roonbug commited on
Commit
7d8065d
·
verified ·
1 Parent(s): ed2c5aa

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 sha256:3b854f1e8a1398f14025f109c082fdd868623b96a430e5dc24ab3f47a02d9918
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 sha256:79690270926722bfb588a916fd005e5d1db6eb9e316ac7cec6c85ea03d03e311
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 sha256:3d0c1e9fb313e7a07053f87beaa90e52ba7e9d308f9d4c554480dcb2f5bbc8b1
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": 0.4,
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": 0.7787420120090246,
14
+ "epoch": 0.004,
15
+ "grad_norm": 72.0,
16
+ "learning_rate": 1.8e-07,
17
+ "loss": 15.7442,
18
+ "mean_token_accuracy": 0.7588483087718487,
19
+ "num_tokens": 89176.0,
20
+ "step": 10
21
+ },
22
+ {
23
+ "entropy": 0.8886257205158472,
24
+ "epoch": 0.008,
25
+ "grad_norm": 73.5,
26
+ "learning_rate": 3.8e-07,
27
+ "loss": 17.8319,
28
+ "mean_token_accuracy": 0.7328463714569807,
29
+ "num_tokens": 173886.0,
30
+ "step": 20
31
+ },
32
+ {
33
+ "entropy": 0.8274001179262995,
34
+ "epoch": 0.012,
35
+ "grad_norm": 97.5,
36
+ "learning_rate": 5.800000000000001e-07,
37
+ "loss": 16.992,
38
+ "mean_token_accuracy": 0.7489429574459792,
39
+ "num_tokens": 258970.0,
40
+ "step": 30
41
+ },
42
+ {
43
+ "entropy": 0.8739964630454778,
44
+ "epoch": 0.016,
45
+ "grad_norm": 122.5,
46
+ "learning_rate": 7.8e-07,
47
+ "loss": 17.5949,
48
+ "mean_token_accuracy": 0.7390060629695654,
49
+ "num_tokens": 338352.0,
50
+ "step": 40
51
+ },
52
+ {
53
+ "entropy": 0.822566282376647,
54
+ "epoch": 0.02,
55
+ "grad_norm": 79.0,
56
+ "learning_rate": 9.800000000000001e-07,
57
+ "loss": 16.3559,
58
+ "mean_token_accuracy": 0.7543706968426704,
59
+ "num_tokens": 424243.0,
60
+ "step": 50
61
+ },
62
+ {
63
+ "entropy": 0.8778181752189994,
64
+ "epoch": 0.024,
65
+ "grad_norm": 90.0,
66
+ "learning_rate": 1.1800000000000001e-06,
67
+ "loss": 17.4115,
68
+ "mean_token_accuracy": 0.7384452436119318,
69
+ "num_tokens": 505613.0,
70
+ "step": 60
71
+ },
72
+ {
73
+ "entropy": 0.8833056677132844,
74
+ "epoch": 0.028,
75
+ "grad_norm": 70.0,
76
+ "learning_rate": 1.3800000000000001e-06,
77
+ "loss": 17.0728,
78
+ "mean_token_accuracy": 0.7415951255708932,
79
+ "num_tokens": 584157.0,
80
+ "step": 70
81
+ },
82
+ {
83
+ "entropy": 0.8511845586821437,
84
+ "epoch": 0.032,
85
+ "grad_norm": 59.0,
86
+ "learning_rate": 1.5800000000000001e-06,
87
+ "loss": 16.0803,
88
+ "mean_token_accuracy": 0.755342447385192,
89
+ "num_tokens": 671193.0,
90
+ "step": 80
91
+ },
92
+ {
93
+ "entropy": 0.9350998625159264,
94
+ "epoch": 0.036,
95
+ "grad_norm": 84.0,
96
+ "learning_rate": 1.7800000000000001e-06,
97
+ "loss": 17.4767,
98
+ "mean_token_accuracy": 0.737629271298647,
99
+ "num_tokens": 753947.0,
100
+ "step": 90
101
+ },
102
+ {
103
+ "entropy": 0.8764035234227776,
104
+ "epoch": 0.04,
105
+ "grad_norm": 65.5,
106
+ "learning_rate": 1.98e-06,
107
+ "loss": 15.8079,
108
+ "mean_token_accuracy": 0.7533309157937765,
109
+ "num_tokens": 835868.0,
110
+ "step": 100
111
+ },
112
+ {
113
+ "epoch": 0.04,
114
+ "eval_biology_entropy": 1.1230690423250198,
115
+ "eval_biology_loss": 1.2683643102645874,
116
+ "eval_biology_mean_token_accuracy": 0.6887624478340149,
117
+ "eval_biology_num_tokens": 835868.0,
118
+ "eval_biology_runtime": 40.1976,
119
+ "eval_biology_samples_per_second": 12.439,
120
+ "eval_biology_steps_per_second": 12.439,
121
+ "step": 100
122
+ },
123
+ {
124
+ "epoch": 0.04,
125
+ "eval_physics_entropy": 0.9141198564767837,
126
+ "eval_physics_loss": 1.0196080207824707,
127
+ "eval_physics_mean_token_accuracy": 0.750413155734539,
128
+ "eval_physics_num_tokens": 835868.0,
129
+ "eval_physics_runtime": 53.5733,
130
+ "eval_physics_samples_per_second": 9.333,
131
+ "eval_physics_steps_per_second": 9.333,
132
+ "step": 100
133
+ },
134
+ {
135
+ "entropy": 0.8583992384374142,
136
+ "epoch": 0.044,
137
+ "grad_norm": 70.0,
138
+ "learning_rate": 2.1800000000000003e-06,
139
+ "loss": 15.0354,
140
+ "mean_token_accuracy": 0.7647321570664645,
141
+ "num_tokens": 917534.0,
142
+ "step": 110
143
+ },
144
+ {
145
+ "entropy": 0.8902238085865974,
146
+ "epoch": 0.048,
147
+ "grad_norm": 66.0,
148
+ "learning_rate": 2.38e-06,
149
+ "loss": 15.3774,
150
+ "mean_token_accuracy": 0.7589590523391962,
151
+ "num_tokens": 1001008.0,
152
+ "step": 120
153
+ },
154
+ {
155
+ "entropy": 0.9226205172017217,
156
+ "epoch": 0.052,
157
+ "grad_norm": 55.75,
158
+ "learning_rate": 2.5800000000000003e-06,
159
+ "loss": 15.5925,
160
+ "mean_token_accuracy": 0.7562482781708241,
161
+ "num_tokens": 1085098.0,
162
+ "step": 130
163
+ },
164
+ {
165
+ "entropy": 0.9374059302732348,
166
+ "epoch": 0.056,
167
+ "grad_norm": 60.0,
168
+ "learning_rate": 2.7800000000000005e-06,
169
+ "loss": 16.0069,
170
+ "mean_token_accuracy": 0.7525967676192522,
171
+ "num_tokens": 1166615.0,
172
+ "step": 140
173
+ },
174
+ {
175
+ "entropy": 0.9370595531538128,
176
+ "epoch": 0.06,
177
+ "grad_norm": 46.0,
178
+ "learning_rate": 2.9800000000000003e-06,
179
+ "loss": 15.4355,
180
+ "mean_token_accuracy": 0.7566674139350653,
181
+ "num_tokens": 1248387.0,
182
+ "step": 150
183
+ },
184
+ {
185
+ "entropy": 0.9721512474119663,
186
+ "epoch": 0.064,
187
+ "grad_norm": 47.75,
188
+ "learning_rate": 3.1800000000000005e-06,
189
+ "loss": 16.3572,
190
+ "mean_token_accuracy": 0.7466333545744419,
191
+ "num_tokens": 1331208.0,
192
+ "step": 160
193
+ },
194
+ {
195
+ "entropy": 0.9271958880126476,
196
+ "epoch": 0.068,
197
+ "grad_norm": 55.25,
198
+ "learning_rate": 3.3800000000000007e-06,
199
+ "loss": 15.062,
200
+ "mean_token_accuracy": 0.7615874823182821,
201
+ "num_tokens": 1416841.0,
202
+ "step": 170
203
+ },
204
+ {
205
+ "entropy": 0.9161595629528165,
206
+ "epoch": 0.072,
207
+ "grad_norm": 49.25,
208
+ "learning_rate": 3.58e-06,
209
+ "loss": 15.0671,
210
+ "mean_token_accuracy": 0.7618400476872921,
211
+ "num_tokens": 1499449.0,
212
+ "step": 180
213
+ },
214
+ {
215
+ "entropy": 0.9304505975916981,
216
+ "epoch": 0.076,
217
+ "grad_norm": 49.5,
218
+ "learning_rate": 3.7800000000000002e-06,
219
+ "loss": 15.1899,
220
+ "mean_token_accuracy": 0.756063225492835,
221
+ "num_tokens": 1585982.0,
222
+ "step": 190
223
+ },
224
+ {
225
+ "entropy": 0.9496602019295096,
226
+ "epoch": 0.08,
227
+ "grad_norm": 43.5,
228
+ "learning_rate": 3.980000000000001e-06,
229
+ "loss": 15.3895,
230
+ "mean_token_accuracy": 0.7551660794764757,
231
+ "num_tokens": 1667532.0,
232
+ "step": 200
233
+ },
234
+ {
235
+ "epoch": 0.08,
236
+ "eval_biology_entropy": 1.1760825806856154,
237
+ "eval_biology_loss": 1.2611405849456787,
238
+ "eval_biology_mean_token_accuracy": 0.6883746695518493,
239
+ "eval_biology_num_tokens": 1667532.0,
240
+ "eval_biology_runtime": 39.9198,
241
+ "eval_biology_samples_per_second": 12.525,
242
+ "eval_biology_steps_per_second": 12.525,
243
+ "step": 200
244
+ },
245
+ {
246
+ "epoch": 0.08,
247
+ "eval_physics_entropy": 0.9386761513948441,
248
+ "eval_physics_loss": 0.9588043093681335,
249
+ "eval_physics_mean_token_accuracy": 0.7587667318582535,
250
+ "eval_physics_num_tokens": 1667532.0,
251
+ "eval_physics_runtime": 53.7591,
252
+ "eval_physics_samples_per_second": 9.301,
253
+ "eval_physics_steps_per_second": 9.301,
254
+ "step": 200
255
+ },
256
+ {
257
+ "entropy": 0.8876599809154868,
258
+ "epoch": 0.084,
259
+ "grad_norm": 47.75,
260
+ "learning_rate": 4.18e-06,
261
+ "loss": 14.5591,
262
+ "mean_token_accuracy": 0.7688302487134934,
263
+ "num_tokens": 1750813.0,
264
+ "step": 210
265
+ },
266
+ {
267
+ "entropy": 0.9550920482724905,
268
+ "epoch": 0.088,
269
+ "grad_norm": 50.25,
270
+ "learning_rate": 4.38e-06,
271
+ "loss": 15.4401,
272
+ "mean_token_accuracy": 0.7583376962691546,
273
+ "num_tokens": 1835489.0,
274
+ "step": 220
275
+ },
276
+ {
277
+ "entropy": 0.9616718722507358,
278
+ "epoch": 0.092,
279
+ "grad_norm": 52.5,
280
+ "learning_rate": 4.58e-06,
281
+ "loss": 15.6379,
282
+ "mean_token_accuracy": 0.7517771814018488,
283
+ "num_tokens": 1915115.0,
284
+ "step": 230
285
+ },
286
+ {
287
+ "entropy": 0.8991832934319973,
288
+ "epoch": 0.096,
289
+ "grad_norm": 59.75,
290
+ "learning_rate": 4.78e-06,
291
+ "loss": 14.4737,
292
+ "mean_token_accuracy": 0.7692118056118489,
293
+ "num_tokens": 2007176.0,
294
+ "step": 240
295
+ },
296
+ {
297
+ "entropy": 0.9369921594858169,
298
+ "epoch": 0.1,
299
+ "grad_norm": 48.25,
300
+ "learning_rate": 4.980000000000001e-06,
301
+ "loss": 15.2105,
302
+ "mean_token_accuracy": 0.7562540877610445,
303
+ "num_tokens": 2088696.0,
304
+ "step": 250
305
+ },
306
+ {
307
+ "entropy": 0.9722234448418021,
308
+ "epoch": 0.104,
309
+ "grad_norm": 54.0,
310
+ "learning_rate": 5.18e-06,
311
+ "loss": 15.7667,
312
+ "mean_token_accuracy": 0.7513973277062178,
313
+ "num_tokens": 2171335.0,
314
+ "step": 260
315
+ },
316
+ {
317
+ "entropy": 0.929803448356688,
318
+ "epoch": 0.108,
319
+ "grad_norm": 59.25,
320
+ "learning_rate": 5.380000000000001e-06,
321
+ "loss": 14.9912,
322
+ "mean_token_accuracy": 0.7627206303179264,
323
+ "num_tokens": 2254168.0,
324
+ "step": 270
325
+ },
326
+ {
327
+ "entropy": 0.9004335077479482,
328
+ "epoch": 0.112,
329
+ "grad_norm": 61.25,
330
+ "learning_rate": 5.580000000000001e-06,
331
+ "loss": 14.405,
332
+ "mean_token_accuracy": 0.76763153411448,
333
+ "num_tokens": 2342259.0,
334
+ "step": 280
335
+ },
336
+ {
337
+ "entropy": 0.9265993487089872,
338
+ "epoch": 0.116,
339
+ "grad_norm": 58.25,
340
+ "learning_rate": 5.78e-06,
341
+ "loss": 14.9674,
342
+ "mean_token_accuracy": 0.7608603104948998,
343
+ "num_tokens": 2426289.0,
344
+ "step": 290
345
+ },
346
+ {
347
+ "entropy": 0.9675430860370398,
348
+ "epoch": 0.12,
349
+ "grad_norm": 44.5,
350
+ "learning_rate": 5.98e-06,
351
+ "loss": 15.4661,
352
+ "mean_token_accuracy": 0.7530908696353436,
353
+ "num_tokens": 2504279.0,
354
+ "step": 300
355
+ },
356
+ {
357
+ "epoch": 0.12,
358
+ "eval_biology_entropy": 1.1881412832140923,
359
+ "eval_biology_loss": 1.2643039226531982,
360
+ "eval_biology_mean_token_accuracy": 0.6870381585955619,
361
+ "eval_biology_num_tokens": 2504279.0,
362
+ "eval_biology_runtime": 40.1357,
363
+ "eval_biology_samples_per_second": 12.458,
364
+ "eval_biology_steps_per_second": 12.458,
365
+ "step": 300
366
+ },
367
+ {
368
+ "epoch": 0.12,
369
+ "eval_physics_entropy": 0.918557298719883,
370
+ "eval_physics_loss": 0.919302761554718,
371
+ "eval_physics_mean_token_accuracy": 0.7652055172920227,
372
+ "eval_physics_num_tokens": 2504279.0,
373
+ "eval_physics_runtime": 53.5764,
374
+ "eval_physics_samples_per_second": 9.332,
375
+ "eval_physics_steps_per_second": 9.332,
376
+ "step": 300
377
+ },
378
+ {
379
+ "entropy": 0.945288896933198,
380
+ "epoch": 0.124,
381
+ "grad_norm": 46.25,
382
+ "learning_rate": 6.18e-06,
383
+ "loss": 15.1972,
384
+ "mean_token_accuracy": 0.7569374658167363,
385
+ "num_tokens": 2583599.0,
386
+ "step": 310
387
+ },
388
+ {
389
+ "entropy": 0.8784681657329202,
390
+ "epoch": 0.128,
391
+ "grad_norm": 46.0,
392
+ "learning_rate": 6.380000000000001e-06,
393
+ "loss": 14.1084,
394
+ "mean_token_accuracy": 0.7712060511112213,
395
+ "num_tokens": 2663987.0,
396
+ "step": 320
397
+ },
398
+ {
399
+ "entropy": 0.9214498041197657,
400
+ "epoch": 0.132,
401
+ "grad_norm": 49.0,
402
+ "learning_rate": 6.5800000000000005e-06,
403
+ "loss": 15.0092,
404
+ "mean_token_accuracy": 0.7606228835880756,
405
+ "num_tokens": 2743933.0,
406
+ "step": 330
407
+ },
408
+ {
409
+ "entropy": 0.9181070065125823,
410
+ "epoch": 0.136,
411
+ "grad_norm": 36.75,
412
+ "learning_rate": 6.780000000000001e-06,
413
+ "loss": 14.8939,
414
+ "mean_token_accuracy": 0.7647624254226685,
415
+ "num_tokens": 2829720.0,
416
+ "step": 340
417
+ },
418
+ {
419
+ "entropy": 0.9362396266311407,
420
+ "epoch": 0.14,
421
+ "grad_norm": 42.75,
422
+ "learning_rate": 6.98e-06,
423
+ "loss": 14.8969,
424
+ "mean_token_accuracy": 0.7599288888275624,
425
+ "num_tokens": 2915814.0,
426
+ "step": 350
427
+ },
428
+ {
429
+ "entropy": 0.896806463971734,
430
+ "epoch": 0.144,
431
+ "grad_norm": 56.0,
432
+ "learning_rate": 7.180000000000001e-06,
433
+ "loss": 14.4447,
434
+ "mean_token_accuracy": 0.7707839358597994,
435
+ "num_tokens": 2997824.0,
436
+ "step": 360
437
+ },
438
+ {
439
+ "entropy": 0.9098270131275058,
440
+ "epoch": 0.148,
441
+ "grad_norm": 47.0,
442
+ "learning_rate": 7.3800000000000005e-06,
443
+ "loss": 14.7097,
444
+ "mean_token_accuracy": 0.763165109232068,
445
+ "num_tokens": 3076422.0,
446
+ "step": 370
447
+ },
448
+ {
449
+ "entropy": 0.8725583601742983,
450
+ "epoch": 0.152,
451
+ "grad_norm": 43.0,
452
+ "learning_rate": 7.58e-06,
453
+ "loss": 13.9897,
454
+ "mean_token_accuracy": 0.7742969345301389,
455
+ "num_tokens": 3159946.0,
456
+ "step": 380
457
+ },
458
+ {
459
+ "entropy": 0.8706006621941924,
460
+ "epoch": 0.156,
461
+ "grad_norm": 49.5,
462
+ "learning_rate": 7.78e-06,
463
+ "loss": 13.8559,
464
+ "mean_token_accuracy": 0.7760756332427263,
465
+ "num_tokens": 3247218.0,
466
+ "step": 390
467
+ },
468
+ {
469
+ "entropy": 0.8869594784453512,
470
+ "epoch": 0.16,
471
+ "grad_norm": 49.25,
472
+ "learning_rate": 7.980000000000002e-06,
473
+ "loss": 14.4662,
474
+ "mean_token_accuracy": 0.767155185341835,
475
+ "num_tokens": 3330597.0,
476
+ "step": 400
477
+ },
478
+ {
479
+ "epoch": 0.16,
480
+ "eval_biology_entropy": 1.197325525343418,
481
+ "eval_biology_loss": 1.2703520059585571,
482
+ "eval_biology_mean_token_accuracy": 0.6873815752267838,
483
+ "eval_biology_num_tokens": 3330597.0,
484
+ "eval_biology_runtime": 40.3991,
485
+ "eval_biology_samples_per_second": 12.377,
486
+ "eval_biology_steps_per_second": 12.377,
487
+ "step": 400
488
+ },
489
+ {
490
+ "epoch": 0.16,
491
+ "eval_physics_entropy": 0.888068499326706,
492
+ "eval_physics_loss": 0.8878387212753296,
493
+ "eval_physics_mean_token_accuracy": 0.7709487853050232,
494
+ "eval_physics_num_tokens": 3330597.0,
495
+ "eval_physics_runtime": 53.4439,
496
+ "eval_physics_samples_per_second": 9.356,
497
+ "eval_physics_steps_per_second": 9.356,
498
+ "step": 400
499
+ },
500
+ {
501
+ "entropy": 0.9179496448487043,
502
+ "epoch": 0.164,
503
+ "grad_norm": 49.0,
504
+ "learning_rate": 8.18e-06,
505
+ "loss": 14.5156,
506
+ "mean_token_accuracy": 0.7636631712317467,
507
+ "num_tokens": 3414445.0,
508
+ "step": 410
509
+ },
510
+ {
511
+ "entropy": 0.8822837853804231,
512
+ "epoch": 0.168,
513
+ "grad_norm": 53.0,
514
+ "learning_rate": 8.380000000000001e-06,
515
+ "loss": 14.148,
516
+ "mean_token_accuracy": 0.7693528279662132,
517
+ "num_tokens": 3496104.0,
518
+ "step": 420
519
+ },
520
+ {
521
+ "entropy": 0.8821934029459954,
522
+ "epoch": 0.172,
523
+ "grad_norm": 45.25,
524
+ "learning_rate": 8.580000000000001e-06,
525
+ "loss": 14.4071,
526
+ "mean_token_accuracy": 0.7662574470043182,
527
+ "num_tokens": 3576979.0,
528
+ "step": 430
529
+ },
530
+ {
531
+ "entropy": 0.8882661890238523,
532
+ "epoch": 0.176,
533
+ "grad_norm": 57.0,
534
+ "learning_rate": 8.78e-06,
535
+ "loss": 14.0855,
536
+ "mean_token_accuracy": 0.7684148631989955,
537
+ "num_tokens": 3658264.0,
538
+ "step": 440
539
+ },
540
+ {
541
+ "entropy": 0.8776636118069291,
542
+ "epoch": 0.18,
543
+ "grad_norm": 50.5,
544
+ "learning_rate": 8.98e-06,
545
+ "loss": 14.1633,
546
+ "mean_token_accuracy": 0.7680283546447754,
547
+ "num_tokens": 3739142.0,
548
+ "step": 450
549
+ },
550
+ {
551
+ "entropy": 0.8854248264804483,
552
+ "epoch": 0.184,
553
+ "grad_norm": 50.25,
554
+ "learning_rate": 9.180000000000002e-06,
555
+ "loss": 14.2742,
556
+ "mean_token_accuracy": 0.7675002634525299,
557
+ "num_tokens": 3821352.0,
558
+ "step": 460
559
+ },
560
+ {
561
+ "entropy": 0.8554168688133359,
562
+ "epoch": 0.188,
563
+ "grad_norm": 43.25,
564
+ "learning_rate": 9.38e-06,
565
+ "loss": 13.6807,
566
+ "mean_token_accuracy": 0.7745012179017067,
567
+ "num_tokens": 3905291.0,
568
+ "step": 470
569
+ },
570
+ {
571
+ "entropy": 0.8508397350087762,
572
+ "epoch": 0.192,
573
+ "grad_norm": 40.25,
574
+ "learning_rate": 9.58e-06,
575
+ "loss": 13.7828,
576
+ "mean_token_accuracy": 0.7750742956995964,
577
+ "num_tokens": 3995568.0,
578
+ "step": 480
579
+ },
580
+ {
581
+ "entropy": 0.8710859406739473,
582
+ "epoch": 0.196,
583
+ "grad_norm": 49.5,
584
+ "learning_rate": 9.780000000000001e-06,
585
+ "loss": 13.9918,
586
+ "mean_token_accuracy": 0.7735175251960754,
587
+ "num_tokens": 4075670.0,
588
+ "step": 490
589
+ },
590
+ {
591
+ "entropy": 0.8420973749831319,
592
+ "epoch": 0.2,
593
+ "grad_norm": 50.0,
594
+ "learning_rate": 9.980000000000001e-06,
595
+ "loss": 13.4998,
596
+ "mean_token_accuracy": 0.779565305262804,
597
+ "num_tokens": 4160918.0,
598
+ "step": 500
599
+ },
600
+ {
601
+ "epoch": 0.2,
602
+ "eval_biology_entropy": 1.1644099119901656,
603
+ "eval_biology_loss": 1.2769672870635986,
604
+ "eval_biology_mean_token_accuracy": 0.6861101101040841,
605
+ "eval_biology_num_tokens": 4160918.0,
606
+ "eval_biology_runtime": 39.8379,
607
+ "eval_biology_samples_per_second": 12.551,
608
+ "eval_biology_steps_per_second": 12.551,
609
+ "step": 500
610
+ },
611
+ {
612
+ "epoch": 0.2,
613
+ "eval_physics_entropy": 0.8531826239228248,
614
+ "eval_physics_loss": 0.8632206916809082,
615
+ "eval_physics_mean_token_accuracy": 0.7755071822404861,
616
+ "eval_physics_num_tokens": 4160918.0,
617
+ "eval_physics_runtime": 55.811,
618
+ "eval_physics_samples_per_second": 8.959,
619
+ "eval_physics_steps_per_second": 8.959,
620
+ "step": 500
621
+ },
622
+ {
623
+ "entropy": 0.8474665513262153,
624
+ "epoch": 0.204,
625
+ "grad_norm": 45.75,
626
+ "learning_rate": 1.018e-05,
627
+ "loss": 13.6166,
628
+ "mean_token_accuracy": 0.7770899929106235,
629
+ "num_tokens": 4238810.0,
630
+ "step": 510
631
+ },
632
+ {
633
+ "entropy": 0.8348175186663866,
634
+ "epoch": 0.208,
635
+ "grad_norm": 45.0,
636
+ "learning_rate": 1.038e-05,
637
+ "loss": 13.4432,
638
+ "mean_token_accuracy": 0.7791284300386906,
639
+ "num_tokens": 4321436.0,
640
+ "step": 520
641
+ },
642
+ {
643
+ "entropy": 0.8515958024188877,
644
+ "epoch": 0.212,
645
+ "grad_norm": 44.0,
646
+ "learning_rate": 1.0580000000000002e-05,
647
+ "loss": 13.5514,
648
+ "mean_token_accuracy": 0.7726950984448194,
649
+ "num_tokens": 4401710.0,
650
+ "step": 530
651
+ },
652
+ {
653
+ "entropy": 0.871463224478066,
654
+ "epoch": 0.216,
655
+ "grad_norm": 43.75,
656
+ "learning_rate": 1.0780000000000002e-05,
657
+ "loss": 14.1078,
658
+ "mean_token_accuracy": 0.7690339071676136,
659
+ "num_tokens": 4482225.0,
660
+ "step": 540
661
+ },
662
+ {
663
+ "entropy": 0.8288272012025117,
664
+ "epoch": 0.22,
665
+ "grad_norm": 46.5,
666
+ "learning_rate": 1.0980000000000002e-05,
667
+ "loss": 13.3444,
668
+ "mean_token_accuracy": 0.7793109815567731,
669
+ "num_tokens": 4564662.0,
670
+ "step": 550
671
+ },
672
+ {
673
+ "entropy": 0.8090431936085224,
674
+ "epoch": 0.224,
675
+ "grad_norm": 40.75,
676
+ "learning_rate": 1.1180000000000001e-05,
677
+ "loss": 13.0281,
678
+ "mean_token_accuracy": 0.7843161709606647,
679
+ "num_tokens": 4648491.0,
680
+ "step": 560
681
+ },
682
+ {
683
+ "entropy": 0.8062232775613666,
684
+ "epoch": 0.228,
685
+ "grad_norm": 38.5,
686
+ "learning_rate": 1.138e-05,
687
+ "loss": 12.9529,
688
+ "mean_token_accuracy": 0.7843800630420447,
689
+ "num_tokens": 4731596.0,
690
+ "step": 570
691
+ },
692
+ {
693
+ "entropy": 0.8517701115459204,
694
+ "epoch": 0.232,
695
+ "grad_norm": 42.25,
696
+ "learning_rate": 1.1580000000000001e-05,
697
+ "loss": 13.6746,
698
+ "mean_token_accuracy": 0.7757678255438805,
699
+ "num_tokens": 4813449.0,
700
+ "step": 580
701
+ },
702
+ {
703
+ "entropy": 0.8369667442515493,
704
+ "epoch": 0.236,
705
+ "grad_norm": 35.5,
706
+ "learning_rate": 1.178e-05,
707
+ "loss": 13.2971,
708
+ "mean_token_accuracy": 0.7806435581296682,
709
+ "num_tokens": 4899144.0,
710
+ "step": 590
711
+ },
712
+ {
713
+ "entropy": 0.8229177813977003,
714
+ "epoch": 0.24,
715
+ "grad_norm": 42.0,
716
+ "learning_rate": 1.198e-05,
717
+ "loss": 13.364,
718
+ "mean_token_accuracy": 0.778663220256567,
719
+ "num_tokens": 4986175.0,
720
+ "step": 600
721
+ },
722
+ {
723
+ "epoch": 0.24,
724
+ "eval_biology_entropy": 1.180227564573288,
725
+ "eval_biology_loss": 1.280086636543274,
726
+ "eval_biology_mean_token_accuracy": 0.6851186206936837,
727
+ "eval_biology_num_tokens": 4986175.0,
728
+ "eval_biology_runtime": 40.1078,
729
+ "eval_biology_samples_per_second": 12.466,
730
+ "eval_biology_steps_per_second": 12.466,
731
+ "step": 600
732
+ },
733
+ {
734
+ "epoch": 0.24,
735
+ "eval_physics_entropy": 0.8273749435544014,
736
+ "eval_physics_loss": 0.8421014547348022,
737
+ "eval_physics_mean_token_accuracy": 0.7796914368867874,
738
+ "eval_physics_num_tokens": 4986175.0,
739
+ "eval_physics_runtime": 53.6333,
740
+ "eval_physics_samples_per_second": 9.323,
741
+ "eval_physics_steps_per_second": 9.323,
742
+ "step": 600
743
+ },
744
+ {
745
+ "entropy": 0.8442536953836679,
746
+ "epoch": 0.244,
747
+ "grad_norm": 42.0,
748
+ "learning_rate": 1.218e-05,
749
+ "loss": 13.4661,
750
+ "mean_token_accuracy": 0.7772998537868261,
751
+ "num_tokens": 5071994.0,
752
+ "step": 610
753
+ },
754
+ {
755
+ "entropy": 0.8010923132300377,
756
+ "epoch": 0.248,
757
+ "grad_norm": 41.75,
758
+ "learning_rate": 1.2380000000000002e-05,
759
+ "loss": 12.8979,
760
+ "mean_token_accuracy": 0.7863831762224436,
761
+ "num_tokens": 5164741.0,
762
+ "step": 620
763
+ },
764
+ {
765
+ "entropy": 0.8448842618614435,
766
+ "epoch": 0.252,
767
+ "grad_norm": 44.75,
768
+ "learning_rate": 1.2580000000000002e-05,
769
+ "loss": 13.612,
770
+ "mean_token_accuracy": 0.7776872050017118,
771
+ "num_tokens": 5245878.0,
772
+ "step": 630
773
+ },
774
+ {
775
+ "entropy": 0.8191616820171476,
776
+ "epoch": 0.256,
777
+ "grad_norm": 46.25,
778
+ "learning_rate": 1.2780000000000001e-05,
779
+ "loss": 13.3478,
780
+ "mean_token_accuracy": 0.7798578917980195,
781
+ "num_tokens": 5329320.0,
782
+ "step": 640
783
+ },
784
+ {
785
+ "entropy": 0.8373894428834319,
786
+ "epoch": 0.26,
787
+ "grad_norm": 43.75,
788
+ "learning_rate": 1.2980000000000001e-05,
789
+ "loss": 13.2408,
790
+ "mean_token_accuracy": 0.780717215128243,
791
+ "num_tokens": 5412557.0,
792
+ "step": 650
793
+ },
794
+ {
795
+ "entropy": 0.8047826003283263,
796
+ "epoch": 0.264,
797
+ "grad_norm": 47.0,
798
+ "learning_rate": 1.3180000000000001e-05,
799
+ "loss": 13.1602,
800
+ "mean_token_accuracy": 0.7808638617396355,
801
+ "num_tokens": 5497259.0,
802
+ "step": 660
803
+ },
804
+ {
805
+ "entropy": 0.8437582038342952,
806
+ "epoch": 0.268,
807
+ "grad_norm": 44.75,
808
+ "learning_rate": 1.3380000000000002e-05,
809
+ "loss": 13.3738,
810
+ "mean_token_accuracy": 0.7791281435638666,
811
+ "num_tokens": 5576721.0,
812
+ "step": 670
813
+ },
814
+ {
815
+ "entropy": 0.8402760226279498,
816
+ "epoch": 0.272,
817
+ "grad_norm": 43.25,
818
+ "learning_rate": 1.3580000000000002e-05,
819
+ "loss": 13.6995,
820
+ "mean_token_accuracy": 0.7730118347331881,
821
+ "num_tokens": 5658796.0,
822
+ "step": 680
823
+ },
824
+ {
825
+ "entropy": 0.8618346724659205,
826
+ "epoch": 0.276,
827
+ "grad_norm": 47.25,
828
+ "learning_rate": 1.378e-05,
829
+ "loss": 13.7742,
830
+ "mean_token_accuracy": 0.7733935236930847,
831
+ "num_tokens": 5737499.0,
832
+ "step": 690
833
+ },
834
+ {
835
+ "entropy": 0.8180074915289879,
836
+ "epoch": 0.28,
837
+ "grad_norm": 45.25,
838
+ "learning_rate": 1.398e-05,
839
+ "loss": 13.2461,
840
+ "mean_token_accuracy": 0.7852540168911218,
841
+ "num_tokens": 5819027.0,
842
+ "step": 700
843
+ },
844
+ {
845
+ "epoch": 0.28,
846
+ "eval_biology_entropy": 1.2050418145656585,
847
+ "eval_biology_loss": 1.2829557657241821,
848
+ "eval_biology_mean_token_accuracy": 0.6858616912961006,
849
+ "eval_biology_num_tokens": 5819027.0,
850
+ "eval_biology_runtime": 40.1656,
851
+ "eval_biology_samples_per_second": 12.448,
852
+ "eval_biology_steps_per_second": 12.448,
853
+ "step": 700
854
+ },
855
+ {
856
+ "epoch": 0.28,
857
+ "eval_physics_entropy": 0.8281588327288628,
858
+ "eval_physics_loss": 0.8266663551330566,
859
+ "eval_physics_mean_token_accuracy": 0.7826227464675903,
860
+ "eval_physics_num_tokens": 5819027.0,
861
+ "eval_physics_runtime": 53.654,
862
+ "eval_physics_samples_per_second": 9.319,
863
+ "eval_physics_steps_per_second": 9.319,
864
+ "step": 700
865
+ },
866
+ {
867
+ "entropy": 0.8689560383558274,
868
+ "epoch": 0.284,
869
+ "grad_norm": 49.0,
870
+ "learning_rate": 1.418e-05,
871
+ "loss": 13.8644,
872
+ "mean_token_accuracy": 0.7714414589107037,
873
+ "num_tokens": 5898693.0,
874
+ "step": 710
875
+ },
876
+ {
877
+ "entropy": 0.804288436844945,
878
+ "epoch": 0.288,
879
+ "grad_norm": 41.0,
880
+ "learning_rate": 1.4380000000000001e-05,
881
+ "loss": 13.0324,
882
+ "mean_token_accuracy": 0.7863233294337988,
883
+ "num_tokens": 5980201.0,
884
+ "step": 720
885
+ },
886
+ {
887
+ "entropy": 0.8008439680561423,
888
+ "epoch": 0.292,
889
+ "grad_norm": 39.75,
890
+ "learning_rate": 1.4580000000000001e-05,
891
+ "loss": 12.8478,
892
+ "mean_token_accuracy": 0.7870619356632232,
893
+ "num_tokens": 6059507.0,
894
+ "step": 730
895
+ },
896
+ {
897
+ "entropy": 0.8202026936225593,
898
+ "epoch": 0.296,
899
+ "grad_norm": 38.75,
900
+ "learning_rate": 1.478e-05,
901
+ "loss": 13.2546,
902
+ "mean_token_accuracy": 0.7811095919460058,
903
+ "num_tokens": 6139830.0,
904
+ "step": 740
905
+ },
906
+ {
907
+ "entropy": 0.8201492626219988,
908
+ "epoch": 0.3,
909
+ "grad_norm": 41.25,
910
+ "learning_rate": 1.498e-05,
911
+ "loss": 13.0217,
912
+ "mean_token_accuracy": 0.7822576854377985,
913
+ "num_tokens": 6218134.0,
914
+ "step": 750
915
+ },
916
+ {
917
+ "entropy": 0.818927645124495,
918
+ "epoch": 0.304,
919
+ "grad_norm": 42.75,
920
+ "learning_rate": 1.5180000000000002e-05,
921
+ "loss": 13.2804,
922
+ "mean_token_accuracy": 0.779948465898633,
923
+ "num_tokens": 6298858.0,
924
+ "step": 760
925
+ },
926
+ {
927
+ "entropy": 0.855727698840201,
928
+ "epoch": 0.308,
929
+ "grad_norm": 52.75,
930
+ "learning_rate": 1.5380000000000002e-05,
931
+ "loss": 13.8917,
932
+ "mean_token_accuracy": 0.7703924782574176,
933
+ "num_tokens": 6375761.0,
934
+ "step": 770
935
+ },
936
+ {
937
+ "entropy": 0.7976011021062732,
938
+ "epoch": 0.312,
939
+ "grad_norm": 36.0,
940
+ "learning_rate": 1.5580000000000003e-05,
941
+ "loss": 12.6345,
942
+ "mean_token_accuracy": 0.7890406355261803,
943
+ "num_tokens": 6456502.0,
944
+ "step": 780
945
+ },
946
+ {
947
+ "entropy": 0.8050644747912884,
948
+ "epoch": 0.316,
949
+ "grad_norm": 44.5,
950
+ "learning_rate": 1.578e-05,
951
+ "loss": 13.1921,
952
+ "mean_token_accuracy": 0.7834381762892008,
953
+ "num_tokens": 6537521.0,
954
+ "step": 790
955
+ },
956
+ {
957
+ "entropy": 0.8325098715722561,
958
+ "epoch": 0.32,
959
+ "grad_norm": 36.25,
960
+ "learning_rate": 1.5980000000000003e-05,
961
+ "loss": 13.3843,
962
+ "mean_token_accuracy": 0.7803130840882659,
963
+ "num_tokens": 6622798.0,
964
+ "step": 800
965
+ },
966
+ {
967
+ "epoch": 0.32,
968
+ "eval_biology_entropy": 1.2224014229774476,
969
+ "eval_biology_loss": 1.2937734127044678,
970
+ "eval_biology_mean_token_accuracy": 0.6835832511186599,
971
+ "eval_biology_num_tokens": 6622798.0,
972
+ "eval_biology_runtime": 40.2146,
973
+ "eval_biology_samples_per_second": 12.433,
974
+ "eval_biology_steps_per_second": 12.433,
975
+ "step": 800
976
+ },
977
+ {
978
+ "epoch": 0.32,
979
+ "eval_physics_entropy": 0.8050925194621086,
980
+ "eval_physics_loss": 0.8124625086784363,
981
+ "eval_physics_mean_token_accuracy": 0.7845232374668122,
982
+ "eval_physics_num_tokens": 6622798.0,
983
+ "eval_physics_runtime": 53.5823,
984
+ "eval_physics_samples_per_second": 9.331,
985
+ "eval_physics_steps_per_second": 9.331,
986
+ "step": 800
987
+ },
988
+ {
989
+ "entropy": 0.8160014390945435,
990
+ "epoch": 0.324,
991
+ "grad_norm": 44.5,
992
+ "learning_rate": 1.618e-05,
993
+ "loss": 13.1114,
994
+ "mean_token_accuracy": 0.7845251262187958,
995
+ "num_tokens": 6699373.0,
996
+ "step": 810
997
+ },
998
+ {
999
+ "entropy": 0.8096626095473767,
1000
+ "epoch": 0.328,
1001
+ "grad_norm": 43.25,
1002
+ "learning_rate": 1.638e-05,
1003
+ "loss": 13.2707,
1004
+ "mean_token_accuracy": 0.7790007971227169,
1005
+ "num_tokens": 6782985.0,
1006
+ "step": 820
1007
+ },
1008
+ {
1009
+ "entropy": 0.8155604779720307,
1010
+ "epoch": 0.332,
1011
+ "grad_norm": 42.75,
1012
+ "learning_rate": 1.658e-05,
1013
+ "loss": 12.8927,
1014
+ "mean_token_accuracy": 0.7841650832444429,
1015
+ "num_tokens": 6868578.0,
1016
+ "step": 830
1017
+ },
1018
+ {
1019
+ "entropy": 0.7713794196024537,
1020
+ "epoch": 0.336,
1021
+ "grad_norm": 35.75,
1022
+ "learning_rate": 1.6780000000000002e-05,
1023
+ "loss": 12.3469,
1024
+ "mean_token_accuracy": 0.7945763759315014,
1025
+ "num_tokens": 6953611.0,
1026
+ "step": 840
1027
+ },
1028
+ {
1029
+ "entropy": 0.805246700718999,
1030
+ "epoch": 0.34,
1031
+ "grad_norm": 45.25,
1032
+ "learning_rate": 1.698e-05,
1033
+ "loss": 13.264,
1034
+ "mean_token_accuracy": 0.7829924900084734,
1035
+ "num_tokens": 7037831.0,
1036
+ "step": 850
1037
+ },
1038
+ {
1039
+ "entropy": 0.7846146428957581,
1040
+ "epoch": 0.344,
1041
+ "grad_norm": 50.25,
1042
+ "learning_rate": 1.718e-05,
1043
+ "loss": 12.5468,
1044
+ "mean_token_accuracy": 0.7915518656373024,
1045
+ "num_tokens": 7121949.0,
1046
+ "step": 860
1047
+ },
1048
+ {
1049
+ "entropy": 0.7968592442572117,
1050
+ "epoch": 0.348,
1051
+ "grad_norm": 40.75,
1052
+ "learning_rate": 1.7380000000000003e-05,
1053
+ "loss": 12.8986,
1054
+ "mean_token_accuracy": 0.7861088421195745,
1055
+ "num_tokens": 7206056.0,
1056
+ "step": 870
1057
+ },
1058
+ {
1059
+ "entropy": 0.8040603714063763,
1060
+ "epoch": 0.352,
1061
+ "grad_norm": 49.5,
1062
+ "learning_rate": 1.758e-05,
1063
+ "loss": 12.9695,
1064
+ "mean_token_accuracy": 0.7847281444817782,
1065
+ "num_tokens": 7290378.0,
1066
+ "step": 880
1067
+ },
1068
+ {
1069
+ "entropy": 0.8249117810279131,
1070
+ "epoch": 0.356,
1071
+ "grad_norm": 35.25,
1072
+ "learning_rate": 1.7780000000000003e-05,
1073
+ "loss": 13.0379,
1074
+ "mean_token_accuracy": 0.7836297146975995,
1075
+ "num_tokens": 7367952.0,
1076
+ "step": 890
1077
+ },
1078
+ {
1079
+ "entropy": 0.7825405385345221,
1080
+ "epoch": 0.36,
1081
+ "grad_norm": 50.75,
1082
+ "learning_rate": 1.798e-05,
1083
+ "loss": 12.7677,
1084
+ "mean_token_accuracy": 0.7906032208353281,
1085
+ "num_tokens": 7452920.0,
1086
+ "step": 900
1087
+ },
1088
+ {
1089
+ "epoch": 0.36,
1090
+ "eval_biology_entropy": 1.260676340520382,
1091
+ "eval_biology_loss": 1.2954092025756836,
1092
+ "eval_biology_mean_token_accuracy": 0.6837160994410515,
1093
+ "eval_biology_num_tokens": 7452920.0,
1094
+ "eval_biology_runtime": 39.7587,
1095
+ "eval_biology_samples_per_second": 12.576,
1096
+ "eval_biology_steps_per_second": 12.576,
1097
+ "step": 900
1098
+ },
1099
+ {
1100
+ "epoch": 0.36,
1101
+ "eval_physics_entropy": 0.8267089626193047,
1102
+ "eval_physics_loss": 0.804395854473114,
1103
+ "eval_physics_mean_token_accuracy": 0.7860386987924576,
1104
+ "eval_physics_num_tokens": 7452920.0,
1105
+ "eval_physics_runtime": 53.385,
1106
+ "eval_physics_samples_per_second": 9.366,
1107
+ "eval_physics_steps_per_second": 9.366,
1108
+ "step": 900
1109
+ },
1110
+ {
1111
+ "entropy": 0.7688560528680682,
1112
+ "epoch": 0.364,
1113
+ "grad_norm": 38.0,
1114
+ "learning_rate": 1.8180000000000002e-05,
1115
+ "loss": 12.4576,
1116
+ "mean_token_accuracy": 0.7921908859163522,
1117
+ "num_tokens": 7535961.0,
1118
+ "step": 910
1119
+ },
1120
+ {
1121
+ "entropy": 0.787941231764853,
1122
+ "epoch": 0.368,
1123
+ "grad_norm": 38.0,
1124
+ "learning_rate": 1.8380000000000004e-05,
1125
+ "loss": 12.5149,
1126
+ "mean_token_accuracy": 0.7924131523817778,
1127
+ "num_tokens": 7621082.0,
1128
+ "step": 920
1129
+ },
1130
+ {
1131
+ "entropy": 0.7637785948812962,
1132
+ "epoch": 0.372,
1133
+ "grad_norm": 39.5,
1134
+ "learning_rate": 1.858e-05,
1135
+ "loss": 12.4069,
1136
+ "mean_token_accuracy": 0.7892315208911895,
1137
+ "num_tokens": 7705651.0,
1138
+ "step": 930
1139
+ },
1140
+ {
1141
+ "entropy": 0.8655371183529497,
1142
+ "epoch": 0.376,
1143
+ "grad_norm": 37.5,
1144
+ "learning_rate": 1.878e-05,
1145
+ "loss": 13.9821,
1146
+ "mean_token_accuracy": 0.767190694063902,
1147
+ "num_tokens": 7787036.0,
1148
+ "step": 940
1149
+ },
1150
+ {
1151
+ "entropy": 0.8061225781217217,
1152
+ "epoch": 0.38,
1153
+ "grad_norm": 43.25,
1154
+ "learning_rate": 1.898e-05,
1155
+ "loss": 12.8783,
1156
+ "mean_token_accuracy": 0.7815280705690384,
1157
+ "num_tokens": 7869437.0,
1158
+ "step": 950
1159
+ },
1160
+ {
1161
+ "entropy": 0.7763598024845123,
1162
+ "epoch": 0.384,
1163
+ "grad_norm": 36.0,
1164
+ "learning_rate": 1.918e-05,
1165
+ "loss": 12.5365,
1166
+ "mean_token_accuracy": 0.7898113425821066,
1167
+ "num_tokens": 7955570.0,
1168
+ "step": 960
1169
+ },
1170
+ {
1171
+ "entropy": 0.7618165619671344,
1172
+ "epoch": 0.388,
1173
+ "grad_norm": 34.25,
1174
+ "learning_rate": 1.938e-05,
1175
+ "loss": 12.2278,
1176
+ "mean_token_accuracy": 0.7936428274959326,
1177
+ "num_tokens": 8038818.0,
1178
+ "step": 970
1179
+ },
1180
+ {
1181
+ "entropy": 0.7720302825793623,
1182
+ "epoch": 0.392,
1183
+ "grad_norm": 33.5,
1184
+ "learning_rate": 1.9580000000000002e-05,
1185
+ "loss": 12.4276,
1186
+ "mean_token_accuracy": 0.7894316878169775,
1187
+ "num_tokens": 8123527.0,
1188
+ "step": 980
1189
+ },
1190
+ {
1191
+ "entropy": 0.7889690216630697,
1192
+ "epoch": 0.396,
1193
+ "grad_norm": 35.75,
1194
+ "learning_rate": 1.978e-05,
1195
+ "loss": 12.6712,
1196
+ "mean_token_accuracy": 0.7906296912580728,
1197
+ "num_tokens": 8206501.0,
1198
+ "step": 990
1199
+ },
1200
+ {
1201
+ "entropy": 0.7546495482325554,
1202
+ "epoch": 0.4,
1203
+ "grad_norm": 30.375,
1204
+ "learning_rate": 1.9980000000000002e-05,
1205
+ "loss": 12.2605,
1206
+ "mean_token_accuracy": 0.7954238150268793,
1207
+ "num_tokens": 8291049.0,
1208
+ "step": 1000
1209
+ },
1210
+ {
1211
+ "epoch": 0.4,
1212
+ "eval_biology_entropy": 1.2203906371593476,
1213
+ "eval_biology_loss": 1.3063849210739136,
1214
+ "eval_biology_mean_token_accuracy": 0.6822154860496521,
1215
+ "eval_biology_num_tokens": 8291049.0,
1216
+ "eval_biology_runtime": 39.6187,
1217
+ "eval_biology_samples_per_second": 12.62,
1218
+ "eval_biology_steps_per_second": 12.62,
1219
+ "step": 1000
1220
+ },
1221
+ {
1222
+ "epoch": 0.4,
1223
+ "eval_physics_entropy": 0.7869865165352822,
1224
+ "eval_physics_loss": 0.7939902544021606,
1225
+ "eval_physics_mean_token_accuracy": 0.7892250344753265,
1226
+ "eval_physics_num_tokens": 8291049.0,
1227
+ "eval_physics_runtime": 53.3297,
1228
+ "eval_physics_samples_per_second": 9.376,
1229
+ "eval_physics_steps_per_second": 9.376,
1230
+ "step": 1000
1231
+ }
1232
+ ],
1233
+ "logging_steps": 10,
1234
+ "max_steps": 10000,
1235
+ "num_input_tokens_seen": 0,
1236
+ "num_train_epochs": 4,
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": 5.559510372210398e+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:b85b206341a2c78fa165a1c6c756efacf9fc2b262c7e937aec9e94d86112ba55
3
+ size 6481