Upload folder using huggingface_hub
Browse files- special_tokens_map.json +1 -2
- tokenizer.json +2 -2
- tokenizer_config.json +0 -2
- trainer_state.json +2031 -0
- training_args.bin +3 -0
special_tokens_map.json
CHANGED
|
@@ -12,6 +12,5 @@
|
|
| 12 |
"normalized": false,
|
| 13 |
"rstrip": false,
|
| 14 |
"single_word": false
|
| 15 |
-
}
|
| 16 |
-
"pad_token": "<|end_of_text|>"
|
| 17 |
}
|
|
|
|
| 12 |
"normalized": false,
|
| 13 |
"rstrip": false,
|
| 14 |
"single_word": false
|
| 15 |
+
}
|
|
|
|
| 16 |
}
|
tokenizer.json
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3c5cf44023714fb39b05e71e425f8d7b92805ff73f7988b083b8c87f0bf87393
|
| 3 |
+
size 17209961
|
tokenizer_config.json
CHANGED
|
@@ -2050,7 +2050,6 @@
|
|
| 2050 |
}
|
| 2051 |
},
|
| 2052 |
"bos_token": "<|begin_of_text|>",
|
| 2053 |
-
"chat_template": "{%- for message in messages %}{%- if message['role'] == 'system' %}## System:{{- message['content'].strip() }}\n{%- elif message['role'] == 'user' %}## Instruction: {{- message['content'].strip() }}\n{%- elif message['role'] == 'assistant' %}## Response: {{- message['content'].strip() }}\n{%- endif %}{%- endfor %}",
|
| 2054 |
"clean_up_tokenization_spaces": true,
|
| 2055 |
"eos_token": "<|end_of_text|>",
|
| 2056 |
"extra_special_tokens": {},
|
|
@@ -2059,6 +2058,5 @@
|
|
| 2059 |
"attention_mask"
|
| 2060 |
],
|
| 2061 |
"model_max_length": 1000000000000000019884624838656,
|
| 2062 |
-
"pad_token": "<|end_of_text|>",
|
| 2063 |
"tokenizer_class": "PreTrainedTokenizer"
|
| 2064 |
}
|
|
|
|
| 2050 |
}
|
| 2051 |
},
|
| 2052 |
"bos_token": "<|begin_of_text|>",
|
|
|
|
| 2053 |
"clean_up_tokenization_spaces": true,
|
| 2054 |
"eos_token": "<|end_of_text|>",
|
| 2055 |
"extra_special_tokens": {},
|
|
|
|
| 2058 |
"attention_mask"
|
| 2059 |
],
|
| 2060 |
"model_max_length": 1000000000000000019884624838656,
|
|
|
|
| 2061 |
"tokenizer_class": "PreTrainedTokenizer"
|
| 2062 |
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2031 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 9250,
|
| 3 |
+
"best_metric": 0.9165179133415222,
|
| 4 |
+
"best_model_checkpoint": "./results/checkpoint-9000",
|
| 5 |
+
"epoch": 4.997752808988764,
|
| 6 |
+
"eval_steps": 250,
|
| 7 |
+
"global_step": 9455,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.026437541308658295,
|
| 14 |
+
"grad_norm": 43.99717712402344,
|
| 15 |
+
"learning_rate": 0.00034625958983852136,
|
| 16 |
+
"loss": 9.1282,
|
| 17 |
+
"mean_token_accuracy": 0.3046248774370179,
|
| 18 |
+
"num_tokens": 1638400.0,
|
| 19 |
+
"step": 50
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"epoch": 0.05287508261731659,
|
| 23 |
+
"grad_norm": 6.144142150878906,
|
| 24 |
+
"learning_rate": 0.000407611186724682,
|
| 25 |
+
"loss": 4.215,
|
| 26 |
+
"mean_token_accuracy": 0.4404753165692091,
|
| 27 |
+
"num_tokens": 3276800.0,
|
| 28 |
+
"step": 100
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 0.07931262392597488,
|
| 32 |
+
"grad_norm": 2.289468288421631,
|
| 33 |
+
"learning_rate": 0.0004434995702624468,
|
| 34 |
+
"loss": 3.3619,
|
| 35 |
+
"mean_token_accuracy": 0.4969379284977913,
|
| 36 |
+
"num_tokens": 4915200.0,
|
| 37 |
+
"step": 150
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.10575016523463318,
|
| 41 |
+
"grad_norm": 1.4314253330230713,
|
| 42 |
+
"learning_rate": 0.0004689627836108426,
|
| 43 |
+
"loss": 2.3482,
|
| 44 |
+
"mean_token_accuracy": 0.6427779817581176,
|
| 45 |
+
"num_tokens": 6553600.0,
|
| 46 |
+
"step": 200
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.13218770654329148,
|
| 50 |
+
"grad_norm": 1.0555275678634644,
|
| 51 |
+
"learning_rate": 0.0004887135863147016,
|
| 52 |
+
"loss": 2.0093,
|
| 53 |
+
"step": 250
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.13218770654329148,
|
| 57 |
+
"eval_loss": 1.941351056098938,
|
| 58 |
+
"eval_mean_token_accuracy": 0.6896467157826616,
|
| 59 |
+
"eval_num_tokens": 8192000.0,
|
| 60 |
+
"eval_runtime": 1710.841,
|
| 61 |
+
"eval_samples_per_second": 4.422,
|
| 62 |
+
"eval_steps_per_second": 0.553,
|
| 63 |
+
"step": 250
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.15862524785194976,
|
| 67 |
+
"grad_norm": 1.0403132438659668,
|
| 68 |
+
"learning_rate": 0.0004991822047759241,
|
| 69 |
+
"loss": 1.8613,
|
| 70 |
+
"mean_token_accuracy": 0.6918438403308391,
|
| 71 |
+
"num_tokens": 9830400.0,
|
| 72 |
+
"step": 300
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.18506278916060806,
|
| 76 |
+
"grad_norm": 1.0463595390319824,
|
| 77 |
+
"learning_rate": 0.0004964562206956711,
|
| 78 |
+
"loss": 1.7529,
|
| 79 |
+
"mean_token_accuracy": 0.7160025656223297,
|
| 80 |
+
"num_tokens": 11468800.0,
|
| 81 |
+
"step": 350
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.21150033046926636,
|
| 85 |
+
"grad_norm": 0.8012986183166504,
|
| 86 |
+
"learning_rate": 0.0004937302366154182,
|
| 87 |
+
"loss": 1.642,
|
| 88 |
+
"mean_token_accuracy": 0.731451002061367,
|
| 89 |
+
"num_tokens": 13107200.0,
|
| 90 |
+
"step": 400
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.23793787177792466,
|
| 94 |
+
"grad_norm": 0.9207384586334229,
|
| 95 |
+
"learning_rate": 0.0004910042525351653,
|
| 96 |
+
"loss": 1.5717,
|
| 97 |
+
"mean_token_accuracy": 0.7387188410758972,
|
| 98 |
+
"num_tokens": 14745600.0,
|
| 99 |
+
"step": 450
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.26437541308658297,
|
| 103 |
+
"grad_norm": 0.6695268750190735,
|
| 104 |
+
"learning_rate": 0.00048827826845491225,
|
| 105 |
+
"loss": 1.5342,
|
| 106 |
+
"step": 500
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 0.26437541308658297,
|
| 110 |
+
"eval_loss": 1.520922064781189,
|
| 111 |
+
"eval_mean_token_accuracy": 0.7428315233982643,
|
| 112 |
+
"eval_num_tokens": 16384000.0,
|
| 113 |
+
"eval_runtime": 1711.3798,
|
| 114 |
+
"eval_samples_per_second": 4.42,
|
| 115 |
+
"eval_steps_per_second": 0.553,
|
| 116 |
+
"step": 500
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.29081295439524124,
|
| 120 |
+
"grad_norm": 0.7166313529014587,
|
| 121 |
+
"learning_rate": 0.0004855522843746593,
|
| 122 |
+
"loss": 1.5115,
|
| 123 |
+
"mean_token_accuracy": 0.7436290304362774,
|
| 124 |
+
"num_tokens": 18022400.0,
|
| 125 |
+
"step": 550
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.3172504957038995,
|
| 129 |
+
"grad_norm": 0.7090346217155457,
|
| 130 |
+
"learning_rate": 0.00048282630029440626,
|
| 131 |
+
"loss": 1.4558,
|
| 132 |
+
"mean_token_accuracy": 0.7508443316817284,
|
| 133 |
+
"num_tokens": 19660800.0,
|
| 134 |
+
"step": 600
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.34368803701255785,
|
| 138 |
+
"grad_norm": 0.8063613176345825,
|
| 139 |
+
"learning_rate": 0.00048010031621415335,
|
| 140 |
+
"loss": 1.4464,
|
| 141 |
+
"mean_token_accuracy": 0.7509249743819236,
|
| 142 |
+
"num_tokens": 21299200.0,
|
| 143 |
+
"step": 650
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.3701255783212161,
|
| 147 |
+
"grad_norm": 1.142205834388733,
|
| 148 |
+
"learning_rate": 0.0004773743321339004,
|
| 149 |
+
"loss": 1.3984,
|
| 150 |
+
"mean_token_accuracy": 0.756721040904522,
|
| 151 |
+
"num_tokens": 22937600.0,
|
| 152 |
+
"step": 700
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.3965631196298744,
|
| 156 |
+
"grad_norm": 0.8467270135879517,
|
| 157 |
+
"learning_rate": 0.00047464834805364736,
|
| 158 |
+
"loss": 1.3935,
|
| 159 |
+
"step": 750
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.3965631196298744,
|
| 163 |
+
"eval_loss": 1.3857780694961548,
|
| 164 |
+
"eval_mean_token_accuracy": 0.7577579059888142,
|
| 165 |
+
"eval_num_tokens": 24576000.0,
|
| 166 |
+
"eval_runtime": 1712.2261,
|
| 167 |
+
"eval_samples_per_second": 4.418,
|
| 168 |
+
"eval_steps_per_second": 0.552,
|
| 169 |
+
"step": 750
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 0.4230006609385327,
|
| 173 |
+
"grad_norm": 0.7939649820327759,
|
| 174 |
+
"learning_rate": 0.0004719223639733944,
|
| 175 |
+
"loss": 1.3699,
|
| 176 |
+
"mean_token_accuracy": 0.7578213591873646,
|
| 177 |
+
"num_tokens": 26214400.0,
|
| 178 |
+
"step": 800
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.449438202247191,
|
| 182 |
+
"grad_norm": 0.7455005645751953,
|
| 183 |
+
"learning_rate": 0.0004691963798931415,
|
| 184 |
+
"loss": 1.3693,
|
| 185 |
+
"mean_token_accuracy": 0.7589296215772628,
|
| 186 |
+
"num_tokens": 27852800.0,
|
| 187 |
+
"step": 850
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 0.47587574355584933,
|
| 191 |
+
"grad_norm": 0.7528285980224609,
|
| 192 |
+
"learning_rate": 0.00046647039581288846,
|
| 193 |
+
"loss": 1.3242,
|
| 194 |
+
"mean_token_accuracy": 0.7669953557848931,
|
| 195 |
+
"num_tokens": 29491200.0,
|
| 196 |
+
"step": 900
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 0.5023132848645075,
|
| 200 |
+
"grad_norm": 0.8056386113166809,
|
| 201 |
+
"learning_rate": 0.0004637444117326355,
|
| 202 |
+
"loss": 1.335,
|
| 203 |
+
"mean_token_accuracy": 0.7646432068943977,
|
| 204 |
+
"num_tokens": 31129600.0,
|
| 205 |
+
"step": 950
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.5287508261731659,
|
| 209 |
+
"grad_norm": 0.7774543166160583,
|
| 210 |
+
"learning_rate": 0.0004610184276523825,
|
| 211 |
+
"loss": 1.3286,
|
| 212 |
+
"step": 1000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.5287508261731659,
|
| 216 |
+
"eval_loss": 1.3367868661880493,
|
| 217 |
+
"eval_mean_token_accuracy": 0.7645143720362706,
|
| 218 |
+
"eval_num_tokens": 32768000.0,
|
| 219 |
+
"eval_runtime": 1713.3241,
|
| 220 |
+
"eval_samples_per_second": 4.415,
|
| 221 |
+
"eval_steps_per_second": 0.552,
|
| 222 |
+
"step": 1000
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.5551883674818242,
|
| 226 |
+
"grad_norm": 1.0932977199554443,
|
| 227 |
+
"learning_rate": 0.00045829244357212956,
|
| 228 |
+
"loss": 1.3357,
|
| 229 |
+
"mean_token_accuracy": 0.7644849714636802,
|
| 230 |
+
"num_tokens": 34406400.0,
|
| 231 |
+
"step": 1050
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.5816259087904825,
|
| 235 |
+
"grad_norm": 0.9064853191375732,
|
| 236 |
+
"learning_rate": 0.0004555664594918766,
|
| 237 |
+
"loss": 1.3309,
|
| 238 |
+
"mean_token_accuracy": 0.7655669602751732,
|
| 239 |
+
"num_tokens": 36044800.0,
|
| 240 |
+
"step": 1100
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.6080634500991408,
|
| 244 |
+
"grad_norm": 1.1222511529922485,
|
| 245 |
+
"learning_rate": 0.0004528404754116236,
|
| 246 |
+
"loss": 1.3088,
|
| 247 |
+
"mean_token_accuracy": 0.7684097030758857,
|
| 248 |
+
"num_tokens": 37683200.0,
|
| 249 |
+
"step": 1150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.634500991407799,
|
| 253 |
+
"grad_norm": 0.8550713658332825,
|
| 254 |
+
"learning_rate": 0.0004501144913313706,
|
| 255 |
+
"loss": 1.3121,
|
| 256 |
+
"mean_token_accuracy": 0.7672879993915558,
|
| 257 |
+
"num_tokens": 39321600.0,
|
| 258 |
+
"step": 1200
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.6609385327164574,
|
| 262 |
+
"grad_norm": 1.1444681882858276,
|
| 263 |
+
"learning_rate": 0.0004473885072511177,
|
| 264 |
+
"loss": 1.3124,
|
| 265 |
+
"step": 1250
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.6609385327164574,
|
| 269 |
+
"eval_loss": 1.3140060901641846,
|
| 270 |
+
"eval_mean_token_accuracy": 0.7666200130988882,
|
| 271 |
+
"eval_num_tokens": 40960000.0,
|
| 272 |
+
"eval_runtime": 1711.9779,
|
| 273 |
+
"eval_samples_per_second": 4.419,
|
| 274 |
+
"eval_steps_per_second": 0.553,
|
| 275 |
+
"step": 1250
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.6873760740251157,
|
| 279 |
+
"grad_norm": 0.9385507702827454,
|
| 280 |
+
"learning_rate": 0.0004446625231708647,
|
| 281 |
+
"loss": 1.299,
|
| 282 |
+
"mean_token_accuracy": 0.7683782380819321,
|
| 283 |
+
"num_tokens": 42598400.0,
|
| 284 |
+
"step": 1300
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.713813615333774,
|
| 288 |
+
"grad_norm": 0.9909027218818665,
|
| 289 |
+
"learning_rate": 0.0004419365390906117,
|
| 290 |
+
"loss": 1.2776,
|
| 291 |
+
"mean_token_accuracy": 0.7721034941077233,
|
| 292 |
+
"num_tokens": 44236800.0,
|
| 293 |
+
"step": 1350
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 0.7402511566424322,
|
| 297 |
+
"grad_norm": 1.0495429039001465,
|
| 298 |
+
"learning_rate": 0.00043921055501035873,
|
| 299 |
+
"loss": 1.2533,
|
| 300 |
+
"mean_token_accuracy": 0.7770307904481888,
|
| 301 |
+
"num_tokens": 45875200.0,
|
| 302 |
+
"step": 1400
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.7666886979510905,
|
| 306 |
+
"grad_norm": 0.9478822350502014,
|
| 307 |
+
"learning_rate": 0.0004364845709301058,
|
| 308 |
+
"loss": 1.2386,
|
| 309 |
+
"mean_token_accuracy": 0.7782954525947571,
|
| 310 |
+
"num_tokens": 47513600.0,
|
| 311 |
+
"step": 1450
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.7931262392597488,
|
| 315 |
+
"grad_norm": 0.9904903173446655,
|
| 316 |
+
"learning_rate": 0.0004337585868498528,
|
| 317 |
+
"loss": 1.2542,
|
| 318 |
+
"step": 1500
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.7931262392597488,
|
| 322 |
+
"eval_loss": 1.2644829750061035,
|
| 323 |
+
"eval_mean_token_accuracy": 0.7736368500027042,
|
| 324 |
+
"eval_num_tokens": 49152000.0,
|
| 325 |
+
"eval_runtime": 1711.2255,
|
| 326 |
+
"eval_samples_per_second": 4.421,
|
| 327 |
+
"eval_steps_per_second": 0.553,
|
| 328 |
+
"step": 1500
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 0.8195637805684072,
|
| 332 |
+
"grad_norm": 0.797590970993042,
|
| 333 |
+
"learning_rate": 0.00043103260276959983,
|
| 334 |
+
"loss": 1.2613,
|
| 335 |
+
"mean_token_accuracy": 0.7753057803213597,
|
| 336 |
+
"num_tokens": 50790400.0,
|
| 337 |
+
"step": 1550
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.8460013218770654,
|
| 341 |
+
"grad_norm": 1.1253471374511719,
|
| 342 |
+
"learning_rate": 0.0004283066186893468,
|
| 343 |
+
"loss": 1.2626,
|
| 344 |
+
"mean_token_accuracy": 0.7741856071352958,
|
| 345 |
+
"num_tokens": 52428800.0,
|
| 346 |
+
"step": 1600
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.8724388631857237,
|
| 350 |
+
"grad_norm": 0.9880785346031189,
|
| 351 |
+
"learning_rate": 0.0004255806346090939,
|
| 352 |
+
"loss": 1.2731,
|
| 353 |
+
"mean_token_accuracy": 0.7714717736840249,
|
| 354 |
+
"num_tokens": 54067200.0,
|
| 355 |
+
"step": 1650
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.898876404494382,
|
| 359 |
+
"grad_norm": 0.7777257561683655,
|
| 360 |
+
"learning_rate": 0.00042285465052884093,
|
| 361 |
+
"loss": 1.2649,
|
| 362 |
+
"mean_token_accuracy": 0.7737677192687988,
|
| 363 |
+
"num_tokens": 55705600.0,
|
| 364 |
+
"step": 1700
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 0.9253139458030403,
|
| 368 |
+
"grad_norm": 0.8958262801170349,
|
| 369 |
+
"learning_rate": 0.0004201286664485879,
|
| 370 |
+
"loss": 1.2498,
|
| 371 |
+
"step": 1750
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.9253139458030403,
|
| 375 |
+
"eval_loss": 1.2289273738861084,
|
| 376 |
+
"eval_mean_token_accuracy": 0.7793115381836639,
|
| 377 |
+
"eval_num_tokens": 57344000.0,
|
| 378 |
+
"eval_runtime": 1711.9208,
|
| 379 |
+
"eval_samples_per_second": 4.419,
|
| 380 |
+
"eval_steps_per_second": 0.553,
|
| 381 |
+
"step": 1750
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.9517514871116987,
|
| 385 |
+
"grad_norm": 0.9251750111579895,
|
| 386 |
+
"learning_rate": 0.00041740268236833495,
|
| 387 |
+
"loss": 1.2212,
|
| 388 |
+
"mean_token_accuracy": 0.7776866452395916,
|
| 389 |
+
"num_tokens": 58982400.0,
|
| 390 |
+
"step": 1800
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 0.9781890284203569,
|
| 394 |
+
"grad_norm": 0.8101162910461426,
|
| 395 |
+
"learning_rate": 0.00041467669828808203,
|
| 396 |
+
"loss": 1.2025,
|
| 397 |
+
"mean_token_accuracy": 0.7838043755292893,
|
| 398 |
+
"num_tokens": 60620800.0,
|
| 399 |
+
"step": 1850
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 1.0042300066093852,
|
| 403 |
+
"grad_norm": 1.5328075885772705,
|
| 404 |
+
"learning_rate": 0.00041200523388943414,
|
| 405 |
+
"loss": 1.2204,
|
| 406 |
+
"mean_token_accuracy": 0.7796021451804843,
|
| 407 |
+
"num_tokens": 62234624.0,
|
| 408 |
+
"step": 1900
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 1.0306675479180436,
|
| 412 |
+
"grad_norm": 0.9993696212768555,
|
| 413 |
+
"learning_rate": 0.0004092792498091811,
|
| 414 |
+
"loss": 1.1622,
|
| 415 |
+
"mean_token_accuracy": 0.7850250500440598,
|
| 416 |
+
"num_tokens": 63873024.0,
|
| 417 |
+
"step": 1950
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 1.057105089226702,
|
| 421 |
+
"grad_norm": 0.9176653027534485,
|
| 422 |
+
"learning_rate": 0.00040655326572892816,
|
| 423 |
+
"loss": 1.1807,
|
| 424 |
+
"step": 2000
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 1.057105089226702,
|
| 428 |
+
"eval_loss": 1.2058873176574707,
|
| 429 |
+
"eval_mean_token_accuracy": 0.7832688034588893,
|
| 430 |
+
"eval_num_tokens": 65511424.0,
|
| 431 |
+
"eval_runtime": 1712.3284,
|
| 432 |
+
"eval_samples_per_second": 4.418,
|
| 433 |
+
"eval_steps_per_second": 0.552,
|
| 434 |
+
"step": 2000
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 1.0835426305353602,
|
| 438 |
+
"grad_norm": 1.060483455657959,
|
| 439 |
+
"learning_rate": 0.00040382728164867513,
|
| 440 |
+
"loss": 1.1635,
|
| 441 |
+
"mean_token_accuracy": 0.7839937689900398,
|
| 442 |
+
"num_tokens": 67149824.0,
|
| 443 |
+
"step": 2050
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 1.1099801718440185,
|
| 447 |
+
"grad_norm": 1.3085092306137085,
|
| 448 |
+
"learning_rate": 0.0004011012975684222,
|
| 449 |
+
"loss": 1.1928,
|
| 450 |
+
"mean_token_accuracy": 0.7807007575035095,
|
| 451 |
+
"num_tokens": 68788224.0,
|
| 452 |
+
"step": 2100
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.1364177131526767,
|
| 456 |
+
"grad_norm": 0.9808939695358276,
|
| 457 |
+
"learning_rate": 0.00039837531348816925,
|
| 458 |
+
"loss": 1.1454,
|
| 459 |
+
"mean_token_accuracy": 0.7877211648225785,
|
| 460 |
+
"num_tokens": 70426624.0,
|
| 461 |
+
"step": 2150
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.162855254461335,
|
| 465 |
+
"grad_norm": 0.8935715556144714,
|
| 466 |
+
"learning_rate": 0.00039564932940791623,
|
| 467 |
+
"loss": 1.1514,
|
| 468 |
+
"mean_token_accuracy": 0.7872568437457085,
|
| 469 |
+
"num_tokens": 72065024.0,
|
| 470 |
+
"step": 2200
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.1892927957699935,
|
| 474 |
+
"grad_norm": 0.9606215357780457,
|
| 475 |
+
"learning_rate": 0.00039292334532766327,
|
| 476 |
+
"loss": 1.1569,
|
| 477 |
+
"step": 2250
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.1892927957699935,
|
| 481 |
+
"eval_loss": 1.1869500875473022,
|
| 482 |
+
"eval_mean_token_accuracy": 0.7853513486037547,
|
| 483 |
+
"eval_num_tokens": 73703424.0,
|
| 484 |
+
"eval_runtime": 1713.6219,
|
| 485 |
+
"eval_samples_per_second": 4.415,
|
| 486 |
+
"eval_steps_per_second": 0.552,
|
| 487 |
+
"step": 2250
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 1.2157303370786516,
|
| 491 |
+
"grad_norm": 1.2232052087783813,
|
| 492 |
+
"learning_rate": 0.00039019736124741035,
|
| 493 |
+
"loss": 1.1651,
|
| 494 |
+
"mean_token_accuracy": 0.785599644035101,
|
| 495 |
+
"num_tokens": 75341824.0,
|
| 496 |
+
"step": 2300
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 1.24216787838731,
|
| 500 |
+
"grad_norm": 1.2846604585647583,
|
| 501 |
+
"learning_rate": 0.00038747137716715733,
|
| 502 |
+
"loss": 1.1691,
|
| 503 |
+
"mean_token_accuracy": 0.7854020059108734,
|
| 504 |
+
"num_tokens": 76980224.0,
|
| 505 |
+
"step": 2350
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 1.2686054196959682,
|
| 509 |
+
"grad_norm": 1.0899465084075928,
|
| 510 |
+
"learning_rate": 0.00038474539308690437,
|
| 511 |
+
"loss": 1.1416,
|
| 512 |
+
"mean_token_accuracy": 0.7897036933898925,
|
| 513 |
+
"num_tokens": 78618624.0,
|
| 514 |
+
"step": 2400
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.2950429610046266,
|
| 518 |
+
"grad_norm": 1.0662829875946045,
|
| 519 |
+
"learning_rate": 0.0003820194090066514,
|
| 520 |
+
"loss": 1.1435,
|
| 521 |
+
"mean_token_accuracy": 0.7887658843398094,
|
| 522 |
+
"num_tokens": 80257024.0,
|
| 523 |
+
"step": 2450
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 1.321480502313285,
|
| 527 |
+
"grad_norm": 0.9844885468482971,
|
| 528 |
+
"learning_rate": 0.00037929342492639843,
|
| 529 |
+
"loss": 1.1374,
|
| 530 |
+
"step": 2500
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 1.321480502313285,
|
| 534 |
+
"eval_loss": 1.1827911138534546,
|
| 535 |
+
"eval_mean_token_accuracy": 0.7856577540930936,
|
| 536 |
+
"eval_num_tokens": 81895424.0,
|
| 537 |
+
"eval_runtime": 1711.6622,
|
| 538 |
+
"eval_samples_per_second": 4.42,
|
| 539 |
+
"eval_steps_per_second": 0.553,
|
| 540 |
+
"step": 2500
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 1.3479180436219431,
|
| 544 |
+
"grad_norm": 0.8380193710327148,
|
| 545 |
+
"learning_rate": 0.00037656744084614547,
|
| 546 |
+
"loss": 1.1353,
|
| 547 |
+
"mean_token_accuracy": 0.790248963534832,
|
| 548 |
+
"num_tokens": 83533824.0,
|
| 549 |
+
"step": 2550
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.3743555849306015,
|
| 553 |
+
"grad_norm": 0.7707766890525818,
|
| 554 |
+
"learning_rate": 0.0003738414567658925,
|
| 555 |
+
"loss": 1.1309,
|
| 556 |
+
"mean_token_accuracy": 0.7910679399967193,
|
| 557 |
+
"num_tokens": 85172224.0,
|
| 558 |
+
"step": 2600
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 1.4007931262392597,
|
| 562 |
+
"grad_norm": 1.3123962879180908,
|
| 563 |
+
"learning_rate": 0.0003711154726856395,
|
| 564 |
+
"loss": 1.119,
|
| 565 |
+
"mean_token_accuracy": 0.7927541556954384,
|
| 566 |
+
"num_tokens": 86810624.0,
|
| 567 |
+
"step": 2650
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.427230667547918,
|
| 571 |
+
"grad_norm": 0.9434394836425781,
|
| 572 |
+
"learning_rate": 0.00036838948860538656,
|
| 573 |
+
"loss": 1.1233,
|
| 574 |
+
"mean_token_accuracy": 0.7917754176259041,
|
| 575 |
+
"num_tokens": 88449024.0,
|
| 576 |
+
"step": 2700
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 1.4536682088565764,
|
| 580 |
+
"grad_norm": 1.0501152276992798,
|
| 581 |
+
"learning_rate": 0.0003656635045251336,
|
| 582 |
+
"loss": 1.1463,
|
| 583 |
+
"step": 2750
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 1.4536682088565764,
|
| 587 |
+
"eval_loss": 1.1497071981430054,
|
| 588 |
+
"eval_mean_token_accuracy": 0.7899814840981119,
|
| 589 |
+
"eval_num_tokens": 90087424.0,
|
| 590 |
+
"eval_runtime": 1712.288,
|
| 591 |
+
"eval_samples_per_second": 4.418,
|
| 592 |
+
"eval_steps_per_second": 0.552,
|
| 593 |
+
"step": 2750
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.4801057501652346,
|
| 597 |
+
"grad_norm": 1.1930551528930664,
|
| 598 |
+
"learning_rate": 0.0003629375204448806,
|
| 599 |
+
"loss": 1.1309,
|
| 600 |
+
"mean_token_accuracy": 0.7891275675594807,
|
| 601 |
+
"num_tokens": 91725824.0,
|
| 602 |
+
"step": 2800
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 1.5065432914738928,
|
| 606 |
+
"grad_norm": 1.3507503271102905,
|
| 607 |
+
"learning_rate": 0.0003602115363646276,
|
| 608 |
+
"loss": 1.1303,
|
| 609 |
+
"mean_token_accuracy": 0.7900684276223182,
|
| 610 |
+
"num_tokens": 93364224.0,
|
| 611 |
+
"step": 2850
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 1.5329808327825512,
|
| 615 |
+
"grad_norm": 0.935932993888855,
|
| 616 |
+
"learning_rate": 0.0003574855522843747,
|
| 617 |
+
"loss": 1.124,
|
| 618 |
+
"mean_token_accuracy": 0.7918371230363845,
|
| 619 |
+
"num_tokens": 95002624.0,
|
| 620 |
+
"step": 2900
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 1.5594183740912095,
|
| 624 |
+
"grad_norm": 1.3527334928512573,
|
| 625 |
+
"learning_rate": 0.0003547595682041217,
|
| 626 |
+
"loss": 1.1103,
|
| 627 |
+
"mean_token_accuracy": 0.7935648819804192,
|
| 628 |
+
"num_tokens": 96641024.0,
|
| 629 |
+
"step": 2950
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.585855915399868,
|
| 633 |
+
"grad_norm": 0.8783284425735474,
|
| 634 |
+
"learning_rate": 0.0003520335841238687,
|
| 635 |
+
"loss": 1.113,
|
| 636 |
+
"step": 3000
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.585855915399868,
|
| 640 |
+
"eval_loss": 1.1312052011489868,
|
| 641 |
+
"eval_mean_token_accuracy": 0.7928862137597913,
|
| 642 |
+
"eval_num_tokens": 98279424.0,
|
| 643 |
+
"eval_runtime": 1712.2465,
|
| 644 |
+
"eval_samples_per_second": 4.418,
|
| 645 |
+
"eval_steps_per_second": 0.552,
|
| 646 |
+
"step": 3000
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 1.612293456708526,
|
| 650 |
+
"grad_norm": 1.079725980758667,
|
| 651 |
+
"learning_rate": 0.00034930760004361574,
|
| 652 |
+
"loss": 1.107,
|
| 653 |
+
"mean_token_accuracy": 0.7933507452905179,
|
| 654 |
+
"num_tokens": 99917824.0,
|
| 655 |
+
"step": 3050
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.6387309980171842,
|
| 659 |
+
"grad_norm": 1.188661813735962,
|
| 660 |
+
"learning_rate": 0.0003465816159633628,
|
| 661 |
+
"loss": 1.1139,
|
| 662 |
+
"mean_token_accuracy": 0.7925458225607872,
|
| 663 |
+
"num_tokens": 101556224.0,
|
| 664 |
+
"step": 3100
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.6651685393258426,
|
| 668 |
+
"grad_norm": 0.983051061630249,
|
| 669 |
+
"learning_rate": 0.0003438556318831098,
|
| 670 |
+
"loss": 1.1198,
|
| 671 |
+
"mean_token_accuracy": 0.7924914485216141,
|
| 672 |
+
"num_tokens": 103194624.0,
|
| 673 |
+
"step": 3150
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 1.691606080634501,
|
| 677 |
+
"grad_norm": 0.9756836891174316,
|
| 678 |
+
"learning_rate": 0.00034112964780285684,
|
| 679 |
+
"loss": 1.1175,
|
| 680 |
+
"mean_token_accuracy": 0.7917662528157234,
|
| 681 |
+
"num_tokens": 104833024.0,
|
| 682 |
+
"step": 3200
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 1.7180436219431594,
|
| 686 |
+
"grad_norm": 1.1230757236480713,
|
| 687 |
+
"learning_rate": 0.0003384036637226039,
|
| 688 |
+
"loss": 1.0988,
|
| 689 |
+
"step": 3250
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 1.7180436219431594,
|
| 693 |
+
"eval_loss": 1.1241850852966309,
|
| 694 |
+
"eval_mean_token_accuracy": 0.793406566215116,
|
| 695 |
+
"eval_num_tokens": 106471424.0,
|
| 696 |
+
"eval_runtime": 1712.6064,
|
| 697 |
+
"eval_samples_per_second": 4.417,
|
| 698 |
+
"eval_steps_per_second": 0.552,
|
| 699 |
+
"step": 3250
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.7444811632518176,
|
| 703 |
+
"grad_norm": 1.3361942768096924,
|
| 704 |
+
"learning_rate": 0.0003356776796423509,
|
| 705 |
+
"loss": 1.0957,
|
| 706 |
+
"mean_token_accuracy": 0.7953433538973331,
|
| 707 |
+
"num_tokens": 108109824.0,
|
| 708 |
+
"step": 3300
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 1.7709187045604757,
|
| 712 |
+
"grad_norm": 0.8606221079826355,
|
| 713 |
+
"learning_rate": 0.00033295169556209794,
|
| 714 |
+
"loss": 1.072,
|
| 715 |
+
"mean_token_accuracy": 0.7988990727066994,
|
| 716 |
+
"num_tokens": 109748224.0,
|
| 717 |
+
"step": 3350
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 1.7973562458691341,
|
| 721 |
+
"grad_norm": 1.1985405683517456,
|
| 722 |
+
"learning_rate": 0.0003302257114818449,
|
| 723 |
+
"loss": 1.0964,
|
| 724 |
+
"mean_token_accuracy": 0.7953415229916573,
|
| 725 |
+
"num_tokens": 111386624.0,
|
| 726 |
+
"step": 3400
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 1.8237937871777925,
|
| 730 |
+
"grad_norm": 1.0375052690505981,
|
| 731 |
+
"learning_rate": 0.000327499727401592,
|
| 732 |
+
"loss": 1.0821,
|
| 733 |
+
"mean_token_accuracy": 0.7982869046926498,
|
| 734 |
+
"num_tokens": 113025024.0,
|
| 735 |
+
"step": 3450
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 1.8502313284864509,
|
| 739 |
+
"grad_norm": 1.037645936012268,
|
| 740 |
+
"learning_rate": 0.00032477374332133904,
|
| 741 |
+
"loss": 1.0854,
|
| 742 |
+
"step": 3500
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 1.8502313284864509,
|
| 746 |
+
"eval_loss": 1.1053054332733154,
|
| 747 |
+
"eval_mean_token_accuracy": 0.7964076004119051,
|
| 748 |
+
"eval_num_tokens": 114663424.0,
|
| 749 |
+
"eval_runtime": 1713.2036,
|
| 750 |
+
"eval_samples_per_second": 4.416,
|
| 751 |
+
"eval_steps_per_second": 0.552,
|
| 752 |
+
"step": 3500
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 1.876668869795109,
|
| 756 |
+
"grad_norm": 0.8609442114830017,
|
| 757 |
+
"learning_rate": 0.000322047759241086,
|
| 758 |
+
"loss": 1.0789,
|
| 759 |
+
"mean_token_accuracy": 0.7976255512237549,
|
| 760 |
+
"num_tokens": 116301824.0,
|
| 761 |
+
"step": 3550
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 1.9031064111037672,
|
| 765 |
+
"grad_norm": 0.7845131158828735,
|
| 766 |
+
"learning_rate": 0.00031932177516083305,
|
| 767 |
+
"loss": 1.1106,
|
| 768 |
+
"mean_token_accuracy": 0.7933571606874465,
|
| 769 |
+
"num_tokens": 117940224.0,
|
| 770 |
+
"step": 3600
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 1.9295439524124256,
|
| 774 |
+
"grad_norm": 0.9003056287765503,
|
| 775 |
+
"learning_rate": 0.00031659579108058014,
|
| 776 |
+
"loss": 1.0636,
|
| 777 |
+
"mean_token_accuracy": 0.7998753663897514,
|
| 778 |
+
"num_tokens": 119578624.0,
|
| 779 |
+
"step": 3650
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 1.955981493721084,
|
| 783 |
+
"grad_norm": 1.202172040939331,
|
| 784 |
+
"learning_rate": 0.0003138698070003271,
|
| 785 |
+
"loss": 1.0809,
|
| 786 |
+
"mean_token_accuracy": 0.7975763711333275,
|
| 787 |
+
"num_tokens": 121217024.0,
|
| 788 |
+
"step": 3700
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 1.9824190350297424,
|
| 792 |
+
"grad_norm": 0.6643933653831482,
|
| 793 |
+
"learning_rate": 0.00031114382292007415,
|
| 794 |
+
"loss": 1.0808,
|
| 795 |
+
"step": 3750
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 1.9824190350297424,
|
| 799 |
+
"eval_loss": 1.0950915813446045,
|
| 800 |
+
"eval_mean_token_accuracy": 0.7972050871833939,
|
| 801 |
+
"eval_num_tokens": 122855424.0,
|
| 802 |
+
"eval_runtime": 1713.9729,
|
| 803 |
+
"eval_samples_per_second": 4.414,
|
| 804 |
+
"eval_steps_per_second": 0.552,
|
| 805 |
+
"step": 3750
|
| 806 |
+
},
|
| 807 |
+
{
|
| 808 |
+
"epoch": 2.0084600132187704,
|
| 809 |
+
"grad_norm": 1.0089055299758911,
|
| 810 |
+
"learning_rate": 0.00030847235852142626,
|
| 811 |
+
"loss": 1.0434,
|
| 812 |
+
"mean_token_accuracy": 0.7995400524860065,
|
| 813 |
+
"num_tokens": 124469248.0,
|
| 814 |
+
"step": 3800
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 2.034897554527429,
|
| 818 |
+
"grad_norm": 1.8495018482208252,
|
| 819 |
+
"learning_rate": 0.00030574637444117324,
|
| 820 |
+
"loss": 1.0069,
|
| 821 |
+
"mean_token_accuracy": 0.8041293996572495,
|
| 822 |
+
"num_tokens": 126107648.0,
|
| 823 |
+
"step": 3850
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 2.061335095836087,
|
| 827 |
+
"grad_norm": 0.9792631268501282,
|
| 828 |
+
"learning_rate": 0.0003030203903609203,
|
| 829 |
+
"loss": 1.0248,
|
| 830 |
+
"mean_token_accuracy": 0.800488149523735,
|
| 831 |
+
"num_tokens": 127746048.0,
|
| 832 |
+
"step": 3900
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 2.0877726371447456,
|
| 836 |
+
"grad_norm": 1.0454398393630981,
|
| 837 |
+
"learning_rate": 0.00030029440628066736,
|
| 838 |
+
"loss": 1.0032,
|
| 839 |
+
"mean_token_accuracy": 0.8048930823802948,
|
| 840 |
+
"num_tokens": 129384448.0,
|
| 841 |
+
"step": 3950
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 2.114210178453404,
|
| 845 |
+
"grad_norm": 1.124090313911438,
|
| 846 |
+
"learning_rate": 0.00029756842220041434,
|
| 847 |
+
"loss": 1.0141,
|
| 848 |
+
"step": 4000
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 2.114210178453404,
|
| 852 |
+
"eval_loss": 1.0894951820373535,
|
| 853 |
+
"eval_mean_token_accuracy": 0.7976183624146604,
|
| 854 |
+
"eval_num_tokens": 131022848.0,
|
| 855 |
+
"eval_runtime": 1713.146,
|
| 856 |
+
"eval_samples_per_second": 4.416,
|
| 857 |
+
"eval_steps_per_second": 0.552,
|
| 858 |
+
"step": 4000
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 2.140647719762062,
|
| 862 |
+
"grad_norm": 1.068744421005249,
|
| 863 |
+
"learning_rate": 0.0002948424381201614,
|
| 864 |
+
"loss": 1.017,
|
| 865 |
+
"mean_token_accuracy": 0.8032655183970928,
|
| 866 |
+
"num_tokens": 132661248.0,
|
| 867 |
+
"step": 4050
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 2.1670852610707203,
|
| 871 |
+
"grad_norm": 0.6453216671943665,
|
| 872 |
+
"learning_rate": 0.0002921164540399084,
|
| 873 |
+
"loss": 1.0173,
|
| 874 |
+
"mean_token_accuracy": 0.8023881965875626,
|
| 875 |
+
"num_tokens": 134299648.0,
|
| 876 |
+
"step": 4100
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 2.1935228023793787,
|
| 880 |
+
"grad_norm": 0.8488343954086304,
|
| 881 |
+
"learning_rate": 0.00028939046995965544,
|
| 882 |
+
"loss": 1.0192,
|
| 883 |
+
"mean_token_accuracy": 0.8028500735759735,
|
| 884 |
+
"num_tokens": 135938048.0,
|
| 885 |
+
"step": 4150
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 2.219960343688037,
|
| 889 |
+
"grad_norm": 1.107086181640625,
|
| 890 |
+
"learning_rate": 0.00028666448587940247,
|
| 891 |
+
"loss": 1.0167,
|
| 892 |
+
"mean_token_accuracy": 0.8023680368065834,
|
| 893 |
+
"num_tokens": 137576448.0,
|
| 894 |
+
"step": 4200
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 2.2463978849966955,
|
| 898 |
+
"grad_norm": 0.9816263914108276,
|
| 899 |
+
"learning_rate": 0.0002839385017991495,
|
| 900 |
+
"loss": 1.026,
|
| 901 |
+
"step": 4250
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 2.2463978849966955,
|
| 905 |
+
"eval_loss": 1.070574402809143,
|
| 906 |
+
"eval_mean_token_accuracy": 0.8008053159738948,
|
| 907 |
+
"eval_num_tokens": 139214848.0,
|
| 908 |
+
"eval_runtime": 1713.5677,
|
| 909 |
+
"eval_samples_per_second": 4.415,
|
| 910 |
+
"eval_steps_per_second": 0.552,
|
| 911 |
+
"step": 4250
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 2.2728354263053534,
|
| 915 |
+
"grad_norm": 1.0816267728805542,
|
| 916 |
+
"learning_rate": 0.00028121251771889654,
|
| 917 |
+
"loss": 1.0271,
|
| 918 |
+
"mean_token_accuracy": 0.8018050470948219,
|
| 919 |
+
"num_tokens": 140853248.0,
|
| 920 |
+
"step": 4300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 2.299272967614012,
|
| 924 |
+
"grad_norm": 1.1499203443527222,
|
| 925 |
+
"learning_rate": 0.00027848653363864357,
|
| 926 |
+
"loss": 1.0008,
|
| 927 |
+
"mean_token_accuracy": 0.8052902013063431,
|
| 928 |
+
"num_tokens": 142491648.0,
|
| 929 |
+
"step": 4350
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 2.32571050892267,
|
| 933 |
+
"grad_norm": 0.9058449864387512,
|
| 934 |
+
"learning_rate": 0.0002757605495583906,
|
| 935 |
+
"loss": 1.002,
|
| 936 |
+
"mean_token_accuracy": 0.806196848154068,
|
| 937 |
+
"num_tokens": 144130048.0,
|
| 938 |
+
"step": 4400
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 2.3521480502313286,
|
| 942 |
+
"grad_norm": 0.8086408972740173,
|
| 943 |
+
"learning_rate": 0.0002730345654781376,
|
| 944 |
+
"loss": 0.9968,
|
| 945 |
+
"mean_token_accuracy": 0.8068004646897315,
|
| 946 |
+
"num_tokens": 145768448.0,
|
| 947 |
+
"step": 4450
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 2.378585591539987,
|
| 951 |
+
"grad_norm": 0.9422939419746399,
|
| 952 |
+
"learning_rate": 0.00027030858139788467,
|
| 953 |
+
"loss": 0.9925,
|
| 954 |
+
"step": 4500
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 2.378585591539987,
|
| 958 |
+
"eval_loss": 1.0708719491958618,
|
| 959 |
+
"eval_mean_token_accuracy": 0.8005553823570872,
|
| 960 |
+
"eval_num_tokens": 147406848.0,
|
| 961 |
+
"eval_runtime": 1714.6525,
|
| 962 |
+
"eval_samples_per_second": 4.412,
|
| 963 |
+
"eval_steps_per_second": 0.552,
|
| 964 |
+
"step": 4500
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 2.405023132848645,
|
| 968 |
+
"grad_norm": 0.9562957882881165,
|
| 969 |
+
"learning_rate": 0.0002675825973176317,
|
| 970 |
+
"loss": 1.0125,
|
| 971 |
+
"mean_token_accuracy": 0.8054704304039478,
|
| 972 |
+
"num_tokens": 149045248.0,
|
| 973 |
+
"step": 4550
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 2.4314606741573033,
|
| 977 |
+
"grad_norm": 0.8408384919166565,
|
| 978 |
+
"learning_rate": 0.0002648566132373787,
|
| 979 |
+
"loss": 1.0194,
|
| 980 |
+
"mean_token_accuracy": 0.8027529340982437,
|
| 981 |
+
"num_tokens": 150683648.0,
|
| 982 |
+
"step": 4600
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 2.4578982154659617,
|
| 986 |
+
"grad_norm": 0.9756256341934204,
|
| 987 |
+
"learning_rate": 0.0002621306291571257,
|
| 988 |
+
"loss": 0.9828,
|
| 989 |
+
"mean_token_accuracy": 0.8083504402637481,
|
| 990 |
+
"num_tokens": 152322048.0,
|
| 991 |
+
"step": 4650
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 2.48433575677462,
|
| 995 |
+
"grad_norm": 1.2137442827224731,
|
| 996 |
+
"learning_rate": 0.0002594046450768728,
|
| 997 |
+
"loss": 1.0021,
|
| 998 |
+
"mean_token_accuracy": 0.8056860953569412,
|
| 999 |
+
"num_tokens": 153960448.0,
|
| 1000 |
+
"step": 4700
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 2.5107732980832784,
|
| 1004 |
+
"grad_norm": 0.7544079422950745,
|
| 1005 |
+
"learning_rate": 0.0002566786609966198,
|
| 1006 |
+
"loss": 1.0089,
|
| 1007 |
+
"step": 4750
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 2.5107732980832784,
|
| 1011 |
+
"eval_loss": 1.0549876689910889,
|
| 1012 |
+
"eval_mean_token_accuracy": 0.8026504306158102,
|
| 1013 |
+
"eval_num_tokens": 155598848.0,
|
| 1014 |
+
"eval_runtime": 1712.6528,
|
| 1015 |
+
"eval_samples_per_second": 4.417,
|
| 1016 |
+
"eval_steps_per_second": 0.552,
|
| 1017 |
+
"step": 4750
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.5372108393919364,
|
| 1021 |
+
"grad_norm": 1.3571584224700928,
|
| 1022 |
+
"learning_rate": 0.0002539526769163668,
|
| 1023 |
+
"loss": 0.9947,
|
| 1024 |
+
"mean_token_accuracy": 0.8055629892647267,
|
| 1025 |
+
"num_tokens": 157237248.0,
|
| 1026 |
+
"step": 4800
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 2.5636483807005948,
|
| 1030 |
+
"grad_norm": 0.9661728739738464,
|
| 1031 |
+
"learning_rate": 0.00025122669283611385,
|
| 1032 |
+
"loss": 0.9934,
|
| 1033 |
+
"mean_token_accuracy": 0.8067454797029495,
|
| 1034 |
+
"num_tokens": 158875648.0,
|
| 1035 |
+
"step": 4850
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.590085922009253,
|
| 1039 |
+
"grad_norm": 0.8960219025611877,
|
| 1040 |
+
"learning_rate": 0.0002485007087558609,
|
| 1041 |
+
"loss": 1.0084,
|
| 1042 |
+
"mean_token_accuracy": 0.8042240959405899,
|
| 1043 |
+
"num_tokens": 160514048.0,
|
| 1044 |
+
"step": 4900
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 2.6165234633179115,
|
| 1048 |
+
"grad_norm": 0.8337807059288025,
|
| 1049 |
+
"learning_rate": 0.0002457747246756079,
|
| 1050 |
+
"loss": 1.0028,
|
| 1051 |
+
"mean_token_accuracy": 0.8062023460865021,
|
| 1052 |
+
"num_tokens": 162152448.0,
|
| 1053 |
+
"step": 4950
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.64296100462657,
|
| 1057 |
+
"grad_norm": 1.2237184047698975,
|
| 1058 |
+
"learning_rate": 0.00024304874059535492,
|
| 1059 |
+
"loss": 0.9907,
|
| 1060 |
+
"step": 5000
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 2.64296100462657,
|
| 1064 |
+
"eval_loss": 1.0443217754364014,
|
| 1065 |
+
"eval_mean_token_accuracy": 0.8038508863706165,
|
| 1066 |
+
"eval_num_tokens": 163790848.0,
|
| 1067 |
+
"eval_runtime": 1716.2002,
|
| 1068 |
+
"eval_samples_per_second": 4.408,
|
| 1069 |
+
"eval_steps_per_second": 0.551,
|
| 1070 |
+
"step": 5000
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 2.669398545935228,
|
| 1074 |
+
"grad_norm": 0.7289232015609741,
|
| 1075 |
+
"learning_rate": 0.00024032275651510195,
|
| 1076 |
+
"loss": 1.0012,
|
| 1077 |
+
"mean_token_accuracy": 0.8063395051658153,
|
| 1078 |
+
"num_tokens": 165429248.0,
|
| 1079 |
+
"step": 5050
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 2.6958360872438862,
|
| 1083 |
+
"grad_norm": 1.0250189304351807,
|
| 1084 |
+
"learning_rate": 0.000237596772434849,
|
| 1085 |
+
"loss": 0.976,
|
| 1086 |
+
"mean_token_accuracy": 0.8098240447044373,
|
| 1087 |
+
"num_tokens": 167067648.0,
|
| 1088 |
+
"step": 5100
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 2.7222736285525446,
|
| 1092 |
+
"grad_norm": 1.0034643411636353,
|
| 1093 |
+
"learning_rate": 0.00023487078835459602,
|
| 1094 |
+
"loss": 0.9848,
|
| 1095 |
+
"mean_token_accuracy": 0.8076087480783463,
|
| 1096 |
+
"num_tokens": 168706048.0,
|
| 1097 |
+
"step": 5150
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 2.748711169861203,
|
| 1101 |
+
"grad_norm": 0.9291382431983948,
|
| 1102 |
+
"learning_rate": 0.00023214480427434303,
|
| 1103 |
+
"loss": 0.9757,
|
| 1104 |
+
"mean_token_accuracy": 0.8101533487439155,
|
| 1105 |
+
"num_tokens": 170344448.0,
|
| 1106 |
+
"step": 5200
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 2.7751487111698614,
|
| 1110 |
+
"grad_norm": 0.6038099527359009,
|
| 1111 |
+
"learning_rate": 0.00022941882019409009,
|
| 1112 |
+
"loss": 0.989,
|
| 1113 |
+
"step": 5250
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 2.7751487111698614,
|
| 1117 |
+
"eval_loss": 1.0298680067062378,
|
| 1118 |
+
"eval_mean_token_accuracy": 0.8057682283584966,
|
| 1119 |
+
"eval_num_tokens": 171982848.0,
|
| 1120 |
+
"eval_runtime": 1715.5734,
|
| 1121 |
+
"eval_samples_per_second": 4.41,
|
| 1122 |
+
"eval_steps_per_second": 0.551,
|
| 1123 |
+
"step": 5250
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 2.8015862524785193,
|
| 1127 |
+
"grad_norm": 0.8782141804695129,
|
| 1128 |
+
"learning_rate": 0.0002266928361138371,
|
| 1129 |
+
"loss": 1.0017,
|
| 1130 |
+
"mean_token_accuracy": 0.8068728642165661,
|
| 1131 |
+
"num_tokens": 173621248.0,
|
| 1132 |
+
"step": 5300
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 2.8280237937871777,
|
| 1136 |
+
"grad_norm": 0.5077300667762756,
|
| 1137 |
+
"learning_rate": 0.00022396685203358413,
|
| 1138 |
+
"loss": 0.9759,
|
| 1139 |
+
"mean_token_accuracy": 0.8097360721230507,
|
| 1140 |
+
"num_tokens": 175259648.0,
|
| 1141 |
+
"step": 5350
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 2.854461335095836,
|
| 1145 |
+
"grad_norm": 0.571225643157959,
|
| 1146 |
+
"learning_rate": 0.00022124086795333116,
|
| 1147 |
+
"loss": 0.9693,
|
| 1148 |
+
"mean_token_accuracy": 0.8103903934359551,
|
| 1149 |
+
"num_tokens": 176898048.0,
|
| 1150 |
+
"step": 5400
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 2.8808988764044945,
|
| 1154 |
+
"grad_norm": 0.9907204508781433,
|
| 1155 |
+
"learning_rate": 0.0002185148838730782,
|
| 1156 |
+
"loss": 0.9783,
|
| 1157 |
+
"mean_token_accuracy": 0.8095271262526512,
|
| 1158 |
+
"num_tokens": 178536448.0,
|
| 1159 |
+
"step": 5450
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 2.907336417713153,
|
| 1163 |
+
"grad_norm": 1.0461844205856323,
|
| 1164 |
+
"learning_rate": 0.0002157888997928252,
|
| 1165 |
+
"loss": 0.9796,
|
| 1166 |
+
"step": 5500
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 2.907336417713153,
|
| 1170 |
+
"eval_loss": 1.0164023637771606,
|
| 1171 |
+
"eval_mean_token_accuracy": 0.8078667395462698,
|
| 1172 |
+
"eval_num_tokens": 180174848.0,
|
| 1173 |
+
"eval_runtime": 1713.163,
|
| 1174 |
+
"eval_samples_per_second": 4.416,
|
| 1175 |
+
"eval_steps_per_second": 0.552,
|
| 1176 |
+
"step": 5500
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 2.933773959021811,
|
| 1180 |
+
"grad_norm": 0.5164626240730286,
|
| 1181 |
+
"learning_rate": 0.00021306291571257226,
|
| 1182 |
+
"loss": 0.9813,
|
| 1183 |
+
"mean_token_accuracy": 0.8092189015448094,
|
| 1184 |
+
"num_tokens": 181813248.0,
|
| 1185 |
+
"step": 5550
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 2.960211500330469,
|
| 1189 |
+
"grad_norm": 0.9014139771461487,
|
| 1190 |
+
"learning_rate": 0.00021033693163231926,
|
| 1191 |
+
"loss": 0.9785,
|
| 1192 |
+
"mean_token_accuracy": 0.808426809310913,
|
| 1193 |
+
"num_tokens": 183451648.0,
|
| 1194 |
+
"step": 5600
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 2.9866490416391276,
|
| 1198 |
+
"grad_norm": 1.2656482458114624,
|
| 1199 |
+
"learning_rate": 0.0002076109475520663,
|
| 1200 |
+
"loss": 0.9546,
|
| 1201 |
+
"mean_token_accuracy": 0.8128830647468567,
|
| 1202 |
+
"num_tokens": 185090048.0,
|
| 1203 |
+
"step": 5650
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 3.012690019828156,
|
| 1207 |
+
"grad_norm": 0.6291442513465881,
|
| 1208 |
+
"learning_rate": 0.0002049394831534184,
|
| 1209 |
+
"loss": 0.9425,
|
| 1210 |
+
"mean_token_accuracy": 0.8123012103405095,
|
| 1211 |
+
"num_tokens": 186703872.0,
|
| 1212 |
+
"step": 5700
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 3.0391275611368145,
|
| 1216 |
+
"grad_norm": 0.9149487614631653,
|
| 1217 |
+
"learning_rate": 0.0002022134990731654,
|
| 1218 |
+
"loss": 0.9036,
|
| 1219 |
+
"step": 5750
|
| 1220 |
+
},
|
| 1221 |
+
{
|
| 1222 |
+
"epoch": 3.0391275611368145,
|
| 1223 |
+
"eval_loss": 1.010271668434143,
|
| 1224 |
+
"eval_mean_token_accuracy": 0.8088774525463959,
|
| 1225 |
+
"eval_num_tokens": 188342272.0,
|
| 1226 |
+
"eval_runtime": 1714.8688,
|
| 1227 |
+
"eval_samples_per_second": 4.411,
|
| 1228 |
+
"eval_steps_per_second": 0.552,
|
| 1229 |
+
"step": 5750
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 3.0655651024454724,
|
| 1233 |
+
"grad_norm": 0.7065662741661072,
|
| 1234 |
+
"learning_rate": 0.00019948751499291245,
|
| 1235 |
+
"loss": 0.9032,
|
| 1236 |
+
"mean_token_accuracy": 0.816273825019598,
|
| 1237 |
+
"num_tokens": 189980672.0,
|
| 1238 |
+
"step": 5800
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 3.092002643754131,
|
| 1242 |
+
"grad_norm": 0.8670871257781982,
|
| 1243 |
+
"learning_rate": 0.00019676153091265948,
|
| 1244 |
+
"loss": 0.9084,
|
| 1245 |
+
"mean_token_accuracy": 0.8148881956934929,
|
| 1246 |
+
"num_tokens": 191619072.0,
|
| 1247 |
+
"step": 5850
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 3.118440185062789,
|
| 1251 |
+
"grad_norm": 0.9667902588844299,
|
| 1252 |
+
"learning_rate": 0.0001940355468324065,
|
| 1253 |
+
"loss": 0.8968,
|
| 1254 |
+
"mean_token_accuracy": 0.8168530049920082,
|
| 1255 |
+
"num_tokens": 193257472.0,
|
| 1256 |
+
"step": 5900
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 3.1448777263714476,
|
| 1260 |
+
"grad_norm": 0.6061888933181763,
|
| 1261 |
+
"learning_rate": 0.00019130956275215352,
|
| 1262 |
+
"loss": 0.9082,
|
| 1263 |
+
"mean_token_accuracy": 0.8162536644935607,
|
| 1264 |
+
"num_tokens": 194895872.0,
|
| 1265 |
+
"step": 5950
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 3.1713152676801055,
|
| 1269 |
+
"grad_norm": 0.8645080924034119,
|
| 1270 |
+
"learning_rate": 0.00018858357867190058,
|
| 1271 |
+
"loss": 0.9014,
|
| 1272 |
+
"step": 6000
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 3.1713152676801055,
|
| 1276 |
+
"eval_loss": 1.0060479640960693,
|
| 1277 |
+
"eval_mean_token_accuracy": 0.8095184696275134,
|
| 1278 |
+
"eval_num_tokens": 196534272.0,
|
| 1279 |
+
"eval_runtime": 1716.7924,
|
| 1280 |
+
"eval_samples_per_second": 4.406,
|
| 1281 |
+
"eval_steps_per_second": 0.551,
|
| 1282 |
+
"step": 6000
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 3.197752808988764,
|
| 1286 |
+
"grad_norm": 0.9978011250495911,
|
| 1287 |
+
"learning_rate": 0.00018585759459164758,
|
| 1288 |
+
"loss": 0.8941,
|
| 1289 |
+
"mean_token_accuracy": 0.817662510573864,
|
| 1290 |
+
"num_tokens": 198172672.0,
|
| 1291 |
+
"step": 6050
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 3.2241903502974223,
|
| 1295 |
+
"grad_norm": 0.610701322555542,
|
| 1296 |
+
"learning_rate": 0.00018313161051139462,
|
| 1297 |
+
"loss": 0.9088,
|
| 1298 |
+
"mean_token_accuracy": 0.8161278122663498,
|
| 1299 |
+
"num_tokens": 199811072.0,
|
| 1300 |
+
"step": 6100
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 3.2506278916060807,
|
| 1304 |
+
"grad_norm": 0.592491626739502,
|
| 1305 |
+
"learning_rate": 0.00018040562643114165,
|
| 1306 |
+
"loss": 0.9111,
|
| 1307 |
+
"mean_token_accuracy": 0.8152193301916122,
|
| 1308 |
+
"num_tokens": 201449472.0,
|
| 1309 |
+
"step": 6150
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 3.277065432914739,
|
| 1313 |
+
"grad_norm": 0.5505239367485046,
|
| 1314 |
+
"learning_rate": 0.00017767964235088868,
|
| 1315 |
+
"loss": 0.894,
|
| 1316 |
+
"mean_token_accuracy": 0.8186956241726875,
|
| 1317 |
+
"num_tokens": 203087872.0,
|
| 1318 |
+
"step": 6200
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 3.303502974223397,
|
| 1322 |
+
"grad_norm": 0.6099046468734741,
|
| 1323 |
+
"learning_rate": 0.0001749536582706357,
|
| 1324 |
+
"loss": 0.9061,
|
| 1325 |
+
"step": 6250
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 3.303502974223397,
|
| 1329 |
+
"eval_loss": 0.9953573942184448,
|
| 1330 |
+
"eval_mean_token_accuracy": 0.8111508759585294,
|
| 1331 |
+
"eval_num_tokens": 204726272.0,
|
| 1332 |
+
"eval_runtime": 1714.1014,
|
| 1333 |
+
"eval_samples_per_second": 4.413,
|
| 1334 |
+
"eval_steps_per_second": 0.552,
|
| 1335 |
+
"step": 6250
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 3.3299405155320554,
|
| 1339 |
+
"grad_norm": 0.6866306066513062,
|
| 1340 |
+
"learning_rate": 0.00017222767419038275,
|
| 1341 |
+
"loss": 0.9023,
|
| 1342 |
+
"mean_token_accuracy": 0.8162371690571308,
|
| 1343 |
+
"num_tokens": 206364672.0,
|
| 1344 |
+
"step": 6300
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 3.3563780568407138,
|
| 1348 |
+
"grad_norm": 0.6457993984222412,
|
| 1349 |
+
"learning_rate": 0.00016950169011012976,
|
| 1350 |
+
"loss": 0.9083,
|
| 1351 |
+
"mean_token_accuracy": 0.8154759269952774,
|
| 1352 |
+
"num_tokens": 208003072.0,
|
| 1353 |
+
"step": 6350
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 3.382815598149372,
|
| 1357 |
+
"grad_norm": 0.5925601124763489,
|
| 1358 |
+
"learning_rate": 0.0001667757060298768,
|
| 1359 |
+
"loss": 0.9009,
|
| 1360 |
+
"mean_token_accuracy": 0.8165108740329743,
|
| 1361 |
+
"num_tokens": 209641472.0,
|
| 1362 |
+
"step": 6400
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 3.4092531394580305,
|
| 1366 |
+
"grad_norm": 0.8631545901298523,
|
| 1367 |
+
"learning_rate": 0.00016404972194962382,
|
| 1368 |
+
"loss": 0.8779,
|
| 1369 |
+
"mean_token_accuracy": 0.8210025626420975,
|
| 1370 |
+
"num_tokens": 211279872.0,
|
| 1371 |
+
"step": 6450
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 3.4356906807666885,
|
| 1375 |
+
"grad_norm": 0.6113960146903992,
|
| 1376 |
+
"learning_rate": 0.00016132373786937086,
|
| 1377 |
+
"loss": 0.8894,
|
| 1378 |
+
"step": 6500
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 3.4356906807666885,
|
| 1382 |
+
"eval_loss": 0.9821568131446838,
|
| 1383 |
+
"eval_mean_token_accuracy": 0.8126549717613809,
|
| 1384 |
+
"eval_num_tokens": 212918272.0,
|
| 1385 |
+
"eval_runtime": 1714.991,
|
| 1386 |
+
"eval_samples_per_second": 4.411,
|
| 1387 |
+
"eval_steps_per_second": 0.552,
|
| 1388 |
+
"step": 6500
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 3.462128222075347,
|
| 1392 |
+
"grad_norm": 0.6569721698760986,
|
| 1393 |
+
"learning_rate": 0.00015859775378911786,
|
| 1394 |
+
"loss": 0.8828,
|
| 1395 |
+
"mean_token_accuracy": 0.8195891354978084,
|
| 1396 |
+
"num_tokens": 214556672.0,
|
| 1397 |
+
"step": 6550
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 3.4885657633840053,
|
| 1401 |
+
"grad_norm": 0.548383891582489,
|
| 1402 |
+
"learning_rate": 0.0001558717697088649,
|
| 1403 |
+
"loss": 0.8922,
|
| 1404 |
+
"mean_token_accuracy": 0.8191379508376122,
|
| 1405 |
+
"num_tokens": 216195072.0,
|
| 1406 |
+
"step": 6600
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 3.5150033046926636,
|
| 1410 |
+
"grad_norm": 0.4519716799259186,
|
| 1411 |
+
"learning_rate": 0.00015314578562861193,
|
| 1412 |
+
"loss": 0.9047,
|
| 1413 |
+
"mean_token_accuracy": 0.8174547863006592,
|
| 1414 |
+
"num_tokens": 217833472.0,
|
| 1415 |
+
"step": 6650
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 3.541440846001322,
|
| 1419 |
+
"grad_norm": 0.4486851692199707,
|
| 1420 |
+
"learning_rate": 0.00015041980154835896,
|
| 1421 |
+
"loss": 0.8812,
|
| 1422 |
+
"mean_token_accuracy": 0.8206359946727753,
|
| 1423 |
+
"num_tokens": 219471872.0,
|
| 1424 |
+
"step": 6700
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 3.56787838730998,
|
| 1428 |
+
"grad_norm": 0.511616587638855,
|
| 1429 |
+
"learning_rate": 0.00014769381746810597,
|
| 1430 |
+
"loss": 0.8754,
|
| 1431 |
+
"step": 6750
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 3.56787838730998,
|
| 1435 |
+
"eval_loss": 0.9736062288284302,
|
| 1436 |
+
"eval_mean_token_accuracy": 0.8140331042997428,
|
| 1437 |
+
"eval_num_tokens": 221110272.0,
|
| 1438 |
+
"eval_runtime": 1716.1615,
|
| 1439 |
+
"eval_samples_per_second": 4.408,
|
| 1440 |
+
"eval_steps_per_second": 0.551,
|
| 1441 |
+
"step": 6750
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 3.5943159286186384,
|
| 1445 |
+
"grad_norm": 0.4776919186115265,
|
| 1446 |
+
"learning_rate": 0.00014496783338785303,
|
| 1447 |
+
"loss": 0.8959,
|
| 1448 |
+
"mean_token_accuracy": 0.8200723953545094,
|
| 1449 |
+
"num_tokens": 222748672.0,
|
| 1450 |
+
"step": 6800
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 3.6207534699272967,
|
| 1454 |
+
"grad_norm": 0.6696462035179138,
|
| 1455 |
+
"learning_rate": 0.00014224184930760003,
|
| 1456 |
+
"loss": 0.8818,
|
| 1457 |
+
"mean_token_accuracy": 0.8205584043264389,
|
| 1458 |
+
"num_tokens": 224387072.0,
|
| 1459 |
+
"step": 6850
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 3.647191011235955,
|
| 1463 |
+
"grad_norm": 0.6305286884307861,
|
| 1464 |
+
"learning_rate": 0.00013951586522734707,
|
| 1465 |
+
"loss": 0.8918,
|
| 1466 |
+
"mean_token_accuracy": 0.8183388301730156,
|
| 1467 |
+
"num_tokens": 226025472.0,
|
| 1468 |
+
"step": 6900
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 3.6736285525446135,
|
| 1472 |
+
"grad_norm": 0.4481205344200134,
|
| 1473 |
+
"learning_rate": 0.0001367898811470941,
|
| 1474 |
+
"loss": 0.8689,
|
| 1475 |
+
"mean_token_accuracy": 0.822255617082119,
|
| 1476 |
+
"num_tokens": 227663872.0,
|
| 1477 |
+
"step": 6950
|
| 1478 |
+
},
|
| 1479 |
+
{
|
| 1480 |
+
"epoch": 3.7000660938532715,
|
| 1481 |
+
"grad_norm": 0.5297748446464539,
|
| 1482 |
+
"learning_rate": 0.00013406389706684113,
|
| 1483 |
+
"loss": 0.891,
|
| 1484 |
+
"step": 7000
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 3.7000660938532715,
|
| 1488 |
+
"eval_loss": 0.9645546078681946,
|
| 1489 |
+
"eval_mean_token_accuracy": 0.8154595721725681,
|
| 1490 |
+
"eval_num_tokens": 229302272.0,
|
| 1491 |
+
"eval_runtime": 1714.7844,
|
| 1492 |
+
"eval_samples_per_second": 4.412,
|
| 1493 |
+
"eval_steps_per_second": 0.552,
|
| 1494 |
+
"step": 7000
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 3.72650363516193,
|
| 1498 |
+
"grad_norm": 0.5066333413124084,
|
| 1499 |
+
"learning_rate": 0.00013133791298658814,
|
| 1500 |
+
"loss": 0.8888,
|
| 1501 |
+
"mean_token_accuracy": 0.8194779419898987,
|
| 1502 |
+
"num_tokens": 230940672.0,
|
| 1503 |
+
"step": 7050
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 3.7529411764705882,
|
| 1507 |
+
"grad_norm": 0.5374875068664551,
|
| 1508 |
+
"learning_rate": 0.0001286119289063352,
|
| 1509 |
+
"loss": 0.8668,
|
| 1510 |
+
"mean_token_accuracy": 0.8228415179252625,
|
| 1511 |
+
"num_tokens": 232579072.0,
|
| 1512 |
+
"step": 7100
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"epoch": 3.7793787177792466,
|
| 1516 |
+
"grad_norm": 0.45081761479377747,
|
| 1517 |
+
"learning_rate": 0.0001258859448260822,
|
| 1518 |
+
"loss": 0.8793,
|
| 1519 |
+
"mean_token_accuracy": 0.8205449622869492,
|
| 1520 |
+
"num_tokens": 234217472.0,
|
| 1521 |
+
"step": 7150
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 3.805816259087905,
|
| 1525 |
+
"grad_norm": 0.4918268620967865,
|
| 1526 |
+
"learning_rate": 0.00012315996074582924,
|
| 1527 |
+
"loss": 0.8733,
|
| 1528 |
+
"mean_token_accuracy": 0.8207551288604736,
|
| 1529 |
+
"num_tokens": 235855872.0,
|
| 1530 |
+
"step": 7200
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 3.832253800396563,
|
| 1534 |
+
"grad_norm": 0.5663712024688721,
|
| 1535 |
+
"learning_rate": 0.00012043397666557627,
|
| 1536 |
+
"loss": 0.8701,
|
| 1537 |
+
"step": 7250
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 3.832253800396563,
|
| 1541 |
+
"eval_loss": 0.9548874497413635,
|
| 1542 |
+
"eval_mean_token_accuracy": 0.8167319189418446,
|
| 1543 |
+
"eval_num_tokens": 237494272.0,
|
| 1544 |
+
"eval_runtime": 1714.4512,
|
| 1545 |
+
"eval_samples_per_second": 4.412,
|
| 1546 |
+
"eval_steps_per_second": 0.552,
|
| 1547 |
+
"step": 7250
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 3.8586913417052213,
|
| 1551 |
+
"grad_norm": 1.051850438117981,
|
| 1552 |
+
"learning_rate": 0.0001177079925853233,
|
| 1553 |
+
"loss": 0.8694,
|
| 1554 |
+
"mean_token_accuracy": 0.8223982758820056,
|
| 1555 |
+
"num_tokens": 239132672.0,
|
| 1556 |
+
"step": 7300
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 3.8851288830138797,
|
| 1560 |
+
"grad_norm": 0.4363590478897095,
|
| 1561 |
+
"learning_rate": 0.00011498200850507034,
|
| 1562 |
+
"loss": 0.8864,
|
| 1563 |
+
"mean_token_accuracy": 0.8197085753083229,
|
| 1564 |
+
"num_tokens": 240771072.0,
|
| 1565 |
+
"step": 7350
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 3.911566424322538,
|
| 1569 |
+
"grad_norm": 1.5718705654144287,
|
| 1570 |
+
"learning_rate": 0.00011225602442481736,
|
| 1571 |
+
"loss": 0.8714,
|
| 1572 |
+
"mean_token_accuracy": 0.8220063516497612,
|
| 1573 |
+
"num_tokens": 242409472.0,
|
| 1574 |
+
"step": 7400
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 3.9380039656311965,
|
| 1578 |
+
"grad_norm": 2.0182573795318604,
|
| 1579 |
+
"learning_rate": 0.00010953004034456439,
|
| 1580 |
+
"loss": 0.8666,
|
| 1581 |
+
"mean_token_accuracy": 0.8229606547951698,
|
| 1582 |
+
"num_tokens": 244047872.0,
|
| 1583 |
+
"step": 7450
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 3.9644415069398544,
|
| 1587 |
+
"grad_norm": 0.5944796800613403,
|
| 1588 |
+
"learning_rate": 0.00010680405626431142,
|
| 1589 |
+
"loss": 0.8553,
|
| 1590 |
+
"step": 7500
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 3.9644415069398544,
|
| 1594 |
+
"eval_loss": 0.9463370442390442,
|
| 1595 |
+
"eval_mean_token_accuracy": 0.8175775335026594,
|
| 1596 |
+
"eval_num_tokens": 245686272.0,
|
| 1597 |
+
"eval_runtime": 1716.9243,
|
| 1598 |
+
"eval_samples_per_second": 4.406,
|
| 1599 |
+
"eval_steps_per_second": 0.551,
|
| 1600 |
+
"step": 7500
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"epoch": 3.990879048248513,
|
| 1604 |
+
"grad_norm": 0.39286720752716064,
|
| 1605 |
+
"learning_rate": 0.00010407807218405844,
|
| 1606 |
+
"loss": 0.8742,
|
| 1607 |
+
"mean_token_accuracy": 0.8231118628382683,
|
| 1608 |
+
"num_tokens": 247324672.0,
|
| 1609 |
+
"step": 7550
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 4.016920026437541,
|
| 1613 |
+
"grad_norm": 0.485441118478775,
|
| 1614 |
+
"learning_rate": 0.00010135208810380548,
|
| 1615 |
+
"loss": 0.8161,
|
| 1616 |
+
"mean_token_accuracy": 0.8290703360199323,
|
| 1617 |
+
"num_tokens": 248938496.0,
|
| 1618 |
+
"step": 7600
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 4.0433575677462,
|
| 1622 |
+
"grad_norm": 0.6562045216560364,
|
| 1623 |
+
"learning_rate": 9.86261040235525e-05,
|
| 1624 |
+
"loss": 0.7963,
|
| 1625 |
+
"mean_token_accuracy": 0.829828929901123,
|
| 1626 |
+
"num_tokens": 250576896.0,
|
| 1627 |
+
"step": 7650
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 4.069795109054858,
|
| 1631 |
+
"grad_norm": 0.4619589149951935,
|
| 1632 |
+
"learning_rate": 9.590011994329953e-05,
|
| 1633 |
+
"loss": 0.7968,
|
| 1634 |
+
"mean_token_accuracy": 0.829875974059105,
|
| 1635 |
+
"num_tokens": 252215296.0,
|
| 1636 |
+
"step": 7700
|
| 1637 |
+
},
|
| 1638 |
+
{
|
| 1639 |
+
"epoch": 4.0962326503635165,
|
| 1640 |
+
"grad_norm": 0.5542292594909668,
|
| 1641 |
+
"learning_rate": 9.317413586304656e-05,
|
| 1642 |
+
"loss": 0.7897,
|
| 1643 |
+
"step": 7750
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 4.0962326503635165,
|
| 1647 |
+
"eval_loss": 0.9498882293701172,
|
| 1648 |
+
"eval_mean_token_accuracy": 0.8178022539136778,
|
| 1649 |
+
"eval_num_tokens": 253853696.0,
|
| 1650 |
+
"eval_runtime": 1713.64,
|
| 1651 |
+
"eval_samples_per_second": 4.415,
|
| 1652 |
+
"eval_steps_per_second": 0.552,
|
| 1653 |
+
"step": 7750
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 4.122670191672174,
|
| 1657 |
+
"grad_norm": 0.40135377645492554,
|
| 1658 |
+
"learning_rate": 9.044815178279358e-05,
|
| 1659 |
+
"loss": 0.7963,
|
| 1660 |
+
"mean_token_accuracy": 0.8303173841536045,
|
| 1661 |
+
"num_tokens": 255492096.0,
|
| 1662 |
+
"step": 7800
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 4.149107732980832,
|
| 1666 |
+
"grad_norm": 0.39157313108444214,
|
| 1667 |
+
"learning_rate": 8.772216770254061e-05,
|
| 1668 |
+
"loss": 0.7846,
|
| 1669 |
+
"mean_token_accuracy": 0.8317717489600182,
|
| 1670 |
+
"num_tokens": 257130496.0,
|
| 1671 |
+
"step": 7850
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 4.175545274289491,
|
| 1675 |
+
"grad_norm": 0.4963982105255127,
|
| 1676 |
+
"learning_rate": 8.499618362228765e-05,
|
| 1677 |
+
"loss": 0.8069,
|
| 1678 |
+
"mean_token_accuracy": 0.8283883157372475,
|
| 1679 |
+
"num_tokens": 258768896.0,
|
| 1680 |
+
"step": 7900
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 4.201982815598149,
|
| 1684 |
+
"grad_norm": 0.3942487835884094,
|
| 1685 |
+
"learning_rate": 8.227019954203467e-05,
|
| 1686 |
+
"loss": 0.803,
|
| 1687 |
+
"mean_token_accuracy": 0.8284738489985466,
|
| 1688 |
+
"num_tokens": 260407296.0,
|
| 1689 |
+
"step": 7950
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 4.228420356906808,
|
| 1693 |
+
"grad_norm": 0.37145310640335083,
|
| 1694 |
+
"learning_rate": 7.95442154617817e-05,
|
| 1695 |
+
"loss": 0.805,
|
| 1696 |
+
"step": 8000
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 4.228420356906808,
|
| 1700 |
+
"eval_loss": 0.9421485066413879,
|
| 1701 |
+
"eval_mean_token_accuracy": 0.8191204303540841,
|
| 1702 |
+
"eval_num_tokens": 262045696.0,
|
| 1703 |
+
"eval_runtime": 1716.291,
|
| 1704 |
+
"eval_samples_per_second": 4.408,
|
| 1705 |
+
"eval_steps_per_second": 0.551,
|
| 1706 |
+
"step": 8000
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 4.254857898215466,
|
| 1710 |
+
"grad_norm": 0.3017653524875641,
|
| 1711 |
+
"learning_rate": 7.681823138152873e-05,
|
| 1712 |
+
"loss": 0.796,
|
| 1713 |
+
"mean_token_accuracy": 0.8288511091470718,
|
| 1714 |
+
"num_tokens": 263684096.0,
|
| 1715 |
+
"step": 8050
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 4.281295439524124,
|
| 1719 |
+
"grad_norm": 0.4065409004688263,
|
| 1720 |
+
"learning_rate": 7.409224730127575e-05,
|
| 1721 |
+
"loss": 0.7812,
|
| 1722 |
+
"mean_token_accuracy": 0.8326673975586891,
|
| 1723 |
+
"num_tokens": 265322496.0,
|
| 1724 |
+
"step": 8100
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 4.307732980832783,
|
| 1728 |
+
"grad_norm": 0.33838245272636414,
|
| 1729 |
+
"learning_rate": 7.136626322102279e-05,
|
| 1730 |
+
"loss": 0.7868,
|
| 1731 |
+
"mean_token_accuracy": 0.8319782489538192,
|
| 1732 |
+
"num_tokens": 266960896.0,
|
| 1733 |
+
"step": 8150
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 4.334170522141441,
|
| 1737 |
+
"grad_norm": 0.39351001381874084,
|
| 1738 |
+
"learning_rate": 6.864027914076983e-05,
|
| 1739 |
+
"loss": 0.7891,
|
| 1740 |
+
"mean_token_accuracy": 0.8313220903277397,
|
| 1741 |
+
"num_tokens": 268599296.0,
|
| 1742 |
+
"step": 8200
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 4.360608063450099,
|
| 1746 |
+
"grad_norm": 0.3555977940559387,
|
| 1747 |
+
"learning_rate": 6.591429506051685e-05,
|
| 1748 |
+
"loss": 0.7858,
|
| 1749 |
+
"step": 8250
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 4.360608063450099,
|
| 1753 |
+
"eval_loss": 0.9359485507011414,
|
| 1754 |
+
"eval_mean_token_accuracy": 0.8200258982483983,
|
| 1755 |
+
"eval_num_tokens": 270237696.0,
|
| 1756 |
+
"eval_runtime": 1713.1543,
|
| 1757 |
+
"eval_samples_per_second": 4.416,
|
| 1758 |
+
"eval_steps_per_second": 0.552,
|
| 1759 |
+
"step": 8250
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 4.387045604758757,
|
| 1763 |
+
"grad_norm": 0.34437137842178345,
|
| 1764 |
+
"learning_rate": 6.318831098026388e-05,
|
| 1765 |
+
"loss": 0.7825,
|
| 1766 |
+
"mean_token_accuracy": 0.8322449275851249,
|
| 1767 |
+
"num_tokens": 271876096.0,
|
| 1768 |
+
"step": 8300
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"epoch": 4.413483146067415,
|
| 1772 |
+
"grad_norm": 0.37046581506729126,
|
| 1773 |
+
"learning_rate": 6.0462326900010904e-05,
|
| 1774 |
+
"loss": 0.7936,
|
| 1775 |
+
"mean_token_accuracy": 0.8298429843783378,
|
| 1776 |
+
"num_tokens": 273514496.0,
|
| 1777 |
+
"step": 8350
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 4.439920687376074,
|
| 1781 |
+
"grad_norm": 0.44303834438323975,
|
| 1782 |
+
"learning_rate": 5.773634281975793e-05,
|
| 1783 |
+
"loss": 0.7891,
|
| 1784 |
+
"mean_token_accuracy": 0.8311491903662681,
|
| 1785 |
+
"num_tokens": 275152896.0,
|
| 1786 |
+
"step": 8400
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 4.466358228684732,
|
| 1790 |
+
"grad_norm": 0.382201611995697,
|
| 1791 |
+
"learning_rate": 5.5010358739504963e-05,
|
| 1792 |
+
"loss": 0.7781,
|
| 1793 |
+
"mean_token_accuracy": 0.8330895602703094,
|
| 1794 |
+
"num_tokens": 276791296.0,
|
| 1795 |
+
"step": 8450
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 4.492795769993391,
|
| 1799 |
+
"grad_norm": 0.39989522099494934,
|
| 1800 |
+
"learning_rate": 5.228437465925199e-05,
|
| 1801 |
+
"loss": 0.7877,
|
| 1802 |
+
"step": 8500
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 4.492795769993391,
|
| 1806 |
+
"eval_loss": 0.9300816059112549,
|
| 1807 |
+
"eval_mean_token_accuracy": 0.8209756191890789,
|
| 1808 |
+
"eval_num_tokens": 278429696.0,
|
| 1809 |
+
"eval_runtime": 1713.4373,
|
| 1810 |
+
"eval_samples_per_second": 4.415,
|
| 1811 |
+
"eval_steps_per_second": 0.552,
|
| 1812 |
+
"step": 8500
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 4.519233311302049,
|
| 1816 |
+
"grad_norm": 0.3832317590713501,
|
| 1817 |
+
"learning_rate": 4.9558390578999016e-05,
|
| 1818 |
+
"loss": 0.7782,
|
| 1819 |
+
"mean_token_accuracy": 0.8327266594767571,
|
| 1820 |
+
"num_tokens": 280068096.0,
|
| 1821 |
+
"step": 8550
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 4.545670852610707,
|
| 1825 |
+
"grad_norm": 0.4667583703994751,
|
| 1826 |
+
"learning_rate": 4.683240649874604e-05,
|
| 1827 |
+
"loss": 0.7849,
|
| 1828 |
+
"mean_token_accuracy": 0.8322770014405251,
|
| 1829 |
+
"num_tokens": 281706496.0,
|
| 1830 |
+
"step": 8600
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 4.572108393919366,
|
| 1834 |
+
"grad_norm": 0.3977579176425934,
|
| 1835 |
+
"learning_rate": 4.4106422418493076e-05,
|
| 1836 |
+
"loss": 0.786,
|
| 1837 |
+
"mean_token_accuracy": 0.83141067892313,
|
| 1838 |
+
"num_tokens": 283344896.0,
|
| 1839 |
+
"step": 8650
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 4.598545935228024,
|
| 1843 |
+
"grad_norm": 0.38620129227638245,
|
| 1844 |
+
"learning_rate": 4.138043833824011e-05,
|
| 1845 |
+
"loss": 0.7865,
|
| 1846 |
+
"mean_token_accuracy": 0.8317729702591896,
|
| 1847 |
+
"num_tokens": 284983296.0,
|
| 1848 |
+
"step": 8700
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 4.624983476536682,
|
| 1852 |
+
"grad_norm": 0.3608716130256653,
|
| 1853 |
+
"learning_rate": 3.8654454257987135e-05,
|
| 1854 |
+
"loss": 0.7905,
|
| 1855 |
+
"step": 8750
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 4.624983476536682,
|
| 1859 |
+
"eval_loss": 0.9239566922187805,
|
| 1860 |
+
"eval_mean_token_accuracy": 0.8218199411607948,
|
| 1861 |
+
"eval_num_tokens": 286621696.0,
|
| 1862 |
+
"eval_runtime": 1715.401,
|
| 1863 |
+
"eval_samples_per_second": 4.41,
|
| 1864 |
+
"eval_steps_per_second": 0.551,
|
| 1865 |
+
"step": 8750
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 4.65142101784534,
|
| 1869 |
+
"grad_norm": 0.30198875069618225,
|
| 1870 |
+
"learning_rate": 3.592847017773417e-05,
|
| 1871 |
+
"loss": 0.7672,
|
| 1872 |
+
"mean_token_accuracy": 0.8327672865986824,
|
| 1873 |
+
"num_tokens": 288260096.0,
|
| 1874 |
+
"step": 8800
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 4.677858559153998,
|
| 1878 |
+
"grad_norm": 0.3867688477039337,
|
| 1879 |
+
"learning_rate": 3.3202486097481194e-05,
|
| 1880 |
+
"loss": 0.7728,
|
| 1881 |
+
"mean_token_accuracy": 0.8339509972929955,
|
| 1882 |
+
"num_tokens": 289898496.0,
|
| 1883 |
+
"step": 8850
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 4.704296100462657,
|
| 1887 |
+
"grad_norm": 0.38331055641174316,
|
| 1888 |
+
"learning_rate": 3.0476502017228217e-05,
|
| 1889 |
+
"loss": 0.7801,
|
| 1890 |
+
"mean_token_accuracy": 0.8332563516497612,
|
| 1891 |
+
"num_tokens": 291536896.0,
|
| 1892 |
+
"step": 8900
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 4.730733641771315,
|
| 1896 |
+
"grad_norm": 0.32032325863838196,
|
| 1897 |
+
"learning_rate": 2.775051793697525e-05,
|
| 1898 |
+
"loss": 0.7896,
|
| 1899 |
+
"mean_token_accuracy": 0.8309176415205002,
|
| 1900 |
+
"num_tokens": 293175296.0,
|
| 1901 |
+
"step": 8950
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 4.757171183079974,
|
| 1905 |
+
"grad_norm": 0.41111549735069275,
|
| 1906 |
+
"learning_rate": 2.502453385672228e-05,
|
| 1907 |
+
"loss": 0.7691,
|
| 1908 |
+
"step": 9000
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 4.757171183079974,
|
| 1912 |
+
"eval_loss": 0.9195617437362671,
|
| 1913 |
+
"eval_mean_token_accuracy": 0.8225174557583025,
|
| 1914 |
+
"eval_num_tokens": 294813696.0,
|
| 1915 |
+
"eval_runtime": 1712.3817,
|
| 1916 |
+
"eval_samples_per_second": 4.418,
|
| 1917 |
+
"eval_steps_per_second": 0.552,
|
| 1918 |
+
"step": 9000
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 4.783608724388632,
|
| 1922 |
+
"grad_norm": 0.32703885436058044,
|
| 1923 |
+
"learning_rate": 2.2298549776469306e-05,
|
| 1924 |
+
"loss": 0.7527,
|
| 1925 |
+
"mean_token_accuracy": 0.8362371681630612,
|
| 1926 |
+
"num_tokens": 296452096.0,
|
| 1927 |
+
"step": 9050
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 4.81004626569729,
|
| 1931 |
+
"grad_norm": 0.915315568447113,
|
| 1932 |
+
"learning_rate": 1.9572565696216336e-05,
|
| 1933 |
+
"loss": 0.7746,
|
| 1934 |
+
"mean_token_accuracy": 0.8344965758919716,
|
| 1935 |
+
"num_tokens": 298090496.0,
|
| 1936 |
+
"step": 9100
|
| 1937 |
+
},
|
| 1938 |
+
{
|
| 1939 |
+
"epoch": 4.836483807005949,
|
| 1940 |
+
"grad_norm": 0.3820320665836334,
|
| 1941 |
+
"learning_rate": 1.6846581615963362e-05,
|
| 1942 |
+
"loss": 0.7714,
|
| 1943 |
+
"mean_token_accuracy": 0.8342699158191681,
|
| 1944 |
+
"num_tokens": 299728896.0,
|
| 1945 |
+
"step": 9150
|
| 1946 |
+
},
|
| 1947 |
+
{
|
| 1948 |
+
"epoch": 4.8629213483146065,
|
| 1949 |
+
"grad_norm": 0.36762210726737976,
|
| 1950 |
+
"learning_rate": 1.4120597535710392e-05,
|
| 1951 |
+
"loss": 0.7768,
|
| 1952 |
+
"mean_token_accuracy": 0.8333681541681289,
|
| 1953 |
+
"num_tokens": 301367296.0,
|
| 1954 |
+
"step": 9200
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 4.889358889623265,
|
| 1958 |
+
"grad_norm": 0.5305154919624329,
|
| 1959 |
+
"learning_rate": 1.139461345545742e-05,
|
| 1960 |
+
"loss": 0.7682,
|
| 1961 |
+
"step": 9250
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 4.889358889623265,
|
| 1965 |
+
"eval_loss": 0.9165179133415222,
|
| 1966 |
+
"eval_mean_token_accuracy": 0.8231746450413105,
|
| 1967 |
+
"eval_num_tokens": 303005696.0,
|
| 1968 |
+
"eval_runtime": 1714.4129,
|
| 1969 |
+
"eval_samples_per_second": 4.413,
|
| 1970 |
+
"eval_steps_per_second": 0.552,
|
| 1971 |
+
"step": 9250
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 4.915796430931923,
|
| 1975 |
+
"grad_norm": 0.37008801102638245,
|
| 1976 |
+
"learning_rate": 8.668629375204448e-06,
|
| 1977 |
+
"loss": 0.7568,
|
| 1978 |
+
"mean_token_accuracy": 0.8357875108718872,
|
| 1979 |
+
"num_tokens": 304644096.0,
|
| 1980 |
+
"step": 9300
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 4.942233972240581,
|
| 1984 |
+
"grad_norm": 0.3372200131416321,
|
| 1985 |
+
"learning_rate": 5.942645294951477e-06,
|
| 1986 |
+
"loss": 0.7791,
|
| 1987 |
+
"mean_token_accuracy": 0.8329618725180626,
|
| 1988 |
+
"num_tokens": 306282496.0,
|
| 1989 |
+
"step": 9350
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 4.96867151354924,
|
| 1993 |
+
"grad_norm": 0.36545583605766296,
|
| 1994 |
+
"learning_rate": 3.2166612146985063e-06,
|
| 1995 |
+
"loss": 0.7812,
|
| 1996 |
+
"mean_token_accuracy": 0.8330565708875656,
|
| 1997 |
+
"num_tokens": 307920896.0,
|
| 1998 |
+
"step": 9400
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 4.995109054857898,
|
| 2002 |
+
"grad_norm": 0.2767314910888672,
|
| 2003 |
+
"learning_rate": 4.906771344455349e-07,
|
| 2004 |
+
"loss": 0.7773,
|
| 2005 |
+
"mean_token_accuracy": 0.8339308369159698,
|
| 2006 |
+
"num_tokens": 309559296.0,
|
| 2007 |
+
"step": 9450
|
| 2008 |
+
}
|
| 2009 |
+
],
|
| 2010 |
+
"logging_steps": 50,
|
| 2011 |
+
"max_steps": 9455,
|
| 2012 |
+
"num_input_tokens_seen": 0,
|
| 2013 |
+
"num_train_epochs": 5,
|
| 2014 |
+
"save_steps": 500,
|
| 2015 |
+
"stateful_callbacks": {
|
| 2016 |
+
"TrainerControl": {
|
| 2017 |
+
"args": {
|
| 2018 |
+
"should_epoch_stop": false,
|
| 2019 |
+
"should_evaluate": false,
|
| 2020 |
+
"should_log": false,
|
| 2021 |
+
"should_save": true,
|
| 2022 |
+
"should_training_stop": true
|
| 2023 |
+
},
|
| 2024 |
+
"attributes": {}
|
| 2025 |
+
}
|
| 2026 |
+
},
|
| 2027 |
+
"total_flos": 494751040143360.0,
|
| 2028 |
+
"train_batch_size": 1,
|
| 2029 |
+
"trial_name": null,
|
| 2030 |
+
"trial_params": null
|
| 2031 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8527608b21d92bfcb521038a595e7b03ec2677df219c7daabc21faa1c4638f48
|
| 3 |
+
size 8120
|