Upload folder using huggingface_hub
Browse files- .gitattributes +1 -0
- checkpoint-2000/added_tokens.json +3 -0
- checkpoint-2000/chat_template.jinja +47 -0
- checkpoint-2000/config.json +110 -0
- checkpoint-2000/generation_config.json +13 -0
- checkpoint-2000/model-00001-of-00005.safetensors +3 -0
- checkpoint-2000/model-00002-of-00005.safetensors +3 -0
- checkpoint-2000/model-00003-of-00005.safetensors +3 -0
- checkpoint-2000/model-00004-of-00005.safetensors +3 -0
- checkpoint-2000/model-00005-of-00005.safetensors +3 -0
- checkpoint-2000/model.safetensors.index.json +0 -0
- checkpoint-2000/optimizer.pt +3 -0
- checkpoint-2000/rng_state.pth +3 -0
- checkpoint-2000/scheduler.pt +3 -0
- checkpoint-2000/special_tokens_map.json +33 -0
- checkpoint-2000/tokenizer.json +3 -0
- checkpoint-2000/tokenizer.model +3 -0
- checkpoint-2000/tokenizer_config.json +0 -0
- checkpoint-2000/trainer_state.json +2914 -0
- checkpoint-2000/training_args.bin +3 -0
.gitattributes
CHANGED
|
@@ -36,3 +36,4 @@ saved_model/**/* 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
|
| 38 |
checkpoint-1500/tokenizer.json 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
|
| 38 |
checkpoint-1500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 39 |
+
checkpoint-2000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
checkpoint-2000/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<image_soft_token>": 262144
|
| 3 |
+
}
|
checkpoint-2000/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-2000/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-2000/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-2000/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-2000/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-2000/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-2000/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-2000/model-00005-of-00005.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fc04c3ab0368e12ebbddd25f565e2fb58025086225569287c654a5af81dad97d
|
| 3 |
+
size 4601000928
|
checkpoint-2000/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-2000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7e36c5e960e1f515fcdcf712cc156c5043c8b82328de61c791588e976903433c
|
| 3 |
+
size 896607944
|
checkpoint-2000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4132949318e028debc59f4db164bc9a537cbb130ee35d892cb08945d6ab6790a
|
| 3 |
+
size 14645
|
checkpoint-2000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7c56cc7d2ab3dfe9edf7a28cb038a85004572c59f841d9bdfd8a137d5d874b19
|
| 3 |
+
size 1465
|
checkpoint-2000/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-2000/tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4667f2089529e8e7657cfb6d1c19910ae71ff5f28aa7ab2ff2763330affad795
|
| 3 |
+
size 33384568
|
checkpoint-2000/tokenizer.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1299c11d7cf632ef3b4e11937501358ada021bbdf7c47638d13c0ee982f2e79c
|
| 3 |
+
size 4689074
|
checkpoint-2000/tokenizer_config.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-2000/trainer_state.json
ADDED
|
@@ -0,0 +1,2914 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 3.2,
|
| 6 |
+
"eval_steps": 100,
|
| 7 |
+
"global_step": 2000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"entropy": 0.9036113740876317,
|
| 14 |
+
"epoch": 0.016,
|
| 15 |
+
"grad_norm": 128.0,
|
| 16 |
+
"learning_rate": 6.000000000000001e-07,
|
| 17 |
+
"loss": 34.907,
|
| 18 |
+
"mean_token_accuracy": 0.6652832090854645,
|
| 19 |
+
"num_tokens": 338352.0,
|
| 20 |
+
"step": 10
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"entropy": 0.9087308555841446,
|
| 24 |
+
"epoch": 0.032,
|
| 25 |
+
"grad_norm": 119.5,
|
| 26 |
+
"learning_rate": 1.2666666666666669e-06,
|
| 27 |
+
"loss": 34.745,
|
| 28 |
+
"mean_token_accuracy": 0.6691140927374363,
|
| 29 |
+
"num_tokens": 671193.0,
|
| 30 |
+
"step": 20
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"entropy": 0.9059555266052485,
|
| 34 |
+
"epoch": 0.048,
|
| 35 |
+
"grad_norm": 109.5,
|
| 36 |
+
"learning_rate": 1.9333333333333336e-06,
|
| 37 |
+
"loss": 34.1928,
|
| 38 |
+
"mean_token_accuracy": 0.6685802672058344,
|
| 39 |
+
"num_tokens": 1001008.0,
|
| 40 |
+
"step": 30
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"entropy": 0.9450886320322752,
|
| 44 |
+
"epoch": 0.064,
|
| 45 |
+
"grad_norm": 102.0,
|
| 46 |
+
"learning_rate": 2.6e-06,
|
| 47 |
+
"loss": 34.8515,
|
| 48 |
+
"mean_token_accuracy": 0.6614982625469565,
|
| 49 |
+
"num_tokens": 1331208.0,
|
| 50 |
+
"step": 40
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"entropy": 0.9593638770282269,
|
| 54 |
+
"epoch": 0.08,
|
| 55 |
+
"grad_norm": 94.5,
|
| 56 |
+
"learning_rate": 3.266666666666667e-06,
|
| 57 |
+
"loss": 33.1046,
|
| 58 |
+
"mean_token_accuracy": 0.6680185409262777,
|
| 59 |
+
"num_tokens": 1667532.0,
|
| 60 |
+
"step": 50
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"entropy": 0.9606023650616408,
|
| 64 |
+
"epoch": 0.096,
|
| 65 |
+
"grad_norm": 80.0,
|
| 66 |
+
"learning_rate": 3.9333333333333335e-06,
|
| 67 |
+
"loss": 31.2876,
|
| 68 |
+
"mean_token_accuracy": 0.678196857497096,
|
| 69 |
+
"num_tokens": 2007176.0,
|
| 70 |
+
"step": 60
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"entropy": 1.069316839799285,
|
| 74 |
+
"epoch": 0.112,
|
| 75 |
+
"grad_norm": 63.25,
|
| 76 |
+
"learning_rate": 4.600000000000001e-06,
|
| 77 |
+
"loss": 30.7512,
|
| 78 |
+
"mean_token_accuracy": 0.6743977405130863,
|
| 79 |
+
"num_tokens": 2342259.0,
|
| 80 |
+
"step": 70
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"entropy": 1.2628379821777345,
|
| 84 |
+
"epoch": 0.128,
|
| 85 |
+
"grad_norm": 48.0,
|
| 86 |
+
"learning_rate": 5.2666666666666665e-06,
|
| 87 |
+
"loss": 30.3935,
|
| 88 |
+
"mean_token_accuracy": 0.6708028431981802,
|
| 89 |
+
"num_tokens": 2663987.0,
|
| 90 |
+
"step": 80
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"entropy": 1.3055987607687711,
|
| 94 |
+
"epoch": 0.144,
|
| 95 |
+
"grad_norm": 32.25,
|
| 96 |
+
"learning_rate": 5.933333333333335e-06,
|
| 97 |
+
"loss": 27.3167,
|
| 98 |
+
"mean_token_accuracy": 0.6920903529971838,
|
| 99 |
+
"num_tokens": 2997824.0,
|
| 100 |
+
"step": 90
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"entropy": 1.3579398691654205,
|
| 104 |
+
"epoch": 0.16,
|
| 105 |
+
"grad_norm": 24.625,
|
| 106 |
+
"learning_rate": 6.600000000000001e-06,
|
| 107 |
+
"loss": 25.5362,
|
| 108 |
+
"mean_token_accuracy": 0.7021430663764476,
|
| 109 |
+
"num_tokens": 3330597.0,
|
| 110 |
+
"step": 100
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.16,
|
| 114 |
+
"eval_biology_entropy": 3.549320913314819,
|
| 115 |
+
"eval_biology_loss": 4.242035388946533,
|
| 116 |
+
"eval_biology_mean_token_accuracy": 0.4143077299594879,
|
| 117 |
+
"eval_biology_num_tokens": 3330597.0,
|
| 118 |
+
"eval_biology_runtime": 55.9019,
|
| 119 |
+
"eval_biology_samples_per_second": 8.944,
|
| 120 |
+
"eval_biology_steps_per_second": 2.236,
|
| 121 |
+
"step": 100
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.16,
|
| 125 |
+
"eval_chemistry_entropy": 1.9644288024902343,
|
| 126 |
+
"eval_chemistry_loss": 2.1523141860961914,
|
| 127 |
+
"eval_chemistry_mean_token_accuracy": 0.635781112909317,
|
| 128 |
+
"eval_chemistry_num_tokens": 3330597.0,
|
| 129 |
+
"eval_chemistry_runtime": 69.6126,
|
| 130 |
+
"eval_chemistry_samples_per_second": 7.183,
|
| 131 |
+
"eval_chemistry_steps_per_second": 1.796,
|
| 132 |
+
"step": 100
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.16,
|
| 136 |
+
"eval_math_entropy": 1.1708643288612366,
|
| 137 |
+
"eval_math_loss": 1.6400102376937866,
|
| 138 |
+
"eval_math_mean_token_accuracy": 0.7028316712379455,
|
| 139 |
+
"eval_math_num_tokens": 3330597.0,
|
| 140 |
+
"eval_math_runtime": 71.6855,
|
| 141 |
+
"eval_math_samples_per_second": 6.975,
|
| 142 |
+
"eval_math_steps_per_second": 1.744,
|
| 143 |
+
"step": 100
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.16,
|
| 147 |
+
"eval_physics_entropy": 1.388730420589447,
|
| 148 |
+
"eval_physics_loss": 1.5701876878738403,
|
| 149 |
+
"eval_physics_mean_token_accuracy": 0.7025446329116821,
|
| 150 |
+
"eval_physics_num_tokens": 3330597.0,
|
| 151 |
+
"eval_physics_runtime": 83.1014,
|
| 152 |
+
"eval_physics_samples_per_second": 6.017,
|
| 153 |
+
"eval_physics_steps_per_second": 1.504,
|
| 154 |
+
"step": 100
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"entropy": 1.3951868526637554,
|
| 158 |
+
"epoch": 0.176,
|
| 159 |
+
"grad_norm": 23.0,
|
| 160 |
+
"learning_rate": 7.266666666666668e-06,
|
| 161 |
+
"loss": 24.797,
|
| 162 |
+
"mean_token_accuracy": 0.7061628632247448,
|
| 163 |
+
"num_tokens": 3658264.0,
|
| 164 |
+
"step": 110
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"entropy": 1.3387951213866471,
|
| 168 |
+
"epoch": 0.192,
|
| 169 |
+
"grad_norm": 15.375,
|
| 170 |
+
"learning_rate": 7.933333333333334e-06,
|
| 171 |
+
"loss": 23.7842,
|
| 172 |
+
"mean_token_accuracy": 0.713041927665472,
|
| 173 |
+
"num_tokens": 3995568.0,
|
| 174 |
+
"step": 120
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"entropy": 1.3034498687833547,
|
| 178 |
+
"epoch": 0.208,
|
| 179 |
+
"grad_norm": 17.5,
|
| 180 |
+
"learning_rate": 8.6e-06,
|
| 181 |
+
"loss": 22.4839,
|
| 182 |
+
"mean_token_accuracy": 0.7252091735601425,
|
| 183 |
+
"num_tokens": 4321436.0,
|
| 184 |
+
"step": 130
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"entropy": 1.2866257604211568,
|
| 188 |
+
"epoch": 0.224,
|
| 189 |
+
"grad_norm": 14.6875,
|
| 190 |
+
"learning_rate": 9.266666666666667e-06,
|
| 191 |
+
"loss": 22.0837,
|
| 192 |
+
"mean_token_accuracy": 0.7270880017429591,
|
| 193 |
+
"num_tokens": 4648491.0,
|
| 194 |
+
"step": 140
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"entropy": 1.268964756652713,
|
| 198 |
+
"epoch": 0.24,
|
| 199 |
+
"grad_norm": 14.25,
|
| 200 |
+
"learning_rate": 9.933333333333334e-06,
|
| 201 |
+
"loss": 21.6129,
|
| 202 |
+
"mean_token_accuracy": 0.7301765695214272,
|
| 203 |
+
"num_tokens": 4986175.0,
|
| 204 |
+
"step": 150
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"entropy": 1.2642716094851494,
|
| 208 |
+
"epoch": 0.256,
|
| 209 |
+
"grad_norm": 15.5625,
|
| 210 |
+
"learning_rate": 1.0600000000000002e-05,
|
| 211 |
+
"loss": 21.2618,
|
| 212 |
+
"mean_token_accuracy": 0.734415826946497,
|
| 213 |
+
"num_tokens": 5329320.0,
|
| 214 |
+
"step": 160
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"entropy": 1.2717196103185415,
|
| 218 |
+
"epoch": 0.272,
|
| 219 |
+
"grad_norm": 17.5,
|
| 220 |
+
"learning_rate": 1.1266666666666668e-05,
|
| 221 |
+
"loss": 21.4982,
|
| 222 |
+
"mean_token_accuracy": 0.7297716859728098,
|
| 223 |
+
"num_tokens": 5658796.0,
|
| 224 |
+
"step": 170
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"entropy": 1.2510493908077478,
|
| 228 |
+
"epoch": 0.288,
|
| 229 |
+
"grad_norm": 14.5625,
|
| 230 |
+
"learning_rate": 1.1933333333333335e-05,
|
| 231 |
+
"loss": 21.0954,
|
| 232 |
+
"mean_token_accuracy": 0.7358391582965851,
|
| 233 |
+
"num_tokens": 5980201.0,
|
| 234 |
+
"step": 180
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"entropy": 1.2463122855871915,
|
| 238 |
+
"epoch": 0.304,
|
| 239 |
+
"grad_norm": 19.5,
|
| 240 |
+
"learning_rate": 1.2600000000000001e-05,
|
| 241 |
+
"loss": 20.8303,
|
| 242 |
+
"mean_token_accuracy": 0.7361553307622671,
|
| 243 |
+
"num_tokens": 6298858.0,
|
| 244 |
+
"step": 190
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"entropy": 1.215079977735877,
|
| 248 |
+
"epoch": 0.32,
|
| 249 |
+
"grad_norm": 15.875,
|
| 250 |
+
"learning_rate": 1.3266666666666668e-05,
|
| 251 |
+
"loss": 20.4326,
|
| 252 |
+
"mean_token_accuracy": 0.7394854992628097,
|
| 253 |
+
"num_tokens": 6622798.0,
|
| 254 |
+
"step": 200
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.32,
|
| 258 |
+
"eval_biology_entropy": 3.1598935775756836,
|
| 259 |
+
"eval_biology_loss": 3.6490557193756104,
|
| 260 |
+
"eval_biology_mean_token_accuracy": 0.45875698828697203,
|
| 261 |
+
"eval_biology_num_tokens": 6622798.0,
|
| 262 |
+
"eval_biology_runtime": 56.3027,
|
| 263 |
+
"eval_biology_samples_per_second": 8.881,
|
| 264 |
+
"eval_biology_steps_per_second": 2.22,
|
| 265 |
+
"step": 200
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.32,
|
| 269 |
+
"eval_chemistry_entropy": 1.6644244351387023,
|
| 270 |
+
"eval_chemistry_loss": 1.768973469734192,
|
| 271 |
+
"eval_chemistry_mean_token_accuracy": 0.6738141541481018,
|
| 272 |
+
"eval_chemistry_num_tokens": 6622798.0,
|
| 273 |
+
"eval_chemistry_runtime": 69.6913,
|
| 274 |
+
"eval_chemistry_samples_per_second": 7.174,
|
| 275 |
+
"eval_chemistry_steps_per_second": 1.794,
|
| 276 |
+
"step": 200
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.32,
|
| 280 |
+
"eval_math_entropy": 1.124363118648529,
|
| 281 |
+
"eval_math_loss": 1.4134961366653442,
|
| 282 |
+
"eval_math_mean_token_accuracy": 0.7205373592376709,
|
| 283 |
+
"eval_math_num_tokens": 6622798.0,
|
| 284 |
+
"eval_math_runtime": 71.6804,
|
| 285 |
+
"eval_math_samples_per_second": 6.975,
|
| 286 |
+
"eval_math_steps_per_second": 1.744,
|
| 287 |
+
"step": 200
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.32,
|
| 291 |
+
"eval_physics_entropy": 1.2113610072135925,
|
| 292 |
+
"eval_physics_loss": 1.2715808153152466,
|
| 293 |
+
"eval_physics_mean_token_accuracy": 0.7393231744766235,
|
| 294 |
+
"eval_physics_num_tokens": 6622798.0,
|
| 295 |
+
"eval_physics_runtime": 83.0035,
|
| 296 |
+
"eval_physics_samples_per_second": 6.024,
|
| 297 |
+
"eval_physics_steps_per_second": 1.506,
|
| 298 |
+
"step": 200
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"entropy": 1.2057966101914643,
|
| 302 |
+
"epoch": 0.336,
|
| 303 |
+
"grad_norm": 13.875,
|
| 304 |
+
"learning_rate": 1.3933333333333334e-05,
|
| 305 |
+
"loss": 20.1637,
|
| 306 |
+
"mean_token_accuracy": 0.740947175398469,
|
| 307 |
+
"num_tokens": 6953611.0,
|
| 308 |
+
"step": 210
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"entropy": 1.1682135988026858,
|
| 312 |
+
"epoch": 0.352,
|
| 313 |
+
"grad_norm": 14.625,
|
| 314 |
+
"learning_rate": 1.46e-05,
|
| 315 |
+
"loss": 19.5323,
|
| 316 |
+
"mean_token_accuracy": 0.748641949146986,
|
| 317 |
+
"num_tokens": 7290378.0,
|
| 318 |
+
"step": 220
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"entropy": 1.1533767595887183,
|
| 322 |
+
"epoch": 0.368,
|
| 323 |
+
"grad_norm": 14.75,
|
| 324 |
+
"learning_rate": 1.5266666666666667e-05,
|
| 325 |
+
"loss": 19.3393,
|
| 326 |
+
"mean_token_accuracy": 0.7496714886277914,
|
| 327 |
+
"num_tokens": 7621082.0,
|
| 328 |
+
"step": 230
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"entropy": 1.2154120814055205,
|
| 332 |
+
"epoch": 0.384,
|
| 333 |
+
"grad_norm": 13.875,
|
| 334 |
+
"learning_rate": 1.5933333333333336e-05,
|
| 335 |
+
"loss": 20.1519,
|
| 336 |
+
"mean_token_accuracy": 0.7383838955312967,
|
| 337 |
+
"num_tokens": 7955570.0,
|
| 338 |
+
"step": 240
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"entropy": 1.162219974398613,
|
| 342 |
+
"epoch": 0.4,
|
| 343 |
+
"grad_norm": 17.0,
|
| 344 |
+
"learning_rate": 1.66e-05,
|
| 345 |
+
"loss": 19.3544,
|
| 346 |
+
"mean_token_accuracy": 0.7475786007940769,
|
| 347 |
+
"num_tokens": 8291049.0,
|
| 348 |
+
"step": 250
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"entropy": 1.1398306384682655,
|
| 352 |
+
"epoch": 0.416,
|
| 353 |
+
"grad_norm": 17.0,
|
| 354 |
+
"learning_rate": 1.726666666666667e-05,
|
| 355 |
+
"loss": 18.9335,
|
| 356 |
+
"mean_token_accuracy": 0.7522856764495373,
|
| 357 |
+
"num_tokens": 8627310.0,
|
| 358 |
+
"step": 260
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"entropy": 1.160587527230382,
|
| 362 |
+
"epoch": 0.432,
|
| 363 |
+
"grad_norm": 14.5,
|
| 364 |
+
"learning_rate": 1.7933333333333333e-05,
|
| 365 |
+
"loss": 19.2248,
|
| 366 |
+
"mean_token_accuracy": 0.7491097755730152,
|
| 367 |
+
"num_tokens": 8958371.0,
|
| 368 |
+
"step": 270
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"entropy": 1.1003841817378999,
|
| 372 |
+
"epoch": 0.448,
|
| 373 |
+
"grad_norm": 14.0,
|
| 374 |
+
"learning_rate": 1.86e-05,
|
| 375 |
+
"loss": 18.3143,
|
| 376 |
+
"mean_token_accuracy": 0.7592806451022625,
|
| 377 |
+
"num_tokens": 9294388.0,
|
| 378 |
+
"step": 280
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"entropy": 1.1066032487899065,
|
| 382 |
+
"epoch": 0.464,
|
| 383 |
+
"grad_norm": 15.375,
|
| 384 |
+
"learning_rate": 1.926666666666667e-05,
|
| 385 |
+
"loss": 18.3092,
|
| 386 |
+
"mean_token_accuracy": 0.7574895836412907,
|
| 387 |
+
"num_tokens": 9620366.0,
|
| 388 |
+
"step": 290
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"entropy": 1.1134729530662297,
|
| 392 |
+
"epoch": 0.48,
|
| 393 |
+
"grad_norm": 14.875,
|
| 394 |
+
"learning_rate": 1.9933333333333334e-05,
|
| 395 |
+
"loss": 18.5931,
|
| 396 |
+
"mean_token_accuracy": 0.7540634006261826,
|
| 397 |
+
"num_tokens": 9955431.0,
|
| 398 |
+
"step": 300
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.48,
|
| 402 |
+
"eval_biology_entropy": 3.058082123756409,
|
| 403 |
+
"eval_biology_loss": 3.4944381713867188,
|
| 404 |
+
"eval_biology_mean_token_accuracy": 0.4677197203636169,
|
| 405 |
+
"eval_biology_num_tokens": 9955431.0,
|
| 406 |
+
"eval_biology_runtime": 56.129,
|
| 407 |
+
"eval_biology_samples_per_second": 8.908,
|
| 408 |
+
"eval_biology_steps_per_second": 2.227,
|
| 409 |
+
"step": 300
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.48,
|
| 413 |
+
"eval_chemistry_entropy": 1.584096737384796,
|
| 414 |
+
"eval_chemistry_loss": 1.6560814380645752,
|
| 415 |
+
"eval_chemistry_mean_token_accuracy": 0.6847201838493348,
|
| 416 |
+
"eval_chemistry_num_tokens": 9955431.0,
|
| 417 |
+
"eval_chemistry_runtime": 69.5947,
|
| 418 |
+
"eval_chemistry_samples_per_second": 7.184,
|
| 419 |
+
"eval_chemistry_steps_per_second": 1.796,
|
| 420 |
+
"step": 300
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.48,
|
| 424 |
+
"eval_math_entropy": 1.1037719421386718,
|
| 425 |
+
"eval_math_loss": 1.3507343530654907,
|
| 426 |
+
"eval_math_mean_token_accuracy": 0.7274009571075439,
|
| 427 |
+
"eval_math_num_tokens": 9955431.0,
|
| 428 |
+
"eval_math_runtime": 71.8212,
|
| 429 |
+
"eval_math_samples_per_second": 6.962,
|
| 430 |
+
"eval_math_steps_per_second": 1.74,
|
| 431 |
+
"step": 300
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.48,
|
| 435 |
+
"eval_physics_entropy": 1.1398421788215638,
|
| 436 |
+
"eval_physics_loss": 1.1733136177062988,
|
| 437 |
+
"eval_physics_mean_token_accuracy": 0.7520353999137879,
|
| 438 |
+
"eval_physics_num_tokens": 9955431.0,
|
| 439 |
+
"eval_physics_runtime": 83.0575,
|
| 440 |
+
"eval_physics_samples_per_second": 6.02,
|
| 441 |
+
"eval_physics_steps_per_second": 1.505,
|
| 442 |
+
"step": 300
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"entropy": 1.1404459465295076,
|
| 446 |
+
"epoch": 0.496,
|
| 447 |
+
"grad_norm": 14.25,
|
| 448 |
+
"learning_rate": 1.9933333333333334e-05,
|
| 449 |
+
"loss": 18.8326,
|
| 450 |
+
"mean_token_accuracy": 0.7513993989676238,
|
| 451 |
+
"num_tokens": 10289883.0,
|
| 452 |
+
"step": 310
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"entropy": 1.114009590819478,
|
| 456 |
+
"epoch": 0.512,
|
| 457 |
+
"grad_norm": 15.4375,
|
| 458 |
+
"learning_rate": 1.985925925925926e-05,
|
| 459 |
+
"loss": 18.526,
|
| 460 |
+
"mean_token_accuracy": 0.7558605384081603,
|
| 461 |
+
"num_tokens": 10619468.0,
|
| 462 |
+
"step": 320
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"entropy": 1.126872155070305,
|
| 466 |
+
"epoch": 0.528,
|
| 467 |
+
"grad_norm": 13.5625,
|
| 468 |
+
"learning_rate": 1.9785185185185187e-05,
|
| 469 |
+
"loss": 18.6944,
|
| 470 |
+
"mean_token_accuracy": 0.75301427654922,
|
| 471 |
+
"num_tokens": 10944904.0,
|
| 472 |
+
"step": 330
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"entropy": 1.11179842017591,
|
| 476 |
+
"epoch": 0.544,
|
| 477 |
+
"grad_norm": 14.625,
|
| 478 |
+
"learning_rate": 1.971111111111111e-05,
|
| 479 |
+
"loss": 18.3887,
|
| 480 |
+
"mean_token_accuracy": 0.7559159055352211,
|
| 481 |
+
"num_tokens": 11266409.0,
|
| 482 |
+
"step": 340
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"entropy": 1.0923905447125435,
|
| 486 |
+
"epoch": 0.56,
|
| 487 |
+
"grad_norm": 13.5,
|
| 488 |
+
"learning_rate": 1.963703703703704e-05,
|
| 489 |
+
"loss": 18.0431,
|
| 490 |
+
"mean_token_accuracy": 0.7601441144943237,
|
| 491 |
+
"num_tokens": 11605544.0,
|
| 492 |
+
"step": 350
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"entropy": 1.075535924360156,
|
| 496 |
+
"epoch": 0.576,
|
| 497 |
+
"grad_norm": 15.375,
|
| 498 |
+
"learning_rate": 1.9562962962962964e-05,
|
| 499 |
+
"loss": 17.9134,
|
| 500 |
+
"mean_token_accuracy": 0.762933773174882,
|
| 501 |
+
"num_tokens": 11931715.0,
|
| 502 |
+
"step": 360
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"entropy": 1.0857708267867565,
|
| 506 |
+
"epoch": 0.592,
|
| 507 |
+
"grad_norm": 16.5,
|
| 508 |
+
"learning_rate": 1.948888888888889e-05,
|
| 509 |
+
"loss": 17.9806,
|
| 510 |
+
"mean_token_accuracy": 0.7590868677943945,
|
| 511 |
+
"num_tokens": 12257243.0,
|
| 512 |
+
"step": 370
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"entropy": 1.077372272312641,
|
| 516 |
+
"epoch": 0.608,
|
| 517 |
+
"grad_norm": 14.25,
|
| 518 |
+
"learning_rate": 1.9414814814814817e-05,
|
| 519 |
+
"loss": 17.8473,
|
| 520 |
+
"mean_token_accuracy": 0.7623418286442757,
|
| 521 |
+
"num_tokens": 12583154.0,
|
| 522 |
+
"step": 380
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"entropy": 1.101408040151,
|
| 526 |
+
"epoch": 0.624,
|
| 527 |
+
"grad_norm": 13.4375,
|
| 528 |
+
"learning_rate": 1.9340740740740743e-05,
|
| 529 |
+
"loss": 18.1987,
|
| 530 |
+
"mean_token_accuracy": 0.7578636299818754,
|
| 531 |
+
"num_tokens": 12905392.0,
|
| 532 |
+
"step": 390
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"entropy": 1.0936459716409446,
|
| 536 |
+
"epoch": 0.64,
|
| 537 |
+
"grad_norm": 15.0,
|
| 538 |
+
"learning_rate": 1.926666666666667e-05,
|
| 539 |
+
"loss": 18.2164,
|
| 540 |
+
"mean_token_accuracy": 0.7581360317766667,
|
| 541 |
+
"num_tokens": 13232198.0,
|
| 542 |
+
"step": 400
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.64,
|
| 546 |
+
"eval_biology_entropy": 2.985183084487915,
|
| 547 |
+
"eval_biology_loss": 3.3934555053710938,
|
| 548 |
+
"eval_biology_mean_token_accuracy": 0.47690209436416625,
|
| 549 |
+
"eval_biology_num_tokens": 13232198.0,
|
| 550 |
+
"eval_biology_runtime": 56.2992,
|
| 551 |
+
"eval_biology_samples_per_second": 8.881,
|
| 552 |
+
"eval_biology_steps_per_second": 2.22,
|
| 553 |
+
"step": 400
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 0.64,
|
| 557 |
+
"eval_chemistry_entropy": 1.5230218110084535,
|
| 558 |
+
"eval_chemistry_loss": 1.5974879264831543,
|
| 559 |
+
"eval_chemistry_mean_token_accuracy": 0.6924252939224244,
|
| 560 |
+
"eval_chemistry_num_tokens": 13232198.0,
|
| 561 |
+
"eval_chemistry_runtime": 69.6914,
|
| 562 |
+
"eval_chemistry_samples_per_second": 7.174,
|
| 563 |
+
"eval_chemistry_steps_per_second": 1.794,
|
| 564 |
+
"step": 400
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.64,
|
| 568 |
+
"eval_math_entropy": 1.0690260066986084,
|
| 569 |
+
"eval_math_loss": 1.3214365243911743,
|
| 570 |
+
"eval_math_mean_token_accuracy": 0.7313346996307373,
|
| 571 |
+
"eval_math_num_tokens": 13232198.0,
|
| 572 |
+
"eval_math_runtime": 71.8154,
|
| 573 |
+
"eval_math_samples_per_second": 6.962,
|
| 574 |
+
"eval_math_steps_per_second": 1.741,
|
| 575 |
+
"step": 400
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.64,
|
| 579 |
+
"eval_physics_entropy": 1.079647916316986,
|
| 580 |
+
"eval_physics_loss": 1.122850775718689,
|
| 581 |
+
"eval_physics_mean_token_accuracy": 0.7587474570274353,
|
| 582 |
+
"eval_physics_num_tokens": 13232198.0,
|
| 583 |
+
"eval_physics_runtime": 83.3676,
|
| 584 |
+
"eval_physics_samples_per_second": 5.998,
|
| 585 |
+
"eval_physics_steps_per_second": 1.499,
|
| 586 |
+
"step": 400
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"entropy": 1.0861666969954968,
|
| 590 |
+
"epoch": 0.656,
|
| 591 |
+
"grad_norm": 14.1875,
|
| 592 |
+
"learning_rate": 1.9192592592592593e-05,
|
| 593 |
+
"loss": 17.9599,
|
| 594 |
+
"mean_token_accuracy": 0.7603602163493633,
|
| 595 |
+
"num_tokens": 13575902.0,
|
| 596 |
+
"step": 410
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"entropy": 1.0595566183328629,
|
| 600 |
+
"epoch": 0.672,
|
| 601 |
+
"grad_norm": 15.5625,
|
| 602 |
+
"learning_rate": 1.911851851851852e-05,
|
| 603 |
+
"loss": 17.6133,
|
| 604 |
+
"mean_token_accuracy": 0.7644100178033113,
|
| 605 |
+
"num_tokens": 13895997.0,
|
| 606 |
+
"step": 420
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"entropy": 1.0779153060168027,
|
| 610 |
+
"epoch": 0.688,
|
| 611 |
+
"grad_norm": 13.625,
|
| 612 |
+
"learning_rate": 1.9044444444444446e-05,
|
| 613 |
+
"loss": 17.708,
|
| 614 |
+
"mean_token_accuracy": 0.7641124926507473,
|
| 615 |
+
"num_tokens": 14234888.0,
|
| 616 |
+
"step": 430
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"entropy": 1.0761012937873602,
|
| 620 |
+
"epoch": 0.704,
|
| 621 |
+
"grad_norm": 14.4375,
|
| 622 |
+
"learning_rate": 1.8970370370370372e-05,
|
| 623 |
+
"loss": 17.9412,
|
| 624 |
+
"mean_token_accuracy": 0.7593970015645027,
|
| 625 |
+
"num_tokens": 14567908.0,
|
| 626 |
+
"step": 440
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"entropy": 1.0635488711297512,
|
| 630 |
+
"epoch": 0.72,
|
| 631 |
+
"grad_norm": 15.5625,
|
| 632 |
+
"learning_rate": 1.8896296296296295e-05,
|
| 633 |
+
"loss": 17.5409,
|
| 634 |
+
"mean_token_accuracy": 0.7655271884053946,
|
| 635 |
+
"num_tokens": 14882927.0,
|
| 636 |
+
"step": 450
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"entropy": 1.0563195772469043,
|
| 640 |
+
"epoch": 0.736,
|
| 641 |
+
"grad_norm": 15.8125,
|
| 642 |
+
"learning_rate": 1.8822222222222225e-05,
|
| 643 |
+
"loss": 17.5759,
|
| 644 |
+
"mean_token_accuracy": 0.7641396790742874,
|
| 645 |
+
"num_tokens": 15217342.0,
|
| 646 |
+
"step": 460
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"entropy": 1.037678936868906,
|
| 650 |
+
"epoch": 0.752,
|
| 651 |
+
"grad_norm": 13.125,
|
| 652 |
+
"learning_rate": 1.874814814814815e-05,
|
| 653 |
+
"loss": 17.1107,
|
| 654 |
+
"mean_token_accuracy": 0.7690521724522114,
|
| 655 |
+
"num_tokens": 15547710.0,
|
| 656 |
+
"step": 470
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"entropy": 1.0838343646377324,
|
| 660 |
+
"epoch": 0.768,
|
| 661 |
+
"grad_norm": 16.75,
|
| 662 |
+
"learning_rate": 1.8674074074074075e-05,
|
| 663 |
+
"loss": 18.0446,
|
| 664 |
+
"mean_token_accuracy": 0.759132880717516,
|
| 665 |
+
"num_tokens": 15877177.0,
|
| 666 |
+
"step": 480
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"entropy": 1.0296639803797007,
|
| 670 |
+
"epoch": 0.784,
|
| 671 |
+
"grad_norm": 12.5,
|
| 672 |
+
"learning_rate": 1.86e-05,
|
| 673 |
+
"loss": 17.1015,
|
| 674 |
+
"mean_token_accuracy": 0.7697366833686828,
|
| 675 |
+
"num_tokens": 16219640.0,
|
| 676 |
+
"step": 490
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"entropy": 1.0563473679125308,
|
| 680 |
+
"epoch": 0.8,
|
| 681 |
+
"grad_norm": 15.5,
|
| 682 |
+
"learning_rate": 1.8525925925925928e-05,
|
| 683 |
+
"loss": 17.5054,
|
| 684 |
+
"mean_token_accuracy": 0.7657136749476194,
|
| 685 |
+
"num_tokens": 16548261.0,
|
| 686 |
+
"step": 500
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 0.8,
|
| 690 |
+
"eval_biology_entropy": 2.9737762069702147,
|
| 691 |
+
"eval_biology_loss": 3.3795254230499268,
|
| 692 |
+
"eval_biology_mean_token_accuracy": 0.48070894145965576,
|
| 693 |
+
"eval_biology_num_tokens": 16548261.0,
|
| 694 |
+
"eval_biology_runtime": 56.18,
|
| 695 |
+
"eval_biology_samples_per_second": 8.9,
|
| 696 |
+
"eval_biology_steps_per_second": 2.225,
|
| 697 |
+
"step": 500
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.8,
|
| 701 |
+
"eval_chemistry_entropy": 1.5115246725082399,
|
| 702 |
+
"eval_chemistry_loss": 1.573546051979065,
|
| 703 |
+
"eval_chemistry_mean_token_accuracy": 0.6963776540756226,
|
| 704 |
+
"eval_chemistry_num_tokens": 16548261.0,
|
| 705 |
+
"eval_chemistry_runtime": 69.0701,
|
| 706 |
+
"eval_chemistry_samples_per_second": 7.239,
|
| 707 |
+
"eval_chemistry_steps_per_second": 1.81,
|
| 708 |
+
"step": 500
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 0.8,
|
| 712 |
+
"eval_math_entropy": 1.0560337038040162,
|
| 713 |
+
"eval_math_loss": 1.306509017944336,
|
| 714 |
+
"eval_math_mean_token_accuracy": 0.7335917987823486,
|
| 715 |
+
"eval_math_num_tokens": 16548261.0,
|
| 716 |
+
"eval_math_runtime": 71.2835,
|
| 717 |
+
"eval_math_samples_per_second": 7.014,
|
| 718 |
+
"eval_math_steps_per_second": 1.754,
|
| 719 |
+
"step": 500
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 0.8,
|
| 723 |
+
"eval_physics_entropy": 1.0595522990226747,
|
| 724 |
+
"eval_physics_loss": 1.0949721336364746,
|
| 725 |
+
"eval_physics_mean_token_accuracy": 0.7634064397811889,
|
| 726 |
+
"eval_physics_num_tokens": 16548261.0,
|
| 727 |
+
"eval_physics_runtime": 83.0981,
|
| 728 |
+
"eval_physics_samples_per_second": 6.017,
|
| 729 |
+
"eval_physics_steps_per_second": 1.504,
|
| 730 |
+
"step": 500
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"entropy": 1.042676868662238,
|
| 734 |
+
"epoch": 0.816,
|
| 735 |
+
"grad_norm": 14.1875,
|
| 736 |
+
"learning_rate": 1.8451851851851855e-05,
|
| 737 |
+
"loss": 17.3757,
|
| 738 |
+
"mean_token_accuracy": 0.7653235342353583,
|
| 739 |
+
"num_tokens": 16870533.0,
|
| 740 |
+
"step": 510
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"entropy": 1.060330757498741,
|
| 744 |
+
"epoch": 0.832,
|
| 745 |
+
"grad_norm": 17.875,
|
| 746 |
+
"learning_rate": 1.8377777777777778e-05,
|
| 747 |
+
"loss": 17.5577,
|
| 748 |
+
"mean_token_accuracy": 0.7640024948865175,
|
| 749 |
+
"num_tokens": 17196371.0,
|
| 750 |
+
"step": 520
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"entropy": 1.045187360420823,
|
| 754 |
+
"epoch": 0.848,
|
| 755 |
+
"grad_norm": 12.75,
|
| 756 |
+
"learning_rate": 1.8303703703703704e-05,
|
| 757 |
+
"loss": 17.2392,
|
| 758 |
+
"mean_token_accuracy": 0.7683587469160557,
|
| 759 |
+
"num_tokens": 17527066.0,
|
| 760 |
+
"step": 530
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"entropy": 1.0516619730740786,
|
| 764 |
+
"epoch": 0.864,
|
| 765 |
+
"grad_norm": 13.9375,
|
| 766 |
+
"learning_rate": 1.822962962962963e-05,
|
| 767 |
+
"loss": 17.4228,
|
| 768 |
+
"mean_token_accuracy": 0.7653108265250921,
|
| 769 |
+
"num_tokens": 17859605.0,
|
| 770 |
+
"step": 540
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"entropy": 1.0745812084525823,
|
| 774 |
+
"epoch": 0.88,
|
| 775 |
+
"grad_norm": 16.0,
|
| 776 |
+
"learning_rate": 1.8155555555555557e-05,
|
| 777 |
+
"loss": 17.7266,
|
| 778 |
+
"mean_token_accuracy": 0.7631923343986273,
|
| 779 |
+
"num_tokens": 18187315.0,
|
| 780 |
+
"step": 550
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"entropy": 0.9957344826310873,
|
| 784 |
+
"epoch": 0.896,
|
| 785 |
+
"grad_norm": 12.625,
|
| 786 |
+
"learning_rate": 1.8081481481481484e-05,
|
| 787 |
+
"loss": 16.4857,
|
| 788 |
+
"mean_token_accuracy": 0.7752151470631361,
|
| 789 |
+
"num_tokens": 18536790.0,
|
| 790 |
+
"step": 560
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"entropy": 1.0575346298515798,
|
| 794 |
+
"epoch": 0.912,
|
| 795 |
+
"grad_norm": 14.9375,
|
| 796 |
+
"learning_rate": 1.800740740740741e-05,
|
| 797 |
+
"loss": 17.4739,
|
| 798 |
+
"mean_token_accuracy": 0.7637861587107182,
|
| 799 |
+
"num_tokens": 18869344.0,
|
| 800 |
+
"step": 570
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"entropy": 1.0460248701274395,
|
| 804 |
+
"epoch": 0.928,
|
| 805 |
+
"grad_norm": 16.125,
|
| 806 |
+
"learning_rate": 1.7933333333333333e-05,
|
| 807 |
+
"loss": 17.3717,
|
| 808 |
+
"mean_token_accuracy": 0.7649744093418122,
|
| 809 |
+
"num_tokens": 19199834.0,
|
| 810 |
+
"step": 580
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"entropy": 1.0241859197616576,
|
| 814 |
+
"epoch": 0.944,
|
| 815 |
+
"grad_norm": 15.5,
|
| 816 |
+
"learning_rate": 1.785925925925926e-05,
|
| 817 |
+
"loss": 16.9257,
|
| 818 |
+
"mean_token_accuracy": 0.7700011480599642,
|
| 819 |
+
"num_tokens": 19525385.0,
|
| 820 |
+
"step": 590
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"entropy": 1.0272901140153408,
|
| 824 |
+
"epoch": 0.96,
|
| 825 |
+
"grad_norm": 17.125,
|
| 826 |
+
"learning_rate": 1.7785185185185186e-05,
|
| 827 |
+
"loss": 17.0788,
|
| 828 |
+
"mean_token_accuracy": 0.7674089819192886,
|
| 829 |
+
"num_tokens": 19856973.0,
|
| 830 |
+
"step": 600
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.96,
|
| 834 |
+
"eval_biology_entropy": 2.8927044954299928,
|
| 835 |
+
"eval_biology_loss": 3.3549082279205322,
|
| 836 |
+
"eval_biology_mean_token_accuracy": 0.48264122939109805,
|
| 837 |
+
"eval_biology_num_tokens": 19856973.0,
|
| 838 |
+
"eval_biology_runtime": 55.8568,
|
| 839 |
+
"eval_biology_samples_per_second": 8.951,
|
| 840 |
+
"eval_biology_steps_per_second": 2.238,
|
| 841 |
+
"step": 600
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 0.96,
|
| 845 |
+
"eval_chemistry_entropy": 1.464129506111145,
|
| 846 |
+
"eval_chemistry_loss": 1.5512925386428833,
|
| 847 |
+
"eval_chemistry_mean_token_accuracy": 0.6986791820526123,
|
| 848 |
+
"eval_chemistry_num_tokens": 19856973.0,
|
| 849 |
+
"eval_chemistry_runtime": 69.6627,
|
| 850 |
+
"eval_chemistry_samples_per_second": 7.177,
|
| 851 |
+
"eval_chemistry_steps_per_second": 1.794,
|
| 852 |
+
"step": 600
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 0.96,
|
| 856 |
+
"eval_math_entropy": 1.0420140266418456,
|
| 857 |
+
"eval_math_loss": 1.297760248184204,
|
| 858 |
+
"eval_math_mean_token_accuracy": 0.7350094408988953,
|
| 859 |
+
"eval_math_num_tokens": 19856973.0,
|
| 860 |
+
"eval_math_runtime": 71.5855,
|
| 861 |
+
"eval_math_samples_per_second": 6.985,
|
| 862 |
+
"eval_math_steps_per_second": 1.746,
|
| 863 |
+
"step": 600
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.96,
|
| 867 |
+
"eval_physics_entropy": 1.0300544242858887,
|
| 868 |
+
"eval_physics_loss": 1.076904058456421,
|
| 869 |
+
"eval_physics_mean_token_accuracy": 0.7661643705368042,
|
| 870 |
+
"eval_physics_num_tokens": 19856973.0,
|
| 871 |
+
"eval_physics_runtime": 83.7599,
|
| 872 |
+
"eval_physics_samples_per_second": 5.969,
|
| 873 |
+
"eval_physics_steps_per_second": 1.492,
|
| 874 |
+
"step": 600
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"entropy": 1.029942188039422,
|
| 878 |
+
"epoch": 0.976,
|
| 879 |
+
"grad_norm": 14.5625,
|
| 880 |
+
"learning_rate": 1.7711111111111113e-05,
|
| 881 |
+
"loss": 17.0833,
|
| 882 |
+
"mean_token_accuracy": 0.7702321965247393,
|
| 883 |
+
"num_tokens": 20190600.0,
|
| 884 |
+
"step": 610
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"entropy": 1.043160342797637,
|
| 888 |
+
"epoch": 0.992,
|
| 889 |
+
"grad_norm": 15.0,
|
| 890 |
+
"learning_rate": 1.763703703703704e-05,
|
| 891 |
+
"loss": 17.2354,
|
| 892 |
+
"mean_token_accuracy": 0.7673168186098337,
|
| 893 |
+
"num_tokens": 20518014.0,
|
| 894 |
+
"step": 620
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"entropy": 1.0153347484767437,
|
| 898 |
+
"epoch": 1.008,
|
| 899 |
+
"grad_norm": 14.8125,
|
| 900 |
+
"learning_rate": 1.7562962962962962e-05,
|
| 901 |
+
"loss": 16.634,
|
| 902 |
+
"mean_token_accuracy": 0.7728456899523735,
|
| 903 |
+
"num_tokens": 20859215.0,
|
| 904 |
+
"step": 630
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"entropy": 1.010590048506856,
|
| 908 |
+
"epoch": 1.024,
|
| 909 |
+
"grad_norm": 15.25,
|
| 910 |
+
"learning_rate": 1.7488888888888892e-05,
|
| 911 |
+
"loss": 16.8635,
|
| 912 |
+
"mean_token_accuracy": 0.770306408405304,
|
| 913 |
+
"num_tokens": 21186658.0,
|
| 914 |
+
"step": 640
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"entropy": 1.0216664642095565,
|
| 918 |
+
"epoch": 1.04,
|
| 919 |
+
"grad_norm": 13.4375,
|
| 920 |
+
"learning_rate": 1.7414814814814815e-05,
|
| 921 |
+
"loss": 16.8228,
|
| 922 |
+
"mean_token_accuracy": 0.771704015135765,
|
| 923 |
+
"num_tokens": 21517237.0,
|
| 924 |
+
"step": 650
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"entropy": 0.9916718799620867,
|
| 928 |
+
"epoch": 1.056,
|
| 929 |
+
"grad_norm": 13.9375,
|
| 930 |
+
"learning_rate": 1.7340740740740742e-05,
|
| 931 |
+
"loss": 16.4977,
|
| 932 |
+
"mean_token_accuracy": 0.7741733457893133,
|
| 933 |
+
"num_tokens": 21852528.0,
|
| 934 |
+
"step": 660
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"entropy": 1.0193660650402308,
|
| 938 |
+
"epoch": 1.072,
|
| 939 |
+
"grad_norm": 12.6875,
|
| 940 |
+
"learning_rate": 1.726666666666667e-05,
|
| 941 |
+
"loss": 16.7942,
|
| 942 |
+
"mean_token_accuracy": 0.7709830049425364,
|
| 943 |
+
"num_tokens": 22192380.0,
|
| 944 |
+
"step": 670
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"entropy": 1.017938245087862,
|
| 948 |
+
"epoch": 1.088,
|
| 949 |
+
"grad_norm": 17.5,
|
| 950 |
+
"learning_rate": 1.7192592592592595e-05,
|
| 951 |
+
"loss": 17.004,
|
| 952 |
+
"mean_token_accuracy": 0.7690784070640803,
|
| 953 |
+
"num_tokens": 22517971.0,
|
| 954 |
+
"step": 680
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"entropy": 1.009276295825839,
|
| 958 |
+
"epoch": 1.104,
|
| 959 |
+
"grad_norm": 13.5625,
|
| 960 |
+
"learning_rate": 1.711851851851852e-05,
|
| 961 |
+
"loss": 16.5871,
|
| 962 |
+
"mean_token_accuracy": 0.773436525464058,
|
| 963 |
+
"num_tokens": 22848694.0,
|
| 964 |
+
"step": 690
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"entropy": 1.0195886611938476,
|
| 968 |
+
"epoch": 1.12,
|
| 969 |
+
"grad_norm": 14.125,
|
| 970 |
+
"learning_rate": 1.7044444444444445e-05,
|
| 971 |
+
"loss": 17.0053,
|
| 972 |
+
"mean_token_accuracy": 0.7680860497057438,
|
| 973 |
+
"num_tokens": 23183889.0,
|
| 974 |
+
"step": 700
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 1.12,
|
| 978 |
+
"eval_biology_entropy": 2.837118970870972,
|
| 979 |
+
"eval_biology_loss": 3.3420016765594482,
|
| 980 |
+
"eval_biology_mean_token_accuracy": 0.48326786851882936,
|
| 981 |
+
"eval_biology_num_tokens": 23183889.0,
|
| 982 |
+
"eval_biology_runtime": 56.3022,
|
| 983 |
+
"eval_biology_samples_per_second": 8.881,
|
| 984 |
+
"eval_biology_steps_per_second": 2.22,
|
| 985 |
+
"step": 700
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 1.12,
|
| 989 |
+
"eval_chemistry_entropy": 1.433508418083191,
|
| 990 |
+
"eval_chemistry_loss": 1.5409785509109497,
|
| 991 |
+
"eval_chemistry_mean_token_accuracy": 0.6995256505012513,
|
| 992 |
+
"eval_chemistry_num_tokens": 23183889.0,
|
| 993 |
+
"eval_chemistry_runtime": 69.4682,
|
| 994 |
+
"eval_chemistry_samples_per_second": 7.198,
|
| 995 |
+
"eval_chemistry_steps_per_second": 1.799,
|
| 996 |
+
"step": 700
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 1.12,
|
| 1000 |
+
"eval_math_entropy": 1.0230936131477355,
|
| 1001 |
+
"eval_math_loss": 1.2898101806640625,
|
| 1002 |
+
"eval_math_mean_token_accuracy": 0.736490508556366,
|
| 1003 |
+
"eval_math_num_tokens": 23183889.0,
|
| 1004 |
+
"eval_math_runtime": 71.817,
|
| 1005 |
+
"eval_math_samples_per_second": 6.962,
|
| 1006 |
+
"eval_math_steps_per_second": 1.741,
|
| 1007 |
+
"step": 700
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 1.12,
|
| 1011 |
+
"eval_physics_entropy": 1.0090678911209106,
|
| 1012 |
+
"eval_physics_loss": 1.062973976135254,
|
| 1013 |
+
"eval_physics_mean_token_accuracy": 0.7677445983886719,
|
| 1014 |
+
"eval_physics_num_tokens": 23183889.0,
|
| 1015 |
+
"eval_physics_runtime": 82.8728,
|
| 1016 |
+
"eval_physics_samples_per_second": 6.033,
|
| 1017 |
+
"eval_physics_steps_per_second": 1.508,
|
| 1018 |
+
"step": 700
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"entropy": 0.9851093973964453,
|
| 1022 |
+
"epoch": 1.1360000000000001,
|
| 1023 |
+
"grad_norm": 13.0625,
|
| 1024 |
+
"learning_rate": 1.697037037037037e-05,
|
| 1025 |
+
"loss": 16.2335,
|
| 1026 |
+
"mean_token_accuracy": 0.7770712476223707,
|
| 1027 |
+
"num_tokens": 23523903.0,
|
| 1028 |
+
"step": 710
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"entropy": 1.0010987296700478,
|
| 1032 |
+
"epoch": 1.152,
|
| 1033 |
+
"grad_norm": 16.375,
|
| 1034 |
+
"learning_rate": 1.6896296296296298e-05,
|
| 1035 |
+
"loss": 16.6749,
|
| 1036 |
+
"mean_token_accuracy": 0.7745790995657444,
|
| 1037 |
+
"num_tokens": 23848848.0,
|
| 1038 |
+
"step": 720
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"entropy": 1.0347776237875224,
|
| 1042 |
+
"epoch": 1.168,
|
| 1043 |
+
"grad_norm": 14.375,
|
| 1044 |
+
"learning_rate": 1.6822222222222224e-05,
|
| 1045 |
+
"loss": 16.9856,
|
| 1046 |
+
"mean_token_accuracy": 0.7672174524515867,
|
| 1047 |
+
"num_tokens": 24174589.0,
|
| 1048 |
+
"step": 730
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"entropy": 1.0073780696839094,
|
| 1052 |
+
"epoch": 1.184,
|
| 1053 |
+
"grad_norm": 16.875,
|
| 1054 |
+
"learning_rate": 1.6748148148148147e-05,
|
| 1055 |
+
"loss": 16.7297,
|
| 1056 |
+
"mean_token_accuracy": 0.7723128654062748,
|
| 1057 |
+
"num_tokens": 24495981.0,
|
| 1058 |
+
"step": 740
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"entropy": 0.9867725741118193,
|
| 1062 |
+
"epoch": 1.2,
|
| 1063 |
+
"grad_norm": 14.25,
|
| 1064 |
+
"learning_rate": 1.6674074074074077e-05,
|
| 1065 |
+
"loss": 16.3695,
|
| 1066 |
+
"mean_token_accuracy": 0.7768875520676375,
|
| 1067 |
+
"num_tokens": 24826245.0,
|
| 1068 |
+
"step": 750
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"entropy": 1.018533943966031,
|
| 1072 |
+
"epoch": 1.216,
|
| 1073 |
+
"grad_norm": 17.875,
|
| 1074 |
+
"learning_rate": 1.66e-05,
|
| 1075 |
+
"loss": 16.7892,
|
| 1076 |
+
"mean_token_accuracy": 0.7702058039605617,
|
| 1077 |
+
"num_tokens": 25147775.0,
|
| 1078 |
+
"step": 760
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"entropy": 0.9930577903985978,
|
| 1082 |
+
"epoch": 1.232,
|
| 1083 |
+
"grad_norm": 16.125,
|
| 1084 |
+
"learning_rate": 1.6525925925925927e-05,
|
| 1085 |
+
"loss": 16.5313,
|
| 1086 |
+
"mean_token_accuracy": 0.7744447194039822,
|
| 1087 |
+
"num_tokens": 25480033.0,
|
| 1088 |
+
"step": 770
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"entropy": 1.0064157444983721,
|
| 1092 |
+
"epoch": 1.248,
|
| 1093 |
+
"grad_norm": 15.8125,
|
| 1094 |
+
"learning_rate": 1.6451851851851853e-05,
|
| 1095 |
+
"loss": 16.6421,
|
| 1096 |
+
"mean_token_accuracy": 0.773406020924449,
|
| 1097 |
+
"num_tokens": 25811978.0,
|
| 1098 |
+
"step": 780
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"entropy": 0.9985887616872787,
|
| 1102 |
+
"epoch": 1.264,
|
| 1103 |
+
"grad_norm": 16.5,
|
| 1104 |
+
"learning_rate": 1.637777777777778e-05,
|
| 1105 |
+
"loss": 16.5153,
|
| 1106 |
+
"mean_token_accuracy": 0.774219336360693,
|
| 1107 |
+
"num_tokens": 26146299.0,
|
| 1108 |
+
"step": 790
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"entropy": 1.0295063827186823,
|
| 1112 |
+
"epoch": 1.28,
|
| 1113 |
+
"grad_norm": 16.125,
|
| 1114 |
+
"learning_rate": 1.6303703703703706e-05,
|
| 1115 |
+
"loss": 17.0668,
|
| 1116 |
+
"mean_token_accuracy": 0.7661171369254589,
|
| 1117 |
+
"num_tokens": 26478207.0,
|
| 1118 |
+
"step": 800
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 1.28,
|
| 1122 |
+
"eval_biology_entropy": 2.898805745124817,
|
| 1123 |
+
"eval_biology_loss": 3.341296911239624,
|
| 1124 |
+
"eval_biology_mean_token_accuracy": 0.48686809062957764,
|
| 1125 |
+
"eval_biology_num_tokens": 26478207.0,
|
| 1126 |
+
"eval_biology_runtime": 56.3495,
|
| 1127 |
+
"eval_biology_samples_per_second": 8.873,
|
| 1128 |
+
"eval_biology_steps_per_second": 2.218,
|
| 1129 |
+
"step": 800
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.28,
|
| 1133 |
+
"eval_chemistry_entropy": 1.4466705160140991,
|
| 1134 |
+
"eval_chemistry_loss": 1.5262336730957031,
|
| 1135 |
+
"eval_chemistry_mean_token_accuracy": 0.7025133318901062,
|
| 1136 |
+
"eval_chemistry_num_tokens": 26478207.0,
|
| 1137 |
+
"eval_chemistry_runtime": 69.3655,
|
| 1138 |
+
"eval_chemistry_samples_per_second": 7.208,
|
| 1139 |
+
"eval_chemistry_steps_per_second": 1.802,
|
| 1140 |
+
"step": 800
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 1.28,
|
| 1144 |
+
"eval_math_entropy": 1.024560465335846,
|
| 1145 |
+
"eval_math_loss": 1.2816721200942993,
|
| 1146 |
+
"eval_math_mean_token_accuracy": 0.737667652130127,
|
| 1147 |
+
"eval_math_num_tokens": 26478207.0,
|
| 1148 |
+
"eval_math_runtime": 71.7104,
|
| 1149 |
+
"eval_math_samples_per_second": 6.972,
|
| 1150 |
+
"eval_math_steps_per_second": 1.743,
|
| 1151 |
+
"step": 800
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 1.28,
|
| 1155 |
+
"eval_physics_entropy": 1.0090858311653137,
|
| 1156 |
+
"eval_physics_loss": 1.0530790090560913,
|
| 1157 |
+
"eval_physics_mean_token_accuracy": 0.7696638231277466,
|
| 1158 |
+
"eval_physics_num_tokens": 26478207.0,
|
| 1159 |
+
"eval_physics_runtime": 83.3567,
|
| 1160 |
+
"eval_physics_samples_per_second": 5.998,
|
| 1161 |
+
"eval_physics_steps_per_second": 1.5,
|
| 1162 |
+
"step": 800
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"entropy": 0.9791497606784105,
|
| 1166 |
+
"epoch": 1.296,
|
| 1167 |
+
"grad_norm": 15.9375,
|
| 1168 |
+
"learning_rate": 1.622962962962963e-05,
|
| 1169 |
+
"loss": 16.2738,
|
| 1170 |
+
"mean_token_accuracy": 0.777652832493186,
|
| 1171 |
+
"num_tokens": 26806477.0,
|
| 1172 |
+
"step": 810
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"entropy": 1.0318988490849734,
|
| 1176 |
+
"epoch": 1.312,
|
| 1177 |
+
"grad_norm": 17.125,
|
| 1178 |
+
"learning_rate": 1.6155555555555556e-05,
|
| 1179 |
+
"loss": 16.9474,
|
| 1180 |
+
"mean_token_accuracy": 0.7686588883399963,
|
| 1181 |
+
"num_tokens": 27137528.0,
|
| 1182 |
+
"step": 820
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"entropy": 1.007080302760005,
|
| 1186 |
+
"epoch": 1.328,
|
| 1187 |
+
"grad_norm": 17.875,
|
| 1188 |
+
"learning_rate": 1.6081481481481482e-05,
|
| 1189 |
+
"loss": 16.6844,
|
| 1190 |
+
"mean_token_accuracy": 0.771523043885827,
|
| 1191 |
+
"num_tokens": 27465859.0,
|
| 1192 |
+
"step": 830
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"entropy": 1.0098473474383354,
|
| 1196 |
+
"epoch": 1.3439999999999999,
|
| 1197 |
+
"grad_norm": 16.5,
|
| 1198 |
+
"learning_rate": 1.600740740740741e-05,
|
| 1199 |
+
"loss": 16.8087,
|
| 1200 |
+
"mean_token_accuracy": 0.7707158491015434,
|
| 1201 |
+
"num_tokens": 27795568.0,
|
| 1202 |
+
"step": 840
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"entropy": 1.0220563992857934,
|
| 1206 |
+
"epoch": 1.3599999999999999,
|
| 1207 |
+
"grad_norm": 15.3125,
|
| 1208 |
+
"learning_rate": 1.5933333333333336e-05,
|
| 1209 |
+
"loss": 16.768,
|
| 1210 |
+
"mean_token_accuracy": 0.770363200828433,
|
| 1211 |
+
"num_tokens": 28129988.0,
|
| 1212 |
+
"step": 850
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"entropy": 0.9859919909387826,
|
| 1216 |
+
"epoch": 1.376,
|
| 1217 |
+
"grad_norm": 17.875,
|
| 1218 |
+
"learning_rate": 1.5859259259259262e-05,
|
| 1219 |
+
"loss": 16.4135,
|
| 1220 |
+
"mean_token_accuracy": 0.7744484882801771,
|
| 1221 |
+
"num_tokens": 28465903.0,
|
| 1222 |
+
"step": 860
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"entropy": 0.988647548854351,
|
| 1226 |
+
"epoch": 1.392,
|
| 1227 |
+
"grad_norm": 17.125,
|
| 1228 |
+
"learning_rate": 1.5785185185185185e-05,
|
| 1229 |
+
"loss": 16.2043,
|
| 1230 |
+
"mean_token_accuracy": 0.7775332607328892,
|
| 1231 |
+
"num_tokens": 28801554.0,
|
| 1232 |
+
"step": 870
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"entropy": 0.9736344590783119,
|
| 1236 |
+
"epoch": 1.408,
|
| 1237 |
+
"grad_norm": 15.25,
|
| 1238 |
+
"learning_rate": 1.571111111111111e-05,
|
| 1239 |
+
"loss": 16.332,
|
| 1240 |
+
"mean_token_accuracy": 0.7759838547557593,
|
| 1241 |
+
"num_tokens": 29125114.0,
|
| 1242 |
+
"step": 880
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"entropy": 1.024445422179997,
|
| 1246 |
+
"epoch": 1.424,
|
| 1247 |
+
"grad_norm": 14.125,
|
| 1248 |
+
"learning_rate": 1.5637037037037038e-05,
|
| 1249 |
+
"loss": 16.9175,
|
| 1250 |
+
"mean_token_accuracy": 0.7679786011576653,
|
| 1251 |
+
"num_tokens": 29454334.0,
|
| 1252 |
+
"step": 890
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"entropy": 0.9635057615116238,
|
| 1256 |
+
"epoch": 1.44,
|
| 1257 |
+
"grad_norm": 15.4375,
|
| 1258 |
+
"learning_rate": 1.5562962962962965e-05,
|
| 1259 |
+
"loss": 15.9192,
|
| 1260 |
+
"mean_token_accuracy": 0.7802809584885836,
|
| 1261 |
+
"num_tokens": 29789253.0,
|
| 1262 |
+
"step": 900
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 1.44,
|
| 1266 |
+
"eval_biology_entropy": 2.8437174711227415,
|
| 1267 |
+
"eval_biology_loss": 3.318575143814087,
|
| 1268 |
+
"eval_biology_mean_token_accuracy": 0.4861623268127441,
|
| 1269 |
+
"eval_biology_num_tokens": 29789253.0,
|
| 1270 |
+
"eval_biology_runtime": 56.4644,
|
| 1271 |
+
"eval_biology_samples_per_second": 8.855,
|
| 1272 |
+
"eval_biology_steps_per_second": 2.214,
|
| 1273 |
+
"step": 900
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 1.44,
|
| 1277 |
+
"eval_chemistry_entropy": 1.420595220565796,
|
| 1278 |
+
"eval_chemistry_loss": 1.5160387754440308,
|
| 1279 |
+
"eval_chemistry_mean_token_accuracy": 0.7028642320632934,
|
| 1280 |
+
"eval_chemistry_num_tokens": 29789253.0,
|
| 1281 |
+
"eval_chemistry_runtime": 69.7177,
|
| 1282 |
+
"eval_chemistry_samples_per_second": 7.172,
|
| 1283 |
+
"eval_chemistry_steps_per_second": 1.793,
|
| 1284 |
+
"step": 900
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.44,
|
| 1288 |
+
"eval_math_entropy": 1.0116905975341797,
|
| 1289 |
+
"eval_math_loss": 1.2779794931411743,
|
| 1290 |
+
"eval_math_mean_token_accuracy": 0.7384266419410705,
|
| 1291 |
+
"eval_math_num_tokens": 29789253.0,
|
| 1292 |
+
"eval_math_runtime": 71.6683,
|
| 1293 |
+
"eval_math_samples_per_second": 6.977,
|
| 1294 |
+
"eval_math_steps_per_second": 1.744,
|
| 1295 |
+
"step": 900
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 1.44,
|
| 1299 |
+
"eval_physics_entropy": 0.9905482964515686,
|
| 1300 |
+
"eval_physics_loss": 1.0435165166854858,
|
| 1301 |
+
"eval_physics_mean_token_accuracy": 0.770782280921936,
|
| 1302 |
+
"eval_physics_num_tokens": 29789253.0,
|
| 1303 |
+
"eval_physics_runtime": 83.2228,
|
| 1304 |
+
"eval_physics_samples_per_second": 6.008,
|
| 1305 |
+
"eval_physics_steps_per_second": 1.502,
|
| 1306 |
+
"step": 900
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"entropy": 0.9581953696906567,
|
| 1310 |
+
"epoch": 1.456,
|
| 1311 |
+
"grad_norm": 13.1875,
|
| 1312 |
+
"learning_rate": 1.548888888888889e-05,
|
| 1313 |
+
"loss": 15.9515,
|
| 1314 |
+
"mean_token_accuracy": 0.7805543716996908,
|
| 1315 |
+
"num_tokens": 30128576.0,
|
| 1316 |
+
"step": 910
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"entropy": 1.0345965854823589,
|
| 1320 |
+
"epoch": 1.472,
|
| 1321 |
+
"grad_norm": 17.125,
|
| 1322 |
+
"learning_rate": 1.5414814814814814e-05,
|
| 1323 |
+
"loss": 17.1026,
|
| 1324 |
+
"mean_token_accuracy": 0.7668049655854702,
|
| 1325 |
+
"num_tokens": 30452719.0,
|
| 1326 |
+
"step": 920
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"entropy": 1.0023833878338337,
|
| 1330 |
+
"epoch": 1.488,
|
| 1331 |
+
"grad_norm": 13.8125,
|
| 1332 |
+
"learning_rate": 1.5340740740740744e-05,
|
| 1333 |
+
"loss": 16.5595,
|
| 1334 |
+
"mean_token_accuracy": 0.7728984329849482,
|
| 1335 |
+
"num_tokens": 30782343.0,
|
| 1336 |
+
"step": 930
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"entropy": 0.9856885317713022,
|
| 1340 |
+
"epoch": 1.504,
|
| 1341 |
+
"grad_norm": 14.1875,
|
| 1342 |
+
"learning_rate": 1.5266666666666667e-05,
|
| 1343 |
+
"loss": 16.267,
|
| 1344 |
+
"mean_token_accuracy": 0.7769489102065563,
|
| 1345 |
+
"num_tokens": 31113238.0,
|
| 1346 |
+
"step": 940
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"entropy": 1.0002378012984992,
|
| 1350 |
+
"epoch": 1.52,
|
| 1351 |
+
"grad_norm": 15.4375,
|
| 1352 |
+
"learning_rate": 1.5192592592592594e-05,
|
| 1353 |
+
"loss": 16.5956,
|
| 1354 |
+
"mean_token_accuracy": 0.7721422206610441,
|
| 1355 |
+
"num_tokens": 31444589.0,
|
| 1356 |
+
"step": 950
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"entropy": 1.0214108165353537,
|
| 1360 |
+
"epoch": 1.536,
|
| 1361 |
+
"grad_norm": 17.625,
|
| 1362 |
+
"learning_rate": 1.5118518518518519e-05,
|
| 1363 |
+
"loss": 16.8896,
|
| 1364 |
+
"mean_token_accuracy": 0.7694170981645584,
|
| 1365 |
+
"num_tokens": 31774137.0,
|
| 1366 |
+
"step": 960
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"entropy": 0.9722279075533151,
|
| 1370 |
+
"epoch": 1.552,
|
| 1371 |
+
"grad_norm": 16.375,
|
| 1372 |
+
"learning_rate": 1.5044444444444445e-05,
|
| 1373 |
+
"loss": 16.0224,
|
| 1374 |
+
"mean_token_accuracy": 0.7783945810049773,
|
| 1375 |
+
"num_tokens": 32101776.0,
|
| 1376 |
+
"step": 970
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"entropy": 0.9741225987672806,
|
| 1380 |
+
"epoch": 1.568,
|
| 1381 |
+
"grad_norm": 15.25,
|
| 1382 |
+
"learning_rate": 1.497037037037037e-05,
|
| 1383 |
+
"loss": 16.145,
|
| 1384 |
+
"mean_token_accuracy": 0.778332832083106,
|
| 1385 |
+
"num_tokens": 32428021.0,
|
| 1386 |
+
"step": 980
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"entropy": 1.0168811423704027,
|
| 1390 |
+
"epoch": 1.584,
|
| 1391 |
+
"grad_norm": 16.5,
|
| 1392 |
+
"learning_rate": 1.4896296296296298e-05,
|
| 1393 |
+
"loss": 16.8197,
|
| 1394 |
+
"mean_token_accuracy": 0.7688806146383286,
|
| 1395 |
+
"num_tokens": 32752735.0,
|
| 1396 |
+
"step": 990
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"entropy": 1.0012462731450795,
|
| 1400 |
+
"epoch": 1.6,
|
| 1401 |
+
"grad_norm": 14.625,
|
| 1402 |
+
"learning_rate": 1.4822222222222225e-05,
|
| 1403 |
+
"loss": 16.6785,
|
| 1404 |
+
"mean_token_accuracy": 0.7718310851603747,
|
| 1405 |
+
"num_tokens": 33075822.0,
|
| 1406 |
+
"step": 1000
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 1.6,
|
| 1410 |
+
"eval_biology_entropy": 2.871693316459656,
|
| 1411 |
+
"eval_biology_loss": 3.2952377796173096,
|
| 1412 |
+
"eval_biology_mean_token_accuracy": 0.48806494069099426,
|
| 1413 |
+
"eval_biology_num_tokens": 33075822.0,
|
| 1414 |
+
"eval_biology_runtime": 56.2343,
|
| 1415 |
+
"eval_biology_samples_per_second": 8.891,
|
| 1416 |
+
"eval_biology_steps_per_second": 2.223,
|
| 1417 |
+
"step": 1000
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 1.6,
|
| 1421 |
+
"eval_chemistry_entropy": 1.435364501476288,
|
| 1422 |
+
"eval_chemistry_loss": 1.5101022720336914,
|
| 1423 |
+
"eval_chemistry_mean_token_accuracy": 0.7037839899063111,
|
| 1424 |
+
"eval_chemistry_num_tokens": 33075822.0,
|
| 1425 |
+
"eval_chemistry_runtime": 69.2005,
|
| 1426 |
+
"eval_chemistry_samples_per_second": 7.225,
|
| 1427 |
+
"eval_chemistry_steps_per_second": 1.806,
|
| 1428 |
+
"step": 1000
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 1.6,
|
| 1432 |
+
"eval_math_entropy": 1.0202706332206726,
|
| 1433 |
+
"eval_math_loss": 1.2738072872161865,
|
| 1434 |
+
"eval_math_mean_token_accuracy": 0.7389369897842407,
|
| 1435 |
+
"eval_math_num_tokens": 33075822.0,
|
| 1436 |
+
"eval_math_runtime": 71.9142,
|
| 1437 |
+
"eval_math_samples_per_second": 6.953,
|
| 1438 |
+
"eval_math_steps_per_second": 1.738,
|
| 1439 |
+
"step": 1000
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 1.6,
|
| 1443 |
+
"eval_physics_entropy": 0.9962421131134033,
|
| 1444 |
+
"eval_physics_loss": 1.0371240377426147,
|
| 1445 |
+
"eval_physics_mean_token_accuracy": 0.7717792973518371,
|
| 1446 |
+
"eval_physics_num_tokens": 33075822.0,
|
| 1447 |
+
"eval_physics_runtime": 83.5174,
|
| 1448 |
+
"eval_physics_samples_per_second": 5.987,
|
| 1449 |
+
"eval_physics_steps_per_second": 1.497,
|
| 1450 |
+
"step": 1000
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"entropy": 0.9602924525737763,
|
| 1454 |
+
"epoch": 1.616,
|
| 1455 |
+
"grad_norm": 16.0,
|
| 1456 |
+
"learning_rate": 1.474814814814815e-05,
|
| 1457 |
+
"loss": 15.8747,
|
| 1458 |
+
"mean_token_accuracy": 0.7811841100454331,
|
| 1459 |
+
"num_tokens": 33411850.0,
|
| 1460 |
+
"step": 1010
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"entropy": 0.984598821774125,
|
| 1464 |
+
"epoch": 1.6320000000000001,
|
| 1465 |
+
"grad_norm": 15.8125,
|
| 1466 |
+
"learning_rate": 1.4674074074074076e-05,
|
| 1467 |
+
"loss": 16.2686,
|
| 1468 |
+
"mean_token_accuracy": 0.7774637959897518,
|
| 1469 |
+
"num_tokens": 33739456.0,
|
| 1470 |
+
"step": 1020
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"entropy": 0.9493789797648787,
|
| 1474 |
+
"epoch": 1.6480000000000001,
|
| 1475 |
+
"grad_norm": 15.875,
|
| 1476 |
+
"learning_rate": 1.46e-05,
|
| 1477 |
+
"loss": 15.7615,
|
| 1478 |
+
"mean_token_accuracy": 0.781728545576334,
|
| 1479 |
+
"num_tokens": 34072022.0,
|
| 1480 |
+
"step": 1030
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"entropy": 0.9813297174870967,
|
| 1484 |
+
"epoch": 1.6640000000000001,
|
| 1485 |
+
"grad_norm": 18.0,
|
| 1486 |
+
"learning_rate": 1.4525925925925927e-05,
|
| 1487 |
+
"loss": 16.2623,
|
| 1488 |
+
"mean_token_accuracy": 0.7778959140181542,
|
| 1489 |
+
"num_tokens": 34398959.0,
|
| 1490 |
+
"step": 1040
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"entropy": 0.9663880743086338,
|
| 1494 |
+
"epoch": 1.6800000000000002,
|
| 1495 |
+
"grad_norm": 14.125,
|
| 1496 |
+
"learning_rate": 1.4451851851851852e-05,
|
| 1497 |
+
"loss": 15.898,
|
| 1498 |
+
"mean_token_accuracy": 0.7800997123122215,
|
| 1499 |
+
"num_tokens": 34732284.0,
|
| 1500 |
+
"step": 1050
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"entropy": 0.9691609870642424,
|
| 1504 |
+
"epoch": 1.696,
|
| 1505 |
+
"grad_norm": 17.375,
|
| 1506 |
+
"learning_rate": 1.4377777777777779e-05,
|
| 1507 |
+
"loss": 16.1995,
|
| 1508 |
+
"mean_token_accuracy": 0.7772838160395622,
|
| 1509 |
+
"num_tokens": 35053712.0,
|
| 1510 |
+
"step": 1060
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"entropy": 1.0014944590628148,
|
| 1514 |
+
"epoch": 1.712,
|
| 1515 |
+
"grad_norm": 19.0,
|
| 1516 |
+
"learning_rate": 1.4303703703703703e-05,
|
| 1517 |
+
"loss": 16.4625,
|
| 1518 |
+
"mean_token_accuracy": 0.773622689396143,
|
| 1519 |
+
"num_tokens": 35393088.0,
|
| 1520 |
+
"step": 1070
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"entropy": 0.9721997212618589,
|
| 1524 |
+
"epoch": 1.728,
|
| 1525 |
+
"grad_norm": 16.125,
|
| 1526 |
+
"learning_rate": 1.4229629629629632e-05,
|
| 1527 |
+
"loss": 16.1251,
|
| 1528 |
+
"mean_token_accuracy": 0.7777348395437003,
|
| 1529 |
+
"num_tokens": 35729007.0,
|
| 1530 |
+
"step": 1080
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"entropy": 0.9720193948596716,
|
| 1534 |
+
"epoch": 1.744,
|
| 1535 |
+
"grad_norm": 16.875,
|
| 1536 |
+
"learning_rate": 1.4155555555555556e-05,
|
| 1537 |
+
"loss": 16.0859,
|
| 1538 |
+
"mean_token_accuracy": 0.7778170190751552,
|
| 1539 |
+
"num_tokens": 36072489.0,
|
| 1540 |
+
"step": 1090
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"entropy": 0.9743611078709364,
|
| 1544 |
+
"epoch": 1.76,
|
| 1545 |
+
"grad_norm": 15.0,
|
| 1546 |
+
"learning_rate": 1.4081481481481483e-05,
|
| 1547 |
+
"loss": 16.0758,
|
| 1548 |
+
"mean_token_accuracy": 0.7777593333274126,
|
| 1549 |
+
"num_tokens": 36398285.0,
|
| 1550 |
+
"step": 1100
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 1.76,
|
| 1554 |
+
"eval_biology_entropy": 2.836181663513184,
|
| 1555 |
+
"eval_biology_loss": 3.3000376224517822,
|
| 1556 |
+
"eval_biology_mean_token_accuracy": 0.4904142751693726,
|
| 1557 |
+
"eval_biology_num_tokens": 36398285.0,
|
| 1558 |
+
"eval_biology_runtime": 56.6192,
|
| 1559 |
+
"eval_biology_samples_per_second": 8.831,
|
| 1560 |
+
"eval_biology_steps_per_second": 2.208,
|
| 1561 |
+
"step": 1100
|
| 1562 |
+
},
|
| 1563 |
+
{
|
| 1564 |
+
"epoch": 1.76,
|
| 1565 |
+
"eval_chemistry_entropy": 1.4143978090286256,
|
| 1566 |
+
"eval_chemistry_loss": 1.5046298503875732,
|
| 1567 |
+
"eval_chemistry_mean_token_accuracy": 0.705154767036438,
|
| 1568 |
+
"eval_chemistry_num_tokens": 36398285.0,
|
| 1569 |
+
"eval_chemistry_runtime": 69.8363,
|
| 1570 |
+
"eval_chemistry_samples_per_second": 7.16,
|
| 1571 |
+
"eval_chemistry_steps_per_second": 1.79,
|
| 1572 |
+
"step": 1100
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 1.76,
|
| 1576 |
+
"eval_math_entropy": 1.0144752063751221,
|
| 1577 |
+
"eval_math_loss": 1.270599603652954,
|
| 1578 |
+
"eval_math_mean_token_accuracy": 0.7394923534393311,
|
| 1579 |
+
"eval_math_num_tokens": 36398285.0,
|
| 1580 |
+
"eval_math_runtime": 71.8526,
|
| 1581 |
+
"eval_math_samples_per_second": 6.959,
|
| 1582 |
+
"eval_math_steps_per_second": 1.74,
|
| 1583 |
+
"step": 1100
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 1.76,
|
| 1587 |
+
"eval_physics_entropy": 0.9861182165145874,
|
| 1588 |
+
"eval_physics_loss": 1.0319875478744507,
|
| 1589 |
+
"eval_physics_mean_token_accuracy": 0.772673840045929,
|
| 1590 |
+
"eval_physics_num_tokens": 36398285.0,
|
| 1591 |
+
"eval_physics_runtime": 82.9698,
|
| 1592 |
+
"eval_physics_samples_per_second": 6.026,
|
| 1593 |
+
"eval_physics_steps_per_second": 1.507,
|
| 1594 |
+
"step": 1100
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"entropy": 1.0112663459032774,
|
| 1598 |
+
"epoch": 1.776,
|
| 1599 |
+
"grad_norm": 15.5,
|
| 1600 |
+
"learning_rate": 1.400740740740741e-05,
|
| 1601 |
+
"loss": 16.7726,
|
| 1602 |
+
"mean_token_accuracy": 0.7704095777124167,
|
| 1603 |
+
"num_tokens": 36730479.0,
|
| 1604 |
+
"step": 1110
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"entropy": 0.9648675471544266,
|
| 1608 |
+
"epoch": 1.792,
|
| 1609 |
+
"grad_norm": 18.0,
|
| 1610 |
+
"learning_rate": 1.3933333333333334e-05,
|
| 1611 |
+
"loss": 16.0369,
|
| 1612 |
+
"mean_token_accuracy": 0.7790144141763449,
|
| 1613 |
+
"num_tokens": 37057425.0,
|
| 1614 |
+
"step": 1120
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"entropy": 0.9587565205991269,
|
| 1618 |
+
"epoch": 1.808,
|
| 1619 |
+
"grad_norm": 14.8125,
|
| 1620 |
+
"learning_rate": 1.385925925925926e-05,
|
| 1621 |
+
"loss": 15.7551,
|
| 1622 |
+
"mean_token_accuracy": 0.7815911706537009,
|
| 1623 |
+
"num_tokens": 37395449.0,
|
| 1624 |
+
"step": 1130
|
| 1625 |
+
},
|
| 1626 |
+
{
|
| 1627 |
+
"entropy": 0.9664028909057378,
|
| 1628 |
+
"epoch": 1.8239999999999998,
|
| 1629 |
+
"grad_norm": 15.1875,
|
| 1630 |
+
"learning_rate": 1.3785185185185186e-05,
|
| 1631 |
+
"loss": 16.0738,
|
| 1632 |
+
"mean_token_accuracy": 0.7792270630598068,
|
| 1633 |
+
"num_tokens": 37729535.0,
|
| 1634 |
+
"step": 1140
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"entropy": 0.9699144329875707,
|
| 1638 |
+
"epoch": 1.8399999999999999,
|
| 1639 |
+
"grad_norm": 16.125,
|
| 1640 |
+
"learning_rate": 1.3711111111111112e-05,
|
| 1641 |
+
"loss": 15.9827,
|
| 1642 |
+
"mean_token_accuracy": 0.7795627668499947,
|
| 1643 |
+
"num_tokens": 38057839.0,
|
| 1644 |
+
"step": 1150
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"entropy": 0.9710695983842015,
|
| 1648 |
+
"epoch": 1.8559999999999999,
|
| 1649 |
+
"grad_norm": 17.75,
|
| 1650 |
+
"learning_rate": 1.3637037037037037e-05,
|
| 1651 |
+
"loss": 16.0651,
|
| 1652 |
+
"mean_token_accuracy": 0.7783871240913868,
|
| 1653 |
+
"num_tokens": 38398033.0,
|
| 1654 |
+
"step": 1160
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"entropy": 0.9714042950421572,
|
| 1658 |
+
"epoch": 1.8719999999999999,
|
| 1659 |
+
"grad_norm": 14.5,
|
| 1660 |
+
"learning_rate": 1.3562962962962965e-05,
|
| 1661 |
+
"loss": 16.0555,
|
| 1662 |
+
"mean_token_accuracy": 0.7771369840949773,
|
| 1663 |
+
"num_tokens": 38729203.0,
|
| 1664 |
+
"step": 1170
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"entropy": 0.9900001417845488,
|
| 1668 |
+
"epoch": 1.888,
|
| 1669 |
+
"grad_norm": 16.25,
|
| 1670 |
+
"learning_rate": 1.3488888888888888e-05,
|
| 1671 |
+
"loss": 16.3823,
|
| 1672 |
+
"mean_token_accuracy": 0.775692193210125,
|
| 1673 |
+
"num_tokens": 39058791.0,
|
| 1674 |
+
"step": 1180
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"entropy": 1.0028734058141708,
|
| 1678 |
+
"epoch": 1.904,
|
| 1679 |
+
"grad_norm": 15.625,
|
| 1680 |
+
"learning_rate": 1.3414814814814817e-05,
|
| 1681 |
+
"loss": 16.6676,
|
| 1682 |
+
"mean_token_accuracy": 0.772062387317419,
|
| 1683 |
+
"num_tokens": 39386170.0,
|
| 1684 |
+
"step": 1190
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"entropy": 1.0105124205350875,
|
| 1688 |
+
"epoch": 1.92,
|
| 1689 |
+
"grad_norm": 15.75,
|
| 1690 |
+
"learning_rate": 1.3340740740740741e-05,
|
| 1691 |
+
"loss": 16.6591,
|
| 1692 |
+
"mean_token_accuracy": 0.7720069587230682,
|
| 1693 |
+
"num_tokens": 39710381.0,
|
| 1694 |
+
"step": 1200
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 1.92,
|
| 1698 |
+
"eval_biology_entropy": 2.8252229404449465,
|
| 1699 |
+
"eval_biology_loss": 3.293790102005005,
|
| 1700 |
+
"eval_biology_mean_token_accuracy": 0.49097717118263245,
|
| 1701 |
+
"eval_biology_num_tokens": 39710381.0,
|
| 1702 |
+
"eval_biology_runtime": 56.3872,
|
| 1703 |
+
"eval_biology_samples_per_second": 8.867,
|
| 1704 |
+
"eval_biology_steps_per_second": 2.217,
|
| 1705 |
+
"step": 1200
|
| 1706 |
+
},
|
| 1707 |
+
{
|
| 1708 |
+
"epoch": 1.92,
|
| 1709 |
+
"eval_chemistry_entropy": 1.4020655665397643,
|
| 1710 |
+
"eval_chemistry_loss": 1.4992702007293701,
|
| 1711 |
+
"eval_chemistry_mean_token_accuracy": 0.7056584782600402,
|
| 1712 |
+
"eval_chemistry_num_tokens": 39710381.0,
|
| 1713 |
+
"eval_chemistry_runtime": 69.4771,
|
| 1714 |
+
"eval_chemistry_samples_per_second": 7.197,
|
| 1715 |
+
"eval_chemistry_steps_per_second": 1.799,
|
| 1716 |
+
"step": 1200
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 1.92,
|
| 1720 |
+
"eval_math_entropy": 1.0065196805000305,
|
| 1721 |
+
"eval_math_loss": 1.2684106826782227,
|
| 1722 |
+
"eval_math_mean_token_accuracy": 0.7399292430877685,
|
| 1723 |
+
"eval_math_num_tokens": 39710381.0,
|
| 1724 |
+
"eval_math_runtime": 71.8266,
|
| 1725 |
+
"eval_math_samples_per_second": 6.961,
|
| 1726 |
+
"eval_math_steps_per_second": 1.74,
|
| 1727 |
+
"step": 1200
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 1.92,
|
| 1731 |
+
"eval_physics_entropy": 0.9763948659896851,
|
| 1732 |
+
"eval_physics_loss": 1.0278587341308594,
|
| 1733 |
+
"eval_physics_mean_token_accuracy": 0.7729829773902893,
|
| 1734 |
+
"eval_physics_num_tokens": 39710381.0,
|
| 1735 |
+
"eval_physics_runtime": 82.9547,
|
| 1736 |
+
"eval_physics_samples_per_second": 6.027,
|
| 1737 |
+
"eval_physics_steps_per_second": 1.507,
|
| 1738 |
+
"step": 1200
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"entropy": 0.972571362182498,
|
| 1742 |
+
"epoch": 1.936,
|
| 1743 |
+
"grad_norm": 16.5,
|
| 1744 |
+
"learning_rate": 1.3266666666666668e-05,
|
| 1745 |
+
"loss": 16.1763,
|
| 1746 |
+
"mean_token_accuracy": 0.7767009306699038,
|
| 1747 |
+
"num_tokens": 40043354.0,
|
| 1748 |
+
"step": 1210
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"entropy": 0.926283460110426,
|
| 1752 |
+
"epoch": 1.952,
|
| 1753 |
+
"grad_norm": 14.5,
|
| 1754 |
+
"learning_rate": 1.3192592592592594e-05,
|
| 1755 |
+
"loss": 15.2868,
|
| 1756 |
+
"mean_token_accuracy": 0.7873167935758829,
|
| 1757 |
+
"num_tokens": 40375790.0,
|
| 1758 |
+
"step": 1220
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"entropy": 0.9751455634832382,
|
| 1762 |
+
"epoch": 1.968,
|
| 1763 |
+
"grad_norm": 15.5625,
|
| 1764 |
+
"learning_rate": 1.311851851851852e-05,
|
| 1765 |
+
"loss": 16.1496,
|
| 1766 |
+
"mean_token_accuracy": 0.7774738963693381,
|
| 1767 |
+
"num_tokens": 40705451.0,
|
| 1768 |
+
"step": 1230
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"entropy": 0.995421402156353,
|
| 1772 |
+
"epoch": 1.984,
|
| 1773 |
+
"grad_norm": 17.375,
|
| 1774 |
+
"learning_rate": 1.3044444444444446e-05,
|
| 1775 |
+
"loss": 16.443,
|
| 1776 |
+
"mean_token_accuracy": 0.7746346723288298,
|
| 1777 |
+
"num_tokens": 41040313.0,
|
| 1778 |
+
"step": 1240
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"entropy": 0.9862227950245142,
|
| 1782 |
+
"epoch": 2.0,
|
| 1783 |
+
"grad_norm": 15.9375,
|
| 1784 |
+
"learning_rate": 1.297037037037037e-05,
|
| 1785 |
+
"loss": 16.3211,
|
| 1786 |
+
"mean_token_accuracy": 0.7764228660613298,
|
| 1787 |
+
"num_tokens": 41373608.0,
|
| 1788 |
+
"step": 1250
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"entropy": 0.9604721080511809,
|
| 1792 |
+
"epoch": 2.016,
|
| 1793 |
+
"grad_norm": 14.0,
|
| 1794 |
+
"learning_rate": 1.2896296296296299e-05,
|
| 1795 |
+
"loss": 15.7374,
|
| 1796 |
+
"mean_token_accuracy": 0.7811667166650296,
|
| 1797 |
+
"num_tokens": 41708226.0,
|
| 1798 |
+
"step": 1260
|
| 1799 |
+
},
|
| 1800 |
+
{
|
| 1801 |
+
"entropy": 0.9497564036399126,
|
| 1802 |
+
"epoch": 2.032,
|
| 1803 |
+
"grad_norm": 14.4375,
|
| 1804 |
+
"learning_rate": 1.2822222222222222e-05,
|
| 1805 |
+
"loss": 15.7967,
|
| 1806 |
+
"mean_token_accuracy": 0.7820551857352257,
|
| 1807 |
+
"num_tokens": 42045433.0,
|
| 1808 |
+
"step": 1270
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"entropy": 0.9955548478290439,
|
| 1812 |
+
"epoch": 2.048,
|
| 1813 |
+
"grad_norm": 16.375,
|
| 1814 |
+
"learning_rate": 1.274814814814815e-05,
|
| 1815 |
+
"loss": 16.499,
|
| 1816 |
+
"mean_token_accuracy": 0.772939319908619,
|
| 1817 |
+
"num_tokens": 42372694.0,
|
| 1818 |
+
"step": 1280
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"entropy": 0.9601335551589727,
|
| 1822 |
+
"epoch": 2.064,
|
| 1823 |
+
"grad_norm": 15.125,
|
| 1824 |
+
"learning_rate": 1.2674074074074075e-05,
|
| 1825 |
+
"loss": 15.7713,
|
| 1826 |
+
"mean_token_accuracy": 0.7813012763857842,
|
| 1827 |
+
"num_tokens": 42710240.0,
|
| 1828 |
+
"step": 1290
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"entropy": 0.9750869631767273,
|
| 1832 |
+
"epoch": 2.08,
|
| 1833 |
+
"grad_norm": 15.0625,
|
| 1834 |
+
"learning_rate": 1.2600000000000001e-05,
|
| 1835 |
+
"loss": 16.2017,
|
| 1836 |
+
"mean_token_accuracy": 0.7777177035808563,
|
| 1837 |
+
"num_tokens": 43037844.0,
|
| 1838 |
+
"step": 1300
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 2.08,
|
| 1842 |
+
"eval_biology_entropy": 2.8672749757766725,
|
| 1843 |
+
"eval_biology_loss": 3.2819793224334717,
|
| 1844 |
+
"eval_biology_mean_token_accuracy": 0.4901486375331879,
|
| 1845 |
+
"eval_biology_num_tokens": 43037844.0,
|
| 1846 |
+
"eval_biology_runtime": 56.3377,
|
| 1847 |
+
"eval_biology_samples_per_second": 8.875,
|
| 1848 |
+
"eval_biology_steps_per_second": 2.219,
|
| 1849 |
+
"step": 1300
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 2.08,
|
| 1853 |
+
"eval_chemistry_entropy": 1.4090923619270326,
|
| 1854 |
+
"eval_chemistry_loss": 1.4964148998260498,
|
| 1855 |
+
"eval_chemistry_mean_token_accuracy": 0.7058250217437744,
|
| 1856 |
+
"eval_chemistry_num_tokens": 43037844.0,
|
| 1857 |
+
"eval_chemistry_runtime": 69.503,
|
| 1858 |
+
"eval_chemistry_samples_per_second": 7.194,
|
| 1859 |
+
"eval_chemistry_steps_per_second": 1.798,
|
| 1860 |
+
"step": 1300
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 2.08,
|
| 1864 |
+
"eval_math_entropy": 1.0088651123046875,
|
| 1865 |
+
"eval_math_loss": 1.2671351432800293,
|
| 1866 |
+
"eval_math_mean_token_accuracy": 0.7401549768447876,
|
| 1867 |
+
"eval_math_num_tokens": 43037844.0,
|
| 1868 |
+
"eval_math_runtime": 71.5652,
|
| 1869 |
+
"eval_math_samples_per_second": 6.987,
|
| 1870 |
+
"eval_math_steps_per_second": 1.747,
|
| 1871 |
+
"step": 1300
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 2.08,
|
| 1875 |
+
"eval_physics_entropy": 0.9760756931304931,
|
| 1876 |
+
"eval_physics_loss": 1.0242512226104736,
|
| 1877 |
+
"eval_physics_mean_token_accuracy": 0.7737720923423768,
|
| 1878 |
+
"eval_physics_num_tokens": 43037844.0,
|
| 1879 |
+
"eval_physics_runtime": 83.4322,
|
| 1880 |
+
"eval_physics_samples_per_second": 5.993,
|
| 1881 |
+
"eval_physics_steps_per_second": 1.498,
|
| 1882 |
+
"step": 1300
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"entropy": 0.9748470045626163,
|
| 1886 |
+
"epoch": 2.096,
|
| 1887 |
+
"grad_norm": 17.0,
|
| 1888 |
+
"learning_rate": 1.2525925925925928e-05,
|
| 1889 |
+
"loss": 16.1191,
|
| 1890 |
+
"mean_token_accuracy": 0.7778720472007989,
|
| 1891 |
+
"num_tokens": 43368675.0,
|
| 1892 |
+
"step": 1310
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"entropy": 0.9635545436292887,
|
| 1896 |
+
"epoch": 2.112,
|
| 1897 |
+
"grad_norm": 15.4375,
|
| 1898 |
+
"learning_rate": 1.2451851851851853e-05,
|
| 1899 |
+
"loss": 15.9131,
|
| 1900 |
+
"mean_token_accuracy": 0.7799594800919294,
|
| 1901 |
+
"num_tokens": 43699544.0,
|
| 1902 |
+
"step": 1320
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"entropy": 0.9748049929738045,
|
| 1906 |
+
"epoch": 2.128,
|
| 1907 |
+
"grad_norm": 16.375,
|
| 1908 |
+
"learning_rate": 1.237777777777778e-05,
|
| 1909 |
+
"loss": 16.1106,
|
| 1910 |
+
"mean_token_accuracy": 0.7764891773462296,
|
| 1911 |
+
"num_tokens": 44012632.0,
|
| 1912 |
+
"step": 1330
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"entropy": 0.9570101447403431,
|
| 1916 |
+
"epoch": 2.144,
|
| 1917 |
+
"grad_norm": 14.25,
|
| 1918 |
+
"learning_rate": 1.2303703703703704e-05,
|
| 1919 |
+
"loss": 15.8491,
|
| 1920 |
+
"mean_token_accuracy": 0.7784212082624435,
|
| 1921 |
+
"num_tokens": 44349732.0,
|
| 1922 |
+
"step": 1340
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"entropy": 0.9998667161911726,
|
| 1926 |
+
"epoch": 2.16,
|
| 1927 |
+
"grad_norm": 16.5,
|
| 1928 |
+
"learning_rate": 1.222962962962963e-05,
|
| 1929 |
+
"loss": 16.5269,
|
| 1930 |
+
"mean_token_accuracy": 0.7714359667152166,
|
| 1931 |
+
"num_tokens": 44680410.0,
|
| 1932 |
+
"step": 1350
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"entropy": 0.95210868306458,
|
| 1936 |
+
"epoch": 2.176,
|
| 1937 |
+
"grad_norm": 14.125,
|
| 1938 |
+
"learning_rate": 1.2155555555555555e-05,
|
| 1939 |
+
"loss": 15.7683,
|
| 1940 |
+
"mean_token_accuracy": 0.7823099143803119,
|
| 1941 |
+
"num_tokens": 45017922.0,
|
| 1942 |
+
"step": 1360
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"entropy": 0.962488704174757,
|
| 1946 |
+
"epoch": 2.192,
|
| 1947 |
+
"grad_norm": 15.9375,
|
| 1948 |
+
"learning_rate": 1.2081481481481484e-05,
|
| 1949 |
+
"loss": 15.7913,
|
| 1950 |
+
"mean_token_accuracy": 0.7814613375812769,
|
| 1951 |
+
"num_tokens": 45348179.0,
|
| 1952 |
+
"step": 1370
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"entropy": 0.9470339801162482,
|
| 1956 |
+
"epoch": 2.208,
|
| 1957 |
+
"grad_norm": 16.125,
|
| 1958 |
+
"learning_rate": 1.2007407407407408e-05,
|
| 1959 |
+
"loss": 15.7356,
|
| 1960 |
+
"mean_token_accuracy": 0.781909292936325,
|
| 1961 |
+
"num_tokens": 45678397.0,
|
| 1962 |
+
"step": 1380
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"entropy": 0.9522435653954744,
|
| 1966 |
+
"epoch": 2.224,
|
| 1967 |
+
"grad_norm": 14.875,
|
| 1968 |
+
"learning_rate": 1.1933333333333335e-05,
|
| 1969 |
+
"loss": 15.802,
|
| 1970 |
+
"mean_token_accuracy": 0.7811244171112776,
|
| 1971 |
+
"num_tokens": 46010052.0,
|
| 1972 |
+
"step": 1390
|
| 1973 |
+
},
|
| 1974 |
+
{
|
| 1975 |
+
"entropy": 0.9532753810286522,
|
| 1976 |
+
"epoch": 2.24,
|
| 1977 |
+
"grad_norm": 14.25,
|
| 1978 |
+
"learning_rate": 1.185925925925926e-05,
|
| 1979 |
+
"loss": 15.7228,
|
| 1980 |
+
"mean_token_accuracy": 0.7819092500954866,
|
| 1981 |
+
"num_tokens": 46347116.0,
|
| 1982 |
+
"step": 1400
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 2.24,
|
| 1986 |
+
"eval_biology_entropy": 2.8466386623382567,
|
| 1987 |
+
"eval_biology_loss": 3.285724401473999,
|
| 1988 |
+
"eval_biology_mean_token_accuracy": 0.4921459174156189,
|
| 1989 |
+
"eval_biology_num_tokens": 46347116.0,
|
| 1990 |
+
"eval_biology_runtime": 56.3045,
|
| 1991 |
+
"eval_biology_samples_per_second": 8.88,
|
| 1992 |
+
"eval_biology_steps_per_second": 2.22,
|
| 1993 |
+
"step": 1400
|
| 1994 |
+
},
|
| 1995 |
+
{
|
| 1996 |
+
"epoch": 2.24,
|
| 1997 |
+
"eval_chemistry_entropy": 1.41047416639328,
|
| 1998 |
+
"eval_chemistry_loss": 1.4948508739471436,
|
| 1999 |
+
"eval_chemistry_mean_token_accuracy": 0.7069983949661255,
|
| 2000 |
+
"eval_chemistry_num_tokens": 46347116.0,
|
| 2001 |
+
"eval_chemistry_runtime": 69.496,
|
| 2002 |
+
"eval_chemistry_samples_per_second": 7.195,
|
| 2003 |
+
"eval_chemistry_steps_per_second": 1.799,
|
| 2004 |
+
"step": 1400
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 2.24,
|
| 2008 |
+
"eval_math_entropy": 1.0100535235404968,
|
| 2009 |
+
"eval_math_loss": 1.2652060985565186,
|
| 2010 |
+
"eval_math_mean_token_accuracy": 0.7402367944717407,
|
| 2011 |
+
"eval_math_num_tokens": 46347116.0,
|
| 2012 |
+
"eval_math_runtime": 71.7995,
|
| 2013 |
+
"eval_math_samples_per_second": 6.964,
|
| 2014 |
+
"eval_math_steps_per_second": 1.741,
|
| 2015 |
+
"step": 1400
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 2.24,
|
| 2019 |
+
"eval_physics_entropy": 0.9783945083618164,
|
| 2020 |
+
"eval_physics_loss": 1.0216351747512817,
|
| 2021 |
+
"eval_physics_mean_token_accuracy": 0.7738124356269837,
|
| 2022 |
+
"eval_physics_num_tokens": 46347116.0,
|
| 2023 |
+
"eval_physics_runtime": 82.9174,
|
| 2024 |
+
"eval_physics_samples_per_second": 6.03,
|
| 2025 |
+
"eval_physics_steps_per_second": 1.508,
|
| 2026 |
+
"step": 1400
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"entropy": 1.0024078279733657,
|
| 2030 |
+
"epoch": 2.2560000000000002,
|
| 2031 |
+
"grad_norm": 16.75,
|
| 2032 |
+
"learning_rate": 1.1785185185185186e-05,
|
| 2033 |
+
"loss": 16.5504,
|
| 2034 |
+
"mean_token_accuracy": 0.7735020380467177,
|
| 2035 |
+
"num_tokens": 46673074.0,
|
| 2036 |
+
"step": 1410
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"entropy": 0.9434977924451232,
|
| 2040 |
+
"epoch": 2.2720000000000002,
|
| 2041 |
+
"grad_norm": 15.0,
|
| 2042 |
+
"learning_rate": 1.1711111111111113e-05,
|
| 2043 |
+
"loss": 15.5742,
|
| 2044 |
+
"mean_token_accuracy": 0.7847557462751865,
|
| 2045 |
+
"num_tokens": 47009386.0,
|
| 2046 |
+
"step": 1420
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"entropy": 0.9876439906656742,
|
| 2050 |
+
"epoch": 2.288,
|
| 2051 |
+
"grad_norm": 15.625,
|
| 2052 |
+
"learning_rate": 1.1637037037037037e-05,
|
| 2053 |
+
"loss": 16.3152,
|
| 2054 |
+
"mean_token_accuracy": 0.7754988320171833,
|
| 2055 |
+
"num_tokens": 47344231.0,
|
| 2056 |
+
"step": 1430
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"entropy": 0.9828802436590195,
|
| 2060 |
+
"epoch": 2.304,
|
| 2061 |
+
"grad_norm": 16.75,
|
| 2062 |
+
"learning_rate": 1.1562962962962964e-05,
|
| 2063 |
+
"loss": 16.153,
|
| 2064 |
+
"mean_token_accuracy": 0.7779621798545122,
|
| 2065 |
+
"num_tokens": 47658392.0,
|
| 2066 |
+
"step": 1440
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"entropy": 0.9514119651168584,
|
| 2070 |
+
"epoch": 2.32,
|
| 2071 |
+
"grad_norm": 16.0,
|
| 2072 |
+
"learning_rate": 1.1488888888888889e-05,
|
| 2073 |
+
"loss": 15.7986,
|
| 2074 |
+
"mean_token_accuracy": 0.7797571372240781,
|
| 2075 |
+
"num_tokens": 47995015.0,
|
| 2076 |
+
"step": 1450
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"entropy": 0.9356315284967422,
|
| 2080 |
+
"epoch": 2.336,
|
| 2081 |
+
"grad_norm": 14.6875,
|
| 2082 |
+
"learning_rate": 1.1414814814814817e-05,
|
| 2083 |
+
"loss": 15.477,
|
| 2084 |
+
"mean_token_accuracy": 0.7844580728560686,
|
| 2085 |
+
"num_tokens": 48336708.0,
|
| 2086 |
+
"step": 1460
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"entropy": 0.9371037229895591,
|
| 2090 |
+
"epoch": 2.352,
|
| 2091 |
+
"grad_norm": 14.3125,
|
| 2092 |
+
"learning_rate": 1.1340740740740742e-05,
|
| 2093 |
+
"loss": 15.4649,
|
| 2094 |
+
"mean_token_accuracy": 0.783124253898859,
|
| 2095 |
+
"num_tokens": 48665311.0,
|
| 2096 |
+
"step": 1470
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"entropy": 0.9581775091588497,
|
| 2100 |
+
"epoch": 2.368,
|
| 2101 |
+
"grad_norm": 14.9375,
|
| 2102 |
+
"learning_rate": 1.1266666666666668e-05,
|
| 2103 |
+
"loss": 15.9657,
|
| 2104 |
+
"mean_token_accuracy": 0.7788518495857716,
|
| 2105 |
+
"num_tokens": 49005334.0,
|
| 2106 |
+
"step": 1480
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"entropy": 0.9918505910784006,
|
| 2110 |
+
"epoch": 2.384,
|
| 2111 |
+
"grad_norm": 17.5,
|
| 2112 |
+
"learning_rate": 1.1192592592592593e-05,
|
| 2113 |
+
"loss": 16.1851,
|
| 2114 |
+
"mean_token_accuracy": 0.7754409562796354,
|
| 2115 |
+
"num_tokens": 49331936.0,
|
| 2116 |
+
"step": 1490
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"entropy": 0.9667624041438103,
|
| 2120 |
+
"epoch": 2.4,
|
| 2121 |
+
"grad_norm": 15.75,
|
| 2122 |
+
"learning_rate": 1.111851851851852e-05,
|
| 2123 |
+
"loss": 16.0425,
|
| 2124 |
+
"mean_token_accuracy": 0.7772644326090813,
|
| 2125 |
+
"num_tokens": 49671048.0,
|
| 2126 |
+
"step": 1500
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 2.4,
|
| 2130 |
+
"eval_biology_entropy": 2.764361233711243,
|
| 2131 |
+
"eval_biology_loss": 3.286792755126953,
|
| 2132 |
+
"eval_biology_mean_token_accuracy": 0.49071564149856567,
|
| 2133 |
+
"eval_biology_num_tokens": 49671048.0,
|
| 2134 |
+
"eval_biology_runtime": 56.1555,
|
| 2135 |
+
"eval_biology_samples_per_second": 8.904,
|
| 2136 |
+
"eval_biology_steps_per_second": 2.226,
|
| 2137 |
+
"step": 1500
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 2.4,
|
| 2141 |
+
"eval_chemistry_entropy": 1.3750218210220337,
|
| 2142 |
+
"eval_chemistry_loss": 1.4918104410171509,
|
| 2143 |
+
"eval_chemistry_mean_token_accuracy": 0.7060160474777222,
|
| 2144 |
+
"eval_chemistry_num_tokens": 49671048.0,
|
| 2145 |
+
"eval_chemistry_runtime": 69.1848,
|
| 2146 |
+
"eval_chemistry_samples_per_second": 7.227,
|
| 2147 |
+
"eval_chemistry_steps_per_second": 1.807,
|
| 2148 |
+
"step": 1500
|
| 2149 |
+
},
|
| 2150 |
+
{
|
| 2151 |
+
"epoch": 2.4,
|
| 2152 |
+
"eval_math_entropy": 0.994632875919342,
|
| 2153 |
+
"eval_math_loss": 1.264724612236023,
|
| 2154 |
+
"eval_math_mean_token_accuracy": 0.7406163611412049,
|
| 2155 |
+
"eval_math_num_tokens": 49671048.0,
|
| 2156 |
+
"eval_math_runtime": 71.7078,
|
| 2157 |
+
"eval_math_samples_per_second": 6.973,
|
| 2158 |
+
"eval_math_steps_per_second": 1.743,
|
| 2159 |
+
"step": 1500
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 2.4,
|
| 2163 |
+
"eval_physics_entropy": 0.9608834567070007,
|
| 2164 |
+
"eval_physics_loss": 1.0192519426345825,
|
| 2165 |
+
"eval_physics_mean_token_accuracy": 0.7745839128494263,
|
| 2166 |
+
"eval_physics_num_tokens": 49671048.0,
|
| 2167 |
+
"eval_physics_runtime": 82.8516,
|
| 2168 |
+
"eval_physics_samples_per_second": 6.035,
|
| 2169 |
+
"eval_physics_steps_per_second": 1.509,
|
| 2170 |
+
"step": 1500
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"entropy": 0.9463426861912012,
|
| 2174 |
+
"epoch": 2.416,
|
| 2175 |
+
"grad_norm": 14.1875,
|
| 2176 |
+
"learning_rate": 1.1044444444444444e-05,
|
| 2177 |
+
"loss": 15.6417,
|
| 2178 |
+
"mean_token_accuracy": 0.7824339278042316,
|
| 2179 |
+
"num_tokens": 50018182.0,
|
| 2180 |
+
"step": 1510
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"entropy": 0.9611309446394444,
|
| 2184 |
+
"epoch": 2.432,
|
| 2185 |
+
"grad_norm": 16.375,
|
| 2186 |
+
"learning_rate": 1.0970370370370371e-05,
|
| 2187 |
+
"loss": 15.8601,
|
| 2188 |
+
"mean_token_accuracy": 0.7820219408720732,
|
| 2189 |
+
"num_tokens": 50341187.0,
|
| 2190 |
+
"step": 1520
|
| 2191 |
+
},
|
| 2192 |
+
{
|
| 2193 |
+
"entropy": 0.9917191587388515,
|
| 2194 |
+
"epoch": 2.448,
|
| 2195 |
+
"grad_norm": 13.875,
|
| 2196 |
+
"learning_rate": 1.0896296296296298e-05,
|
| 2197 |
+
"loss": 16.3874,
|
| 2198 |
+
"mean_token_accuracy": 0.7728928077965975,
|
| 2199 |
+
"num_tokens": 50667019.0,
|
| 2200 |
+
"step": 1530
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"entropy": 0.9828215155750513,
|
| 2204 |
+
"epoch": 2.464,
|
| 2205 |
+
"grad_norm": 16.375,
|
| 2206 |
+
"learning_rate": 1.0822222222222222e-05,
|
| 2207 |
+
"loss": 16.1547,
|
| 2208 |
+
"mean_token_accuracy": 0.7776284907013178,
|
| 2209 |
+
"num_tokens": 50991113.0,
|
| 2210 |
+
"step": 1540
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"entropy": 0.9663635555654764,
|
| 2214 |
+
"epoch": 2.48,
|
| 2215 |
+
"grad_norm": 15.6875,
|
| 2216 |
+
"learning_rate": 1.074814814814815e-05,
|
| 2217 |
+
"loss": 16.0741,
|
| 2218 |
+
"mean_token_accuracy": 0.7778375498950482,
|
| 2219 |
+
"num_tokens": 51319624.0,
|
| 2220 |
+
"step": 1550
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"entropy": 0.9860180348157883,
|
| 2224 |
+
"epoch": 2.496,
|
| 2225 |
+
"grad_norm": 14.9375,
|
| 2226 |
+
"learning_rate": 1.0674074074074074e-05,
|
| 2227 |
+
"loss": 16.1857,
|
| 2228 |
+
"mean_token_accuracy": 0.7756810195744037,
|
| 2229 |
+
"num_tokens": 51654633.0,
|
| 2230 |
+
"step": 1560
|
| 2231 |
+
},
|
| 2232 |
+
{
|
| 2233 |
+
"entropy": 0.9566800069063902,
|
| 2234 |
+
"epoch": 2.512,
|
| 2235 |
+
"grad_norm": 16.875,
|
| 2236 |
+
"learning_rate": 1.0600000000000002e-05,
|
| 2237 |
+
"loss": 15.8687,
|
| 2238 |
+
"mean_token_accuracy": 0.7806913875043392,
|
| 2239 |
+
"num_tokens": 51980051.0,
|
| 2240 |
+
"step": 1570
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"entropy": 0.9594778981059789,
|
| 2244 |
+
"epoch": 2.528,
|
| 2245 |
+
"grad_norm": 14.75,
|
| 2246 |
+
"learning_rate": 1.0525925925925927e-05,
|
| 2247 |
+
"loss": 15.8059,
|
| 2248 |
+
"mean_token_accuracy": 0.7803925272077322,
|
| 2249 |
+
"num_tokens": 52316109.0,
|
| 2250 |
+
"step": 1580
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"entropy": 0.9729896225035191,
|
| 2254 |
+
"epoch": 2.544,
|
| 2255 |
+
"grad_norm": 18.125,
|
| 2256 |
+
"learning_rate": 1.0451851851851853e-05,
|
| 2257 |
+
"loss": 16.1791,
|
| 2258 |
+
"mean_token_accuracy": 0.7766127910465002,
|
| 2259 |
+
"num_tokens": 52634528.0,
|
| 2260 |
+
"step": 1590
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"entropy": 0.9532167240977287,
|
| 2264 |
+
"epoch": 2.56,
|
| 2265 |
+
"grad_norm": 15.4375,
|
| 2266 |
+
"learning_rate": 1.0377777777777778e-05,
|
| 2267 |
+
"loss": 15.801,
|
| 2268 |
+
"mean_token_accuracy": 0.7799606509506702,
|
| 2269 |
+
"num_tokens": 52964035.0,
|
| 2270 |
+
"step": 1600
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 2.56,
|
| 2274 |
+
"eval_biology_entropy": 2.7770239686965943,
|
| 2275 |
+
"eval_biology_loss": 3.2758572101593018,
|
| 2276 |
+
"eval_biology_mean_token_accuracy": 0.4915564143657684,
|
| 2277 |
+
"eval_biology_num_tokens": 52964035.0,
|
| 2278 |
+
"eval_biology_runtime": 56.3152,
|
| 2279 |
+
"eval_biology_samples_per_second": 8.879,
|
| 2280 |
+
"eval_biology_steps_per_second": 2.22,
|
| 2281 |
+
"step": 1600
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 2.56,
|
| 2285 |
+
"eval_chemistry_entropy": 1.3856621465682983,
|
| 2286 |
+
"eval_chemistry_loss": 1.4901551008224487,
|
| 2287 |
+
"eval_chemistry_mean_token_accuracy": 0.7066429214477539,
|
| 2288 |
+
"eval_chemistry_num_tokens": 52964035.0,
|
| 2289 |
+
"eval_chemistry_runtime": 69.4464,
|
| 2290 |
+
"eval_chemistry_samples_per_second": 7.2,
|
| 2291 |
+
"eval_chemistry_steps_per_second": 1.8,
|
| 2292 |
+
"step": 1600
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 2.56,
|
| 2296 |
+
"eval_math_entropy": 1.0012836136817933,
|
| 2297 |
+
"eval_math_loss": 1.2628675699234009,
|
| 2298 |
+
"eval_math_mean_token_accuracy": 0.7406582078933716,
|
| 2299 |
+
"eval_math_num_tokens": 52964035.0,
|
| 2300 |
+
"eval_math_runtime": 71.8017,
|
| 2301 |
+
"eval_math_samples_per_second": 6.964,
|
| 2302 |
+
"eval_math_steps_per_second": 1.741,
|
| 2303 |
+
"step": 1600
|
| 2304 |
+
},
|
| 2305 |
+
{
|
| 2306 |
+
"epoch": 2.56,
|
| 2307 |
+
"eval_physics_entropy": 0.9688046803474426,
|
| 2308 |
+
"eval_physics_loss": 1.0173183679580688,
|
| 2309 |
+
"eval_physics_mean_token_accuracy": 0.775043842792511,
|
| 2310 |
+
"eval_physics_num_tokens": 52964035.0,
|
| 2311 |
+
"eval_physics_runtime": 82.9445,
|
| 2312 |
+
"eval_physics_samples_per_second": 6.028,
|
| 2313 |
+
"eval_physics_steps_per_second": 1.507,
|
| 2314 |
+
"step": 1600
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"entropy": 0.9521009422838688,
|
| 2318 |
+
"epoch": 2.576,
|
| 2319 |
+
"grad_norm": 16.5,
|
| 2320 |
+
"learning_rate": 1.0303703703703705e-05,
|
| 2321 |
+
"loss": 15.6752,
|
| 2322 |
+
"mean_token_accuracy": 0.7822756599634886,
|
| 2323 |
+
"num_tokens": 53293345.0,
|
| 2324 |
+
"step": 1610
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"entropy": 0.964577104896307,
|
| 2328 |
+
"epoch": 2.592,
|
| 2329 |
+
"grad_norm": 13.75,
|
| 2330 |
+
"learning_rate": 1.0229629629629631e-05,
|
| 2331 |
+
"loss": 15.9694,
|
| 2332 |
+
"mean_token_accuracy": 0.7799866251647473,
|
| 2333 |
+
"num_tokens": 53620896.0,
|
| 2334 |
+
"step": 1620
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"entropy": 0.9362829696387053,
|
| 2338 |
+
"epoch": 2.608,
|
| 2339 |
+
"grad_norm": 14.1875,
|
| 2340 |
+
"learning_rate": 1.0155555555555556e-05,
|
| 2341 |
+
"loss": 15.4772,
|
| 2342 |
+
"mean_token_accuracy": 0.7848034355789423,
|
| 2343 |
+
"num_tokens": 53950989.0,
|
| 2344 |
+
"step": 1630
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"entropy": 0.9582307059317827,
|
| 2348 |
+
"epoch": 2.624,
|
| 2349 |
+
"grad_norm": 13.75,
|
| 2350 |
+
"learning_rate": 1.0081481481481484e-05,
|
| 2351 |
+
"loss": 15.8442,
|
| 2352 |
+
"mean_token_accuracy": 0.7801676865667104,
|
| 2353 |
+
"num_tokens": 54282732.0,
|
| 2354 |
+
"step": 1640
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"entropy": 0.954668253660202,
|
| 2358 |
+
"epoch": 2.64,
|
| 2359 |
+
"grad_norm": 16.125,
|
| 2360 |
+
"learning_rate": 1.0007407407407407e-05,
|
| 2361 |
+
"loss": 15.783,
|
| 2362 |
+
"mean_token_accuracy": 0.7808378152549267,
|
| 2363 |
+
"num_tokens": 54609748.0,
|
| 2364 |
+
"step": 1650
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"entropy": 0.9885619092732668,
|
| 2368 |
+
"epoch": 2.656,
|
| 2369 |
+
"grad_norm": 16.625,
|
| 2370 |
+
"learning_rate": 9.933333333333334e-06,
|
| 2371 |
+
"loss": 16.4165,
|
| 2372 |
+
"mean_token_accuracy": 0.7741628363728523,
|
| 2373 |
+
"num_tokens": 54953583.0,
|
| 2374 |
+
"step": 1660
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"entropy": 0.9665027219802141,
|
| 2378 |
+
"epoch": 2.672,
|
| 2379 |
+
"grad_norm": 14.875,
|
| 2380 |
+
"learning_rate": 9.85925925925926e-06,
|
| 2381 |
+
"loss": 15.971,
|
| 2382 |
+
"mean_token_accuracy": 0.7774426110088826,
|
| 2383 |
+
"num_tokens": 55281945.0,
|
| 2384 |
+
"step": 1670
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"entropy": 0.9464025635272264,
|
| 2388 |
+
"epoch": 2.6879999999999997,
|
| 2389 |
+
"grad_norm": 16.25,
|
| 2390 |
+
"learning_rate": 9.785185185185187e-06,
|
| 2391 |
+
"loss": 15.5004,
|
| 2392 |
+
"mean_token_accuracy": 0.7852077301591635,
|
| 2393 |
+
"num_tokens": 55612391.0,
|
| 2394 |
+
"step": 1680
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"entropy": 0.9382720027118922,
|
| 2398 |
+
"epoch": 2.7039999999999997,
|
| 2399 |
+
"grad_norm": 14.375,
|
| 2400 |
+
"learning_rate": 9.711111111111111e-06,
|
| 2401 |
+
"loss": 15.6926,
|
| 2402 |
+
"mean_token_accuracy": 0.7824352242052555,
|
| 2403 |
+
"num_tokens": 55949791.0,
|
| 2404 |
+
"step": 1690
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"entropy": 0.9608704566955566,
|
| 2408 |
+
"epoch": 2.7199999999999998,
|
| 2409 |
+
"grad_norm": 18.375,
|
| 2410 |
+
"learning_rate": 9.637037037037038e-06,
|
| 2411 |
+
"loss": 15.794,
|
| 2412 |
+
"mean_token_accuracy": 0.7801640577614307,
|
| 2413 |
+
"num_tokens": 56282204.0,
|
| 2414 |
+
"step": 1700
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 2.7199999999999998,
|
| 2418 |
+
"eval_biology_entropy": 2.8133376455307006,
|
| 2419 |
+
"eval_biology_loss": 3.2727437019348145,
|
| 2420 |
+
"eval_biology_mean_token_accuracy": 0.4918837146759033,
|
| 2421 |
+
"eval_biology_num_tokens": 56282204.0,
|
| 2422 |
+
"eval_biology_runtime": 56.3533,
|
| 2423 |
+
"eval_biology_samples_per_second": 8.873,
|
| 2424 |
+
"eval_biology_steps_per_second": 2.218,
|
| 2425 |
+
"step": 1700
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 2.7199999999999998,
|
| 2429 |
+
"eval_chemistry_entropy": 1.3927633690834045,
|
| 2430 |
+
"eval_chemistry_loss": 1.4875998497009277,
|
| 2431 |
+
"eval_chemistry_mean_token_accuracy": 0.7068242344856263,
|
| 2432 |
+
"eval_chemistry_num_tokens": 56282204.0,
|
| 2433 |
+
"eval_chemistry_runtime": 69.3593,
|
| 2434 |
+
"eval_chemistry_samples_per_second": 7.209,
|
| 2435 |
+
"eval_chemistry_steps_per_second": 1.802,
|
| 2436 |
+
"step": 1700
|
| 2437 |
+
},
|
| 2438 |
+
{
|
| 2439 |
+
"epoch": 2.7199999999999998,
|
| 2440 |
+
"eval_math_entropy": 1.0047017326354981,
|
| 2441 |
+
"eval_math_loss": 1.262471079826355,
|
| 2442 |
+
"eval_math_mean_token_accuracy": 0.7409610171318054,
|
| 2443 |
+
"eval_math_num_tokens": 56282204.0,
|
| 2444 |
+
"eval_math_runtime": 71.9414,
|
| 2445 |
+
"eval_math_samples_per_second": 6.95,
|
| 2446 |
+
"eval_math_steps_per_second": 1.738,
|
| 2447 |
+
"step": 1700
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 2.7199999999999998,
|
| 2451 |
+
"eval_physics_entropy": 0.9688171024322509,
|
| 2452 |
+
"eval_physics_loss": 1.0153486728668213,
|
| 2453 |
+
"eval_physics_mean_token_accuracy": 0.7751585001945496,
|
| 2454 |
+
"eval_physics_num_tokens": 56282204.0,
|
| 2455 |
+
"eval_physics_runtime": 83.3388,
|
| 2456 |
+
"eval_physics_samples_per_second": 6.0,
|
| 2457 |
+
"eval_physics_steps_per_second": 1.5,
|
| 2458 |
+
"step": 1700
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"entropy": 0.9965044487267732,
|
| 2462 |
+
"epoch": 2.7359999999999998,
|
| 2463 |
+
"grad_norm": 16.125,
|
| 2464 |
+
"learning_rate": 9.562962962962965e-06,
|
| 2465 |
+
"loss": 16.4901,
|
| 2466 |
+
"mean_token_accuracy": 0.7732316367328167,
|
| 2467 |
+
"num_tokens": 56605478.0,
|
| 2468 |
+
"step": 1710
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"entropy": 0.9614631816744804,
|
| 2472 |
+
"epoch": 2.752,
|
| 2473 |
+
"grad_norm": 14.4375,
|
| 2474 |
+
"learning_rate": 9.48888888888889e-06,
|
| 2475 |
+
"loss": 15.9918,
|
| 2476 |
+
"mean_token_accuracy": 0.7788697842508554,
|
| 2477 |
+
"num_tokens": 56935874.0,
|
| 2478 |
+
"step": 1720
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"entropy": 0.9439939312636852,
|
| 2482 |
+
"epoch": 2.768,
|
| 2483 |
+
"grad_norm": 15.4375,
|
| 2484 |
+
"learning_rate": 9.414814814814816e-06,
|
| 2485 |
+
"loss": 15.5518,
|
| 2486 |
+
"mean_token_accuracy": 0.7843724682927131,
|
| 2487 |
+
"num_tokens": 57270552.0,
|
| 2488 |
+
"step": 1730
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"entropy": 0.9652665428817272,
|
| 2492 |
+
"epoch": 2.784,
|
| 2493 |
+
"grad_norm": 16.5,
|
| 2494 |
+
"learning_rate": 9.34074074074074e-06,
|
| 2495 |
+
"loss": 15.954,
|
| 2496 |
+
"mean_token_accuracy": 0.7781353067606688,
|
| 2497 |
+
"num_tokens": 57602968.0,
|
| 2498 |
+
"step": 1740
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"entropy": 1.0173277705907822,
|
| 2502 |
+
"epoch": 2.8,
|
| 2503 |
+
"grad_norm": 17.0,
|
| 2504 |
+
"learning_rate": 9.266666666666667e-06,
|
| 2505 |
+
"loss": 16.8998,
|
| 2506 |
+
"mean_token_accuracy": 0.7688637807965278,
|
| 2507 |
+
"num_tokens": 57925886.0,
|
| 2508 |
+
"step": 1750
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"entropy": 0.9443410014733672,
|
| 2512 |
+
"epoch": 2.816,
|
| 2513 |
+
"grad_norm": 15.625,
|
| 2514 |
+
"learning_rate": 9.192592592592594e-06,
|
| 2515 |
+
"loss": 15.4248,
|
| 2516 |
+
"mean_token_accuracy": 0.7846918229013682,
|
| 2517 |
+
"num_tokens": 58262960.0,
|
| 2518 |
+
"step": 1760
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"entropy": 0.9699663460254669,
|
| 2522 |
+
"epoch": 2.832,
|
| 2523 |
+
"grad_norm": 16.375,
|
| 2524 |
+
"learning_rate": 9.118518518518518e-06,
|
| 2525 |
+
"loss": 16.0995,
|
| 2526 |
+
"mean_token_accuracy": 0.7767021499574185,
|
| 2527 |
+
"num_tokens": 58586324.0,
|
| 2528 |
+
"step": 1770
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"entropy": 0.9516465954482556,
|
| 2532 |
+
"epoch": 2.848,
|
| 2533 |
+
"grad_norm": 16.25,
|
| 2534 |
+
"learning_rate": 9.044444444444445e-06,
|
| 2535 |
+
"loss": 15.8066,
|
| 2536 |
+
"mean_token_accuracy": 0.7798173315823078,
|
| 2537 |
+
"num_tokens": 58912949.0,
|
| 2538 |
+
"step": 1780
|
| 2539 |
+
},
|
| 2540 |
+
{
|
| 2541 |
+
"entropy": 0.9439001478254795,
|
| 2542 |
+
"epoch": 2.864,
|
| 2543 |
+
"grad_norm": 13.6875,
|
| 2544 |
+
"learning_rate": 8.970370370370372e-06,
|
| 2545 |
+
"loss": 15.5107,
|
| 2546 |
+
"mean_token_accuracy": 0.7841162201017141,
|
| 2547 |
+
"num_tokens": 59252494.0,
|
| 2548 |
+
"step": 1790
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"entropy": 0.9630456361919641,
|
| 2552 |
+
"epoch": 2.88,
|
| 2553 |
+
"grad_norm": 18.75,
|
| 2554 |
+
"learning_rate": 8.896296296296298e-06,
|
| 2555 |
+
"loss": 15.9262,
|
| 2556 |
+
"mean_token_accuracy": 0.7800596509128809,
|
| 2557 |
+
"num_tokens": 59574951.0,
|
| 2558 |
+
"step": 1800
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 2.88,
|
| 2562 |
+
"eval_biology_entropy": 2.808165202140808,
|
| 2563 |
+
"eval_biology_loss": 3.258385181427002,
|
| 2564 |
+
"eval_biology_mean_token_accuracy": 0.4932124676704407,
|
| 2565 |
+
"eval_biology_num_tokens": 59574951.0,
|
| 2566 |
+
"eval_biology_runtime": 56.2773,
|
| 2567 |
+
"eval_biology_samples_per_second": 8.885,
|
| 2568 |
+
"eval_biology_steps_per_second": 2.221,
|
| 2569 |
+
"step": 1800
|
| 2570 |
+
},
|
| 2571 |
+
{
|
| 2572 |
+
"epoch": 2.88,
|
| 2573 |
+
"eval_chemistry_entropy": 1.3842170772552491,
|
| 2574 |
+
"eval_chemistry_loss": 1.4859730005264282,
|
| 2575 |
+
"eval_chemistry_mean_token_accuracy": 0.7075357518196106,
|
| 2576 |
+
"eval_chemistry_num_tokens": 59574951.0,
|
| 2577 |
+
"eval_chemistry_runtime": 69.4805,
|
| 2578 |
+
"eval_chemistry_samples_per_second": 7.196,
|
| 2579 |
+
"eval_chemistry_steps_per_second": 1.799,
|
| 2580 |
+
"step": 1800
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 2.88,
|
| 2584 |
+
"eval_math_entropy": 0.9982392830848694,
|
| 2585 |
+
"eval_math_loss": 1.2617205381393433,
|
| 2586 |
+
"eval_math_mean_token_accuracy": 0.7408777174949646,
|
| 2587 |
+
"eval_math_num_tokens": 59574951.0,
|
| 2588 |
+
"eval_math_runtime": 71.8482,
|
| 2589 |
+
"eval_math_samples_per_second": 6.959,
|
| 2590 |
+
"eval_math_steps_per_second": 1.74,
|
| 2591 |
+
"step": 1800
|
| 2592 |
+
},
|
| 2593 |
+
{
|
| 2594 |
+
"epoch": 2.88,
|
| 2595 |
+
"eval_physics_entropy": 0.9607987327575683,
|
| 2596 |
+
"eval_physics_loss": 1.014259696006775,
|
| 2597 |
+
"eval_physics_mean_token_accuracy": 0.7754597988128662,
|
| 2598 |
+
"eval_physics_num_tokens": 59574951.0,
|
| 2599 |
+
"eval_physics_runtime": 83.0286,
|
| 2600 |
+
"eval_physics_samples_per_second": 6.022,
|
| 2601 |
+
"eval_physics_steps_per_second": 1.506,
|
| 2602 |
+
"step": 1800
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"entropy": 0.9467792255803943,
|
| 2606 |
+
"epoch": 2.896,
|
| 2607 |
+
"grad_norm": 16.625,
|
| 2608 |
+
"learning_rate": 8.822222222222223e-06,
|
| 2609 |
+
"loss": 15.7767,
|
| 2610 |
+
"mean_token_accuracy": 0.7813689064234495,
|
| 2611 |
+
"num_tokens": 59903274.0,
|
| 2612 |
+
"step": 1810
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"entropy": 0.9469530638307333,
|
| 2616 |
+
"epoch": 2.912,
|
| 2617 |
+
"grad_norm": 15.375,
|
| 2618 |
+
"learning_rate": 8.74814814814815e-06,
|
| 2619 |
+
"loss": 15.6649,
|
| 2620 |
+
"mean_token_accuracy": 0.7820369627326726,
|
| 2621 |
+
"num_tokens": 60234203.0,
|
| 2622 |
+
"step": 1820
|
| 2623 |
+
},
|
| 2624 |
+
{
|
| 2625 |
+
"entropy": 0.9534095168113709,
|
| 2626 |
+
"epoch": 2.928,
|
| 2627 |
+
"grad_norm": 15.875,
|
| 2628 |
+
"learning_rate": 8.674074074074074e-06,
|
| 2629 |
+
"loss": 15.6803,
|
| 2630 |
+
"mean_token_accuracy": 0.7829261634498834,
|
| 2631 |
+
"num_tokens": 60559123.0,
|
| 2632 |
+
"step": 1830
|
| 2633 |
+
},
|
| 2634 |
+
{
|
| 2635 |
+
"entropy": 0.9459656484425067,
|
| 2636 |
+
"epoch": 2.944,
|
| 2637 |
+
"grad_norm": 18.0,
|
| 2638 |
+
"learning_rate": 8.6e-06,
|
| 2639 |
+
"loss": 15.5886,
|
| 2640 |
+
"mean_token_accuracy": 0.7832438927143812,
|
| 2641 |
+
"num_tokens": 60893423.0,
|
| 2642 |
+
"step": 1840
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"entropy": 0.9542483527213335,
|
| 2646 |
+
"epoch": 2.96,
|
| 2647 |
+
"grad_norm": 14.5,
|
| 2648 |
+
"learning_rate": 8.525925925925927e-06,
|
| 2649 |
+
"loss": 15.9044,
|
| 2650 |
+
"mean_token_accuracy": 0.7797330930829048,
|
| 2651 |
+
"num_tokens": 61228692.0,
|
| 2652 |
+
"step": 1850
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"entropy": 0.9721855409443378,
|
| 2656 |
+
"epoch": 2.976,
|
| 2657 |
+
"grad_norm": 15.1875,
|
| 2658 |
+
"learning_rate": 8.451851851851852e-06,
|
| 2659 |
+
"loss": 15.9622,
|
| 2660 |
+
"mean_token_accuracy": 0.7788714807480573,
|
| 2661 |
+
"num_tokens": 61560852.0,
|
| 2662 |
+
"step": 1860
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"entropy": 0.924789821729064,
|
| 2666 |
+
"epoch": 2.992,
|
| 2667 |
+
"grad_norm": 15.6875,
|
| 2668 |
+
"learning_rate": 8.377777777777779e-06,
|
| 2669 |
+
"loss": 15.3714,
|
| 2670 |
+
"mean_token_accuracy": 0.7856392968446017,
|
| 2671 |
+
"num_tokens": 61898951.0,
|
| 2672 |
+
"step": 1870
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"entropy": 0.9437004335224628,
|
| 2676 |
+
"epoch": 3.008,
|
| 2677 |
+
"grad_norm": 14.3125,
|
| 2678 |
+
"learning_rate": 8.303703703703705e-06,
|
| 2679 |
+
"loss": 15.5514,
|
| 2680 |
+
"mean_token_accuracy": 0.7843418989330531,
|
| 2681 |
+
"num_tokens": 62223875.0,
|
| 2682 |
+
"step": 1880
|
| 2683 |
+
},
|
| 2684 |
+
{
|
| 2685 |
+
"entropy": 0.9536178600043058,
|
| 2686 |
+
"epoch": 3.024,
|
| 2687 |
+
"grad_norm": 15.3125,
|
| 2688 |
+
"learning_rate": 8.229629629629632e-06,
|
| 2689 |
+
"loss": 15.7275,
|
| 2690 |
+
"mean_token_accuracy": 0.7818392261862754,
|
| 2691 |
+
"num_tokens": 62553166.0,
|
| 2692 |
+
"step": 1890
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"entropy": 0.9350160963833332,
|
| 2696 |
+
"epoch": 3.04,
|
| 2697 |
+
"grad_norm": 15.25,
|
| 2698 |
+
"learning_rate": 8.155555555555556e-06,
|
| 2699 |
+
"loss": 15.4048,
|
| 2700 |
+
"mean_token_accuracy": 0.7846107162535191,
|
| 2701 |
+
"num_tokens": 62893113.0,
|
| 2702 |
+
"step": 1900
|
| 2703 |
+
},
|
| 2704 |
+
{
|
| 2705 |
+
"epoch": 3.04,
|
| 2706 |
+
"eval_biology_entropy": 2.8041762866973876,
|
| 2707 |
+
"eval_biology_loss": 3.261282205581665,
|
| 2708 |
+
"eval_biology_mean_token_accuracy": 0.4926483144760132,
|
| 2709 |
+
"eval_biology_num_tokens": 62893113.0,
|
| 2710 |
+
"eval_biology_runtime": 56.1352,
|
| 2711 |
+
"eval_biology_samples_per_second": 8.907,
|
| 2712 |
+
"eval_biology_steps_per_second": 2.227,
|
| 2713 |
+
"step": 1900
|
| 2714 |
+
},
|
| 2715 |
+
{
|
| 2716 |
+
"epoch": 3.04,
|
| 2717 |
+
"eval_chemistry_entropy": 1.383084843158722,
|
| 2718 |
+
"eval_chemistry_loss": 1.4848686456680298,
|
| 2719 |
+
"eval_chemistry_mean_token_accuracy": 0.7074742059707642,
|
| 2720 |
+
"eval_chemistry_num_tokens": 62893113.0,
|
| 2721 |
+
"eval_chemistry_runtime": 69.368,
|
| 2722 |
+
"eval_chemistry_samples_per_second": 7.208,
|
| 2723 |
+
"eval_chemistry_steps_per_second": 1.802,
|
| 2724 |
+
"step": 1900
|
| 2725 |
+
},
|
| 2726 |
+
{
|
| 2727 |
+
"epoch": 3.04,
|
| 2728 |
+
"eval_math_entropy": 0.9970106234550477,
|
| 2729 |
+
"eval_math_loss": 1.2620422840118408,
|
| 2730 |
+
"eval_math_mean_token_accuracy": 0.7409479322433472,
|
| 2731 |
+
"eval_math_num_tokens": 62893113.0,
|
| 2732 |
+
"eval_math_runtime": 71.729,
|
| 2733 |
+
"eval_math_samples_per_second": 6.971,
|
| 2734 |
+
"eval_math_steps_per_second": 1.743,
|
| 2735 |
+
"step": 1900
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 3.04,
|
| 2739 |
+
"eval_physics_entropy": 0.9595280365943909,
|
| 2740 |
+
"eval_physics_loss": 1.0132101774215698,
|
| 2741 |
+
"eval_physics_mean_token_accuracy": 0.7754246773719787,
|
| 2742 |
+
"eval_physics_num_tokens": 62893113.0,
|
| 2743 |
+
"eval_physics_runtime": 83.6484,
|
| 2744 |
+
"eval_physics_samples_per_second": 5.977,
|
| 2745 |
+
"eval_physics_steps_per_second": 1.494,
|
| 2746 |
+
"step": 1900
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"entropy": 0.96522372700274,
|
| 2750 |
+
"epoch": 3.056,
|
| 2751 |
+
"grad_norm": 16.25,
|
| 2752 |
+
"learning_rate": 8.081481481481483e-06,
|
| 2753 |
+
"loss": 16.0574,
|
| 2754 |
+
"mean_token_accuracy": 0.778254572674632,
|
| 2755 |
+
"num_tokens": 63218088.0,
|
| 2756 |
+
"step": 1910
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"entropy": 0.9615779049694538,
|
| 2760 |
+
"epoch": 3.072,
|
| 2761 |
+
"grad_norm": 16.625,
|
| 2762 |
+
"learning_rate": 8.007407407407408e-06,
|
| 2763 |
+
"loss": 15.8044,
|
| 2764 |
+
"mean_token_accuracy": 0.7792242676019668,
|
| 2765 |
+
"num_tokens": 63551201.0,
|
| 2766 |
+
"step": 1920
|
| 2767 |
+
},
|
| 2768 |
+
{
|
| 2769 |
+
"entropy": 0.9705667292699218,
|
| 2770 |
+
"epoch": 3.088,
|
| 2771 |
+
"grad_norm": 17.0,
|
| 2772 |
+
"learning_rate": 7.933333333333334e-06,
|
| 2773 |
+
"loss": 16.0257,
|
| 2774 |
+
"mean_token_accuracy": 0.7777945622801781,
|
| 2775 |
+
"num_tokens": 63876362.0,
|
| 2776 |
+
"step": 1930
|
| 2777 |
+
},
|
| 2778 |
+
{
|
| 2779 |
+
"entropy": 0.9308646315708756,
|
| 2780 |
+
"epoch": 3.104,
|
| 2781 |
+
"grad_norm": 17.0,
|
| 2782 |
+
"learning_rate": 7.859259259259259e-06,
|
| 2783 |
+
"loss": 15.3968,
|
| 2784 |
+
"mean_token_accuracy": 0.7849573034793138,
|
| 2785 |
+
"num_tokens": 64208171.0,
|
| 2786 |
+
"step": 1940
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"entropy": 0.9403708964586258,
|
| 2790 |
+
"epoch": 3.12,
|
| 2791 |
+
"grad_norm": 15.5625,
|
| 2792 |
+
"learning_rate": 7.785185185185185e-06,
|
| 2793 |
+
"loss": 15.3807,
|
| 2794 |
+
"mean_token_accuracy": 0.7861266512423754,
|
| 2795 |
+
"num_tokens": 64544493.0,
|
| 2796 |
+
"step": 1950
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"entropy": 0.9759513843804598,
|
| 2800 |
+
"epoch": 3.136,
|
| 2801 |
+
"grad_norm": 16.625,
|
| 2802 |
+
"learning_rate": 7.711111111111112e-06,
|
| 2803 |
+
"loss": 16.1287,
|
| 2804 |
+
"mean_token_accuracy": 0.7761076722294092,
|
| 2805 |
+
"num_tokens": 64872488.0,
|
| 2806 |
+
"step": 1960
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"entropy": 0.92836042791605,
|
| 2810 |
+
"epoch": 3.152,
|
| 2811 |
+
"grad_norm": 15.875,
|
| 2812 |
+
"learning_rate": 7.637037037037037e-06,
|
| 2813 |
+
"loss": 15.3907,
|
| 2814 |
+
"mean_token_accuracy": 0.785225062072277,
|
| 2815 |
+
"num_tokens": 65201124.0,
|
| 2816 |
+
"step": 1970
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"entropy": 0.9509770900011063,
|
| 2820 |
+
"epoch": 3.168,
|
| 2821 |
+
"grad_norm": 13.75,
|
| 2822 |
+
"learning_rate": 7.562962962962963e-06,
|
| 2823 |
+
"loss": 15.8103,
|
| 2824 |
+
"mean_token_accuracy": 0.7816420182585716,
|
| 2825 |
+
"num_tokens": 65525868.0,
|
| 2826 |
+
"step": 1980
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"entropy": 0.9470858953893184,
|
| 2830 |
+
"epoch": 3.184,
|
| 2831 |
+
"grad_norm": 16.5,
|
| 2832 |
+
"learning_rate": 7.48888888888889e-06,
|
| 2833 |
+
"loss": 15.6067,
|
| 2834 |
+
"mean_token_accuracy": 0.7848717745393514,
|
| 2835 |
+
"num_tokens": 65861603.0,
|
| 2836 |
+
"step": 1990
|
| 2837 |
+
},
|
| 2838 |
+
{
|
| 2839 |
+
"entropy": 0.9574329257011414,
|
| 2840 |
+
"epoch": 3.2,
|
| 2841 |
+
"grad_norm": 14.8125,
|
| 2842 |
+
"learning_rate": 7.4148148148148155e-06,
|
| 2843 |
+
"loss": 15.7322,
|
| 2844 |
+
"mean_token_accuracy": 0.7814668208360672,
|
| 2845 |
+
"num_tokens": 66193675.0,
|
| 2846 |
+
"step": 2000
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 3.2,
|
| 2850 |
+
"eval_biology_entropy": 2.7968916082382203,
|
| 2851 |
+
"eval_biology_loss": 3.266221284866333,
|
| 2852 |
+
"eval_biology_mean_token_accuracy": 0.49423834228515623,
|
| 2853 |
+
"eval_biology_num_tokens": 66193675.0,
|
| 2854 |
+
"eval_biology_runtime": 56.2653,
|
| 2855 |
+
"eval_biology_samples_per_second": 8.886,
|
| 2856 |
+
"eval_biology_steps_per_second": 2.222,
|
| 2857 |
+
"step": 2000
|
| 2858 |
+
},
|
| 2859 |
+
{
|
| 2860 |
+
"epoch": 3.2,
|
| 2861 |
+
"eval_chemistry_entropy": 1.3809385676383972,
|
| 2862 |
+
"eval_chemistry_loss": 1.4836703538894653,
|
| 2863 |
+
"eval_chemistry_mean_token_accuracy": 0.7080655236244202,
|
| 2864 |
+
"eval_chemistry_num_tokens": 66193675.0,
|
| 2865 |
+
"eval_chemistry_runtime": 69.0507,
|
| 2866 |
+
"eval_chemistry_samples_per_second": 7.241,
|
| 2867 |
+
"eval_chemistry_steps_per_second": 1.81,
|
| 2868 |
+
"step": 2000
|
| 2869 |
+
},
|
| 2870 |
+
{
|
| 2871 |
+
"epoch": 3.2,
|
| 2872 |
+
"eval_math_entropy": 0.9969238524436951,
|
| 2873 |
+
"eval_math_loss": 1.2609111070632935,
|
| 2874 |
+
"eval_math_mean_token_accuracy": 0.7411160726547241,
|
| 2875 |
+
"eval_math_num_tokens": 66193675.0,
|
| 2876 |
+
"eval_math_runtime": 72.1408,
|
| 2877 |
+
"eval_math_samples_per_second": 6.931,
|
| 2878 |
+
"eval_math_steps_per_second": 1.733,
|
| 2879 |
+
"step": 2000
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 3.2,
|
| 2883 |
+
"eval_physics_entropy": 0.9590741763114929,
|
| 2884 |
+
"eval_physics_loss": 1.0124115943908691,
|
| 2885 |
+
"eval_physics_mean_token_accuracy": 0.7754848318099976,
|
| 2886 |
+
"eval_physics_num_tokens": 66193675.0,
|
| 2887 |
+
"eval_physics_runtime": 83.1629,
|
| 2888 |
+
"eval_physics_samples_per_second": 6.012,
|
| 2889 |
+
"eval_physics_steps_per_second": 1.503,
|
| 2890 |
+
"step": 2000
|
| 2891 |
+
}
|
| 2892 |
+
],
|
| 2893 |
+
"logging_steps": 10,
|
| 2894 |
+
"max_steps": 3000,
|
| 2895 |
+
"num_input_tokens_seen": 0,
|
| 2896 |
+
"num_train_epochs": 5,
|
| 2897 |
+
"save_steps": 500,
|
| 2898 |
+
"stateful_callbacks": {
|
| 2899 |
+
"TrainerControl": {
|
| 2900 |
+
"args": {
|
| 2901 |
+
"should_epoch_stop": false,
|
| 2902 |
+
"should_evaluate": false,
|
| 2903 |
+
"should_log": false,
|
| 2904 |
+
"should_save": true,
|
| 2905 |
+
"should_training_stop": false
|
| 2906 |
+
},
|
| 2907 |
+
"attributes": {}
|
| 2908 |
+
}
|
| 2909 |
+
},
|
| 2910 |
+
"total_flos": 6.614072648666411e+18,
|
| 2911 |
+
"train_batch_size": 4,
|
| 2912 |
+
"trial_name": null,
|
| 2913 |
+
"trial_params": null
|
| 2914 |
+
}
|
checkpoint-2000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:99fae199773974506471525072c20d6088b07c2300f704c09a94a2b4fb6d778d
|
| 3 |
+
size 6417
|