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 sha256:75335bcd8885306655075aa11c840d7410aa53cd73a23cfa7193d33cd445eb87
|
| 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 sha256:8d52b5085f4bdd124008bb1819095cfbfeb940221bfd2798d627b394a76d7477
|
| 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 sha256:28f2e77b1bbddbb70c473d5c52c52c2acf49239291cd241a37fc66dfaaf3a29e
|
| 3 |
+
size 5009508771
|
checkpoint-2000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6b4f2d699ceb4a776b3f18cbb9b5e404aee9e23719242e3e6fb8bacc03b0a2c1
|
| 3 |
+
size 14645
|
checkpoint-2000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e7da898c7cde13cf4778089e6bb91871291cb3f0f1041b88eca5bd72175ec3c4
|
| 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": 0.5395457262173295,
|
| 14 |
+
"epoch": 0.016,
|
| 15 |
+
"grad_norm": 3200.0,
|
| 16 |
+
"learning_rate": 4.5000000000000003e-07,
|
| 17 |
+
"loss": 92.5745,
|
| 18 |
+
"mean_token_accuracy": 0.4849457062780857,
|
| 19 |
+
"num_tokens": 195524.0,
|
| 20 |
+
"step": 10
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"entropy": 0.5515131525695324,
|
| 24 |
+
"epoch": 0.032,
|
| 25 |
+
"grad_norm": 2912.0,
|
| 26 |
+
"learning_rate": 9.500000000000001e-07,
|
| 27 |
+
"loss": 88.931,
|
| 28 |
+
"mean_token_accuracy": 0.4874983370304108,
|
| 29 |
+
"num_tokens": 390903.0,
|
| 30 |
+
"step": 20
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"entropy": 0.6169477274641395,
|
| 34 |
+
"epoch": 0.048,
|
| 35 |
+
"grad_norm": 2912.0,
|
| 36 |
+
"learning_rate": 1.45e-06,
|
| 37 |
+
"loss": 78.3695,
|
| 38 |
+
"mean_token_accuracy": 0.4966813003644347,
|
| 39 |
+
"num_tokens": 589868.0,
|
| 40 |
+
"step": 30
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"entropy": 0.8348326332867145,
|
| 44 |
+
"epoch": 0.064,
|
| 45 |
+
"grad_norm": 1160.0,
|
| 46 |
+
"learning_rate": 1.9500000000000004e-06,
|
| 47 |
+
"loss": 60.9248,
|
| 48 |
+
"mean_token_accuracy": 0.505222667939961,
|
| 49 |
+
"num_tokens": 791190.0,
|
| 50 |
+
"step": 40
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"entropy": 1.3063700653612613,
|
| 54 |
+
"epoch": 0.08,
|
| 55 |
+
"grad_norm": 247.0,
|
| 56 |
+
"learning_rate": 2.4500000000000003e-06,
|
| 57 |
+
"loss": 46.7906,
|
| 58 |
+
"mean_token_accuracy": 0.5185849336907268,
|
| 59 |
+
"num_tokens": 989860.0,
|
| 60 |
+
"step": 50
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"entropy": 1.959995786100626,
|
| 64 |
+
"epoch": 0.096,
|
| 65 |
+
"grad_norm": 81.0,
|
| 66 |
+
"learning_rate": 2.95e-06,
|
| 67 |
+
"loss": 39.2759,
|
| 68 |
+
"mean_token_accuracy": 0.5406044816598297,
|
| 69 |
+
"num_tokens": 1181777.0,
|
| 70 |
+
"step": 60
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"entropy": 2.191839651763439,
|
| 74 |
+
"epoch": 0.112,
|
| 75 |
+
"grad_norm": 57.0,
|
| 76 |
+
"learning_rate": 3.45e-06,
|
| 77 |
+
"loss": 34.17,
|
| 78 |
+
"mean_token_accuracy": 0.5669999368488788,
|
| 79 |
+
"num_tokens": 1385513.0,
|
| 80 |
+
"step": 70
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"entropy": 1.863330227136612,
|
| 84 |
+
"epoch": 0.128,
|
| 85 |
+
"grad_norm": 48.25,
|
| 86 |
+
"learning_rate": 3.95e-06,
|
| 87 |
+
"loss": 31.3942,
|
| 88 |
+
"mean_token_accuracy": 0.5891917455941439,
|
| 89 |
+
"num_tokens": 1582368.0,
|
| 90 |
+
"step": 80
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"entropy": 1.8903757609426974,
|
| 94 |
+
"epoch": 0.144,
|
| 95 |
+
"grad_norm": 39.25,
|
| 96 |
+
"learning_rate": 4.450000000000001e-06,
|
| 97 |
+
"loss": 29.9275,
|
| 98 |
+
"mean_token_accuracy": 0.6045342192053795,
|
| 99 |
+
"num_tokens": 1773764.0,
|
| 100 |
+
"step": 90
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"entropy": 1.8238266378641128,
|
| 104 |
+
"epoch": 0.16,
|
| 105 |
+
"grad_norm": 36.25,
|
| 106 |
+
"learning_rate": 4.95e-06,
|
| 107 |
+
"loss": 29.3315,
|
| 108 |
+
"mean_token_accuracy": 0.6066708620637655,
|
| 109 |
+
"num_tokens": 1970077.0,
|
| 110 |
+
"step": 100
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.16,
|
| 114 |
+
"eval_biology_entropy": 1.7597194662094116,
|
| 115 |
+
"eval_biology_loss": 1.708297848701477,
|
| 116 |
+
"eval_biology_mean_token_accuracy": 0.6260299117565155,
|
| 117 |
+
"eval_biology_num_tokens": 1970077.0,
|
| 118 |
+
"eval_biology_runtime": 36.35,
|
| 119 |
+
"eval_biology_samples_per_second": 13.755,
|
| 120 |
+
"eval_biology_steps_per_second": 3.439,
|
| 121 |
+
"step": 100
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.16,
|
| 125 |
+
"eval_chemistry_entropy": 1.278447434425354,
|
| 126 |
+
"eval_chemistry_loss": 1.24350905418396,
|
| 127 |
+
"eval_chemistry_mean_token_accuracy": 0.717205593585968,
|
| 128 |
+
"eval_chemistry_num_tokens": 1970077.0,
|
| 129 |
+
"eval_chemistry_runtime": 44.9852,
|
| 130 |
+
"eval_chemistry_samples_per_second": 11.115,
|
| 131 |
+
"eval_chemistry_steps_per_second": 2.779,
|
| 132 |
+
"step": 100
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.16,
|
| 136 |
+
"eval_math_entropy": 0.9187946987152099,
|
| 137 |
+
"eval_math_loss": 1.206222414970398,
|
| 138 |
+
"eval_math_mean_token_accuracy": 0.7349825048446655,
|
| 139 |
+
"eval_math_num_tokens": 1970077.0,
|
| 140 |
+
"eval_math_runtime": 46.1658,
|
| 141 |
+
"eval_math_samples_per_second": 10.831,
|
| 142 |
+
"eval_math_steps_per_second": 2.708,
|
| 143 |
+
"step": 100
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.16,
|
| 147 |
+
"eval_physics_entropy": 1.2150983595848084,
|
| 148 |
+
"eval_physics_loss": 1.2188719511032104,
|
| 149 |
+
"eval_physics_mean_token_accuracy": 0.727249002456665,
|
| 150 |
+
"eval_physics_num_tokens": 1970077.0,
|
| 151 |
+
"eval_physics_runtime": 52.9985,
|
| 152 |
+
"eval_physics_samples_per_second": 9.434,
|
| 153 |
+
"eval_physics_steps_per_second": 2.359,
|
| 154 |
+
"step": 100
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"entropy": 1.7616284050047397,
|
| 158 |
+
"epoch": 0.176,
|
| 159 |
+
"grad_norm": 33.25,
|
| 160 |
+
"learning_rate": 5.450000000000001e-06,
|
| 161 |
+
"loss": 27.9906,
|
| 162 |
+
"mean_token_accuracy": 0.6186339979991317,
|
| 163 |
+
"num_tokens": 2168354.0,
|
| 164 |
+
"step": 110
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"entropy": 1.7145348988473414,
|
| 168 |
+
"epoch": 0.192,
|
| 169 |
+
"grad_norm": 36.25,
|
| 170 |
+
"learning_rate": 5.950000000000001e-06,
|
| 171 |
+
"loss": 27.271,
|
| 172 |
+
"mean_token_accuracy": 0.6236628353595733,
|
| 173 |
+
"num_tokens": 2365822.0,
|
| 174 |
+
"step": 120
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"entropy": 1.7000788144767285,
|
| 178 |
+
"epoch": 0.208,
|
| 179 |
+
"grad_norm": 34.75,
|
| 180 |
+
"learning_rate": 6.450000000000001e-06,
|
| 181 |
+
"loss": 26.7483,
|
| 182 |
+
"mean_token_accuracy": 0.6299795724451542,
|
| 183 |
+
"num_tokens": 2558762.0,
|
| 184 |
+
"step": 130
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"entropy": 1.6474159061908722,
|
| 188 |
+
"epoch": 0.224,
|
| 189 |
+
"grad_norm": 30.875,
|
| 190 |
+
"learning_rate": 6.95e-06,
|
| 191 |
+
"loss": 25.9891,
|
| 192 |
+
"mean_token_accuracy": 0.6365062229335308,
|
| 193 |
+
"num_tokens": 2755347.0,
|
| 194 |
+
"step": 140
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"entropy": 1.6314032576978206,
|
| 198 |
+
"epoch": 0.24,
|
| 199 |
+
"grad_norm": 31.125,
|
| 200 |
+
"learning_rate": 7.450000000000001e-06,
|
| 201 |
+
"loss": 25.1799,
|
| 202 |
+
"mean_token_accuracy": 0.64104891307652,
|
| 203 |
+
"num_tokens": 2947346.0,
|
| 204 |
+
"step": 150
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"entropy": 1.609092117100954,
|
| 208 |
+
"epoch": 0.256,
|
| 209 |
+
"grad_norm": 28.5,
|
| 210 |
+
"learning_rate": 7.950000000000002e-06,
|
| 211 |
+
"loss": 24.976,
|
| 212 |
+
"mean_token_accuracy": 0.6415446903556585,
|
| 213 |
+
"num_tokens": 3139957.0,
|
| 214 |
+
"step": 160
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"entropy": 1.5253454238176345,
|
| 218 |
+
"epoch": 0.272,
|
| 219 |
+
"grad_norm": 28.875,
|
| 220 |
+
"learning_rate": 8.45e-06,
|
| 221 |
+
"loss": 24.4431,
|
| 222 |
+
"mean_token_accuracy": 0.6438875071704387,
|
| 223 |
+
"num_tokens": 3335951.0,
|
| 224 |
+
"step": 170
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"entropy": 1.5003480702638625,
|
| 228 |
+
"epoch": 0.288,
|
| 229 |
+
"grad_norm": 31.5,
|
| 230 |
+
"learning_rate": 8.95e-06,
|
| 231 |
+
"loss": 23.9226,
|
| 232 |
+
"mean_token_accuracy": 0.6480991121381521,
|
| 233 |
+
"num_tokens": 3539731.0,
|
| 234 |
+
"step": 180
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"entropy": 1.4835937578231095,
|
| 238 |
+
"epoch": 0.304,
|
| 239 |
+
"grad_norm": 29.375,
|
| 240 |
+
"learning_rate": 9.450000000000001e-06,
|
| 241 |
+
"loss": 23.7553,
|
| 242 |
+
"mean_token_accuracy": 0.6533681858330965,
|
| 243 |
+
"num_tokens": 3733488.0,
|
| 244 |
+
"step": 190
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"entropy": 1.501522469520569,
|
| 248 |
+
"epoch": 0.32,
|
| 249 |
+
"grad_norm": 31.625,
|
| 250 |
+
"learning_rate": 9.950000000000001e-06,
|
| 251 |
+
"loss": 24.0263,
|
| 252 |
+
"mean_token_accuracy": 0.6483922835439444,
|
| 253 |
+
"num_tokens": 3920545.0,
|
| 254 |
+
"step": 200
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.32,
|
| 258 |
+
"eval_biology_entropy": 1.430639804840088,
|
| 259 |
+
"eval_biology_loss": 1.4240669012069702,
|
| 260 |
+
"eval_biology_mean_token_accuracy": 0.6643721928596497,
|
| 261 |
+
"eval_biology_num_tokens": 3920545.0,
|
| 262 |
+
"eval_biology_runtime": 36.631,
|
| 263 |
+
"eval_biology_samples_per_second": 13.65,
|
| 264 |
+
"eval_biology_steps_per_second": 3.412,
|
| 265 |
+
"step": 200
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.32,
|
| 269 |
+
"eval_chemistry_entropy": 1.068233790397644,
|
| 270 |
+
"eval_chemistry_loss": 1.0446134805679321,
|
| 271 |
+
"eval_chemistry_mean_token_accuracy": 0.7419922127723694,
|
| 272 |
+
"eval_chemistry_num_tokens": 3920545.0,
|
| 273 |
+
"eval_chemistry_runtime": 45.0089,
|
| 274 |
+
"eval_chemistry_samples_per_second": 11.109,
|
| 275 |
+
"eval_chemistry_steps_per_second": 2.777,
|
| 276 |
+
"step": 200
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.32,
|
| 280 |
+
"eval_math_entropy": 0.8016578803062439,
|
| 281 |
+
"eval_math_loss": 1.064239740371704,
|
| 282 |
+
"eval_math_mean_token_accuracy": 0.7532512550354004,
|
| 283 |
+
"eval_math_num_tokens": 3920545.0,
|
| 284 |
+
"eval_math_runtime": 46.1998,
|
| 285 |
+
"eval_math_samples_per_second": 10.823,
|
| 286 |
+
"eval_math_steps_per_second": 2.706,
|
| 287 |
+
"step": 200
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.32,
|
| 291 |
+
"eval_physics_entropy": 1.0390429463386535,
|
| 292 |
+
"eval_physics_loss": 1.033566951751709,
|
| 293 |
+
"eval_physics_mean_token_accuracy": 0.751745885848999,
|
| 294 |
+
"eval_physics_num_tokens": 3920545.0,
|
| 295 |
+
"eval_physics_runtime": 53.0382,
|
| 296 |
+
"eval_physics_samples_per_second": 9.427,
|
| 297 |
+
"eval_physics_steps_per_second": 2.357,
|
| 298 |
+
"step": 200
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"entropy": 1.4409266009926796,
|
| 302 |
+
"epoch": 0.336,
|
| 303 |
+
"grad_norm": 30.5,
|
| 304 |
+
"learning_rate": 1.045e-05,
|
| 305 |
+
"loss": 23.1469,
|
| 306 |
+
"mean_token_accuracy": 0.6563692390918732,
|
| 307 |
+
"num_tokens": 4114077.0,
|
| 308 |
+
"step": 210
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"entropy": 1.4634413696825503,
|
| 312 |
+
"epoch": 0.352,
|
| 313 |
+
"grad_norm": 29.875,
|
| 314 |
+
"learning_rate": 1.095e-05,
|
| 315 |
+
"loss": 23.4319,
|
| 316 |
+
"mean_token_accuracy": 0.6559967070817947,
|
| 317 |
+
"num_tokens": 4306949.0,
|
| 318 |
+
"step": 220
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"entropy": 1.4390382423996926,
|
| 322 |
+
"epoch": 0.368,
|
| 323 |
+
"grad_norm": 29.5,
|
| 324 |
+
"learning_rate": 1.145e-05,
|
| 325 |
+
"loss": 23.1058,
|
| 326 |
+
"mean_token_accuracy": 0.6596628129482269,
|
| 327 |
+
"num_tokens": 4504001.0,
|
| 328 |
+
"step": 230
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"entropy": 1.4329517252743245,
|
| 332 |
+
"epoch": 0.384,
|
| 333 |
+
"grad_norm": 30.625,
|
| 334 |
+
"learning_rate": 1.195e-05,
|
| 335 |
+
"loss": 23.0101,
|
| 336 |
+
"mean_token_accuracy": 0.6596732661128044,
|
| 337 |
+
"num_tokens": 4693812.0,
|
| 338 |
+
"step": 240
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"entropy": 1.4356985121965409,
|
| 342 |
+
"epoch": 0.4,
|
| 343 |
+
"grad_norm": 26.0,
|
| 344 |
+
"learning_rate": 1.2450000000000003e-05,
|
| 345 |
+
"loss": 22.9403,
|
| 346 |
+
"mean_token_accuracy": 0.6609152596443891,
|
| 347 |
+
"num_tokens": 4887094.0,
|
| 348 |
+
"step": 250
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"entropy": 1.445911981910467,
|
| 352 |
+
"epoch": 0.416,
|
| 353 |
+
"grad_norm": 28.0,
|
| 354 |
+
"learning_rate": 1.295e-05,
|
| 355 |
+
"loss": 23.2032,
|
| 356 |
+
"mean_token_accuracy": 0.6575265824794769,
|
| 357 |
+
"num_tokens": 5085369.0,
|
| 358 |
+
"step": 260
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"entropy": 1.4276192136108876,
|
| 362 |
+
"epoch": 0.432,
|
| 363 |
+
"grad_norm": 30.25,
|
| 364 |
+
"learning_rate": 1.3450000000000002e-05,
|
| 365 |
+
"loss": 22.9807,
|
| 366 |
+
"mean_token_accuracy": 0.6605408634990454,
|
| 367 |
+
"num_tokens": 5271275.0,
|
| 368 |
+
"step": 270
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"entropy": 1.4254673644900322,
|
| 372 |
+
"epoch": 0.448,
|
| 373 |
+
"grad_norm": 32.0,
|
| 374 |
+
"learning_rate": 1.3950000000000002e-05,
|
| 375 |
+
"loss": 22.9563,
|
| 376 |
+
"mean_token_accuracy": 0.6602664466947317,
|
| 377 |
+
"num_tokens": 5460559.0,
|
| 378 |
+
"step": 280
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"entropy": 1.3786304764449597,
|
| 382 |
+
"epoch": 0.464,
|
| 383 |
+
"grad_norm": 28.375,
|
| 384 |
+
"learning_rate": 1.4450000000000002e-05,
|
| 385 |
+
"loss": 22.0498,
|
| 386 |
+
"mean_token_accuracy": 0.6699710045009851,
|
| 387 |
+
"num_tokens": 5653809.0,
|
| 388 |
+
"step": 290
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"entropy": 1.3980159699916839,
|
| 392 |
+
"epoch": 0.48,
|
| 393 |
+
"grad_norm": 26.0,
|
| 394 |
+
"learning_rate": 1.4950000000000003e-05,
|
| 395 |
+
"loss": 22.5067,
|
| 396 |
+
"mean_token_accuracy": 0.6643361520022154,
|
| 397 |
+
"num_tokens": 5850176.0,
|
| 398 |
+
"step": 300
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.48,
|
| 402 |
+
"eval_biology_entropy": 1.378262495994568,
|
| 403 |
+
"eval_biology_loss": 1.34170663356781,
|
| 404 |
+
"eval_biology_mean_token_accuracy": 0.6778094511032104,
|
| 405 |
+
"eval_biology_num_tokens": 5850176.0,
|
| 406 |
+
"eval_biology_runtime": 36.5741,
|
| 407 |
+
"eval_biology_samples_per_second": 13.671,
|
| 408 |
+
"eval_biology_steps_per_second": 3.418,
|
| 409 |
+
"step": 300
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.48,
|
| 413 |
+
"eval_chemistry_entropy": 1.0265604996681212,
|
| 414 |
+
"eval_chemistry_loss": 0.9883458018302917,
|
| 415 |
+
"eval_chemistry_mean_token_accuracy": 0.7520135464668274,
|
| 416 |
+
"eval_chemistry_num_tokens": 5850176.0,
|
| 417 |
+
"eval_chemistry_runtime": 44.9771,
|
| 418 |
+
"eval_chemistry_samples_per_second": 11.117,
|
| 419 |
+
"eval_chemistry_steps_per_second": 2.779,
|
| 420 |
+
"step": 300
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.48,
|
| 424 |
+
"eval_math_entropy": 0.7964297127723694,
|
| 425 |
+
"eval_math_loss": 1.0260037183761597,
|
| 426 |
+
"eval_math_mean_token_accuracy": 0.759199161529541,
|
| 427 |
+
"eval_math_num_tokens": 5850176.0,
|
| 428 |
+
"eval_math_runtime": 46.2056,
|
| 429 |
+
"eval_math_samples_per_second": 10.821,
|
| 430 |
+
"eval_math_steps_per_second": 2.705,
|
| 431 |
+
"step": 300
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.48,
|
| 435 |
+
"eval_physics_entropy": 1.0048614959716797,
|
| 436 |
+
"eval_physics_loss": 0.9760251045227051,
|
| 437 |
+
"eval_physics_mean_token_accuracy": 0.761619930267334,
|
| 438 |
+
"eval_physics_num_tokens": 5850176.0,
|
| 439 |
+
"eval_physics_runtime": 53.0249,
|
| 440 |
+
"eval_physics_samples_per_second": 9.43,
|
| 441 |
+
"eval_physics_steps_per_second": 2.357,
|
| 442 |
+
"step": 300
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"entropy": 1.3288711309432983,
|
| 446 |
+
"epoch": 0.496,
|
| 447 |
+
"grad_norm": 26.375,
|
| 448 |
+
"learning_rate": 1.545e-05,
|
| 449 |
+
"loss": 21.2235,
|
| 450 |
+
"mean_token_accuracy": 0.6774784430861474,
|
| 451 |
+
"num_tokens": 6046503.0,
|
| 452 |
+
"step": 310
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"entropy": 1.385315388813615,
|
| 456 |
+
"epoch": 0.512,
|
| 457 |
+
"grad_norm": 27.125,
|
| 458 |
+
"learning_rate": 1.595e-05,
|
| 459 |
+
"loss": 22.2247,
|
| 460 |
+
"mean_token_accuracy": 0.6680108610540628,
|
| 461 |
+
"num_tokens": 6240456.0,
|
| 462 |
+
"step": 320
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"entropy": 1.3709059737622737,
|
| 466 |
+
"epoch": 0.528,
|
| 467 |
+
"grad_norm": 24.75,
|
| 468 |
+
"learning_rate": 1.645e-05,
|
| 469 |
+
"loss": 22.1079,
|
| 470 |
+
"mean_token_accuracy": 0.6704635102301836,
|
| 471 |
+
"num_tokens": 6430555.0,
|
| 472 |
+
"step": 330
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"entropy": 1.3751415818929673,
|
| 476 |
+
"epoch": 0.544,
|
| 477 |
+
"grad_norm": 26.25,
|
| 478 |
+
"learning_rate": 1.6950000000000002e-05,
|
| 479 |
+
"loss": 21.961,
|
| 480 |
+
"mean_token_accuracy": 0.6710947025567293,
|
| 481 |
+
"num_tokens": 6626006.0,
|
| 482 |
+
"step": 340
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"entropy": 1.3602795064449311,
|
| 486 |
+
"epoch": 0.56,
|
| 487 |
+
"grad_norm": 30.625,
|
| 488 |
+
"learning_rate": 1.7450000000000004e-05,
|
| 489 |
+
"loss": 21.9474,
|
| 490 |
+
"mean_token_accuracy": 0.6700355879962444,
|
| 491 |
+
"num_tokens": 6820754.0,
|
| 492 |
+
"step": 350
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"entropy": 1.344116109609604,
|
| 496 |
+
"epoch": 0.576,
|
| 497 |
+
"grad_norm": 26.375,
|
| 498 |
+
"learning_rate": 1.795e-05,
|
| 499 |
+
"loss": 21.5572,
|
| 500 |
+
"mean_token_accuracy": 0.6739088613539934,
|
| 501 |
+
"num_tokens": 7021844.0,
|
| 502 |
+
"step": 360
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"entropy": 1.3490931887179614,
|
| 506 |
+
"epoch": 0.592,
|
| 507 |
+
"grad_norm": 26.125,
|
| 508 |
+
"learning_rate": 1.845e-05,
|
| 509 |
+
"loss": 21.7554,
|
| 510 |
+
"mean_token_accuracy": 0.6715496938675642,
|
| 511 |
+
"num_tokens": 7213951.0,
|
| 512 |
+
"step": 370
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"entropy": 1.3264936488121748,
|
| 516 |
+
"epoch": 0.608,
|
| 517 |
+
"grad_norm": 27.25,
|
| 518 |
+
"learning_rate": 1.8950000000000003e-05,
|
| 519 |
+
"loss": 21.3421,
|
| 520 |
+
"mean_token_accuracy": 0.6760978367179632,
|
| 521 |
+
"num_tokens": 7416773.0,
|
| 522 |
+
"step": 380
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"entropy": 1.348380234092474,
|
| 526 |
+
"epoch": 0.624,
|
| 527 |
+
"grad_norm": 28.125,
|
| 528 |
+
"learning_rate": 1.9450000000000002e-05,
|
| 529 |
+
"loss": 21.6293,
|
| 530 |
+
"mean_token_accuracy": 0.6746726848185063,
|
| 531 |
+
"num_tokens": 7612843.0,
|
| 532 |
+
"step": 390
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"entropy": 1.3465636018663645,
|
| 536 |
+
"epoch": 0.64,
|
| 537 |
+
"grad_norm": 26.25,
|
| 538 |
+
"learning_rate": 1.9950000000000004e-05,
|
| 539 |
+
"loss": 21.6324,
|
| 540 |
+
"mean_token_accuracy": 0.6738044265657663,
|
| 541 |
+
"num_tokens": 7801633.0,
|
| 542 |
+
"step": 400
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.64,
|
| 546 |
+
"eval_biology_entropy": 1.3016406693458558,
|
| 547 |
+
"eval_biology_loss": 1.2880380153656006,
|
| 548 |
+
"eval_biology_mean_token_accuracy": 0.6870926094055175,
|
| 549 |
+
"eval_biology_num_tokens": 7801633.0,
|
| 550 |
+
"eval_biology_runtime": 36.6216,
|
| 551 |
+
"eval_biology_samples_per_second": 13.653,
|
| 552 |
+
"eval_biology_steps_per_second": 3.413,
|
| 553 |
+
"step": 400
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 0.64,
|
| 557 |
+
"eval_chemistry_entropy": 0.9742186889648438,
|
| 558 |
+
"eval_chemistry_loss": 0.954098641872406,
|
| 559 |
+
"eval_chemistry_mean_token_accuracy": 0.7588451662063599,
|
| 560 |
+
"eval_chemistry_num_tokens": 7801633.0,
|
| 561 |
+
"eval_chemistry_runtime": 44.9361,
|
| 562 |
+
"eval_chemistry_samples_per_second": 11.127,
|
| 563 |
+
"eval_chemistry_steps_per_second": 2.782,
|
| 564 |
+
"step": 400
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.64,
|
| 568 |
+
"eval_math_entropy": 0.7798485422134399,
|
| 569 |
+
"eval_math_loss": 1.0040466785430908,
|
| 570 |
+
"eval_math_mean_token_accuracy": 0.7620619239807129,
|
| 571 |
+
"eval_math_num_tokens": 7801633.0,
|
| 572 |
+
"eval_math_runtime": 46.2072,
|
| 573 |
+
"eval_math_samples_per_second": 10.821,
|
| 574 |
+
"eval_math_steps_per_second": 2.705,
|
| 575 |
+
"step": 400
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.64,
|
| 579 |
+
"eval_physics_entropy": 0.9594037117958069,
|
| 580 |
+
"eval_physics_loss": 0.9417626261711121,
|
| 581 |
+
"eval_physics_mean_token_accuracy": 0.7674873929023742,
|
| 582 |
+
"eval_physics_num_tokens": 7801633.0,
|
| 583 |
+
"eval_physics_runtime": 53.0463,
|
| 584 |
+
"eval_physics_samples_per_second": 9.426,
|
| 585 |
+
"eval_physics_steps_per_second": 2.356,
|
| 586 |
+
"step": 400
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"entropy": 1.3527625199407338,
|
| 590 |
+
"epoch": 0.656,
|
| 591 |
+
"grad_norm": 27.875,
|
| 592 |
+
"learning_rate": 1.9950000000000004e-05,
|
| 593 |
+
"loss": 21.651,
|
| 594 |
+
"mean_token_accuracy": 0.6724541377276182,
|
| 595 |
+
"num_tokens": 7995843.0,
|
| 596 |
+
"step": 410
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"entropy": 1.3500655848532914,
|
| 600 |
+
"epoch": 0.672,
|
| 601 |
+
"grad_norm": 27.125,
|
| 602 |
+
"learning_rate": 1.9894444444444446e-05,
|
| 603 |
+
"loss": 21.6896,
|
| 604 |
+
"mean_token_accuracy": 0.673105026781559,
|
| 605 |
+
"num_tokens": 8183103.0,
|
| 606 |
+
"step": 420
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"entropy": 1.3231504444032907,
|
| 610 |
+
"epoch": 0.688,
|
| 611 |
+
"grad_norm": 23.5,
|
| 612 |
+
"learning_rate": 1.983888888888889e-05,
|
| 613 |
+
"loss": 21.3727,
|
| 614 |
+
"mean_token_accuracy": 0.6781029276549816,
|
| 615 |
+
"num_tokens": 8385976.0,
|
| 616 |
+
"step": 430
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"entropy": 1.3333053421229124,
|
| 620 |
+
"epoch": 0.704,
|
| 621 |
+
"grad_norm": 23.5,
|
| 622 |
+
"learning_rate": 1.9783333333333334e-05,
|
| 623 |
+
"loss": 21.2722,
|
| 624 |
+
"mean_token_accuracy": 0.6781596153974533,
|
| 625 |
+
"num_tokens": 8578431.0,
|
| 626 |
+
"step": 440
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"entropy": 1.2669135373085738,
|
| 630 |
+
"epoch": 0.72,
|
| 631 |
+
"grad_norm": 23.375,
|
| 632 |
+
"learning_rate": 1.972777777777778e-05,
|
| 633 |
+
"loss": 20.3907,
|
| 634 |
+
"mean_token_accuracy": 0.6872118715196848,
|
| 635 |
+
"num_tokens": 8781342.0,
|
| 636 |
+
"step": 450
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"entropy": 1.2877357687801123,
|
| 640 |
+
"epoch": 0.736,
|
| 641 |
+
"grad_norm": 23.875,
|
| 642 |
+
"learning_rate": 1.9672222222222222e-05,
|
| 643 |
+
"loss": 20.6999,
|
| 644 |
+
"mean_token_accuracy": 0.6825123306363821,
|
| 645 |
+
"num_tokens": 8977918.0,
|
| 646 |
+
"step": 460
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"entropy": 1.3120017491281033,
|
| 650 |
+
"epoch": 0.752,
|
| 651 |
+
"grad_norm": 24.5,
|
| 652 |
+
"learning_rate": 1.9616666666666668e-05,
|
| 653 |
+
"loss": 21.1222,
|
| 654 |
+
"mean_token_accuracy": 0.6796863108873368,
|
| 655 |
+
"num_tokens": 9169322.0,
|
| 656 |
+
"step": 470
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"entropy": 1.263747826963663,
|
| 660 |
+
"epoch": 0.768,
|
| 661 |
+
"grad_norm": 26.375,
|
| 662 |
+
"learning_rate": 1.9561111111111114e-05,
|
| 663 |
+
"loss": 20.3726,
|
| 664 |
+
"mean_token_accuracy": 0.6869186259806156,
|
| 665 |
+
"num_tokens": 9368141.0,
|
| 666 |
+
"step": 480
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"entropy": 1.3065229691565037,
|
| 670 |
+
"epoch": 0.784,
|
| 671 |
+
"grad_norm": 26.5,
|
| 672 |
+
"learning_rate": 1.9505555555555556e-05,
|
| 673 |
+
"loss": 20.8536,
|
| 674 |
+
"mean_token_accuracy": 0.683413552865386,
|
| 675 |
+
"num_tokens": 9565236.0,
|
| 676 |
+
"step": 490
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"entropy": 1.3189398009330033,
|
| 680 |
+
"epoch": 0.8,
|
| 681 |
+
"grad_norm": 25.5,
|
| 682 |
+
"learning_rate": 1.9450000000000002e-05,
|
| 683 |
+
"loss": 21.2576,
|
| 684 |
+
"mean_token_accuracy": 0.6789898939430714,
|
| 685 |
+
"num_tokens": 9761227.0,
|
| 686 |
+
"step": 500
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 0.8,
|
| 690 |
+
"eval_biology_entropy": 1.2551557726860045,
|
| 691 |
+
"eval_biology_loss": 1.2525163888931274,
|
| 692 |
+
"eval_biology_mean_token_accuracy": 0.6931379227638245,
|
| 693 |
+
"eval_biology_num_tokens": 9761227.0,
|
| 694 |
+
"eval_biology_runtime": 36.6065,
|
| 695 |
+
"eval_biology_samples_per_second": 13.659,
|
| 696 |
+
"eval_biology_steps_per_second": 3.415,
|
| 697 |
+
"step": 500
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.8,
|
| 701 |
+
"eval_chemistry_entropy": 0.9375595455169677,
|
| 702 |
+
"eval_chemistry_loss": 0.9276740550994873,
|
| 703 |
+
"eval_chemistry_mean_token_accuracy": 0.7639822545051574,
|
| 704 |
+
"eval_chemistry_num_tokens": 9761227.0,
|
| 705 |
+
"eval_chemistry_runtime": 44.5343,
|
| 706 |
+
"eval_chemistry_samples_per_second": 11.227,
|
| 707 |
+
"eval_chemistry_steps_per_second": 2.807,
|
| 708 |
+
"step": 500
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 0.8,
|
| 712 |
+
"eval_math_entropy": 0.7748257055282592,
|
| 713 |
+
"eval_math_loss": 0.9847215414047241,
|
| 714 |
+
"eval_math_mean_token_accuracy": 0.7653367323875427,
|
| 715 |
+
"eval_math_num_tokens": 9761227.0,
|
| 716 |
+
"eval_math_runtime": 45.9419,
|
| 717 |
+
"eval_math_samples_per_second": 10.883,
|
| 718 |
+
"eval_math_steps_per_second": 2.721,
|
| 719 |
+
"step": 500
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 0.8,
|
| 723 |
+
"eval_physics_entropy": 0.9384980492591858,
|
| 724 |
+
"eval_physics_loss": 0.9171136021614075,
|
| 725 |
+
"eval_physics_mean_token_accuracy": 0.7720685515403748,
|
| 726 |
+
"eval_physics_num_tokens": 9761227.0,
|
| 727 |
+
"eval_physics_runtime": 52.888,
|
| 728 |
+
"eval_physics_samples_per_second": 9.454,
|
| 729 |
+
"eval_physics_steps_per_second": 2.363,
|
| 730 |
+
"step": 500
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"entropy": 1.2767818640917539,
|
| 734 |
+
"epoch": 0.816,
|
| 735 |
+
"grad_norm": 25.5,
|
| 736 |
+
"learning_rate": 1.9394444444444448e-05,
|
| 737 |
+
"loss": 20.4527,
|
| 738 |
+
"mean_token_accuracy": 0.6852765031158924,
|
| 739 |
+
"num_tokens": 9958727.0,
|
| 740 |
+
"step": 510
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"entropy": 1.2603609178215265,
|
| 744 |
+
"epoch": 0.832,
|
| 745 |
+
"grad_norm": 23.0,
|
| 746 |
+
"learning_rate": 1.933888888888889e-05,
|
| 747 |
+
"loss": 20.3411,
|
| 748 |
+
"mean_token_accuracy": 0.6862579733133316,
|
| 749 |
+
"num_tokens": 10155771.0,
|
| 750 |
+
"step": 520
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"entropy": 1.2630852952599525,
|
| 754 |
+
"epoch": 0.848,
|
| 755 |
+
"grad_norm": 26.375,
|
| 756 |
+
"learning_rate": 1.9283333333333332e-05,
|
| 757 |
+
"loss": 20.316,
|
| 758 |
+
"mean_token_accuracy": 0.6886717110872269,
|
| 759 |
+
"num_tokens": 10357721.0,
|
| 760 |
+
"step": 530
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"entropy": 1.2910415723919868,
|
| 764 |
+
"epoch": 0.864,
|
| 765 |
+
"grad_norm": 25.0,
|
| 766 |
+
"learning_rate": 1.9227777777777778e-05,
|
| 767 |
+
"loss": 20.7324,
|
| 768 |
+
"mean_token_accuracy": 0.6839874558150768,
|
| 769 |
+
"num_tokens": 10552495.0,
|
| 770 |
+
"step": 540
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"entropy": 1.2793519891798497,
|
| 774 |
+
"epoch": 0.88,
|
| 775 |
+
"grad_norm": 26.625,
|
| 776 |
+
"learning_rate": 1.9172222222222224e-05,
|
| 777 |
+
"loss": 20.4504,
|
| 778 |
+
"mean_token_accuracy": 0.6864676021039486,
|
| 779 |
+
"num_tokens": 10748749.0,
|
| 780 |
+
"step": 550
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"entropy": 1.2906367748975753,
|
| 784 |
+
"epoch": 0.896,
|
| 785 |
+
"grad_norm": 24.875,
|
| 786 |
+
"learning_rate": 1.911666666666667e-05,
|
| 787 |
+
"loss": 20.7703,
|
| 788 |
+
"mean_token_accuracy": 0.6835371732711792,
|
| 789 |
+
"num_tokens": 10943319.0,
|
| 790 |
+
"step": 560
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"entropy": 1.2809050466865302,
|
| 794 |
+
"epoch": 0.912,
|
| 795 |
+
"grad_norm": 26.0,
|
| 796 |
+
"learning_rate": 1.9061111111111112e-05,
|
| 797 |
+
"loss": 20.5319,
|
| 798 |
+
"mean_token_accuracy": 0.6867790374904871,
|
| 799 |
+
"num_tokens": 11136935.0,
|
| 800 |
+
"step": 570
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"entropy": 1.283418832719326,
|
| 804 |
+
"epoch": 0.928,
|
| 805 |
+
"grad_norm": 23.875,
|
| 806 |
+
"learning_rate": 1.9005555555555557e-05,
|
| 807 |
+
"loss": 20.6537,
|
| 808 |
+
"mean_token_accuracy": 0.6832854464650154,
|
| 809 |
+
"num_tokens": 11331098.0,
|
| 810 |
+
"step": 580
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"entropy": 1.2303649988025427,
|
| 814 |
+
"epoch": 0.944,
|
| 815 |
+
"grad_norm": 21.5,
|
| 816 |
+
"learning_rate": 1.8950000000000003e-05,
|
| 817 |
+
"loss": 19.7901,
|
| 818 |
+
"mean_token_accuracy": 0.6927386663854123,
|
| 819 |
+
"num_tokens": 11530550.0,
|
| 820 |
+
"step": 590
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"entropy": 1.2655960384756326,
|
| 824 |
+
"epoch": 0.96,
|
| 825 |
+
"grad_norm": 26.125,
|
| 826 |
+
"learning_rate": 1.8894444444444446e-05,
|
| 827 |
+
"loss": 20.362,
|
| 828 |
+
"mean_token_accuracy": 0.6886723652482033,
|
| 829 |
+
"num_tokens": 11729645.0,
|
| 830 |
+
"step": 600
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.96,
|
| 834 |
+
"eval_biology_entropy": 1.2511886906623841,
|
| 835 |
+
"eval_biology_loss": 1.2273433208465576,
|
| 836 |
+
"eval_biology_mean_token_accuracy": 0.6978591928482055,
|
| 837 |
+
"eval_biology_num_tokens": 11729645.0,
|
| 838 |
+
"eval_biology_runtime": 36.8245,
|
| 839 |
+
"eval_biology_samples_per_second": 13.578,
|
| 840 |
+
"eval_biology_steps_per_second": 3.394,
|
| 841 |
+
"step": 600
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 0.96,
|
| 845 |
+
"eval_chemistry_entropy": 0.9398184275627136,
|
| 846 |
+
"eval_chemistry_loss": 0.9139939546585083,
|
| 847 |
+
"eval_chemistry_mean_token_accuracy": 0.7667500405311585,
|
| 848 |
+
"eval_chemistry_num_tokens": 11729645.0,
|
| 849 |
+
"eval_chemistry_runtime": 45.002,
|
| 850 |
+
"eval_chemistry_samples_per_second": 11.111,
|
| 851 |
+
"eval_chemistry_steps_per_second": 2.778,
|
| 852 |
+
"step": 600
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 0.96,
|
| 856 |
+
"eval_math_entropy": 0.7878352689743042,
|
| 857 |
+
"eval_math_loss": 0.9694741368293762,
|
| 858 |
+
"eval_math_mean_token_accuracy": 0.767361653804779,
|
| 859 |
+
"eval_math_num_tokens": 11729645.0,
|
| 860 |
+
"eval_math_runtime": 46.2016,
|
| 861 |
+
"eval_math_samples_per_second": 10.822,
|
| 862 |
+
"eval_math_steps_per_second": 2.706,
|
| 863 |
+
"step": 600
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.96,
|
| 867 |
+
"eval_physics_entropy": 0.9437616720199585,
|
| 868 |
+
"eval_physics_loss": 0.9044761061668396,
|
| 869 |
+
"eval_physics_mean_token_accuracy": 0.7742767133712769,
|
| 870 |
+
"eval_physics_num_tokens": 11729645.0,
|
| 871 |
+
"eval_physics_runtime": 53.0008,
|
| 872 |
+
"eval_physics_samples_per_second": 9.434,
|
| 873 |
+
"eval_physics_steps_per_second": 2.358,
|
| 874 |
+
"step": 600
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"entropy": 1.2491024732589722,
|
| 878 |
+
"epoch": 0.976,
|
| 879 |
+
"grad_norm": 25.875,
|
| 880 |
+
"learning_rate": 1.883888888888889e-05,
|
| 881 |
+
"loss": 19.9555,
|
| 882 |
+
"mean_token_accuracy": 0.6900846060365439,
|
| 883 |
+
"num_tokens": 11924644.0,
|
| 884 |
+
"step": 610
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"entropy": 1.2545549362897872,
|
| 888 |
+
"epoch": 0.992,
|
| 889 |
+
"grad_norm": 23.0,
|
| 890 |
+
"learning_rate": 1.8783333333333337e-05,
|
| 891 |
+
"loss": 20.317,
|
| 892 |
+
"mean_token_accuracy": 0.686628095433116,
|
| 893 |
+
"num_tokens": 12123059.0,
|
| 894 |
+
"step": 620
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"entropy": 1.2396365478634834,
|
| 898 |
+
"epoch": 1.008,
|
| 899 |
+
"grad_norm": 22.5,
|
| 900 |
+
"learning_rate": 1.872777777777778e-05,
|
| 901 |
+
"loss": 19.6736,
|
| 902 |
+
"mean_token_accuracy": 0.6955156117677689,
|
| 903 |
+
"num_tokens": 12319366.0,
|
| 904 |
+
"step": 630
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"entropy": 1.143751036375761,
|
| 908 |
+
"epoch": 1.024,
|
| 909 |
+
"grad_norm": 24.5,
|
| 910 |
+
"learning_rate": 1.8672222222222222e-05,
|
| 911 |
+
"loss": 18.51,
|
| 912 |
+
"mean_token_accuracy": 0.7079542689025402,
|
| 913 |
+
"num_tokens": 12524183.0,
|
| 914 |
+
"step": 640
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"entropy": 1.1911311563104392,
|
| 918 |
+
"epoch": 1.04,
|
| 919 |
+
"grad_norm": 25.25,
|
| 920 |
+
"learning_rate": 1.8616666666666667e-05,
|
| 921 |
+
"loss": 19.0515,
|
| 922 |
+
"mean_token_accuracy": 0.6993662625551224,
|
| 923 |
+
"num_tokens": 12718593.0,
|
| 924 |
+
"step": 650
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"entropy": 1.1548987552523613,
|
| 928 |
+
"epoch": 1.056,
|
| 929 |
+
"grad_norm": 25.0,
|
| 930 |
+
"learning_rate": 1.8561111111111113e-05,
|
| 931 |
+
"loss": 18.5231,
|
| 932 |
+
"mean_token_accuracy": 0.707820488885045,
|
| 933 |
+
"num_tokens": 12917803.0,
|
| 934 |
+
"step": 660
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"entropy": 1.1886027652770281,
|
| 938 |
+
"epoch": 1.072,
|
| 939 |
+
"grad_norm": 26.0,
|
| 940 |
+
"learning_rate": 1.8505555555555556e-05,
|
| 941 |
+
"loss": 18.9314,
|
| 942 |
+
"mean_token_accuracy": 0.7034936733543873,
|
| 943 |
+
"num_tokens": 13105826.0,
|
| 944 |
+
"step": 670
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"entropy": 1.1603233687579633,
|
| 948 |
+
"epoch": 1.088,
|
| 949 |
+
"grad_norm": 24.5,
|
| 950 |
+
"learning_rate": 1.845e-05,
|
| 951 |
+
"loss": 18.681,
|
| 952 |
+
"mean_token_accuracy": 0.7059972729533911,
|
| 953 |
+
"num_tokens": 13298619.0,
|
| 954 |
+
"step": 680
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"entropy": 1.1554178480058908,
|
| 958 |
+
"epoch": 1.104,
|
| 959 |
+
"grad_norm": 24.0,
|
| 960 |
+
"learning_rate": 1.8394444444444447e-05,
|
| 961 |
+
"loss": 18.6001,
|
| 962 |
+
"mean_token_accuracy": 0.7077722396701575,
|
| 963 |
+
"num_tokens": 13491486.0,
|
| 964 |
+
"step": 690
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"entropy": 1.1779750995337963,
|
| 968 |
+
"epoch": 1.12,
|
| 969 |
+
"grad_norm": 26.125,
|
| 970 |
+
"learning_rate": 1.833888888888889e-05,
|
| 971 |
+
"loss": 18.886,
|
| 972 |
+
"mean_token_accuracy": 0.7039439618587494,
|
| 973 |
+
"num_tokens": 13674663.0,
|
| 974 |
+
"step": 700
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 1.12,
|
| 978 |
+
"eval_biology_entropy": 1.115936914920807,
|
| 979 |
+
"eval_biology_loss": 1.2197552919387817,
|
| 980 |
+
"eval_biology_mean_token_accuracy": 0.6998812022209168,
|
| 981 |
+
"eval_biology_num_tokens": 13674663.0,
|
| 982 |
+
"eval_biology_runtime": 36.6382,
|
| 983 |
+
"eval_biology_samples_per_second": 13.647,
|
| 984 |
+
"eval_biology_steps_per_second": 3.412,
|
| 985 |
+
"step": 700
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 1.12,
|
| 989 |
+
"eval_chemistry_entropy": 0.8371734113693238,
|
| 990 |
+
"eval_chemistry_loss": 0.9145113229751587,
|
| 991 |
+
"eval_chemistry_mean_token_accuracy": 0.7663986759185791,
|
| 992 |
+
"eval_chemistry_num_tokens": 13674663.0,
|
| 993 |
+
"eval_chemistry_runtime": 45.0064,
|
| 994 |
+
"eval_chemistry_samples_per_second": 11.11,
|
| 995 |
+
"eval_chemistry_steps_per_second": 2.777,
|
| 996 |
+
"step": 700
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 1.12,
|
| 1000 |
+
"eval_math_entropy": 0.7305963957309722,
|
| 1001 |
+
"eval_math_loss": 0.9764987826347351,
|
| 1002 |
+
"eval_math_mean_token_accuracy": 0.766656831741333,
|
| 1003 |
+
"eval_math_num_tokens": 13674663.0,
|
| 1004 |
+
"eval_math_runtime": 46.2254,
|
| 1005 |
+
"eval_math_samples_per_second": 10.817,
|
| 1006 |
+
"eval_math_steps_per_second": 2.704,
|
| 1007 |
+
"step": 700
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 1.12,
|
| 1011 |
+
"eval_physics_entropy": 0.852522144317627,
|
| 1012 |
+
"eval_physics_loss": 0.9007405638694763,
|
| 1013 |
+
"eval_physics_mean_token_accuracy": 0.7750241093635559,
|
| 1014 |
+
"eval_physics_num_tokens": 13674663.0,
|
| 1015 |
+
"eval_physics_runtime": 53.0213,
|
| 1016 |
+
"eval_physics_samples_per_second": 9.43,
|
| 1017 |
+
"eval_physics_steps_per_second": 2.358,
|
| 1018 |
+
"step": 700
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"entropy": 1.1570473805069923,
|
| 1022 |
+
"epoch": 1.1360000000000001,
|
| 1023 |
+
"grad_norm": 30.875,
|
| 1024 |
+
"learning_rate": 1.8283333333333335e-05,
|
| 1025 |
+
"loss": 18.7252,
|
| 1026 |
+
"mean_token_accuracy": 0.7053195346146822,
|
| 1027 |
+
"num_tokens": 13869134.0,
|
| 1028 |
+
"step": 710
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"entropy": 1.153097653761506,
|
| 1032 |
+
"epoch": 1.152,
|
| 1033 |
+
"grad_norm": 23.125,
|
| 1034 |
+
"learning_rate": 1.822777777777778e-05,
|
| 1035 |
+
"loss": 18.3474,
|
| 1036 |
+
"mean_token_accuracy": 0.7079035714268684,
|
| 1037 |
+
"num_tokens": 14078365.0,
|
| 1038 |
+
"step": 720
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"entropy": 1.1558022141456603,
|
| 1042 |
+
"epoch": 1.168,
|
| 1043 |
+
"grad_norm": 25.375,
|
| 1044 |
+
"learning_rate": 1.8172222222222223e-05,
|
| 1045 |
+
"loss": 18.6158,
|
| 1046 |
+
"mean_token_accuracy": 0.706475806236267,
|
| 1047 |
+
"num_tokens": 14266831.0,
|
| 1048 |
+
"step": 730
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"entropy": 1.1689991034567355,
|
| 1052 |
+
"epoch": 1.184,
|
| 1053 |
+
"grad_norm": 24.625,
|
| 1054 |
+
"learning_rate": 1.811666666666667e-05,
|
| 1055 |
+
"loss": 18.7482,
|
| 1056 |
+
"mean_token_accuracy": 0.7053946666419506,
|
| 1057 |
+
"num_tokens": 14465660.0,
|
| 1058 |
+
"step": 740
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"entropy": 1.1589057791978121,
|
| 1062 |
+
"epoch": 1.2,
|
| 1063 |
+
"grad_norm": 26.125,
|
| 1064 |
+
"learning_rate": 1.806111111111111e-05,
|
| 1065 |
+
"loss": 18.4761,
|
| 1066 |
+
"mean_token_accuracy": 0.7090217288583517,
|
| 1067 |
+
"num_tokens": 14653228.0,
|
| 1068 |
+
"step": 750
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"entropy": 1.1286497369408608,
|
| 1072 |
+
"epoch": 1.216,
|
| 1073 |
+
"grad_norm": 24.625,
|
| 1074 |
+
"learning_rate": 1.8005555555555557e-05,
|
| 1075 |
+
"loss": 18.0937,
|
| 1076 |
+
"mean_token_accuracy": 0.710760310664773,
|
| 1077 |
+
"num_tokens": 14857782.0,
|
| 1078 |
+
"step": 760
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"entropy": 1.1454383581876755,
|
| 1082 |
+
"epoch": 1.232,
|
| 1083 |
+
"grad_norm": 23.0,
|
| 1084 |
+
"learning_rate": 1.795e-05,
|
| 1085 |
+
"loss": 18.3666,
|
| 1086 |
+
"mean_token_accuracy": 0.7111793085932732,
|
| 1087 |
+
"num_tokens": 15047356.0,
|
| 1088 |
+
"step": 770
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"entropy": 1.1489330865442753,
|
| 1092 |
+
"epoch": 1.248,
|
| 1093 |
+
"grad_norm": 24.5,
|
| 1094 |
+
"learning_rate": 1.7894444444444445e-05,
|
| 1095 |
+
"loss": 18.4807,
|
| 1096 |
+
"mean_token_accuracy": 0.7070541866123676,
|
| 1097 |
+
"num_tokens": 15241098.0,
|
| 1098 |
+
"step": 780
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"entropy": 1.1616758443415165,
|
| 1102 |
+
"epoch": 1.264,
|
| 1103 |
+
"grad_norm": 23.0,
|
| 1104 |
+
"learning_rate": 1.783888888888889e-05,
|
| 1105 |
+
"loss": 18.6418,
|
| 1106 |
+
"mean_token_accuracy": 0.7069107063114644,
|
| 1107 |
+
"num_tokens": 15437657.0,
|
| 1108 |
+
"step": 790
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"entropy": 1.1660578336566687,
|
| 1112 |
+
"epoch": 1.28,
|
| 1113 |
+
"grad_norm": 26.875,
|
| 1114 |
+
"learning_rate": 1.7783333333333333e-05,
|
| 1115 |
+
"loss": 18.8491,
|
| 1116 |
+
"mean_token_accuracy": 0.7026976983994245,
|
| 1117 |
+
"num_tokens": 15630795.0,
|
| 1118 |
+
"step": 800
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 1.28,
|
| 1122 |
+
"eval_biology_entropy": 1.1199735798835755,
|
| 1123 |
+
"eval_biology_loss": 1.2088388204574585,
|
| 1124 |
+
"eval_biology_mean_token_accuracy": 0.7015871267318725,
|
| 1125 |
+
"eval_biology_num_tokens": 15630795.0,
|
| 1126 |
+
"eval_biology_runtime": 36.9629,
|
| 1127 |
+
"eval_biology_samples_per_second": 13.527,
|
| 1128 |
+
"eval_biology_steps_per_second": 3.382,
|
| 1129 |
+
"step": 800
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.28,
|
| 1133 |
+
"eval_chemistry_entropy": 0.8543037304878235,
|
| 1134 |
+
"eval_chemistry_loss": 0.9098371863365173,
|
| 1135 |
+
"eval_chemistry_mean_token_accuracy": 0.7672378678321838,
|
| 1136 |
+
"eval_chemistry_num_tokens": 15630795.0,
|
| 1137 |
+
"eval_chemistry_runtime": 44.9998,
|
| 1138 |
+
"eval_chemistry_samples_per_second": 11.111,
|
| 1139 |
+
"eval_chemistry_steps_per_second": 2.778,
|
| 1140 |
+
"step": 800
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 1.28,
|
| 1144 |
+
"eval_math_entropy": 0.7368179533481598,
|
| 1145 |
+
"eval_math_loss": 0.9706032872200012,
|
| 1146 |
+
"eval_math_mean_token_accuracy": 0.7682779974937439,
|
| 1147 |
+
"eval_math_num_tokens": 15630795.0,
|
| 1148 |
+
"eval_math_runtime": 46.2307,
|
| 1149 |
+
"eval_math_samples_per_second": 10.815,
|
| 1150 |
+
"eval_math_steps_per_second": 2.704,
|
| 1151 |
+
"step": 800
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 1.28,
|
| 1155 |
+
"eval_physics_entropy": 0.8641948227882386,
|
| 1156 |
+
"eval_physics_loss": 0.8950499296188354,
|
| 1157 |
+
"eval_physics_mean_token_accuracy": 0.7758663225173951,
|
| 1158 |
+
"eval_physics_num_tokens": 15630795.0,
|
| 1159 |
+
"eval_physics_runtime": 53.0442,
|
| 1160 |
+
"eval_physics_samples_per_second": 9.426,
|
| 1161 |
+
"eval_physics_steps_per_second": 2.357,
|
| 1162 |
+
"step": 800
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"entropy": 1.1806952260434627,
|
| 1166 |
+
"epoch": 1.296,
|
| 1167 |
+
"grad_norm": 26.125,
|
| 1168 |
+
"learning_rate": 1.772777777777778e-05,
|
| 1169 |
+
"loss": 18.9295,
|
| 1170 |
+
"mean_token_accuracy": 0.70291638225317,
|
| 1171 |
+
"num_tokens": 15827105.0,
|
| 1172 |
+
"step": 810
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"entropy": 1.1602427676320075,
|
| 1176 |
+
"epoch": 1.312,
|
| 1177 |
+
"grad_norm": 27.625,
|
| 1178 |
+
"learning_rate": 1.7672222222222224e-05,
|
| 1179 |
+
"loss": 18.7911,
|
| 1180 |
+
"mean_token_accuracy": 0.7041201893240213,
|
| 1181 |
+
"num_tokens": 16019645.0,
|
| 1182 |
+
"step": 820
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"entropy": 1.1715985681861638,
|
| 1186 |
+
"epoch": 1.328,
|
| 1187 |
+
"grad_norm": 24.375,
|
| 1188 |
+
"learning_rate": 1.761666666666667e-05,
|
| 1189 |
+
"loss": 18.8262,
|
| 1190 |
+
"mean_token_accuracy": 0.7034816723316908,
|
| 1191 |
+
"num_tokens": 16221726.0,
|
| 1192 |
+
"step": 830
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"entropy": 1.1461844369769096,
|
| 1196 |
+
"epoch": 1.3439999999999999,
|
| 1197 |
+
"grad_norm": 25.5,
|
| 1198 |
+
"learning_rate": 1.7561111111111113e-05,
|
| 1199 |
+
"loss": 18.2642,
|
| 1200 |
+
"mean_token_accuracy": 0.7110818449407816,
|
| 1201 |
+
"num_tokens": 16427594.0,
|
| 1202 |
+
"step": 840
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"entropy": 1.1582913786172866,
|
| 1206 |
+
"epoch": 1.3599999999999999,
|
| 1207 |
+
"grad_norm": 25.25,
|
| 1208 |
+
"learning_rate": 1.7505555555555558e-05,
|
| 1209 |
+
"loss": 18.5772,
|
| 1210 |
+
"mean_token_accuracy": 0.7078793905675411,
|
| 1211 |
+
"num_tokens": 16621605.0,
|
| 1212 |
+
"step": 850
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"entropy": 1.1942507456988096,
|
| 1216 |
+
"epoch": 1.376,
|
| 1217 |
+
"grad_norm": 27.875,
|
| 1218 |
+
"learning_rate": 1.7450000000000004e-05,
|
| 1219 |
+
"loss": 19.0764,
|
| 1220 |
+
"mean_token_accuracy": 0.701082956790924,
|
| 1221 |
+
"num_tokens": 16813444.0,
|
| 1222 |
+
"step": 860
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"entropy": 1.1594038642942905,
|
| 1226 |
+
"epoch": 1.392,
|
| 1227 |
+
"grad_norm": 24.125,
|
| 1228 |
+
"learning_rate": 1.7394444444444446e-05,
|
| 1229 |
+
"loss": 18.7411,
|
| 1230 |
+
"mean_token_accuracy": 0.7061580993235111,
|
| 1231 |
+
"num_tokens": 17006509.0,
|
| 1232 |
+
"step": 870
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"entropy": 1.1641225919127465,
|
| 1236 |
+
"epoch": 1.408,
|
| 1237 |
+
"grad_norm": 21.5,
|
| 1238 |
+
"learning_rate": 1.733888888888889e-05,
|
| 1239 |
+
"loss": 18.5028,
|
| 1240 |
+
"mean_token_accuracy": 0.7065019059926272,
|
| 1241 |
+
"num_tokens": 17197870.0,
|
| 1242 |
+
"step": 880
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"entropy": 1.153414449468255,
|
| 1246 |
+
"epoch": 1.424,
|
| 1247 |
+
"grad_norm": 24.125,
|
| 1248 |
+
"learning_rate": 1.7283333333333334e-05,
|
| 1249 |
+
"loss": 18.5571,
|
| 1250 |
+
"mean_token_accuracy": 0.7072413295507431,
|
| 1251 |
+
"num_tokens": 17394390.0,
|
| 1252 |
+
"step": 890
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"entropy": 1.1581655487418174,
|
| 1256 |
+
"epoch": 1.44,
|
| 1257 |
+
"grad_norm": 28.5,
|
| 1258 |
+
"learning_rate": 1.722777777777778e-05,
|
| 1259 |
+
"loss": 18.5762,
|
| 1260 |
+
"mean_token_accuracy": 0.7053147312253714,
|
| 1261 |
+
"num_tokens": 17587777.0,
|
| 1262 |
+
"step": 900
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 1.44,
|
| 1266 |
+
"eval_biology_entropy": 1.1328148250579835,
|
| 1267 |
+
"eval_biology_loss": 1.1959257125854492,
|
| 1268 |
+
"eval_biology_mean_token_accuracy": 0.7035633249282837,
|
| 1269 |
+
"eval_biology_num_tokens": 17587777.0,
|
| 1270 |
+
"eval_biology_runtime": 38.7737,
|
| 1271 |
+
"eval_biology_samples_per_second": 12.895,
|
| 1272 |
+
"eval_biology_steps_per_second": 3.224,
|
| 1273 |
+
"step": 900
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 1.44,
|
| 1277 |
+
"eval_chemistry_entropy": 0.8504280371665954,
|
| 1278 |
+
"eval_chemistry_loss": 0.9063042402267456,
|
| 1279 |
+
"eval_chemistry_mean_token_accuracy": 0.7678593263626099,
|
| 1280 |
+
"eval_chemistry_num_tokens": 17587777.0,
|
| 1281 |
+
"eval_chemistry_runtime": 44.9861,
|
| 1282 |
+
"eval_chemistry_samples_per_second": 11.115,
|
| 1283 |
+
"eval_chemistry_steps_per_second": 2.779,
|
| 1284 |
+
"step": 900
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.44,
|
| 1288 |
+
"eval_math_entropy": 0.7459685401916504,
|
| 1289 |
+
"eval_math_loss": 0.9690774083137512,
|
| 1290 |
+
"eval_math_mean_token_accuracy": 0.7676572284698486,
|
| 1291 |
+
"eval_math_num_tokens": 17587777.0,
|
| 1292 |
+
"eval_math_runtime": 82.6785,
|
| 1293 |
+
"eval_math_samples_per_second": 6.048,
|
| 1294 |
+
"eval_math_steps_per_second": 1.512,
|
| 1295 |
+
"step": 900
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 1.44,
|
| 1299 |
+
"eval_physics_entropy": 0.8569722015857697,
|
| 1300 |
+
"eval_physics_loss": 0.8932082653045654,
|
| 1301 |
+
"eval_physics_mean_token_accuracy": 0.7766451034545898,
|
| 1302 |
+
"eval_physics_num_tokens": 17587777.0,
|
| 1303 |
+
"eval_physics_runtime": 52.7335,
|
| 1304 |
+
"eval_physics_samples_per_second": 9.482,
|
| 1305 |
+
"eval_physics_steps_per_second": 2.37,
|
| 1306 |
+
"step": 900
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"entropy": 1.1505886979401112,
|
| 1310 |
+
"epoch": 1.456,
|
| 1311 |
+
"grad_norm": 24.25,
|
| 1312 |
+
"learning_rate": 1.7172222222222223e-05,
|
| 1313 |
+
"loss": 18.5378,
|
| 1314 |
+
"mean_token_accuracy": 0.7084831334650517,
|
| 1315 |
+
"num_tokens": 17788456.0,
|
| 1316 |
+
"step": 910
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"entropy": 1.1548077858984471,
|
| 1320 |
+
"epoch": 1.472,
|
| 1321 |
+
"grad_norm": 24.125,
|
| 1322 |
+
"learning_rate": 1.7116666666666668e-05,
|
| 1323 |
+
"loss": 18.4342,
|
| 1324 |
+
"mean_token_accuracy": 0.7085961502045393,
|
| 1325 |
+
"num_tokens": 17984063.0,
|
| 1326 |
+
"step": 920
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"entropy": 1.1482487738132476,
|
| 1330 |
+
"epoch": 1.488,
|
| 1331 |
+
"grad_norm": 24.0,
|
| 1332 |
+
"learning_rate": 1.7061111111111114e-05,
|
| 1333 |
+
"loss": 18.4515,
|
| 1334 |
+
"mean_token_accuracy": 0.7099530510604382,
|
| 1335 |
+
"num_tokens": 18175640.0,
|
| 1336 |
+
"step": 930
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"entropy": 1.1528287250548601,
|
| 1340 |
+
"epoch": 1.504,
|
| 1341 |
+
"grad_norm": 23.375,
|
| 1342 |
+
"learning_rate": 1.7005555555555556e-05,
|
| 1343 |
+
"loss": 18.5572,
|
| 1344 |
+
"mean_token_accuracy": 0.7075925685465336,
|
| 1345 |
+
"num_tokens": 18367857.0,
|
| 1346 |
+
"step": 940
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"entropy": 1.1533676456660031,
|
| 1350 |
+
"epoch": 1.52,
|
| 1351 |
+
"grad_norm": 23.5,
|
| 1352 |
+
"learning_rate": 1.6950000000000002e-05,
|
| 1353 |
+
"loss": 18.4716,
|
| 1354 |
+
"mean_token_accuracy": 0.7079793553799391,
|
| 1355 |
+
"num_tokens": 18569146.0,
|
| 1356 |
+
"step": 950
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"entropy": 1.1715267714112998,
|
| 1360 |
+
"epoch": 1.536,
|
| 1361 |
+
"grad_norm": 26.25,
|
| 1362 |
+
"learning_rate": 1.6894444444444448e-05,
|
| 1363 |
+
"loss": 18.754,
|
| 1364 |
+
"mean_token_accuracy": 0.7050810724496841,
|
| 1365 |
+
"num_tokens": 18755079.0,
|
| 1366 |
+
"step": 960
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"entropy": 1.1350098505616188,
|
| 1370 |
+
"epoch": 1.552,
|
| 1371 |
+
"grad_norm": 26.0,
|
| 1372 |
+
"learning_rate": 1.683888888888889e-05,
|
| 1373 |
+
"loss": 18.4393,
|
| 1374 |
+
"mean_token_accuracy": 0.7091891258955002,
|
| 1375 |
+
"num_tokens": 18956248.0,
|
| 1376 |
+
"step": 970
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"entropy": 1.1361894980072975,
|
| 1380 |
+
"epoch": 1.568,
|
| 1381 |
+
"grad_norm": 25.625,
|
| 1382 |
+
"learning_rate": 1.6783333333333332e-05,
|
| 1383 |
+
"loss": 18.2272,
|
| 1384 |
+
"mean_token_accuracy": 0.711289395019412,
|
| 1385 |
+
"num_tokens": 19150315.0,
|
| 1386 |
+
"step": 980
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"entropy": 1.1439239967614412,
|
| 1390 |
+
"epoch": 1.584,
|
| 1391 |
+
"grad_norm": 26.125,
|
| 1392 |
+
"learning_rate": 1.6727777777777778e-05,
|
| 1393 |
+
"loss": 18.3302,
|
| 1394 |
+
"mean_token_accuracy": 0.7097273994237184,
|
| 1395 |
+
"num_tokens": 19344260.0,
|
| 1396 |
+
"step": 990
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"entropy": 1.1727315995842218,
|
| 1400 |
+
"epoch": 1.6,
|
| 1401 |
+
"grad_norm": 23.625,
|
| 1402 |
+
"learning_rate": 1.6672222222222224e-05,
|
| 1403 |
+
"loss": 18.7555,
|
| 1404 |
+
"mean_token_accuracy": 0.7035728722810746,
|
| 1405 |
+
"num_tokens": 19532552.0,
|
| 1406 |
+
"step": 1000
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 1.6,
|
| 1410 |
+
"eval_biology_entropy": 1.0895025281906128,
|
| 1411 |
+
"eval_biology_loss": 1.1883716583251953,
|
| 1412 |
+
"eval_biology_mean_token_accuracy": 0.705087408065796,
|
| 1413 |
+
"eval_biology_num_tokens": 19532552.0,
|
| 1414 |
+
"eval_biology_runtime": 36.662,
|
| 1415 |
+
"eval_biology_samples_per_second": 13.638,
|
| 1416 |
+
"eval_biology_steps_per_second": 3.41,
|
| 1417 |
+
"step": 1000
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 1.6,
|
| 1421 |
+
"eval_chemistry_entropy": 0.8066284563541413,
|
| 1422 |
+
"eval_chemistry_loss": 0.9020781517028809,
|
| 1423 |
+
"eval_chemistry_mean_token_accuracy": 0.7695949592590332,
|
| 1424 |
+
"eval_chemistry_num_tokens": 19532552.0,
|
| 1425 |
+
"eval_chemistry_runtime": 44.602,
|
| 1426 |
+
"eval_chemistry_samples_per_second": 11.21,
|
| 1427 |
+
"eval_chemistry_steps_per_second": 2.803,
|
| 1428 |
+
"step": 1000
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 1.6,
|
| 1432 |
+
"eval_math_entropy": 0.6928402831554413,
|
| 1433 |
+
"eval_math_loss": 0.9739019870758057,
|
| 1434 |
+
"eval_math_mean_token_accuracy": 0.7690589771270752,
|
| 1435 |
+
"eval_math_num_tokens": 19532552.0,
|
| 1436 |
+
"eval_math_runtime": 46.1135,
|
| 1437 |
+
"eval_math_samples_per_second": 10.843,
|
| 1438 |
+
"eval_math_steps_per_second": 2.711,
|
| 1439 |
+
"step": 1000
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 1.6,
|
| 1443 |
+
"eval_physics_entropy": 0.8003968651294708,
|
| 1444 |
+
"eval_physics_loss": 0.889771044254303,
|
| 1445 |
+
"eval_physics_mean_token_accuracy": 0.7777439093589783,
|
| 1446 |
+
"eval_physics_num_tokens": 19532552.0,
|
| 1447 |
+
"eval_physics_runtime": 53.0769,
|
| 1448 |
+
"eval_physics_samples_per_second": 9.42,
|
| 1449 |
+
"eval_physics_steps_per_second": 2.355,
|
| 1450 |
+
"step": 1000
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"entropy": 1.1351401790976525,
|
| 1454 |
+
"epoch": 1.616,
|
| 1455 |
+
"grad_norm": 23.375,
|
| 1456 |
+
"learning_rate": 1.6616666666666666e-05,
|
| 1457 |
+
"loss": 18.25,
|
| 1458 |
+
"mean_token_accuracy": 0.7113314840942622,
|
| 1459 |
+
"num_tokens": 19732719.0,
|
| 1460 |
+
"step": 1010
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"entropy": 1.1716969449073076,
|
| 1464 |
+
"epoch": 1.6320000000000001,
|
| 1465 |
+
"grad_norm": 26.75,
|
| 1466 |
+
"learning_rate": 1.6561111111111112e-05,
|
| 1467 |
+
"loss": 18.7782,
|
| 1468 |
+
"mean_token_accuracy": 0.7055815242230892,
|
| 1469 |
+
"num_tokens": 19926830.0,
|
| 1470 |
+
"step": 1020
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"entropy": 1.1589280366897583,
|
| 1474 |
+
"epoch": 1.6480000000000001,
|
| 1475 |
+
"grad_norm": 22.5,
|
| 1476 |
+
"learning_rate": 1.6505555555555558e-05,
|
| 1477 |
+
"loss": 18.5404,
|
| 1478 |
+
"mean_token_accuracy": 0.7074035726487636,
|
| 1479 |
+
"num_tokens": 20118800.0,
|
| 1480 |
+
"step": 1030
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"entropy": 1.1163798790425061,
|
| 1484 |
+
"epoch": 1.6640000000000001,
|
| 1485 |
+
"grad_norm": 22.25,
|
| 1486 |
+
"learning_rate": 1.645e-05,
|
| 1487 |
+
"loss": 18.0098,
|
| 1488 |
+
"mean_token_accuracy": 0.7143486425280571,
|
| 1489 |
+
"num_tokens": 20320511.0,
|
| 1490 |
+
"step": 1040
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"entropy": 1.143330230936408,
|
| 1494 |
+
"epoch": 1.6800000000000002,
|
| 1495 |
+
"grad_norm": 24.625,
|
| 1496 |
+
"learning_rate": 1.6394444444444446e-05,
|
| 1497 |
+
"loss": 18.3667,
|
| 1498 |
+
"mean_token_accuracy": 0.7101939171552658,
|
| 1499 |
+
"num_tokens": 20513393.0,
|
| 1500 |
+
"step": 1050
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"entropy": 1.1441194768995047,
|
| 1504 |
+
"epoch": 1.696,
|
| 1505 |
+
"grad_norm": 24.625,
|
| 1506 |
+
"learning_rate": 1.633888888888889e-05,
|
| 1507 |
+
"loss": 18.3598,
|
| 1508 |
+
"mean_token_accuracy": 0.7109571672976017,
|
| 1509 |
+
"num_tokens": 20707237.0,
|
| 1510 |
+
"step": 1060
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"entropy": 1.1500865355134011,
|
| 1514 |
+
"epoch": 1.712,
|
| 1515 |
+
"grad_norm": 25.125,
|
| 1516 |
+
"learning_rate": 1.6283333333333334e-05,
|
| 1517 |
+
"loss": 18.3793,
|
| 1518 |
+
"mean_token_accuracy": 0.7098943468183279,
|
| 1519 |
+
"num_tokens": 20910419.0,
|
| 1520 |
+
"step": 1070
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"entropy": 1.151506431400776,
|
| 1524 |
+
"epoch": 1.728,
|
| 1525 |
+
"grad_norm": 23.875,
|
| 1526 |
+
"learning_rate": 1.622777777777778e-05,
|
| 1527 |
+
"loss": 18.6016,
|
| 1528 |
+
"mean_token_accuracy": 0.706824541464448,
|
| 1529 |
+
"num_tokens": 21107498.0,
|
| 1530 |
+
"step": 1080
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"entropy": 1.1211910132318734,
|
| 1534 |
+
"epoch": 1.744,
|
| 1535 |
+
"grad_norm": 23.125,
|
| 1536 |
+
"learning_rate": 1.6172222222222222e-05,
|
| 1537 |
+
"loss": 17.943,
|
| 1538 |
+
"mean_token_accuracy": 0.7139425069093704,
|
| 1539 |
+
"num_tokens": 21303955.0,
|
| 1540 |
+
"step": 1090
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"entropy": 1.135467940941453,
|
| 1544 |
+
"epoch": 1.76,
|
| 1545 |
+
"grad_norm": 26.625,
|
| 1546 |
+
"learning_rate": 1.6116666666666668e-05,
|
| 1547 |
+
"loss": 18.3203,
|
| 1548 |
+
"mean_token_accuracy": 0.711310513690114,
|
| 1549 |
+
"num_tokens": 21499572.0,
|
| 1550 |
+
"step": 1100
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 1.76,
|
| 1554 |
+
"eval_biology_entropy": 1.128408634185791,
|
| 1555 |
+
"eval_biology_loss": 1.1783721446990967,
|
| 1556 |
+
"eval_biology_mean_token_accuracy": 0.7069245343208312,
|
| 1557 |
+
"eval_biology_num_tokens": 21499572.0,
|
| 1558 |
+
"eval_biology_runtime": 36.7197,
|
| 1559 |
+
"eval_biology_samples_per_second": 13.617,
|
| 1560 |
+
"eval_biology_steps_per_second": 3.404,
|
| 1561 |
+
"step": 1100
|
| 1562 |
+
},
|
| 1563 |
+
{
|
| 1564 |
+
"epoch": 1.76,
|
| 1565 |
+
"eval_chemistry_entropy": 0.8473764100074768,
|
| 1566 |
+
"eval_chemistry_loss": 0.8974772691726685,
|
| 1567 |
+
"eval_chemistry_mean_token_accuracy": 0.7699919505119324,
|
| 1568 |
+
"eval_chemistry_num_tokens": 21499572.0,
|
| 1569 |
+
"eval_chemistry_runtime": 45.1422,
|
| 1570 |
+
"eval_chemistry_samples_per_second": 11.076,
|
| 1571 |
+
"eval_chemistry_steps_per_second": 2.769,
|
| 1572 |
+
"step": 1100
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 1.76,
|
| 1576 |
+
"eval_math_entropy": 0.7425255510807037,
|
| 1577 |
+
"eval_math_loss": 0.964664101600647,
|
| 1578 |
+
"eval_math_mean_token_accuracy": 0.7689636731147766,
|
| 1579 |
+
"eval_math_num_tokens": 21499572.0,
|
| 1580 |
+
"eval_math_runtime": 46.3198,
|
| 1581 |
+
"eval_math_samples_per_second": 10.795,
|
| 1582 |
+
"eval_math_steps_per_second": 2.699,
|
| 1583 |
+
"step": 1100
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 1.76,
|
| 1587 |
+
"eval_physics_entropy": 0.8533617742061615,
|
| 1588 |
+
"eval_physics_loss": 0.8856339454650879,
|
| 1589 |
+
"eval_physics_mean_token_accuracy": 0.7777909712791443,
|
| 1590 |
+
"eval_physics_num_tokens": 21499572.0,
|
| 1591 |
+
"eval_physics_runtime": 53.4691,
|
| 1592 |
+
"eval_physics_samples_per_second": 9.351,
|
| 1593 |
+
"eval_physics_steps_per_second": 2.338,
|
| 1594 |
+
"step": 1100
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"entropy": 1.1444791179150342,
|
| 1598 |
+
"epoch": 1.776,
|
| 1599 |
+
"grad_norm": 25.125,
|
| 1600 |
+
"learning_rate": 1.6061111111111113e-05,
|
| 1601 |
+
"loss": 18.4154,
|
| 1602 |
+
"mean_token_accuracy": 0.7087540179491043,
|
| 1603 |
+
"num_tokens": 21692804.0,
|
| 1604 |
+
"step": 1110
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"entropy": 1.1155106097459793,
|
| 1608 |
+
"epoch": 1.792,
|
| 1609 |
+
"grad_norm": 26.75,
|
| 1610 |
+
"learning_rate": 1.6005555555555556e-05,
|
| 1611 |
+
"loss": 17.8771,
|
| 1612 |
+
"mean_token_accuracy": 0.7151219062507153,
|
| 1613 |
+
"num_tokens": 21894218.0,
|
| 1614 |
+
"step": 1120
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"entropy": 1.1439542584121227,
|
| 1618 |
+
"epoch": 1.808,
|
| 1619 |
+
"grad_norm": 26.0,
|
| 1620 |
+
"learning_rate": 1.595e-05,
|
| 1621 |
+
"loss": 18.2808,
|
| 1622 |
+
"mean_token_accuracy": 0.7120830215513706,
|
| 1623 |
+
"num_tokens": 22082522.0,
|
| 1624 |
+
"step": 1130
|
| 1625 |
+
},
|
| 1626 |
+
{
|
| 1627 |
+
"entropy": 1.164598362520337,
|
| 1628 |
+
"epoch": 1.8239999999999998,
|
| 1629 |
+
"grad_norm": 24.875,
|
| 1630 |
+
"learning_rate": 1.5894444444444447e-05,
|
| 1631 |
+
"loss": 18.6609,
|
| 1632 |
+
"mean_token_accuracy": 0.7053720869123936,
|
| 1633 |
+
"num_tokens": 22278933.0,
|
| 1634 |
+
"step": 1140
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"entropy": 1.1762568525969983,
|
| 1638 |
+
"epoch": 1.8399999999999999,
|
| 1639 |
+
"grad_norm": 25.875,
|
| 1640 |
+
"learning_rate": 1.583888888888889e-05,
|
| 1641 |
+
"loss": 18.8552,
|
| 1642 |
+
"mean_token_accuracy": 0.7038435664027929,
|
| 1643 |
+
"num_tokens": 22473801.0,
|
| 1644 |
+
"step": 1150
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"entropy": 1.124564103782177,
|
| 1648 |
+
"epoch": 1.8559999999999999,
|
| 1649 |
+
"grad_norm": 24.625,
|
| 1650 |
+
"learning_rate": 1.5783333333333335e-05,
|
| 1651 |
+
"loss": 18.1974,
|
| 1652 |
+
"mean_token_accuracy": 0.7101016439497471,
|
| 1653 |
+
"num_tokens": 22677853.0,
|
| 1654 |
+
"step": 1160
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"entropy": 1.1182593822479248,
|
| 1658 |
+
"epoch": 1.8719999999999999,
|
| 1659 |
+
"grad_norm": 24.75,
|
| 1660 |
+
"learning_rate": 1.572777777777778e-05,
|
| 1661 |
+
"loss": 17.8834,
|
| 1662 |
+
"mean_token_accuracy": 0.7157654833048582,
|
| 1663 |
+
"num_tokens": 22874965.0,
|
| 1664 |
+
"step": 1170
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"entropy": 1.1652609888464212,
|
| 1668 |
+
"epoch": 1.888,
|
| 1669 |
+
"grad_norm": 26.375,
|
| 1670 |
+
"learning_rate": 1.5672222222222223e-05,
|
| 1671 |
+
"loss": 18.6854,
|
| 1672 |
+
"mean_token_accuracy": 0.7054859042167664,
|
| 1673 |
+
"num_tokens": 23068892.0,
|
| 1674 |
+
"step": 1180
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"entropy": 1.1525317531079053,
|
| 1678 |
+
"epoch": 1.904,
|
| 1679 |
+
"grad_norm": 25.5,
|
| 1680 |
+
"learning_rate": 1.561666666666667e-05,
|
| 1681 |
+
"loss": 18.4917,
|
| 1682 |
+
"mean_token_accuracy": 0.7093732491135597,
|
| 1683 |
+
"num_tokens": 23263827.0,
|
| 1684 |
+
"step": 1190
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"entropy": 1.1493665751069784,
|
| 1688 |
+
"epoch": 1.92,
|
| 1689 |
+
"grad_norm": 23.25,
|
| 1690 |
+
"learning_rate": 1.556111111111111e-05,
|
| 1691 |
+
"loss": 18.5319,
|
| 1692 |
+
"mean_token_accuracy": 0.7083175975829363,
|
| 1693 |
+
"num_tokens": 23463627.0,
|
| 1694 |
+
"step": 1200
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 1.92,
|
| 1698 |
+
"eval_biology_entropy": 1.1280886573791504,
|
| 1699 |
+
"eval_biology_loss": 1.168996810913086,
|
| 1700 |
+
"eval_biology_mean_token_accuracy": 0.7084329080581665,
|
| 1701 |
+
"eval_biology_num_tokens": 23463627.0,
|
| 1702 |
+
"eval_biology_runtime": 36.7535,
|
| 1703 |
+
"eval_biology_samples_per_second": 13.604,
|
| 1704 |
+
"eval_biology_steps_per_second": 3.401,
|
| 1705 |
+
"step": 1200
|
| 1706 |
+
},
|
| 1707 |
+
{
|
| 1708 |
+
"epoch": 1.92,
|
| 1709 |
+
"eval_chemistry_entropy": 0.8494512581825256,
|
| 1710 |
+
"eval_chemistry_loss": 0.8931704759597778,
|
| 1711 |
+
"eval_chemistry_mean_token_accuracy": 0.7704657254219055,
|
| 1712 |
+
"eval_chemistry_num_tokens": 23463627.0,
|
| 1713 |
+
"eval_chemistry_runtime": 45.1599,
|
| 1714 |
+
"eval_chemistry_samples_per_second": 11.072,
|
| 1715 |
+
"eval_chemistry_steps_per_second": 2.768,
|
| 1716 |
+
"step": 1200
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 1.92,
|
| 1720 |
+
"eval_math_entropy": 0.7389046318531036,
|
| 1721 |
+
"eval_math_loss": 0.9631180167198181,
|
| 1722 |
+
"eval_math_mean_token_accuracy": 0.7691067876815796,
|
| 1723 |
+
"eval_math_num_tokens": 23463627.0,
|
| 1724 |
+
"eval_math_runtime": 46.3688,
|
| 1725 |
+
"eval_math_samples_per_second": 10.783,
|
| 1726 |
+
"eval_math_steps_per_second": 2.696,
|
| 1727 |
+
"step": 1200
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 1.92,
|
| 1731 |
+
"eval_physics_entropy": 0.8575768961906433,
|
| 1732 |
+
"eval_physics_loss": 0.8806874752044678,
|
| 1733 |
+
"eval_physics_mean_token_accuracy": 0.7783133115768432,
|
| 1734 |
+
"eval_physics_num_tokens": 23463627.0,
|
| 1735 |
+
"eval_physics_runtime": 53.2169,
|
| 1736 |
+
"eval_physics_samples_per_second": 9.396,
|
| 1737 |
+
"eval_physics_steps_per_second": 2.349,
|
| 1738 |
+
"step": 1200
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"entropy": 1.119696567952633,
|
| 1742 |
+
"epoch": 1.936,
|
| 1743 |
+
"grad_norm": 25.875,
|
| 1744 |
+
"learning_rate": 1.5505555555555557e-05,
|
| 1745 |
+
"loss": 17.8962,
|
| 1746 |
+
"mean_token_accuracy": 0.7145485125482083,
|
| 1747 |
+
"num_tokens": 23660418.0,
|
| 1748 |
+
"step": 1210
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"entropy": 1.1398029860109091,
|
| 1752 |
+
"epoch": 1.952,
|
| 1753 |
+
"grad_norm": 24.25,
|
| 1754 |
+
"learning_rate": 1.545e-05,
|
| 1755 |
+
"loss": 18.2843,
|
| 1756 |
+
"mean_token_accuracy": 0.7102924332022666,
|
| 1757 |
+
"num_tokens": 23858145.0,
|
| 1758 |
+
"step": 1220
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"entropy": 1.1528537224978208,
|
| 1762 |
+
"epoch": 1.968,
|
| 1763 |
+
"grad_norm": 25.0,
|
| 1764 |
+
"learning_rate": 1.5394444444444445e-05,
|
| 1765 |
+
"loss": 18.4096,
|
| 1766 |
+
"mean_token_accuracy": 0.7092894580215215,
|
| 1767 |
+
"num_tokens": 24053364.0,
|
| 1768 |
+
"step": 1230
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"entropy": 1.130320394039154,
|
| 1772 |
+
"epoch": 1.984,
|
| 1773 |
+
"grad_norm": 24.0,
|
| 1774 |
+
"learning_rate": 1.533888888888889e-05,
|
| 1775 |
+
"loss": 18.2555,
|
| 1776 |
+
"mean_token_accuracy": 0.7108469780534505,
|
| 1777 |
+
"num_tokens": 24249465.0,
|
| 1778 |
+
"step": 1240
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"entropy": 1.1447110261768103,
|
| 1782 |
+
"epoch": 2.0,
|
| 1783 |
+
"grad_norm": 25.375,
|
| 1784 |
+
"learning_rate": 1.5283333333333333e-05,
|
| 1785 |
+
"loss": 18.3602,
|
| 1786 |
+
"mean_token_accuracy": 0.7106993485242128,
|
| 1787 |
+
"num_tokens": 24442582.0,
|
| 1788 |
+
"step": 1250
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"entropy": 1.018304006755352,
|
| 1792 |
+
"epoch": 2.016,
|
| 1793 |
+
"grad_norm": 32.25,
|
| 1794 |
+
"learning_rate": 1.5227777777777779e-05,
|
| 1795 |
+
"loss": 16.0943,
|
| 1796 |
+
"mean_token_accuracy": 0.736331307515502,
|
| 1797 |
+
"num_tokens": 24632353.0,
|
| 1798 |
+
"step": 1260
|
| 1799 |
+
},
|
| 1800 |
+
{
|
| 1801 |
+
"entropy": 1.0125610791146755,
|
| 1802 |
+
"epoch": 2.032,
|
| 1803 |
+
"grad_norm": 29.25,
|
| 1804 |
+
"learning_rate": 1.5172222222222223e-05,
|
| 1805 |
+
"loss": 16.1426,
|
| 1806 |
+
"mean_token_accuracy": 0.7360313270241022,
|
| 1807 |
+
"num_tokens": 24822715.0,
|
| 1808 |
+
"step": 1270
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"entropy": 0.9830857388675213,
|
| 1812 |
+
"epoch": 2.048,
|
| 1813 |
+
"grad_norm": 30.0,
|
| 1814 |
+
"learning_rate": 1.5116666666666669e-05,
|
| 1815 |
+
"loss": 15.7118,
|
| 1816 |
+
"mean_token_accuracy": 0.7416322931647301,
|
| 1817 |
+
"num_tokens": 25023118.0,
|
| 1818 |
+
"step": 1280
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"entropy": 0.9931403815746307,
|
| 1822 |
+
"epoch": 2.064,
|
| 1823 |
+
"grad_norm": 29.25,
|
| 1824 |
+
"learning_rate": 1.5061111111111113e-05,
|
| 1825 |
+
"loss": 15.8433,
|
| 1826 |
+
"mean_token_accuracy": 0.7399392265826463,
|
| 1827 |
+
"num_tokens": 25217414.0,
|
| 1828 |
+
"step": 1290
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"entropy": 0.9739402402192354,
|
| 1832 |
+
"epoch": 2.08,
|
| 1833 |
+
"grad_norm": 28.875,
|
| 1834 |
+
"learning_rate": 1.5005555555555557e-05,
|
| 1835 |
+
"loss": 15.654,
|
| 1836 |
+
"mean_token_accuracy": 0.7406358070671558,
|
| 1837 |
+
"num_tokens": 25408961.0,
|
| 1838 |
+
"step": 1300
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 2.08,
|
| 1842 |
+
"eval_biology_entropy": 1.0040340127944947,
|
| 1843 |
+
"eval_biology_loss": 1.2068090438842773,
|
| 1844 |
+
"eval_biology_mean_token_accuracy": 0.7046228342056274,
|
| 1845 |
+
"eval_biology_num_tokens": 25408961.0,
|
| 1846 |
+
"eval_biology_runtime": 36.7953,
|
| 1847 |
+
"eval_biology_samples_per_second": 13.589,
|
| 1848 |
+
"eval_biology_steps_per_second": 3.397,
|
| 1849 |
+
"step": 1300
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 2.08,
|
| 1853 |
+
"eval_chemistry_entropy": 0.7516332559585571,
|
| 1854 |
+
"eval_chemistry_loss": 0.9387209415435791,
|
| 1855 |
+
"eval_chemistry_mean_token_accuracy": 0.7650035581588746,
|
| 1856 |
+
"eval_chemistry_num_tokens": 25408961.0,
|
| 1857 |
+
"eval_chemistry_runtime": 45.2093,
|
| 1858 |
+
"eval_chemistry_samples_per_second": 11.06,
|
| 1859 |
+
"eval_chemistry_steps_per_second": 2.765,
|
| 1860 |
+
"step": 1300
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 2.08,
|
| 1864 |
+
"eval_math_entropy": 0.6802132661342621,
|
| 1865 |
+
"eval_math_loss": 1.0038195848464966,
|
| 1866 |
+
"eval_math_mean_token_accuracy": 0.7657432436943055,
|
| 1867 |
+
"eval_math_num_tokens": 25408961.0,
|
| 1868 |
+
"eval_math_runtime": 46.4381,
|
| 1869 |
+
"eval_math_samples_per_second": 10.767,
|
| 1870 |
+
"eval_math_steps_per_second": 2.692,
|
| 1871 |
+
"step": 1300
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 2.08,
|
| 1875 |
+
"eval_physics_entropy": 0.7668055052757263,
|
| 1876 |
+
"eval_physics_loss": 0.9190961718559265,
|
| 1877 |
+
"eval_physics_mean_token_accuracy": 0.774013831615448,
|
| 1878 |
+
"eval_physics_num_tokens": 25408961.0,
|
| 1879 |
+
"eval_physics_runtime": 53.1969,
|
| 1880 |
+
"eval_physics_samples_per_second": 9.399,
|
| 1881 |
+
"eval_physics_steps_per_second": 2.35,
|
| 1882 |
+
"step": 1300
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"entropy": 0.9855834160000085,
|
| 1886 |
+
"epoch": 2.096,
|
| 1887 |
+
"grad_norm": 30.125,
|
| 1888 |
+
"learning_rate": 1.4950000000000003e-05,
|
| 1889 |
+
"loss": 15.7302,
|
| 1890 |
+
"mean_token_accuracy": 0.7416285384446383,
|
| 1891 |
+
"num_tokens": 25600511.0,
|
| 1892 |
+
"step": 1310
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"entropy": 0.9560547668486834,
|
| 1896 |
+
"epoch": 2.112,
|
| 1897 |
+
"grad_norm": 30.0,
|
| 1898 |
+
"learning_rate": 1.4894444444444445e-05,
|
| 1899 |
+
"loss": 15.2923,
|
| 1900 |
+
"mean_token_accuracy": 0.7467750016599893,
|
| 1901 |
+
"num_tokens": 25796565.0,
|
| 1902 |
+
"step": 1320
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"entropy": 0.9734839156270028,
|
| 1906 |
+
"epoch": 2.128,
|
| 1907 |
+
"grad_norm": 28.375,
|
| 1908 |
+
"learning_rate": 1.4838888888888889e-05,
|
| 1909 |
+
"loss": 15.4659,
|
| 1910 |
+
"mean_token_accuracy": 0.7434854444116354,
|
| 1911 |
+
"num_tokens": 25991156.0,
|
| 1912 |
+
"step": 1330
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"entropy": 0.9701934043318033,
|
| 1916 |
+
"epoch": 2.144,
|
| 1917 |
+
"grad_norm": 32.0,
|
| 1918 |
+
"learning_rate": 1.4783333333333335e-05,
|
| 1919 |
+
"loss": 15.6079,
|
| 1920 |
+
"mean_token_accuracy": 0.742589833214879,
|
| 1921 |
+
"num_tokens": 26193237.0,
|
| 1922 |
+
"step": 1340
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"entropy": 0.9971341345459223,
|
| 1926 |
+
"epoch": 2.16,
|
| 1927 |
+
"grad_norm": 30.875,
|
| 1928 |
+
"learning_rate": 1.4727777777777779e-05,
|
| 1929 |
+
"loss": 15.9608,
|
| 1930 |
+
"mean_token_accuracy": 0.7389167688786984,
|
| 1931 |
+
"num_tokens": 26387993.0,
|
| 1932 |
+
"step": 1350
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"entropy": 1.0051845651119948,
|
| 1936 |
+
"epoch": 2.176,
|
| 1937 |
+
"grad_norm": 30.125,
|
| 1938 |
+
"learning_rate": 1.4672222222222223e-05,
|
| 1939 |
+
"loss": 16.0894,
|
| 1940 |
+
"mean_token_accuracy": 0.736288172006607,
|
| 1941 |
+
"num_tokens": 26585269.0,
|
| 1942 |
+
"step": 1360
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"entropy": 0.956790243089199,
|
| 1946 |
+
"epoch": 2.192,
|
| 1947 |
+
"grad_norm": 27.0,
|
| 1948 |
+
"learning_rate": 1.4616666666666668e-05,
|
| 1949 |
+
"loss": 15.321,
|
| 1950 |
+
"mean_token_accuracy": 0.7460867285728454,
|
| 1951 |
+
"num_tokens": 26776318.0,
|
| 1952 |
+
"step": 1370
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"entropy": 0.9826645214110613,
|
| 1956 |
+
"epoch": 2.208,
|
| 1957 |
+
"grad_norm": 29.0,
|
| 1958 |
+
"learning_rate": 1.4561111111111112e-05,
|
| 1959 |
+
"loss": 15.6504,
|
| 1960 |
+
"mean_token_accuracy": 0.7417031954973936,
|
| 1961 |
+
"num_tokens": 26974420.0,
|
| 1962 |
+
"step": 1380
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"entropy": 1.0038280066102743,
|
| 1966 |
+
"epoch": 2.224,
|
| 1967 |
+
"grad_norm": 28.375,
|
| 1968 |
+
"learning_rate": 1.4505555555555557e-05,
|
| 1969 |
+
"loss": 16.0751,
|
| 1970 |
+
"mean_token_accuracy": 0.7366515222936869,
|
| 1971 |
+
"num_tokens": 27167577.0,
|
| 1972 |
+
"step": 1390
|
| 1973 |
+
},
|
| 1974 |
+
{
|
| 1975 |
+
"entropy": 0.9645163491368294,
|
| 1976 |
+
"epoch": 2.24,
|
| 1977 |
+
"grad_norm": 29.625,
|
| 1978 |
+
"learning_rate": 1.4450000000000002e-05,
|
| 1979 |
+
"loss": 15.4475,
|
| 1980 |
+
"mean_token_accuracy": 0.7438073411583901,
|
| 1981 |
+
"num_tokens": 27364189.0,
|
| 1982 |
+
"step": 1400
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 2.24,
|
| 1986 |
+
"eval_biology_entropy": 0.981437744140625,
|
| 1987 |
+
"eval_biology_loss": 1.207616925239563,
|
| 1988 |
+
"eval_biology_mean_token_accuracy": 0.704860345363617,
|
| 1989 |
+
"eval_biology_num_tokens": 27364189.0,
|
| 1990 |
+
"eval_biology_runtime": 37.0146,
|
| 1991 |
+
"eval_biology_samples_per_second": 13.508,
|
| 1992 |
+
"eval_biology_steps_per_second": 3.377,
|
| 1993 |
+
"step": 1400
|
| 1994 |
+
},
|
| 1995 |
+
{
|
| 1996 |
+
"epoch": 2.24,
|
| 1997 |
+
"eval_chemistry_entropy": 0.7361237547397613,
|
| 1998 |
+
"eval_chemistry_loss": 0.9404470324516296,
|
| 1999 |
+
"eval_chemistry_mean_token_accuracy": 0.7649986271858216,
|
| 2000 |
+
"eval_chemistry_num_tokens": 27364189.0,
|
| 2001 |
+
"eval_chemistry_runtime": 45.8584,
|
| 2002 |
+
"eval_chemistry_samples_per_second": 10.903,
|
| 2003 |
+
"eval_chemistry_steps_per_second": 2.726,
|
| 2004 |
+
"step": 1400
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 2.24,
|
| 2008 |
+
"eval_math_entropy": 0.6734062578678132,
|
| 2009 |
+
"eval_math_loss": 1.008291482925415,
|
| 2010 |
+
"eval_math_mean_token_accuracy": 0.7654335222244263,
|
| 2011 |
+
"eval_math_num_tokens": 27364189.0,
|
| 2012 |
+
"eval_math_runtime": 46.5839,
|
| 2013 |
+
"eval_math_samples_per_second": 10.733,
|
| 2014 |
+
"eval_math_steps_per_second": 2.683,
|
| 2015 |
+
"step": 1400
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 2.24,
|
| 2019 |
+
"eval_physics_entropy": 0.7516936962604522,
|
| 2020 |
+
"eval_physics_loss": 0.9206180572509766,
|
| 2021 |
+
"eval_physics_mean_token_accuracy": 0.7742253971099854,
|
| 2022 |
+
"eval_physics_num_tokens": 27364189.0,
|
| 2023 |
+
"eval_physics_runtime": 53.4062,
|
| 2024 |
+
"eval_physics_samples_per_second": 9.362,
|
| 2025 |
+
"eval_physics_steps_per_second": 2.341,
|
| 2026 |
+
"step": 1400
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"entropy": 0.95200383849442,
|
| 2030 |
+
"epoch": 2.2560000000000002,
|
| 2031 |
+
"grad_norm": 28.25,
|
| 2032 |
+
"learning_rate": 1.4394444444444446e-05,
|
| 2033 |
+
"loss": 15.29,
|
| 2034 |
+
"mean_token_accuracy": 0.7460251439362764,
|
| 2035 |
+
"num_tokens": 27557387.0,
|
| 2036 |
+
"step": 1410
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"entropy": 0.9579467125236988,
|
| 2040 |
+
"epoch": 2.2720000000000002,
|
| 2041 |
+
"grad_norm": 29.75,
|
| 2042 |
+
"learning_rate": 1.433888888888889e-05,
|
| 2043 |
+
"loss": 15.3949,
|
| 2044 |
+
"mean_token_accuracy": 0.7428190916776657,
|
| 2045 |
+
"num_tokens": 27755725.0,
|
| 2046 |
+
"step": 1420
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"entropy": 1.0049206741154193,
|
| 2050 |
+
"epoch": 2.288,
|
| 2051 |
+
"grad_norm": 29.375,
|
| 2052 |
+
"learning_rate": 1.4283333333333333e-05,
|
| 2053 |
+
"loss": 16.0165,
|
| 2054 |
+
"mean_token_accuracy": 0.7357172459363938,
|
| 2055 |
+
"num_tokens": 27950694.0,
|
| 2056 |
+
"step": 1430
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"entropy": 1.005023618414998,
|
| 2060 |
+
"epoch": 2.304,
|
| 2061 |
+
"grad_norm": 27.375,
|
| 2062 |
+
"learning_rate": 1.4227777777777778e-05,
|
| 2063 |
+
"loss": 16.031,
|
| 2064 |
+
"mean_token_accuracy": 0.7371877241879702,
|
| 2065 |
+
"num_tokens": 28150719.0,
|
| 2066 |
+
"step": 1440
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"entropy": 0.9545973517000675,
|
| 2070 |
+
"epoch": 2.32,
|
| 2071 |
+
"grad_norm": 27.375,
|
| 2072 |
+
"learning_rate": 1.4172222222222222e-05,
|
| 2073 |
+
"loss": 15.3343,
|
| 2074 |
+
"mean_token_accuracy": 0.7450725197792053,
|
| 2075 |
+
"num_tokens": 28348652.0,
|
| 2076 |
+
"step": 1450
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"entropy": 0.9779899045825005,
|
| 2080 |
+
"epoch": 2.336,
|
| 2081 |
+
"grad_norm": 29.125,
|
| 2082 |
+
"learning_rate": 1.4116666666666668e-05,
|
| 2083 |
+
"loss": 15.6932,
|
| 2084 |
+
"mean_token_accuracy": 0.7409903839230537,
|
| 2085 |
+
"num_tokens": 28542945.0,
|
| 2086 |
+
"step": 1460
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"entropy": 0.9990467399358749,
|
| 2090 |
+
"epoch": 2.352,
|
| 2091 |
+
"grad_norm": 30.0,
|
| 2092 |
+
"learning_rate": 1.4061111111111112e-05,
|
| 2093 |
+
"loss": 16.0289,
|
| 2094 |
+
"mean_token_accuracy": 0.737147618085146,
|
| 2095 |
+
"num_tokens": 28731927.0,
|
| 2096 |
+
"step": 1470
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"entropy": 0.9743069138377904,
|
| 2100 |
+
"epoch": 2.368,
|
| 2101 |
+
"grad_norm": 28.25,
|
| 2102 |
+
"learning_rate": 1.4005555555555556e-05,
|
| 2103 |
+
"loss": 15.5588,
|
| 2104 |
+
"mean_token_accuracy": 0.7433720890432596,
|
| 2105 |
+
"num_tokens": 28929298.0,
|
| 2106 |
+
"step": 1480
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"entropy": 0.9916814502328635,
|
| 2110 |
+
"epoch": 2.384,
|
| 2111 |
+
"grad_norm": 30.875,
|
| 2112 |
+
"learning_rate": 1.3950000000000002e-05,
|
| 2113 |
+
"loss": 15.7992,
|
| 2114 |
+
"mean_token_accuracy": 0.738391650095582,
|
| 2115 |
+
"num_tokens": 29121142.0,
|
| 2116 |
+
"step": 1490
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"entropy": 0.9762042462825775,
|
| 2120 |
+
"epoch": 2.4,
|
| 2121 |
+
"grad_norm": 27.75,
|
| 2122 |
+
"learning_rate": 1.3894444444444446e-05,
|
| 2123 |
+
"loss": 15.6866,
|
| 2124 |
+
"mean_token_accuracy": 0.7411531064659357,
|
| 2125 |
+
"num_tokens": 29317919.0,
|
| 2126 |
+
"step": 1500
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 2.4,
|
| 2130 |
+
"eval_biology_entropy": 0.9761668286323547,
|
| 2131 |
+
"eval_biology_loss": 1.204858422279358,
|
| 2132 |
+
"eval_biology_mean_token_accuracy": 0.7061952071189881,
|
| 2133 |
+
"eval_biology_num_tokens": 29317919.0,
|
| 2134 |
+
"eval_biology_runtime": 36.8411,
|
| 2135 |
+
"eval_biology_samples_per_second": 13.572,
|
| 2136 |
+
"eval_biology_steps_per_second": 3.393,
|
| 2137 |
+
"step": 1500
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 2.4,
|
| 2141 |
+
"eval_chemistry_entropy": 0.7322783548831939,
|
| 2142 |
+
"eval_chemistry_loss": 0.9399828314781189,
|
| 2143 |
+
"eval_chemistry_mean_token_accuracy": 0.7658818020820618,
|
| 2144 |
+
"eval_chemistry_num_tokens": 29317919.0,
|
| 2145 |
+
"eval_chemistry_runtime": 44.9217,
|
| 2146 |
+
"eval_chemistry_samples_per_second": 11.13,
|
| 2147 |
+
"eval_chemistry_steps_per_second": 2.783,
|
| 2148 |
+
"step": 1500
|
| 2149 |
+
},
|
| 2150 |
+
{
|
| 2151 |
+
"epoch": 2.4,
|
| 2152 |
+
"eval_math_entropy": 0.6698096024990082,
|
| 2153 |
+
"eval_math_loss": 1.0045055150985718,
|
| 2154 |
+
"eval_math_mean_token_accuracy": 0.7665768723487854,
|
| 2155 |
+
"eval_math_num_tokens": 29317919.0,
|
| 2156 |
+
"eval_math_runtime": 46.3425,
|
| 2157 |
+
"eval_math_samples_per_second": 10.789,
|
| 2158 |
+
"eval_math_steps_per_second": 2.697,
|
| 2159 |
+
"step": 1500
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 2.4,
|
| 2163 |
+
"eval_physics_entropy": 0.7526926681995392,
|
| 2164 |
+
"eval_physics_loss": 0.9185658097267151,
|
| 2165 |
+
"eval_physics_mean_token_accuracy": 0.7750369157791138,
|
| 2166 |
+
"eval_physics_num_tokens": 29317919.0,
|
| 2167 |
+
"eval_physics_runtime": 53.26,
|
| 2168 |
+
"eval_physics_samples_per_second": 9.388,
|
| 2169 |
+
"eval_physics_steps_per_second": 2.347,
|
| 2170 |
+
"step": 1500
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"entropy": 0.9726948961615562,
|
| 2174 |
+
"epoch": 2.416,
|
| 2175 |
+
"grad_norm": 30.75,
|
| 2176 |
+
"learning_rate": 1.383888888888889e-05,
|
| 2177 |
+
"loss": 15.6556,
|
| 2178 |
+
"mean_token_accuracy": 0.7424319490790368,
|
| 2179 |
+
"num_tokens": 29518541.0,
|
| 2180 |
+
"step": 1510
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"entropy": 0.976997298002243,
|
| 2184 |
+
"epoch": 2.432,
|
| 2185 |
+
"grad_norm": 32.0,
|
| 2186 |
+
"learning_rate": 1.3783333333333336e-05,
|
| 2187 |
+
"loss": 15.6639,
|
| 2188 |
+
"mean_token_accuracy": 0.7415778063237667,
|
| 2189 |
+
"num_tokens": 29720768.0,
|
| 2190 |
+
"step": 1520
|
| 2191 |
+
},
|
| 2192 |
+
{
|
| 2193 |
+
"entropy": 1.0026599559932947,
|
| 2194 |
+
"epoch": 2.448,
|
| 2195 |
+
"grad_norm": 31.5,
|
| 2196 |
+
"learning_rate": 1.372777777777778e-05,
|
| 2197 |
+
"loss": 16.037,
|
| 2198 |
+
"mean_token_accuracy": 0.7365970708429813,
|
| 2199 |
+
"num_tokens": 29916619.0,
|
| 2200 |
+
"step": 1530
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"entropy": 0.9530468691140414,
|
| 2204 |
+
"epoch": 2.464,
|
| 2205 |
+
"grad_norm": 31.75,
|
| 2206 |
+
"learning_rate": 1.3672222222222222e-05,
|
| 2207 |
+
"loss": 15.2741,
|
| 2208 |
+
"mean_token_accuracy": 0.7459127757698297,
|
| 2209 |
+
"num_tokens": 30121198.0,
|
| 2210 |
+
"step": 1540
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"entropy": 0.9706476211547852,
|
| 2214 |
+
"epoch": 2.48,
|
| 2215 |
+
"grad_norm": 28.875,
|
| 2216 |
+
"learning_rate": 1.3616666666666666e-05,
|
| 2217 |
+
"loss": 15.4971,
|
| 2218 |
+
"mean_token_accuracy": 0.7433239962905646,
|
| 2219 |
+
"num_tokens": 30325508.0,
|
| 2220 |
+
"step": 1550
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"entropy": 0.9861213160678745,
|
| 2224 |
+
"epoch": 2.496,
|
| 2225 |
+
"grad_norm": 28.625,
|
| 2226 |
+
"learning_rate": 1.3561111111111112e-05,
|
| 2227 |
+
"loss": 15.7964,
|
| 2228 |
+
"mean_token_accuracy": 0.7396648865193128,
|
| 2229 |
+
"num_tokens": 30517445.0,
|
| 2230 |
+
"step": 1560
|
| 2231 |
+
},
|
| 2232 |
+
{
|
| 2233 |
+
"entropy": 0.9939322113990784,
|
| 2234 |
+
"epoch": 2.512,
|
| 2235 |
+
"grad_norm": 33.5,
|
| 2236 |
+
"learning_rate": 1.3505555555555556e-05,
|
| 2237 |
+
"loss": 15.9554,
|
| 2238 |
+
"mean_token_accuracy": 0.7374067515134811,
|
| 2239 |
+
"num_tokens": 30714049.0,
|
| 2240 |
+
"step": 1570
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"entropy": 0.9829558525234461,
|
| 2244 |
+
"epoch": 2.528,
|
| 2245 |
+
"grad_norm": 29.375,
|
| 2246 |
+
"learning_rate": 1.3450000000000002e-05,
|
| 2247 |
+
"loss": 15.7034,
|
| 2248 |
+
"mean_token_accuracy": 0.7400100238621234,
|
| 2249 |
+
"num_tokens": 30911317.0,
|
| 2250 |
+
"step": 1580
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"entropy": 0.97488385848701,
|
| 2254 |
+
"epoch": 2.544,
|
| 2255 |
+
"grad_norm": 28.125,
|
| 2256 |
+
"learning_rate": 1.3394444444444446e-05,
|
| 2257 |
+
"loss": 15.7159,
|
| 2258 |
+
"mean_token_accuracy": 0.7403594095259904,
|
| 2259 |
+
"num_tokens": 31108983.0,
|
| 2260 |
+
"step": 1590
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"entropy": 0.9890310749411583,
|
| 2264 |
+
"epoch": 2.56,
|
| 2265 |
+
"grad_norm": 30.125,
|
| 2266 |
+
"learning_rate": 1.333888888888889e-05,
|
| 2267 |
+
"loss": 15.8407,
|
| 2268 |
+
"mean_token_accuracy": 0.7394855204969645,
|
| 2269 |
+
"num_tokens": 31298994.0,
|
| 2270 |
+
"step": 1600
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 2.56,
|
| 2274 |
+
"eval_biology_entropy": 0.9767315549850464,
|
| 2275 |
+
"eval_biology_loss": 1.1978344917297363,
|
| 2276 |
+
"eval_biology_mean_token_accuracy": 0.7075373558998108,
|
| 2277 |
+
"eval_biology_num_tokens": 31298994.0,
|
| 2278 |
+
"eval_biology_runtime": 36.8405,
|
| 2279 |
+
"eval_biology_samples_per_second": 13.572,
|
| 2280 |
+
"eval_biology_steps_per_second": 3.393,
|
| 2281 |
+
"step": 1600
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 2.56,
|
| 2285 |
+
"eval_chemistry_entropy": 0.7359724206924438,
|
| 2286 |
+
"eval_chemistry_loss": 0.9340105652809143,
|
| 2287 |
+
"eval_chemistry_mean_token_accuracy": 0.7665223875045777,
|
| 2288 |
+
"eval_chemistry_num_tokens": 31298994.0,
|
| 2289 |
+
"eval_chemistry_runtime": 45.3527,
|
| 2290 |
+
"eval_chemistry_samples_per_second": 11.025,
|
| 2291 |
+
"eval_chemistry_steps_per_second": 2.756,
|
| 2292 |
+
"step": 1600
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 2.56,
|
| 2296 |
+
"eval_math_entropy": 0.6756473398208618,
|
| 2297 |
+
"eval_math_loss": 0.998451292514801,
|
| 2298 |
+
"eval_math_mean_token_accuracy": 0.7671175785064698,
|
| 2299 |
+
"eval_math_num_tokens": 31298994.0,
|
| 2300 |
+
"eval_math_runtime": 46.427,
|
| 2301 |
+
"eval_math_samples_per_second": 10.77,
|
| 2302 |
+
"eval_math_steps_per_second": 2.692,
|
| 2303 |
+
"step": 1600
|
| 2304 |
+
},
|
| 2305 |
+
{
|
| 2306 |
+
"epoch": 2.56,
|
| 2307 |
+
"eval_physics_entropy": 0.7543728523254395,
|
| 2308 |
+
"eval_physics_loss": 0.9128937721252441,
|
| 2309 |
+
"eval_physics_mean_token_accuracy": 0.7758728971481323,
|
| 2310 |
+
"eval_physics_num_tokens": 31298994.0,
|
| 2311 |
+
"eval_physics_runtime": 53.705,
|
| 2312 |
+
"eval_physics_samples_per_second": 9.31,
|
| 2313 |
+
"eval_physics_steps_per_second": 2.328,
|
| 2314 |
+
"step": 1600
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"entropy": 0.9974417526274919,
|
| 2318 |
+
"epoch": 2.576,
|
| 2319 |
+
"grad_norm": 30.0,
|
| 2320 |
+
"learning_rate": 1.3283333333333335e-05,
|
| 2321 |
+
"loss": 15.8827,
|
| 2322 |
+
"mean_token_accuracy": 0.7387637589126825,
|
| 2323 |
+
"num_tokens": 31488512.0,
|
| 2324 |
+
"step": 1610
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"entropy": 1.0062089327722787,
|
| 2328 |
+
"epoch": 2.592,
|
| 2329 |
+
"grad_norm": 29.875,
|
| 2330 |
+
"learning_rate": 1.322777777777778e-05,
|
| 2331 |
+
"loss": 16.2346,
|
| 2332 |
+
"mean_token_accuracy": 0.7342662557959556,
|
| 2333 |
+
"num_tokens": 31684651.0,
|
| 2334 |
+
"step": 1620
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"entropy": 1.0037442568689585,
|
| 2338 |
+
"epoch": 2.608,
|
| 2339 |
+
"grad_norm": 29.875,
|
| 2340 |
+
"learning_rate": 1.3172222222222224e-05,
|
| 2341 |
+
"loss": 16.0345,
|
| 2342 |
+
"mean_token_accuracy": 0.7362603198736906,
|
| 2343 |
+
"num_tokens": 31876008.0,
|
| 2344 |
+
"step": 1630
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"entropy": 0.9704636950045824,
|
| 2348 |
+
"epoch": 2.624,
|
| 2349 |
+
"grad_norm": 32.0,
|
| 2350 |
+
"learning_rate": 1.311666666666667e-05,
|
| 2351 |
+
"loss": 15.619,
|
| 2352 |
+
"mean_token_accuracy": 0.7409613560885191,
|
| 2353 |
+
"num_tokens": 32069617.0,
|
| 2354 |
+
"step": 1640
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"entropy": 1.0037952404469253,
|
| 2358 |
+
"epoch": 2.64,
|
| 2359 |
+
"grad_norm": 29.0,
|
| 2360 |
+
"learning_rate": 1.3061111111111112e-05,
|
| 2361 |
+
"loss": 16.0021,
|
| 2362 |
+
"mean_token_accuracy": 0.736137805506587,
|
| 2363 |
+
"num_tokens": 32262932.0,
|
| 2364 |
+
"step": 1650
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"entropy": 0.9621167879551649,
|
| 2368 |
+
"epoch": 2.656,
|
| 2369 |
+
"grad_norm": 28.625,
|
| 2370 |
+
"learning_rate": 1.3005555555555556e-05,
|
| 2371 |
+
"loss": 15.384,
|
| 2372 |
+
"mean_token_accuracy": 0.7444965790957212,
|
| 2373 |
+
"num_tokens": 32465495.0,
|
| 2374 |
+
"step": 1660
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"entropy": 0.9599390238523483,
|
| 2378 |
+
"epoch": 2.672,
|
| 2379 |
+
"grad_norm": 30.25,
|
| 2380 |
+
"learning_rate": 1.295e-05,
|
| 2381 |
+
"loss": 15.5228,
|
| 2382 |
+
"mean_token_accuracy": 0.7436116680502891,
|
| 2383 |
+
"num_tokens": 32665458.0,
|
| 2384 |
+
"step": 1670
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"entropy": 0.9774678956717253,
|
| 2388 |
+
"epoch": 2.6879999999999997,
|
| 2389 |
+
"grad_norm": 27.875,
|
| 2390 |
+
"learning_rate": 1.2894444444444445e-05,
|
| 2391 |
+
"loss": 15.5753,
|
| 2392 |
+
"mean_token_accuracy": 0.7433817163109779,
|
| 2393 |
+
"num_tokens": 32867135.0,
|
| 2394 |
+
"step": 1680
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"entropy": 0.9774248816072941,
|
| 2398 |
+
"epoch": 2.7039999999999997,
|
| 2399 |
+
"grad_norm": 30.125,
|
| 2400 |
+
"learning_rate": 1.283888888888889e-05,
|
| 2401 |
+
"loss": 15.7087,
|
| 2402 |
+
"mean_token_accuracy": 0.7409440513700247,
|
| 2403 |
+
"num_tokens": 33070366.0,
|
| 2404 |
+
"step": 1690
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"entropy": 0.9762023743242025,
|
| 2408 |
+
"epoch": 2.7199999999999998,
|
| 2409 |
+
"grad_norm": 33.75,
|
| 2410 |
+
"learning_rate": 1.2783333333333333e-05,
|
| 2411 |
+
"loss": 15.6031,
|
| 2412 |
+
"mean_token_accuracy": 0.7418079309165477,
|
| 2413 |
+
"num_tokens": 33269122.0,
|
| 2414 |
+
"step": 1700
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 2.7199999999999998,
|
| 2418 |
+
"eval_biology_entropy": 0.9562785272598266,
|
| 2419 |
+
"eval_biology_loss": 1.2001115083694458,
|
| 2420 |
+
"eval_biology_mean_token_accuracy": 0.7079961094856262,
|
| 2421 |
+
"eval_biology_num_tokens": 33269122.0,
|
| 2422 |
+
"eval_biology_runtime": 36.8873,
|
| 2423 |
+
"eval_biology_samples_per_second": 13.555,
|
| 2424 |
+
"eval_biology_steps_per_second": 3.389,
|
| 2425 |
+
"step": 1700
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 2.7199999999999998,
|
| 2429 |
+
"eval_chemistry_entropy": 0.7180265989303589,
|
| 2430 |
+
"eval_chemistry_loss": 0.9368557333946228,
|
| 2431 |
+
"eval_chemistry_mean_token_accuracy": 0.7668515272140503,
|
| 2432 |
+
"eval_chemistry_num_tokens": 33269122.0,
|
| 2433 |
+
"eval_chemistry_runtime": 45.2091,
|
| 2434 |
+
"eval_chemistry_samples_per_second": 11.06,
|
| 2435 |
+
"eval_chemistry_steps_per_second": 2.765,
|
| 2436 |
+
"step": 1700
|
| 2437 |
+
},
|
| 2438 |
+
{
|
| 2439 |
+
"epoch": 2.7199999999999998,
|
| 2440 |
+
"eval_math_entropy": 0.6551018316745758,
|
| 2441 |
+
"eval_math_loss": 1.0031293630599976,
|
| 2442 |
+
"eval_math_mean_token_accuracy": 0.7675854053497314,
|
| 2443 |
+
"eval_math_num_tokens": 33269122.0,
|
| 2444 |
+
"eval_math_runtime": 46.4714,
|
| 2445 |
+
"eval_math_samples_per_second": 10.759,
|
| 2446 |
+
"eval_math_steps_per_second": 2.69,
|
| 2447 |
+
"step": 1700
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 2.7199999999999998,
|
| 2451 |
+
"eval_physics_entropy": 0.7281462309360505,
|
| 2452 |
+
"eval_physics_loss": 0.9170361161231995,
|
| 2453 |
+
"eval_physics_mean_token_accuracy": 0.7760837173461914,
|
| 2454 |
+
"eval_physics_num_tokens": 33269122.0,
|
| 2455 |
+
"eval_physics_runtime": 53.6243,
|
| 2456 |
+
"eval_physics_samples_per_second": 9.324,
|
| 2457 |
+
"eval_physics_steps_per_second": 2.331,
|
| 2458 |
+
"step": 1700
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"entropy": 1.0005533527582884,
|
| 2462 |
+
"epoch": 2.7359999999999998,
|
| 2463 |
+
"grad_norm": 27.5,
|
| 2464 |
+
"learning_rate": 1.272777777777778e-05,
|
| 2465 |
+
"loss": 15.9595,
|
| 2466 |
+
"mean_token_accuracy": 0.7374090943485498,
|
| 2467 |
+
"num_tokens": 33456592.0,
|
| 2468 |
+
"step": 1710
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"entropy": 0.9925159979611635,
|
| 2472 |
+
"epoch": 2.752,
|
| 2473 |
+
"grad_norm": 30.5,
|
| 2474 |
+
"learning_rate": 1.2672222222222223e-05,
|
| 2475 |
+
"loss": 15.9212,
|
| 2476 |
+
"mean_token_accuracy": 0.7368109248578548,
|
| 2477 |
+
"num_tokens": 33645862.0,
|
| 2478 |
+
"step": 1720
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"entropy": 1.0035406082868576,
|
| 2482 |
+
"epoch": 2.768,
|
| 2483 |
+
"grad_norm": 29.125,
|
| 2484 |
+
"learning_rate": 1.2616666666666669e-05,
|
| 2485 |
+
"loss": 16.0741,
|
| 2486 |
+
"mean_token_accuracy": 0.7346370995044709,
|
| 2487 |
+
"num_tokens": 33835740.0,
|
| 2488 |
+
"step": 1730
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"entropy": 0.9657413493841887,
|
| 2492 |
+
"epoch": 2.784,
|
| 2493 |
+
"grad_norm": 27.875,
|
| 2494 |
+
"learning_rate": 1.2561111111111113e-05,
|
| 2495 |
+
"loss": 15.5009,
|
| 2496 |
+
"mean_token_accuracy": 0.7434774361550808,
|
| 2497 |
+
"num_tokens": 34034702.0,
|
| 2498 |
+
"step": 1740
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"entropy": 0.9740934349596501,
|
| 2502 |
+
"epoch": 2.8,
|
| 2503 |
+
"grad_norm": 28.125,
|
| 2504 |
+
"learning_rate": 1.2505555555555557e-05,
|
| 2505 |
+
"loss": 15.5637,
|
| 2506 |
+
"mean_token_accuracy": 0.7438909105956555,
|
| 2507 |
+
"num_tokens": 34235603.0,
|
| 2508 |
+
"step": 1750
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"entropy": 0.9742557104676962,
|
| 2512 |
+
"epoch": 2.816,
|
| 2513 |
+
"grad_norm": 29.875,
|
| 2514 |
+
"learning_rate": 1.2450000000000003e-05,
|
| 2515 |
+
"loss": 15.6336,
|
| 2516 |
+
"mean_token_accuracy": 0.7424697510898113,
|
| 2517 |
+
"num_tokens": 34434833.0,
|
| 2518 |
+
"step": 1760
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"entropy": 1.0037282764911652,
|
| 2522 |
+
"epoch": 2.832,
|
| 2523 |
+
"grad_norm": 31.625,
|
| 2524 |
+
"learning_rate": 1.2394444444444445e-05,
|
| 2525 |
+
"loss": 16.0886,
|
| 2526 |
+
"mean_token_accuracy": 0.7369506523013115,
|
| 2527 |
+
"num_tokens": 34625581.0,
|
| 2528 |
+
"step": 1770
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"entropy": 0.9910386119037866,
|
| 2532 |
+
"epoch": 2.848,
|
| 2533 |
+
"grad_norm": 32.0,
|
| 2534 |
+
"learning_rate": 1.2338888888888889e-05,
|
| 2535 |
+
"loss": 15.8463,
|
| 2536 |
+
"mean_token_accuracy": 0.7397053975611925,
|
| 2537 |
+
"num_tokens": 34816521.0,
|
| 2538 |
+
"step": 1780
|
| 2539 |
+
},
|
| 2540 |
+
{
|
| 2541 |
+
"entropy": 0.9725664932280779,
|
| 2542 |
+
"epoch": 2.864,
|
| 2543 |
+
"grad_norm": 30.5,
|
| 2544 |
+
"learning_rate": 1.2283333333333333e-05,
|
| 2545 |
+
"loss": 15.7461,
|
| 2546 |
+
"mean_token_accuracy": 0.7406340092420578,
|
| 2547 |
+
"num_tokens": 35013631.0,
|
| 2548 |
+
"step": 1790
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"entropy": 1.00489255823195,
|
| 2552 |
+
"epoch": 2.88,
|
| 2553 |
+
"grad_norm": 32.0,
|
| 2554 |
+
"learning_rate": 1.2227777777777779e-05,
|
| 2555 |
+
"loss": 15.9662,
|
| 2556 |
+
"mean_token_accuracy": 0.7373752295970917,
|
| 2557 |
+
"num_tokens": 35207800.0,
|
| 2558 |
+
"step": 1800
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 2.88,
|
| 2562 |
+
"eval_biology_entropy": 0.9769085459709167,
|
| 2563 |
+
"eval_biology_loss": 1.194201946258545,
|
| 2564 |
+
"eval_biology_mean_token_accuracy": 0.7088253183364868,
|
| 2565 |
+
"eval_biology_num_tokens": 35207800.0,
|
| 2566 |
+
"eval_biology_runtime": 36.6978,
|
| 2567 |
+
"eval_biology_samples_per_second": 13.625,
|
| 2568 |
+
"eval_biology_steps_per_second": 3.406,
|
| 2569 |
+
"step": 1800
|
| 2570 |
+
},
|
| 2571 |
+
{
|
| 2572 |
+
"epoch": 2.88,
|
| 2573 |
+
"eval_chemistry_entropy": 0.724792768239975,
|
| 2574 |
+
"eval_chemistry_loss": 0.9388518333435059,
|
| 2575 |
+
"eval_chemistry_mean_token_accuracy": 0.7662976326942443,
|
| 2576 |
+
"eval_chemistry_num_tokens": 35207800.0,
|
| 2577 |
+
"eval_chemistry_runtime": 45.0996,
|
| 2578 |
+
"eval_chemistry_samples_per_second": 11.087,
|
| 2579 |
+
"eval_chemistry_steps_per_second": 2.772,
|
| 2580 |
+
"step": 1800
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 2.88,
|
| 2584 |
+
"eval_math_entropy": 0.6623785650730133,
|
| 2585 |
+
"eval_math_loss": 1.0053212642669678,
|
| 2586 |
+
"eval_math_mean_token_accuracy": 0.7665670537948608,
|
| 2587 |
+
"eval_math_num_tokens": 35207800.0,
|
| 2588 |
+
"eval_math_runtime": 46.2886,
|
| 2589 |
+
"eval_math_samples_per_second": 10.802,
|
| 2590 |
+
"eval_math_steps_per_second": 2.7,
|
| 2591 |
+
"step": 1800
|
| 2592 |
+
},
|
| 2593 |
+
{
|
| 2594 |
+
"epoch": 2.88,
|
| 2595 |
+
"eval_physics_entropy": 0.7309899318218231,
|
| 2596 |
+
"eval_physics_loss": 0.9189478754997253,
|
| 2597 |
+
"eval_physics_mean_token_accuracy": 0.7758140830993653,
|
| 2598 |
+
"eval_physics_num_tokens": 35207800.0,
|
| 2599 |
+
"eval_physics_runtime": 53.3951,
|
| 2600 |
+
"eval_physics_samples_per_second": 9.364,
|
| 2601 |
+
"eval_physics_steps_per_second": 2.341,
|
| 2602 |
+
"step": 1800
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"entropy": 0.9926395244896412,
|
| 2606 |
+
"epoch": 2.896,
|
| 2607 |
+
"grad_norm": 29.875,
|
| 2608 |
+
"learning_rate": 1.2172222222222223e-05,
|
| 2609 |
+
"loss": 15.8639,
|
| 2610 |
+
"mean_token_accuracy": 0.7381080884486437,
|
| 2611 |
+
"num_tokens": 35402995.0,
|
| 2612 |
+
"step": 1810
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"entropy": 1.0010335687547922,
|
| 2616 |
+
"epoch": 2.912,
|
| 2617 |
+
"grad_norm": 33.0,
|
| 2618 |
+
"learning_rate": 1.2116666666666667e-05,
|
| 2619 |
+
"loss": 16.0889,
|
| 2620 |
+
"mean_token_accuracy": 0.7379562932997942,
|
| 2621 |
+
"num_tokens": 35595456.0,
|
| 2622 |
+
"step": 1820
|
| 2623 |
+
},
|
| 2624 |
+
{
|
| 2625 |
+
"entropy": 0.9953980639576911,
|
| 2626 |
+
"epoch": 2.928,
|
| 2627 |
+
"grad_norm": 32.75,
|
| 2628 |
+
"learning_rate": 1.2061111111111113e-05,
|
| 2629 |
+
"loss": 15.9704,
|
| 2630 |
+
"mean_token_accuracy": 0.7373797506093979,
|
| 2631 |
+
"num_tokens": 35787309.0,
|
| 2632 |
+
"step": 1830
|
| 2633 |
+
},
|
| 2634 |
+
{
|
| 2635 |
+
"entropy": 0.9820944830775261,
|
| 2636 |
+
"epoch": 2.944,
|
| 2637 |
+
"grad_norm": 27.875,
|
| 2638 |
+
"learning_rate": 1.2005555555555557e-05,
|
| 2639 |
+
"loss": 15.7722,
|
| 2640 |
+
"mean_token_accuracy": 0.740547452121973,
|
| 2641 |
+
"num_tokens": 35982686.0,
|
| 2642 |
+
"step": 1840
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"entropy": 0.9821060415357351,
|
| 2646 |
+
"epoch": 2.96,
|
| 2647 |
+
"grad_norm": 31.625,
|
| 2648 |
+
"learning_rate": 1.195e-05,
|
| 2649 |
+
"loss": 15.6224,
|
| 2650 |
+
"mean_token_accuracy": 0.7415538590401412,
|
| 2651 |
+
"num_tokens": 36179520.0,
|
| 2652 |
+
"step": 1850
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"entropy": 0.9771638073027133,
|
| 2656 |
+
"epoch": 2.976,
|
| 2657 |
+
"grad_norm": 28.5,
|
| 2658 |
+
"learning_rate": 1.1894444444444447e-05,
|
| 2659 |
+
"loss": 15.6643,
|
| 2660 |
+
"mean_token_accuracy": 0.7411342523992062,
|
| 2661 |
+
"num_tokens": 36373689.0,
|
| 2662 |
+
"step": 1860
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"entropy": 0.9860306549817324,
|
| 2666 |
+
"epoch": 2.992,
|
| 2667 |
+
"grad_norm": 33.5,
|
| 2668 |
+
"learning_rate": 1.183888888888889e-05,
|
| 2669 |
+
"loss": 15.7875,
|
| 2670 |
+
"mean_token_accuracy": 0.7410633962601423,
|
| 2671 |
+
"num_tokens": 36569612.0,
|
| 2672 |
+
"step": 1870
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"entropy": 0.9603324174880982,
|
| 2676 |
+
"epoch": 3.008,
|
| 2677 |
+
"grad_norm": 37.25,
|
| 2678 |
+
"learning_rate": 1.1783333333333333e-05,
|
| 2679 |
+
"loss": 14.9135,
|
| 2680 |
+
"mean_token_accuracy": 0.7520980384200812,
|
| 2681 |
+
"num_tokens": 36759194.0,
|
| 2682 |
+
"step": 1880
|
| 2683 |
+
},
|
| 2684 |
+
{
|
| 2685 |
+
"entropy": 0.8478883054107428,
|
| 2686 |
+
"epoch": 3.024,
|
| 2687 |
+
"grad_norm": 34.75,
|
| 2688 |
+
"learning_rate": 1.1727777777777779e-05,
|
| 2689 |
+
"loss": 13.7212,
|
| 2690 |
+
"mean_token_accuracy": 0.7676899693906307,
|
| 2691 |
+
"num_tokens": 36947293.0,
|
| 2692 |
+
"step": 1890
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"entropy": 0.8174442987889051,
|
| 2696 |
+
"epoch": 3.04,
|
| 2697 |
+
"grad_norm": 37.75,
|
| 2698 |
+
"learning_rate": 1.1672222222222223e-05,
|
| 2699 |
+
"loss": 12.9695,
|
| 2700 |
+
"mean_token_accuracy": 0.7781106177717447,
|
| 2701 |
+
"num_tokens": 37136901.0,
|
| 2702 |
+
"step": 1900
|
| 2703 |
+
},
|
| 2704 |
+
{
|
| 2705 |
+
"epoch": 3.04,
|
| 2706 |
+
"eval_biology_entropy": 0.8307687420845031,
|
| 2707 |
+
"eval_biology_loss": 1.3037042617797852,
|
| 2708 |
+
"eval_biology_mean_token_accuracy": 0.7012130594253541,
|
| 2709 |
+
"eval_biology_num_tokens": 37136901.0,
|
| 2710 |
+
"eval_biology_runtime": 36.7227,
|
| 2711 |
+
"eval_biology_samples_per_second": 13.616,
|
| 2712 |
+
"eval_biology_steps_per_second": 3.404,
|
| 2713 |
+
"step": 1900
|
| 2714 |
+
},
|
| 2715 |
+
{
|
| 2716 |
+
"epoch": 3.04,
|
| 2717 |
+
"eval_chemistry_entropy": 0.6082699120044708,
|
| 2718 |
+
"eval_chemistry_loss": 1.0519765615463257,
|
| 2719 |
+
"eval_chemistry_mean_token_accuracy": 0.7578758845329284,
|
| 2720 |
+
"eval_chemistry_num_tokens": 37136901.0,
|
| 2721 |
+
"eval_chemistry_runtime": 45.1209,
|
| 2722 |
+
"eval_chemistry_samples_per_second": 11.081,
|
| 2723 |
+
"eval_chemistry_steps_per_second": 2.77,
|
| 2724 |
+
"step": 1900
|
| 2725 |
+
},
|
| 2726 |
+
{
|
| 2727 |
+
"epoch": 3.04,
|
| 2728 |
+
"eval_math_entropy": 0.5899905214309692,
|
| 2729 |
+
"eval_math_loss": 1.0844311714172363,
|
| 2730 |
+
"eval_math_mean_token_accuracy": 0.7623905053138733,
|
| 2731 |
+
"eval_math_num_tokens": 37136901.0,
|
| 2732 |
+
"eval_math_runtime": 46.314,
|
| 2733 |
+
"eval_math_samples_per_second": 10.796,
|
| 2734 |
+
"eval_math_steps_per_second": 2.699,
|
| 2735 |
+
"step": 1900
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 3.04,
|
| 2739 |
+
"eval_physics_entropy": 0.6185110075473785,
|
| 2740 |
+
"eval_physics_loss": 1.0144134759902954,
|
| 2741 |
+
"eval_physics_mean_token_accuracy": 0.7686533064842224,
|
| 2742 |
+
"eval_physics_num_tokens": 37136901.0,
|
| 2743 |
+
"eval_physics_runtime": 53.1821,
|
| 2744 |
+
"eval_physics_samples_per_second": 9.402,
|
| 2745 |
+
"eval_physics_steps_per_second": 2.35,
|
| 2746 |
+
"step": 1900
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"entropy": 0.8187011457979679,
|
| 2750 |
+
"epoch": 3.056,
|
| 2751 |
+
"grad_norm": 37.25,
|
| 2752 |
+
"learning_rate": 1.1616666666666667e-05,
|
| 2753 |
+
"loss": 13.1003,
|
| 2754 |
+
"mean_token_accuracy": 0.7760679375380277,
|
| 2755 |
+
"num_tokens": 37331588.0,
|
| 2756 |
+
"step": 1910
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"entropy": 0.8076904013752937,
|
| 2760 |
+
"epoch": 3.072,
|
| 2761 |
+
"grad_norm": 37.5,
|
| 2762 |
+
"learning_rate": 1.1561111111111112e-05,
|
| 2763 |
+
"loss": 12.8615,
|
| 2764 |
+
"mean_token_accuracy": 0.781244232878089,
|
| 2765 |
+
"num_tokens": 37521211.0,
|
| 2766 |
+
"step": 1920
|
| 2767 |
+
},
|
| 2768 |
+
{
|
| 2769 |
+
"entropy": 0.82519664876163,
|
| 2770 |
+
"epoch": 3.088,
|
| 2771 |
+
"grad_norm": 37.0,
|
| 2772 |
+
"learning_rate": 1.1505555555555556e-05,
|
| 2773 |
+
"loss": 13.0928,
|
| 2774 |
+
"mean_token_accuracy": 0.7768605265766382,
|
| 2775 |
+
"num_tokens": 37713328.0,
|
| 2776 |
+
"step": 1930
|
| 2777 |
+
},
|
| 2778 |
+
{
|
| 2779 |
+
"entropy": 0.8123729152604937,
|
| 2780 |
+
"epoch": 3.104,
|
| 2781 |
+
"grad_norm": 33.75,
|
| 2782 |
+
"learning_rate": 1.145e-05,
|
| 2783 |
+
"loss": 12.9409,
|
| 2784 |
+
"mean_token_accuracy": 0.780734346807003,
|
| 2785 |
+
"num_tokens": 37913470.0,
|
| 2786 |
+
"step": 1940
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"entropy": 0.8159066649153829,
|
| 2790 |
+
"epoch": 3.12,
|
| 2791 |
+
"grad_norm": 37.0,
|
| 2792 |
+
"learning_rate": 1.1394444444444446e-05,
|
| 2793 |
+
"loss": 13.0653,
|
| 2794 |
+
"mean_token_accuracy": 0.7791258115321398,
|
| 2795 |
+
"num_tokens": 38106718.0,
|
| 2796 |
+
"step": 1950
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"entropy": 0.8273790344595909,
|
| 2800 |
+
"epoch": 3.136,
|
| 2801 |
+
"grad_norm": 40.75,
|
| 2802 |
+
"learning_rate": 1.133888888888889e-05,
|
| 2803 |
+
"loss": 13.1437,
|
| 2804 |
+
"mean_token_accuracy": 0.7766635850071907,
|
| 2805 |
+
"num_tokens": 38295701.0,
|
| 2806 |
+
"step": 1960
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"entropy": 0.81372896656394,
|
| 2810 |
+
"epoch": 3.152,
|
| 2811 |
+
"grad_norm": 37.5,
|
| 2812 |
+
"learning_rate": 1.1283333333333334e-05,
|
| 2813 |
+
"loss": 12.9402,
|
| 2814 |
+
"mean_token_accuracy": 0.7791113253682852,
|
| 2815 |
+
"num_tokens": 38493951.0,
|
| 2816 |
+
"step": 1970
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"entropy": 0.8379074672237039,
|
| 2820 |
+
"epoch": 3.168,
|
| 2821 |
+
"grad_norm": 39.0,
|
| 2822 |
+
"learning_rate": 1.122777777777778e-05,
|
| 2823 |
+
"loss": 13.4123,
|
| 2824 |
+
"mean_token_accuracy": 0.7723229706287384,
|
| 2825 |
+
"num_tokens": 38688390.0,
|
| 2826 |
+
"step": 1980
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"entropy": 0.8237785773351789,
|
| 2830 |
+
"epoch": 3.184,
|
| 2831 |
+
"grad_norm": 34.25,
|
| 2832 |
+
"learning_rate": 1.1172222222222222e-05,
|
| 2833 |
+
"loss": 13.1071,
|
| 2834 |
+
"mean_token_accuracy": 0.7771053221076727,
|
| 2835 |
+
"num_tokens": 38880157.0,
|
| 2836 |
+
"step": 1990
|
| 2837 |
+
},
|
| 2838 |
+
{
|
| 2839 |
+
"entropy": 0.8222423151135445,
|
| 2840 |
+
"epoch": 3.2,
|
| 2841 |
+
"grad_norm": 35.75,
|
| 2842 |
+
"learning_rate": 1.1116666666666666e-05,
|
| 2843 |
+
"loss": 13.2091,
|
| 2844 |
+
"mean_token_accuracy": 0.7758903849869967,
|
| 2845 |
+
"num_tokens": 39081782.0,
|
| 2846 |
+
"step": 2000
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 3.2,
|
| 2850 |
+
"eval_biology_entropy": 0.8496361360549927,
|
| 2851 |
+
"eval_biology_loss": 1.2998476028442383,
|
| 2852 |
+
"eval_biology_mean_token_accuracy": 0.7012469072341919,
|
| 2853 |
+
"eval_biology_num_tokens": 39081782.0,
|
| 2854 |
+
"eval_biology_runtime": 36.6144,
|
| 2855 |
+
"eval_biology_samples_per_second": 13.656,
|
| 2856 |
+
"eval_biology_steps_per_second": 3.414,
|
| 2857 |
+
"step": 2000
|
| 2858 |
+
},
|
| 2859 |
+
{
|
| 2860 |
+
"epoch": 3.2,
|
| 2861 |
+
"eval_chemistry_entropy": 0.6230687329769135,
|
| 2862 |
+
"eval_chemistry_loss": 1.0478407144546509,
|
| 2863 |
+
"eval_chemistry_mean_token_accuracy": 0.7584491949081421,
|
| 2864 |
+
"eval_chemistry_num_tokens": 39081782.0,
|
| 2865 |
+
"eval_chemistry_runtime": 44.6898,
|
| 2866 |
+
"eval_chemistry_samples_per_second": 11.188,
|
| 2867 |
+
"eval_chemistry_steps_per_second": 2.797,
|
| 2868 |
+
"step": 2000
|
| 2869 |
+
},
|
| 2870 |
+
{
|
| 2871 |
+
"epoch": 3.2,
|
| 2872 |
+
"eval_math_entropy": 0.6067575147151947,
|
| 2873 |
+
"eval_math_loss": 1.082250714302063,
|
| 2874 |
+
"eval_math_mean_token_accuracy": 0.7613544554710389,
|
| 2875 |
+
"eval_math_num_tokens": 39081782.0,
|
| 2876 |
+
"eval_math_runtime": 46.1231,
|
| 2877 |
+
"eval_math_samples_per_second": 10.841,
|
| 2878 |
+
"eval_math_steps_per_second": 2.71,
|
| 2879 |
+
"step": 2000
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 3.2,
|
| 2883 |
+
"eval_physics_entropy": 0.6437831990718842,
|
| 2884 |
+
"eval_physics_loss": 1.007546067237854,
|
| 2885 |
+
"eval_physics_mean_token_accuracy": 0.7685807209014892,
|
| 2886 |
+
"eval_physics_num_tokens": 39081782.0,
|
| 2887 |
+
"eval_physics_runtime": 53.088,
|
| 2888 |
+
"eval_physics_samples_per_second": 9.418,
|
| 2889 |
+
"eval_physics_steps_per_second": 2.355,
|
| 2890 |
+
"step": 2000
|
| 2891 |
+
}
|
| 2892 |
+
],
|
| 2893 |
+
"logging_steps": 10,
|
| 2894 |
+
"max_steps": 4000,
|
| 2895 |
+
"num_input_tokens_seen": 0,
|
| 2896 |
+
"num_train_epochs": 7,
|
| 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": 4.4082567723215933e+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 sha256:33e28e0345bba7de386d9ec79fc07cb58111c3c47c3486fae7abe0d391dd53ad
|
| 3 |
+
size 6481
|