0.184,
641
+ "grad_norm": 53.25,
642
+ "learning_rate": 1.9672222222222222e-05,
643
+ "loss": 24.1044,
644
+ "mean_token_accuracy": 0.6528054753318429,
645
+ "num_tokens": 2270617.0,
646
+ "step": 460
647
+ },
648
+ {
649
+ "entropy": 1.405641496926546,
650
+ "epoch": 0.188,
651
+ "grad_norm": 49.0,
652
+ "learning_rate": 1.9616666666666668e-05,
653
+ "loss": 22.6821,
654
+ "mean_token_accuracy": 0.6653487289324402,
655
+ "num_tokens": 2322034.0,
656
+ "step": 470
657
+ },
658
+ {
659
+ "entropy": 1.5283996649086475,
660
+ "epoch": 0.192,
661
+ "grad_norm": 56.5,
662
+ "learning_rate": 1.9561111111111114e-05,
663
+ "loss": 24.8943,
664
+ "mean_token_accuracy": 0.6409956673160195,
665
+ "num_tokens": 2365822.0,
666
+ "step": 480
667
+ },
668
+ {
669
+ "entropy": 1.4112088453024625,
670
+ "epoch": 0.196,
671
+ "grad_norm": 56.5,
672
+ "learning_rate": 1.9505555555555556e-05,
673
+ "loss": 22.9679,
674
+ "mean_token_accuracy": 0.6614947421476245,
675
+ "num_tokens": 2413853.0,
676
+ "step": 490
677
+ },
678
+ {
679
+ "entropy": 1.528968420997262,
680
+ "epoch": 0.2,
681
+ "grad_norm": 52.25,
682
+ "learning_rate": 1.9450000000000002e-05,
683
+ "loss": 24.7093,
684
+ "mean_token_accuracy": 0.6438989890739322,
685
+ "num_tokens": 2459876.0,
686
+ "step": 500
687
+ },
688
+ {
689
+ "epoch": 0.2,
690
+ "eval_biology_entropy": 1.4363987457752228,
691
+ "eval_biology_loss": 1.4445728063583374,
692
+ "eval_biology_mean_token_accuracy": 0.6621700811982155,
693
+ "eval_biology_num_tokens": 2459876.0,
694
+ "eval_biology_runtime": 48.7652,
695
+ "eval_biology_samples_per_second": 10.253,
696
+ "eval_biology_steps_per_second": 10.253,
697
+ "step": 500
698
+ },
699
+ {
700
+ "epoch": 0.2,
701
+ "eval_chemistry_entropy": 1.0501885437369347,
702
+ "eval_chemistry_loss": 1.0399649143218994,
703
+ "eval_chemistry_mean_token_accuracy": 0.7431609622240066,
704
+ "eval_chemistry_num_tokens": 2459876.0,
705
+ "eval_chemistry_runtime": 56.6012,
706
+ "eval_chemistry_samples_per_second": 8.834,
707
+ "eval_chemistry_steps_per_second": 8.834,
708
+ "step": 500
709
+ },
710
+ {
711
+ "epoch": 0.2,
712
+ "eval_math_entropy": 0.829749099612236,
713
+ "eval_math_loss": 1.0390610694885254,
714
+ "eval_math_mean_token_accuracy": 0.7532211091518403,
715
+ "eval_math_num_tokens": 2459876.0,
716
+ "eval_math_runtime": 57.9592,
717
+ "eval_math_samples_per_second": 8.627,
718
+ "eval_math_steps_per_second": 8.627,
719
+ "step": 500
720
+ },
721
+ {
722
+ "epoch": 0.2,
723
+ "eval_physics_entropy": 0.9729321378469468,
724
+ "eval_physics_loss": 0.9892891049385071,
725
+ "eval_physics_mean_token_accuracy": 0.7594477337598801,
726
+ "eval_physics_num_tokens": 2459876.0,
727
+ "eval_physics_runtime": 64.8275,
728
+ "eval_physics_samples_per_second": 7.713,
729
+ "eval_physics_steps_per_second": 7.713,
730
+ "step": 500
731
+ },
732
+ {
733
+ "entropy": 1.454107392206788,
734
+ "epoch": 0.204,
735
+ "grad_norm": 44.5,
736
+ "learning_rate": 1.9394444444444448e-05,
737
+ "loss": 23.3348,
738
+ "mean_token_accuracy": 0.6589183287695051,
739
+ "num_tokens": 2509187.0,
740
+ "step": 510
741
+ },
742
+ {
743
+ "entropy": 1.4750641040503978,
744
+ "epoch": 0.208,
745
+ "grad_norm": 50.75,
746
+ "learning_rate": 1.933888888888889e-05,
747
+ "loss": 23.9095,
748
+ "mean_token_accuracy": 0.6498184770345687,
749
+ "num_tokens": 2558762.0,
750
+ "step": 520
751
+ },
752
+ {
753
+ "entropy": 1.4242557927966117,
754
+ "epoch": 0.212,
755
+ "grad_norm": 45.75,
756
+ "learning_rate": 1.9283333333333332e-05,
757
+ "loss": 23.3345,
758
+ "mean_token_accuracy": 0.660492136143148,
759
+ "num_tokens": 2608616.0,
760
+ "step": 530
761
+ },
762
+ {
763
+ "entropy": 1.4807924892753364,
764
+ "epoch": 0.216,
765
+ "grad_norm": 45.75,
766
+ "learning_rate": 1.9227777777777778e-05,
767
+ "loss": 23.5843,
768
+ "mean_token_accuracy": 0.6566273784264922,
769
+ "num_tokens": 2654953.0,
770
+ "step": 540
771
+ },
772
+ {
773
+ "entropy": 1.4450799100100995,
774
+ "epoch": 0.22,
775
+ "grad_norm": 43.75,
776
+ "learning_rate": 1.9172222222222224e-05,
777
+ "loss": 23.4185,
778
+ "mean_token_accuracy": 0.6577624851837754,
779
+ "num_tokens": 2704686.0,
780
+ "step": 550
781
+ },
782
+ {
783
+ "entropy": 1.4568904381245376,
784
+ "epoch": 0.224,
785
+ "grad_norm": 43.75,
786
+ "learning_rate": 1.911666666666667e-05,
787
+ "loss": 23.6667,
788
+ "mean_token_accuracy": 0.6560516567900777,
789
+ "num_tokens": 2755347.0,
790
+ "step": 560
791
+ },
792
+ {
793
+ "entropy": 1.4337828256189824,
794
+ "epoch": 0.228,
795
+ "grad_norm": 41.5,
796
+ "learning_rate": 1.9061111111111112e-05,
797
+ "loss": 22.9651,
798
+ "mean_token_accuracy": 0.6626851694658399,
799
+ "num_tokens": 2804340.0,
800
+ "step": 570
801
+ },
802
+ {
803
+ "entropy": 1.420975561439991,
804
+ "epoch": 0.232,
805
+ "grad_norm": 42.0,
806
+ "learning_rate": 1.9005555555555557e-05,
807
+ "loss": 23.1634,
808
+ "mean_token_accuracy": 0.6576923074200749,
809
+ "num_tokens": 2852298.0,
810
+ "step": 580
811
+ },
812
+ {
813
+ "entropy": 1.4190348222851754,
814
+ "epoch": 0.236,
815
+ "grad_norm": 51.0,
816
+ "learning_rate": 1.8950000000000003e-05,
817
+ "loss": 22.8248,
818
+ "mean_token_accuracy": 0.6662792712450027,
819
+ "num_tokens": 2901227.0,
820
+ "step": 590
821
+ },
822
+ {
823
+ "entropy": 1.460960115492344,
824
+ "epoch": 0.24,
825
+ "grad_norm": 50.0,
826
+ "learning_rate": 1.8894444444444446e-05,
827
+ "loss": 23.6843,
828
+ "mean_token_accuracy": 0.6580705074593425,
829
+ "num_tokens": 2947346.0,
830
+ "step": 600
831
+ },
832
+ {
833
+ "epoch": 0.24,
834
+ "eval_biology_entropy": 1.3783106073141098,
835
+ "eval_biology_loss": 1.4261852502822876,
836
+ "eval_biology_mean_token_accuracy": 0.6656758678555489,
837
+ "eval_biology_num_tokens": 2947346.0,
838
+ "eval_biology_runtime": 48.582,
839
+ "eval_biology_samples_per_second": 10.292,
840
+ "eval_biology_steps_per_second": 10.292,
841
+ "step": 600
842
+ },
843
+ {
844
+ "epoch": 0.24,
845
+ "eval_chemistry_entropy": 1.0091136819124222,
846
+ "eval_chemistry_loss": 1.0272349119186401,
847
+ "eval_chemistry_mean_token_accuracy": 0.7455875177383423,
848
+ "eval_chemistry_num_tokens": 2947346.0,
849
+ "eval_chemistry_runtime": 56.154,
850
+ "eval_chemistry_samples_per_second": 8.904,
851
+ "eval_chemistry_steps_per_second": 8.904,
852
+ "step": 600
853
+ },
854
+ {
855
+ "epoch": 0.24,
856
+ "eval_math_entropy": 0.8054847376346588,
857
+ "eval_math_loss": 1.035906195640564,
858
+ "eval_math_mean_token_accuracy": 0.7551211569309234,
859
+ "eval_math_num_tokens": 2947346.0,
860
+ "eval_math_runtime": 57.8264,
861
+ "eval_math_samples_per_second": 8.647,
862
+ "eval_math_steps_per_second": 8.647,
863
+ "step": 600
864
+ },
865
+ {
866
+ "epoch": 0.24,
867
+ "eval_physics_entropy": 0.9419542922973633,
868
+ "eval_physics_loss": 0.9735694527626038,
869
+ "eval_physics_mean_token_accuracy": 0.7620592492818833,
870
+ "eval_physics_num_tokens": 2947346.0,
871
+ "eval_physics_runtime": 64.7503,
872
+ "eval_physics_samples_per_second": 7.722,
873
+ "eval_physics_steps_per_second": 7.722,
874
+ "step": 600
875
+ },
876
+ {
877
+ "entropy": 1.4665087036788464,
878
+ "epoch": 0.244,
879
+ "grad_norm": 48.25,
880
+ "learning_rate": 1.883888888888889e-05,
881
+ "loss": 23.7145,
882
+ "mean_token_accuracy": 0.6519139897078275,
883
+ "num_tokens": 2996175.0,
884
+ "step": 610
885
+ },
886
+ {
887
+ "entropy": 1.474304298311472,
888
+ "epoch": 0.248,
889
+ "grad_norm": 43.5,
890
+ "learning_rate": 1.8783333333333337e-05,
891
+ "loss": 24.0101,
892
+ "mean_token_accuracy": 0.6545385546982289,
893
+ "num_tokens": 3043128.0,
894
+ "step": 620
895
+ },
896
+ {
897
+ "entropy": 1.4359863974153995,
898
+ "epoch": 0.252,
899
+ "grad_norm": 48.0,
900
+ "learning_rate": 1.872777777777778e-05,
901
+ "loss": 23.1418,
902
+ "mean_token_accuracy": 0.658623912371695,
903
+ "num_tokens": 3090827.0,
904
+ "step": 630
905
+ },
906
+ {
907
+ "entropy": 1.3753734774887563,
908
+ "epoch": 0.256,
909
+ "grad_norm": 45.5,
910
+ "learning_rate": 1.8672222222222222e-05,
911
+ "loss": 22.5106,
912
+ "mean_token_accuracy": 0.668263440579176,
913
+ "num_tokens": 3139957.0,
914
+ "step": 640
915
+ },
916
+ {
917
+ "entropy": 1.4551277089864016,
918
+ "epoch": 0.26,
919
+ "grad_norm": 48.5,
920
+ "learning_rate": 1.8616666666666667e-05,
921
+ "loss": 23.4763,
922
+ "mean_token_accuracy": 0.6564172597602009,
923
+ "num_tokens": 3183188.0,
924
+ "step": 650
925
+ },
926
+ {
927
+ "entropy": 1.4548598907887935,
928
+ "epoch": 0.264,
929
+ "grad_norm": 49.75,
930
+ "learning_rate": 1.8561111111111113e-05,
931
+ "loss": 23.4802,
932
+ "mean_token_accuracy": 0.659644197113812,
933
+ "num_tokens": 3232159.0,
934
+ "step": 660
935
+ },
936
+ {
937
+ "entropy": 1.4130937982350589,
938
+ "epoch": 0.268,
939
+ "grad_norm": 45.0,
940
+ "learning_rate": 1.8505555555555556e-05,
941
+ "loss": 22.738,
942
+ "mean_token_accuracy": 0.6623035183176398,
943
+ "num_tokens": 3283908.0,
944
+ "step": 670
945
+ },
946
+ {
947
+ "entropy": 1.389145464077592,
948
+ "epoch": 0.272,
949
+ "grad_norm": 53.5,
950
+ "learning_rate": 1.845e-05,
951
+ "loss": 22.7338,
952
+ "mean_token_accuracy": 0.664409993775189,
953
+ "num_tokens": 3335951.0,
954
+ "step": 680
955
+ },
956
+ {
957
+ "entropy": 1.4257318537682295,
958
+ "epoch": 0.276,
959
+ "grad_norm": 50.0,
960
+ "learning_rate": 1.8394444444444447e-05,
961
+ "loss": 23.0298,
962
+ "mean_token_accuracy": 0.6590598113834858,
963
+ "num_tokens": 3387755.0,
964
+ "step": 690
965
+ },
966
+ {
967
+ "entropy": 1.414656887575984,
968
+ "epoch": 0.28,
969
+ "grad_norm": 54.25,
970
+ "learning_rate": 1.833888888888889e-05,
971
+ "loss": 22.8344,
972
+ "mean_token_accuracy": 0.6620526265352964,
973
+ "num_tokens": 3437213.0,
974
+ "step": 700
975
+ },
976
+ {
977
+ "epoch": 0.28,
978
+ "eval_biology_entropy": 1.4020620829463004,
979
+ "eval_biology_loss": 1.4081073999404907,
980
+ "eval_biology_mean_token_accuracy": 0.6694456328749656,
981
+ "eval_biology_num_tokens": 3437213.0,
982
+ "eval_biology_runtime": 48.5754,
983
+ "eval_biology_samples_per_second": 10.293,
984
+ "eval_biology_steps_per_second": 10.293,
985
+ "step": 700
986
+ },
987
+ {
988
+ "epoch": 0.28,
989
+ "eval_chemistry_entropy": 1.040973679125309,
990
+ "eval_chemistry_loss": 1.0141925811767578,
991
+ "eval_chemistry_mean_token_accuracy": 0.7476384826302528,
992
+ "eval_chemistry_num_tokens": 3437213.0,
993
+ "eval_chemistry_runtime": 56.1787,
994
+ "eval_chemistry_samples_per_second": 8.9,
995
+ "eval_chemistry_steps_per_second": 8.9,
996
+ "step": 700
997
+ },
998
+ {
999
+ "epoch": 0.28,
1000
+ "eval_math_entropy": 0.8242663496732712,
1001
+ "eval_math_loss": 1.0219008922576904,
1002
+ "eval_math_mean_token_accuracy": 0.7559223830103874,
1003
+ "eval_math_num_tokens": 3437213.0,
1004
+ "eval_math_runtime": 57.8767,
1005
+ "eval_math_samples_per_second": 8.639,
1006
+ "eval_math_steps_per_second": 8.639,
1007
+ "step": 700
1008
+ },
1009
+ {
1010
+ "epoch": 0.28,
1011
+ "eval_physics_entropy": 0.9719093123674393,
1012
+ "eval_physics_loss": 0.9615101218223572,
1013
+ "eval_physics_mean_token_accuracy": 0.7637101773023606,
1014
+ "eval_physics_num_tokens": 3437213.0,
1015
+ "eval_physics_runtime": 64.8419,
1016
+ "eval_physics_samples_per_second": 7.711,
1017
+ "eval_physics_steps_per_second": 7.711,
1018
+ "step": 700
1019
+ },
1020
+ {
1021
+ "entropy": 1.368546247854829,
1022
+ "epoch": 0.284,
1023
+ "grad_norm": 37.25,
1024
+ "learning_rate": 1.8283333333333335e-05,
1025
+ "loss": 21.8042,
1026
+ "mean_token_accuracy": 0.6742603346705437,
1027
+ "num_tokens": 3489088.0,
1028
+ "step": 710
1029
+ },
1030
+ {
1031
+ "entropy": 1.3972204247489572,
1032
+ "epoch": 0.288,
1033
+ "grad_norm": 42.75,
1034
+ "learning_rate": 1.822777777777778e-05,
1035
+ "loss": 23.1509,
1036
+ "mean_token_accuracy": 0.6594159681349993,
1037
+ "num_tokens": 3539731.0,
1038
+ "step": 720
1039
+ },
1040
+ {
1041
+ "entropy": 1.4230724427849055,
1042
+ "epoch": 0.292,
1043
+ "grad_norm": 49.5,
1044
+ "learning_rate": 1.8172222222222223e-05,
1045
+ "loss": 22.696,
1046
+ "mean_token_accuracy": 0.6633614806458354,
1047
+ "num_tokens": 3586433.0,
1048
+ "step": 730
1049
+ },
1050
+ {
1051
+ "entropy": 1.4538289617747069,
1052
+ "epoch": 0.296,
1053
+ "grad_norm": 47.75,
1054
+ "learning_rate": 1.811666666666667e-05,
1055
+ "loss": 23.4747,
1056
+ "mean_token_accuracy": 0.6588025886565447,
1057
+ "num_tokens": 3633448.0,
1058
+ "step": 740
1059
+ },
1060
+ {
1061
+ "entropy": 1.349214567989111,
1062
+ "epoch": 0.3,
1063
+ "grad_norm": 45.75,
1064
+ "learning_rate": 1.806111111111111e-05,
1065
+ "loss": 21.967,
1066
+ "mean_token_accuracy": 0.6792342875152826,
1067
+ "num_tokens": 3685080.0,
1068
+ "step": 750
1069
+ },
1070
+ {
1071
+ "entropy": 1.3578782331198453,
1072
+ "epoch": 0.304,
1073
+ "grad_norm": 44.25,
1074
+ "learning_rate": 1.8005555555555557e-05,
1075
+ "loss": 22.2315,
1076
+ "mean_token_accuracy": 0.668846612982452,
1077
+ "num_tokens": 3733488.0,
1078
+ "step": 760
1079
+ },
1080
+ {
1081
+ "entropy": 1.459796354547143,
1082
+ "epoch": 0.308,
1083
+ "grad_norm": 44.25,
1084
+ "learning_rate": 1.795e-05,
1085
+ "loss": 23.1602,
1086
+ "mean_token_accuracy": 0.6614330835640431,
1087
+ "num_tokens": 3779708.0,
1088
+ "step": 770
1089
+ },
1090
+ {
1091
+ "entropy": 1.4126090794801711,
1092
+ "epoch": 0.312,
1093
+ "grad_norm": 48.0,
1094
+ "learning_rate": 1.7894444444444445e-05,
1095
+ "loss": 23.0122,
1096
+ "mean_token_accuracy": 0.6606836277991534,
1097
+ "num_tokens": 3825212.0,
1098
+ "step": 780
1099
+ },
1100
+ {
1101
+ "entropy": 1.4439950220286846,
1102
+ "epoch": 0.316,
1103
+ "grad_norm": 60.5,
1104
+ "learning_rate": 1.783888888888889e-05,
1105
+ "loss": 23.4831,
1106
+ "mean_token_accuracy": 0.6551693486049771,
1107
+ "num_tokens": 3870912.0,
1108
+ "step": 790
1109
+ },
1110
+ {
1111
+ "entropy": 1.4275024197995663,
1112
+ "epoch": 0.32,
1113
+ "grad_norm": 54.25,
1114
+ "learning_rate": 1.7783333333333333e-05,
1115
+ "loss": 22.8627,
1116
+ "mean_token_accuracy": 0.6635510809719563,
1117
+ "num_tokens": 3920545.0,
1118
+ "step": 800
1119
+ },
1120
+ {
1121
+ "epoch": 0.32,
1122
+ "eval_biology_entropy": 1.391498780965805,
1123
+ "eval_biology_loss": 1.3933708667755127,
1124
+ "eval_biology_mean_token_accuracy": 0.6710453909635544,
1125
+ "eval_biology_num_tokens": 3920545.0,
1126
+ "eval_biology_runtime": 48.5173,
1127
+ "eval_biology_samples_per_second": 10.306,
1128
+ "eval_biology_steps_per_second": 10.306,
1129
+ "step": 800
1130
+ },
1131
+ {
1132
+ "epoch": 0.32,
1133
+ "eval_chemistry_entropy": 1.033769123196602,
1134
+ "eval_chemistry_loss": 1.00752592086792,
1135
+ "eval_chemistry_mean_token_accuracy": 0.7497823033332824,
1136
+ "eval_chemistry_num_tokens": 3920545.0,
1137
+ "eval_chemistry_runtime": 56.1437,
1138
+ "eval_chemistry_samples_per_second": 8.906,
1139
+ "eval_chemistry_steps_per_second": 8.906,
1140
+ "step": 800
1141
+ },
1142
+ {
1143
+ "epoch": 0.32,
1144
+ "eval_math_entropy": 0.8434388428926468,
1145
+ "eval_math_loss": 1.0179753303527832,
1146
+ "eval_math_mean_token_accuracy": 0.7559520703554153,
1147
+ "eval_math_num_tokens": 3920545.0,
1148
+ "eval_math_runtime": 57.8344,
1149
+ "eval_math_samples_per_second": 8.645,
1150
+ "eval_math_steps_per_second": 8.645,
1151
+ "step": 800
1152
+ },
1153
+ {
1154
+ "epoch": 0.32,
1155
+ "eval_physics_entropy": 0.9763550131320954,
1156
+ "eval_physics_loss": 0.9549967050552368,
1157
+ "eval_physics_mean_token_accuracy": 0.765521013379097,
1158
+ "eval_physics_num_tokens": 3920545.0,
1159
+ "eval_physics_runtime": 64.8184,
1160
+ "eval_physics_samples_per_second": 7.714,
1161
+ "eval_physics_steps_per_second": 7.714,
1162
+ "step": 800
1163
+ },
1164
+ {
1165
+ "entropy": 1.418811386078596,
1166
+ "epoch": 0.324,
1167
+ "grad_norm": 43.0,
1168
+ "learning_rate": 1.772777777777778e-05,
1169
+ "loss": 22.9286,
1170
+ "mean_token_accuracy": 0.6607653006911278,
1171
+ "num_tokens": 3967499.0,
1172
+ "step": 810
1173
+ },
1174
+ {
1175
+ "entropy": 1.3724840827286244,
1176
+ "epoch": 0.328,
1177
+ "grad_norm": 44.25,
1178
+ "learning_rate": 1.7672222222222224e-05,
1179
+ "loss": 22.2047,
1180
+ "mean_token_accuracy": 0.6701890625059604,
1181
+ "num_tokens": 4016673.0,
1182
+ "step": 820
1183
+ },
1184
+ {
1185
+ "entropy": 1.4031333830207586,
1186
+ "epoch": 0.332,
1187
+ "grad_norm": 47.25,
1188
+ "learning_rate": 1.761666666666667e-05,
1189
+ "loss": 23.0022,
1190
+ "mean_token_accuracy": 0.660862791724503,
1191
+ "num_tokens": 4063859.0,
1192
+ "step": 830
1193
+ },
1194
+ {
1195
+ "entropy": 1.4056745786219835,
1196
+ "epoch": 0.336,
1197
+ "grad_norm": 44.0,
1198
+ "learning_rate": 1.7561111111111113e-05,
1199
+ "loss": 22.5703,
1200
+ "mean_token_accuracy": 0.6621862905099988,
1201
+ "num_tokens": 4114077.0,
1202
+ "step": 840
1203
+ },
1204
+ {
1205
+ "entropy": 1.4290663667023182,
1206
+ "epoch": 0.34,
1207
+ "grad_norm": 42.0,
1208
+ "learning_rate": 1.7505555555555558e-05,
1209
+ "loss": 22.956,
1210
+ "mean_token_accuracy": 0.6598267763853073,
1211
+ "num_tokens": 4160168.0,
1212
+ "step": 850
1213
+ },
1214
+ {
1215
+ "entropy": 1.3684453016147018,
1216
+ "epoch": 0.344,
1217
+ "grad_norm": 50.25,
1218
+ "learning_rate": 1.7450000000000004e-05,
1219
+ "loss": 22.5233,
1220
+ "mean_token_accuracy": 0.670460031926632,
1221
+ "num_tokens": 4210403.0,
1222
+ "step": 860
1223
+ },
1224
+ {
1225
+ "entropy": 1.3586817767471076,
1226
+ "epoch": 0.348,
1227
+ "grad_norm": 42.5,
1228
+ "learning_rate": 1.7394444444444446e-05,
1229
+ "loss": 21.8543,
1230
+ "mean_token_accuracy": 0.672089285030961,
1231
+ "num_tokens": 4256944.0,
1232
+ "step": 870
1233
+ },
1234
+ {
1235
+ "entropy": 1.4272807750850915,
1236
+ "epoch": 0.352,
1237
+ "grad_norm": 42.5,
1238
+ "learning_rate": 1.733888888888889e-05,
1239
+ "loss": 23.0583,
1240
+ "mean_token_accuracy": 0.6652814907953143,
1241
+ "num_tokens": 4306949.0,
1242
+ "step": 880
1243
+ },
1244
+ {
1245
+ "entropy": 1.3964965641498566,
1246
+ "epoch": 0.356,
1247
+ "grad_norm": 42.25,
1248
+ "learning_rate": 1.7283333333333334e-05,
1249
+ "loss": 22.6552,
1250
+ "mean_token_accuracy": 0.6640587879344821,
1251
+ "num_tokens": 4355070.0,
1252
+ "step": 890
1253
+ },
1254
+ {
1255
+ "entropy": 1.4213988482952118,
1256
+ "epoch": 0.36,
1257
+ "grad_norm": 34.0,
1258
+ "learning_rate": 1.722777777777778e-05,
1259
+ "loss": 22.4917,
1260
+ "mean_token_accuracy": 0.6680676110088826,
1261
+ "num_tokens": 4406104.0,
1262
+ "step": 900
1263
+ },
1264
+ {
1265
+ "epoch": 0.36,
1266
+ "eval_biology_entropy": 1.3215562723875045,
1267
+ "eval_biology_loss": 1.3810561895370483,
1268
+ "eval_biology_mean_token_accuracy": 0.6744087367653847,
1269
+ "eval_biology_num_tokens": 4406104.0,
1270
+ "eval_biology_runtime": 48.5466,
1271
+ "eval_biology_samples_per_second": 10.299,
1272
+ "eval_biology_steps_per_second": 10.299,
1273
+ "step": 900
1274
+ },
1275
+ {
1276
+ "epoch": 0.36,
1277
+ "eval_chemistry_entropy": 0.989146444439888,
1278
+ "eval_chemistry_loss": 0.9984143972396851,
1279
+ "eval_chemistry_mean_token_accuracy": 0.7511961939930916,
1280
+ "eval_chemistry_num_tokens": 4406104.0,
1281
+ "eval_chemistry_runtime": 56.171,
1282
+ "eval_chemistry_samples_per_second": 8.901,
1283
+ "eval_chemistry_steps_per_second": 8.901,
1284
+ "step": 900
1285
+ },
1286
+ {
1287
+ "epoch": 0.36,
1288
+ "eval_math_entropy": 0.799599610209465,
1289
+ "eval_math_loss": 1.0206644535064697,
1290
+ "eval_math_mean_token_accuracy": 0.7562864592075348,
1291
+ "eval_math_num_tokens": 4406104.0,
1292
+ "eval_math_runtime": 57.8568,
1293
+ "eval_math_samples_per_second": 8.642,
1294
+ "eval_math_steps_per_second": 8.642,
1295
+ "step": 900
1296
+ },
1297
+ {
1298
+ "epoch": 0.36,
1299
+ "eval_physics_entropy": 0.9270223636031151,
1300
+ "eval_physics_loss": 0.9444345831871033,
1301
+ "eval_physics_mean_token_accuracy": 0.7675528969168663,
1302
+ "eval_physics_num_tokens": 4406104.0,
1303
+ "eval_physics_runtime": 64.7993,
1304
+ "eval_physics_samples_per_second": 7.716,
1305
+ "eval_physics_steps_per_second": 7.716,
1306
+ "step": 900
1307
+ },
1308
+ {
1309
+ "entropy": 1.3845266662538052,
1310
+ "epoch": 0.364,
1311
+ "grad_norm": 44.5,
1312
+ "learning_rate": 1.7172222222222223e-05,
1313
+ "loss": 22.7876,
1314
+ "mean_token_accuracy": 0.6699303101748229,
1315
+ "num_tokens": 4455015.0,
1316
+ "step": 910
1317
+ },
1318
+ {
1319
+ "entropy": 1.3507468841969967,
1320
+ "epoch": 0.368,
1321
+ "grad_norm": 54.25,
1322
+ "learning_rate": 1.7116666666666668e-05,
1323
+ "loss": 22.1191,
1324
+ "mean_token_accuracy": 0.6698823075741529,
1325
+ "num_tokens": 4504001.0,
1326
+ "step": 920
1327
+ },
1328
+ {
1329
+ "entropy": 1.4101964194327592,
1330
+ "epoch": 0.372,
1331
+ "grad_norm": 46.0,
1332
+ "learning_rate": 1.7061111111111114e-05,
1333
+ "loss": 22.5848,
1334
+ "mean_token_accuracy": 0.6687647785991431,
1335
+ "num_tokens": 4550269.0,
1336
+ "step": 930
1337
+ },
1338
+ {
1339
+ "entropy": 1.357139458693564,
1340
+ "epoch": 0.376,
1341
+ "grad_norm": 47.5,
1342
+ "learning_rate": 1.7005555555555556e-05,
1343
+ "loss": 22.1243,
1344
+ "mean_token_accuracy": 0.6716644866392016,
1345
+ "num_tokens": 4597559.0,
1346
+ "step": 940
1347
+ },
1348
+ {
1349
+ "entropy": 1.3879809100180864,
1350
+ "epoch": 0.38,
1351
+ "grad_norm": 47.5,
1352
+ "learning_rate": 1.6950000000000002e-05,
1353
+ "loss": 22.0727,
1354
+ "mean_token_accuracy": 0.6727892020717263,
1355
+ "num_tokens": 4646903.0,
1356
+ "step": 950
1357
+ },
1358
+ {
1359
+ "entropy": 1.3612046051770448,
1360
+ "epoch": 0.384,
1361
+ "grad_norm": 42.75,
1362
+ "learning_rate": 1.6894444444444448e-05,
1363
+ "loss": 22.3926,
1364
+ "mean_token_accuracy": 0.6669768806546926,
1365
+ "num_tokens": 4693812.0,
1366
+ "step": 960
1367
+ },
1368
+ {
1369
+ "entropy": 1.4104670461267232,
1370
+ "epoch": 0.388,
1371
+ "grad_norm": 48.25,
1372
+ "learning_rate": 1.683888888888889e-05,
1373
+ "loss": 22.6604,
1374
+ "mean_token_accuracy": 0.6653323493897915,
1375
+ "num_tokens": 4740309.0,
1376
+ "step": 970
1377
+ },
1378
+ {
1379
+ "entropy": 1.4341064432635904,
1380
+ "epoch": 0.392,
1381
+ "grad_norm": 42.5,
1382
+ "learning_rate": 1.6783333333333332e-05,
1383
+ "loss": 22.8327,
1384
+ "mean_token_accuracy": 0.6646358489990234,
1385
+ "num_tokens": 4790965.0,
1386
+ "step": 980
1387
+ },
1388
+ {
1389
+ "entropy": 1.3664294127374887,
1390
+ "epoch": 0.396,
1391
+ "grad_norm": 43.0,
1392
+ "learning_rate": 1.6727777777777778e-05,
1393
+ "loss": 22.4088,
1394
+ "mean_token_accuracy": 0.6667625535279512,
1395
+ "num_tokens": 4839186.0,
1396
+ "step": 990
1397
+ },
1398
+ {
1399
+ "entropy": 1.4091253302991391,
1400
+ "epoch": 0.4,
1401
+ "grad_norm": 53.25,
1402
+ "learning_rate": 1.6672222222222224e-05,
1403
+ "loss": 22.6754,
1404
+ "mean_token_accuracy": 0.6634825209155679,
1405
+ "num_tokens": 4887094.0,
1406
+ "step": 1000
1407
+ },
1408
+ {
1409
+ "epoch": 0.4,
1410
+ "eval_biology_entropy": 1.3544746099114418,
1411
+ "eval_biology_loss": 1.3718540668487549,
1412
+ "eval_biology_mean_token_accuracy": 0.6754276250600815,
1413
+ "eval_biology_num_tokens": 4887094.0,
1414
+ "eval_biology_runtime": 48.4485,
1415
+ "eval_biology_samples_per_second": 10.32,
1416
+ "eval_biology_steps_per_second": 10.32,
1417
+ "step": 1000
1418
+ },
1419
+ {
1420
+ "epoch": 0.4,
1421
+ "eval_chemistry_entropy": 0.9990591505169868,
1422
+ "eval_chemistry_loss": 0.9907504916191101,
1423
+ "eval_chemistry_mean_token_accuracy": 0.7525870882868767,
1424
+ "eval_chemistry_num_tokens": 4887094.0,
1425
+ "eval_chemistry_runtime": 62.8637,
1426
+ "eval_chemistry_samples_per_second": 7.954,
1427
+ "eval_chemistry_steps_per_second": 7.954,
1428
+ "step": 1000
1429
+ },
1430
+ {
1431
+ "epoch": 0.4,
1432
+ "eval_math_entropy": 0.8055654801130295,
1433
+ "eval_math_loss": 1.0127589702606201,
1434
+ "eval_math_mean_token_accuracy": 0.7569557039737701,
1435
+ "eval_math_num_tokens": 4887094.0,
1436
+ "eval_math_runtime": 57.8462,
1437
+ "eval_math_samples_per_second": 8.644,
1438
+ "eval_math_steps_per_second": 8.644,
1439
+ "step": 1000
1440
+ },
1441
+ {
1442
+ "epoch": 0.4,
1443
+ "eval_physics_entropy": 0.9393544716238975,
1444
+ "eval_physics_loss": 0.9393035769462585,
1445
+ "eval_physics_mean_token_accuracy": 0.7682145276069641,
1446
+ "eval_physics_num_tokens": 4887094.0,
1447
+ "eval_physics_runtime": 65.1801,
1448
+ "eval_physics_samples_per_second": 7.671,
1449
+ "eval_physics_steps_per_second": 7.671,
1450
+ "step": 1000
1451
+ }
1452
+ ],
1453
+ "logging_steps": 10,
1454
+ "max_steps": 4000,
1455
+ "num_input_tokens_seen": 0,
1456
+ "num_train_epochs": 2,
1457
+ "save_steps": 500,
1458
+ "stateful_callbacks": {
1459
+ "TrainerControl": {
1460
+ "args": {
1461
+ "should_epoch_stop": false,
1462
+ "should_evaluate": false,
1463
+ "should_log": false,
1464
+ "should_save": true,
1465
+ "should_training_stop": false
1466
+ },
1467
+ "attributes": {}
1468
+ }
1469
+ },
1470
+ "total_flos": 3.277009915508544e+17,
1471
+ "train_batch_size": 1,
1472
+ "trial_name": null,
1473
+ "trial_params": null
1474
+ }
checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2e85f444fc413f87330ed2f5d5a2733a198022ec5f4ab5972e71dda08fb42186
3
+ size 6481