Intermediate checkpoint upload step=90 (answerer_train)
Browse files- .gitattributes +1 -0
- self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/chat_template.jinja +54 -0
- self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/config.json +57 -0
- self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/generation_config.json +13 -0
- self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/model.safetensors +3 -0
- self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/optimizer.pt +3 -0
- self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/rng_state.pth +3 -0
- self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/scheduler.pt +3 -0
- self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/tokenizer.json +3 -0
- self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/tokenizer_config.json +32 -0
- self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/trainer_state.json +1683 -0
- self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/training_args.bin +3 -0
.gitattributes
CHANGED
|
@@ -49,3 +49,4 @@ self_play_hf_l40s_full/round_001/generator_train/checkpoint-120/tokenizer.json f
|
|
| 49 |
self_play_hf_l40s_full/round_001/generator_train/final_model/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 50 |
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-30/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 51 |
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-60/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 49 |
self_play_hf_l40s_full/round_001/generator_train/final_model/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 50 |
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-30/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 51 |
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-60/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 52 |
+
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/chat_template.jinja
ADDED
|
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{%- if tools %}
|
| 2 |
+
{{- '<|im_start|>system\n' }}
|
| 3 |
+
{%- if messages[0]['role'] == 'system' %}
|
| 4 |
+
{{- messages[0]['content'] }}
|
| 5 |
+
{%- else %}
|
| 6 |
+
{{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
|
| 7 |
+
{%- endif %}
|
| 8 |
+
{{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
|
| 9 |
+
{%- for tool in tools %}
|
| 10 |
+
{{- "\n" }}
|
| 11 |
+
{{- tool | tojson }}
|
| 12 |
+
{%- endfor %}
|
| 13 |
+
{{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
|
| 14 |
+
{%- else %}
|
| 15 |
+
{%- if messages[0]['role'] == 'system' %}
|
| 16 |
+
{{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
|
| 17 |
+
{%- else %}
|
| 18 |
+
{{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
|
| 19 |
+
{%- endif %}
|
| 20 |
+
{%- endif %}
|
| 21 |
+
{%- for message in messages %}
|
| 22 |
+
{%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
|
| 23 |
+
{{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
|
| 24 |
+
{%- elif message.role == "assistant" %}
|
| 25 |
+
{{- '<|im_start|>' + message.role }}
|
| 26 |
+
{%- if message.content %}
|
| 27 |
+
{{- '\n' + message.content }}
|
| 28 |
+
{%- endif %}
|
| 29 |
+
{%- for tool_call in message.tool_calls %}
|
| 30 |
+
{%- if tool_call.function is defined %}
|
| 31 |
+
{%- set tool_call = tool_call.function %}
|
| 32 |
+
{%- endif %}
|
| 33 |
+
{{- '\n<tool_call>\n{"name": "' }}
|
| 34 |
+
{{- tool_call.name }}
|
| 35 |
+
{{- '", "arguments": ' }}
|
| 36 |
+
{{- tool_call.arguments | tojson }}
|
| 37 |
+
{{- '}\n</tool_call>' }}
|
| 38 |
+
{%- endfor %}
|
| 39 |
+
{{- '<|im_end|>\n' }}
|
| 40 |
+
{%- elif message.role == "tool" %}
|
| 41 |
+
{%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
|
| 42 |
+
{{- '<|im_start|>user' }}
|
| 43 |
+
{%- endif %}
|
| 44 |
+
{{- '\n<tool_response>\n' }}
|
| 45 |
+
{{- message.content }}
|
| 46 |
+
{{- '\n</tool_response>' }}
|
| 47 |
+
{%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
|
| 48 |
+
{{- '<|im_end|>\n' }}
|
| 49 |
+
{%- endif %}
|
| 50 |
+
{%- endif %}
|
| 51 |
+
{%- endfor %}
|
| 52 |
+
{%- if add_generation_prompt %}
|
| 53 |
+
{{- '<|im_start|>assistant\n' }}
|
| 54 |
+
{%- endif %}
|
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"Qwen2ForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_dropout": 0.0,
|
| 6 |
+
"bos_token_id": null,
|
| 7 |
+
"dtype": "float32",
|
| 8 |
+
"eos_token_id": 151645,
|
| 9 |
+
"hidden_act": "silu",
|
| 10 |
+
"hidden_size": 896,
|
| 11 |
+
"initializer_range": 0.02,
|
| 12 |
+
"intermediate_size": 4864,
|
| 13 |
+
"layer_types": [
|
| 14 |
+
"full_attention",
|
| 15 |
+
"full_attention",
|
| 16 |
+
"full_attention",
|
| 17 |
+
"full_attention",
|
| 18 |
+
"full_attention",
|
| 19 |
+
"full_attention",
|
| 20 |
+
"full_attention",
|
| 21 |
+
"full_attention",
|
| 22 |
+
"full_attention",
|
| 23 |
+
"full_attention",
|
| 24 |
+
"full_attention",
|
| 25 |
+
"full_attention",
|
| 26 |
+
"full_attention",
|
| 27 |
+
"full_attention",
|
| 28 |
+
"full_attention",
|
| 29 |
+
"full_attention",
|
| 30 |
+
"full_attention",
|
| 31 |
+
"full_attention",
|
| 32 |
+
"full_attention",
|
| 33 |
+
"full_attention",
|
| 34 |
+
"full_attention",
|
| 35 |
+
"full_attention",
|
| 36 |
+
"full_attention",
|
| 37 |
+
"full_attention"
|
| 38 |
+
],
|
| 39 |
+
"max_position_embeddings": 32768,
|
| 40 |
+
"max_window_layers": 21,
|
| 41 |
+
"model_type": "qwen2",
|
| 42 |
+
"num_attention_heads": 14,
|
| 43 |
+
"num_hidden_layers": 24,
|
| 44 |
+
"num_key_value_heads": 2,
|
| 45 |
+
"pad_token_id": 151643,
|
| 46 |
+
"rms_norm_eps": 1e-06,
|
| 47 |
+
"rope_parameters": {
|
| 48 |
+
"rope_theta": 1000000.0,
|
| 49 |
+
"rope_type": "default"
|
| 50 |
+
},
|
| 51 |
+
"sliding_window": null,
|
| 52 |
+
"tie_word_embeddings": true,
|
| 53 |
+
"transformers_version": "5.6.2",
|
| 54 |
+
"use_cache": false,
|
| 55 |
+
"use_sliding_window": false,
|
| 56 |
+
"vocab_size": 151936
|
| 57 |
+
}
|
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/generation_config.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_sample": true,
|
| 3 |
+
"eos_token_id": [
|
| 4 |
+
151645,
|
| 5 |
+
151643
|
| 6 |
+
],
|
| 7 |
+
"pad_token_id": 151643,
|
| 8 |
+
"repetition_penalty": 1.1,
|
| 9 |
+
"temperature": 0.7,
|
| 10 |
+
"top_k": 20,
|
| 11 |
+
"top_p": 0.8,
|
| 12 |
+
"transformers_version": "5.6.2"
|
| 13 |
+
}
|
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:23dc90aefe564e9c7604a8542a412d2f1fe65a6bede8c236a9eed2da8cd21b55
|
| 3 |
+
size 1976163472
|
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0633c45ef6207f18d620dfb0751694265900bc831960189d1786f9a4adb8f65d
|
| 3 |
+
size 3952509771
|
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:26f6240389ad2111a606d4aef5c6832d0136df7debe9f496847f25e20ecf886b
|
| 3 |
+
size 14645
|
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:716b80a1d03fa94e9f2454097632e924f71019b15d068a8837fcbbbce1910893
|
| 3 |
+
size 1465
|
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
|
| 3 |
+
size 11421892
|
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/tokenizer_config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": null,
|
| 5 |
+
"clean_up_tokenization_spaces": false,
|
| 6 |
+
"eos_token": "<|im_end|>",
|
| 7 |
+
"errors": "replace",
|
| 8 |
+
"extra_special_tokens": [
|
| 9 |
+
"<|im_start|>",
|
| 10 |
+
"<|im_end|>",
|
| 11 |
+
"<|object_ref_start|>",
|
| 12 |
+
"<|object_ref_end|>",
|
| 13 |
+
"<|box_start|>",
|
| 14 |
+
"<|box_end|>",
|
| 15 |
+
"<|quad_start|>",
|
| 16 |
+
"<|quad_end|>",
|
| 17 |
+
"<|vision_start|>",
|
| 18 |
+
"<|vision_end|>",
|
| 19 |
+
"<|vision_pad|>",
|
| 20 |
+
"<|image_pad|>",
|
| 21 |
+
"<|video_pad|>"
|
| 22 |
+
],
|
| 23 |
+
"is_local": true,
|
| 24 |
+
"local_files_only": false,
|
| 25 |
+
"model_max_length": 131072,
|
| 26 |
+
"pad_token": "<|endoftext|>",
|
| 27 |
+
"padding_side": "left",
|
| 28 |
+
"split_special_tokens": false,
|
| 29 |
+
"tokenizer_class": "Qwen2Tokenizer",
|
| 30 |
+
"truncation_side": "left",
|
| 31 |
+
"unk_token": null
|
| 32 |
+
}
|
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/trainer_state.json
ADDED
|
@@ -0,0 +1,1683 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.25,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 90,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"clip_ratio/high_max": 0.0,
|
| 14 |
+
"clip_ratio/high_mean": 0.0,
|
| 15 |
+
"clip_ratio/low_mean": 0.0,
|
| 16 |
+
"clip_ratio/low_min": 0.0,
|
| 17 |
+
"clip_ratio/region_mean": 0.0,
|
| 18 |
+
"completions/clipped_ratio": 1.0,
|
| 19 |
+
"completions/max_length": 256.0,
|
| 20 |
+
"completions/max_terminated_length": 0.0,
|
| 21 |
+
"completions/mean_length": 256.0,
|
| 22 |
+
"completions/mean_terminated_length": 0.0,
|
| 23 |
+
"completions/min_length": 256.0,
|
| 24 |
+
"completions/min_terminated_length": 0.0,
|
| 25 |
+
"entropy": 1.0711654424667358,
|
| 26 |
+
"epoch": 0.025,
|
| 27 |
+
"frac_reward_zero_std": 0.0,
|
| 28 |
+
"grad_norm": 2.947098731994629,
|
| 29 |
+
"kl": 0.0,
|
| 30 |
+
"learning_rate": 3e-06,
|
| 31 |
+
"loss": 0.06321408599615097,
|
| 32 |
+
"num_tokens": 9896.0,
|
| 33 |
+
"reward": 0.5316684246063232,
|
| 34 |
+
"reward_std": 0.11483359336853027,
|
| 35 |
+
"rewards/AnswererRewardFunction/mean": 0.5316684246063232,
|
| 36 |
+
"rewards/AnswererRewardFunction/std": 0.11483359336853027,
|
| 37 |
+
"step": 1,
|
| 38 |
+
"step_time": 8.027100405000056
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"clip_ratio/high_max": 0.033203125,
|
| 42 |
+
"clip_ratio/high_mean": 0.033203125,
|
| 43 |
+
"clip_ratio/low_mean": 0.0224609375,
|
| 44 |
+
"clip_ratio/low_min": 0.0224609375,
|
| 45 |
+
"clip_ratio/region_mean": 0.0556640625,
|
| 46 |
+
"entropy": 0.8880369663238525,
|
| 47 |
+
"epoch": 0.05,
|
| 48 |
+
"grad_norm": 2.81626296043396,
|
| 49 |
+
"kl": 0.04245668277144432,
|
| 50 |
+
"learning_rate": 2.9750000000000003e-06,
|
| 51 |
+
"loss": -0.07583755254745483,
|
| 52 |
+
"step": 2,
|
| 53 |
+
"step_time": 0.12222773600001346
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"clip_ratio/high_max": 0.0166015625,
|
| 57 |
+
"clip_ratio/high_mean": 0.0166015625,
|
| 58 |
+
"clip_ratio/low_mean": 0.05078125,
|
| 59 |
+
"clip_ratio/low_min": 0.05078125,
|
| 60 |
+
"clip_ratio/region_mean": 0.0673828125,
|
| 61 |
+
"entropy": 1.0095021724700928,
|
| 62 |
+
"epoch": 0.075,
|
| 63 |
+
"grad_norm": 5.157853603363037,
|
| 64 |
+
"kl": 0.016499709337949753,
|
| 65 |
+
"learning_rate": 2.9499999999999997e-06,
|
| 66 |
+
"loss": 0.18797507882118225,
|
| 67 |
+
"step": 3,
|
| 68 |
+
"step_time": 0.11191487599990069
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"clip_ratio/high_max": 0.0546875,
|
| 72 |
+
"clip_ratio/high_mean": 0.0546875,
|
| 73 |
+
"clip_ratio/low_mean": 0.025390625,
|
| 74 |
+
"clip_ratio/low_min": 0.025390625,
|
| 75 |
+
"clip_ratio/region_mean": 0.080078125,
|
| 76 |
+
"entropy": 0.8205338716506958,
|
| 77 |
+
"epoch": 0.1,
|
| 78 |
+
"grad_norm": 4.075566291809082,
|
| 79 |
+
"kl": 0.05314125120639801,
|
| 80 |
+
"learning_rate": 2.925e-06,
|
| 81 |
+
"loss": -0.16264761984348297,
|
| 82 |
+
"step": 4,
|
| 83 |
+
"step_time": 0.11004377599988402
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"clip_ratio/high_max": 0.0009765625,
|
| 87 |
+
"clip_ratio/high_mean": 0.0009765625,
|
| 88 |
+
"clip_ratio/low_mean": 0.0,
|
| 89 |
+
"clip_ratio/low_min": 0.0,
|
| 90 |
+
"clip_ratio/region_mean": 0.0009765625,
|
| 91 |
+
"completions/clipped_ratio": 1.0,
|
| 92 |
+
"completions/max_length": 256.0,
|
| 93 |
+
"completions/max_terminated_length": 0.0,
|
| 94 |
+
"completions/mean_length": 256.0,
|
| 95 |
+
"completions/mean_terminated_length": 0.0,
|
| 96 |
+
"completions/min_length": 256.0,
|
| 97 |
+
"completions/min_terminated_length": 0.0,
|
| 98 |
+
"entropy": 0.751447856426239,
|
| 99 |
+
"epoch": 0.125,
|
| 100 |
+
"frac_reward_zero_std": 0.0,
|
| 101 |
+
"grad_norm": 3.1406376361846924,
|
| 102 |
+
"kl": 0.021290192380547523,
|
| 103 |
+
"learning_rate": 2.9e-06,
|
| 104 |
+
"loss": -0.0005250873509794474,
|
| 105 |
+
"num_tokens": 19632.0,
|
| 106 |
+
"reward": 0.543575644493103,
|
| 107 |
+
"reward_std": 0.11543858796358109,
|
| 108 |
+
"rewards/AnswererRewardFunction/mean": 0.543575644493103,
|
| 109 |
+
"rewards/AnswererRewardFunction/std": 0.11543858796358109,
|
| 110 |
+
"step": 5,
|
| 111 |
+
"step_time": 8.046513188000063
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"clip_ratio/high_max": 0.00390625,
|
| 115 |
+
"clip_ratio/high_mean": 0.00390625,
|
| 116 |
+
"clip_ratio/low_mean": 0.0078125,
|
| 117 |
+
"clip_ratio/low_min": 0.0078125,
|
| 118 |
+
"clip_ratio/region_mean": 0.01171875,
|
| 119 |
+
"entropy": 0.6940755248069763,
|
| 120 |
+
"epoch": 0.15,
|
| 121 |
+
"grad_norm": 3.9180352687835693,
|
| 122 |
+
"kl": 0.02370832860469818,
|
| 123 |
+
"learning_rate": 2.875e-06,
|
| 124 |
+
"loss": 0.061240844428539276,
|
| 125 |
+
"step": 6,
|
| 126 |
+
"step_time": 0.1118731530000332
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"clip_ratio/high_max": 0.0087890625,
|
| 130 |
+
"clip_ratio/high_mean": 0.0087890625,
|
| 131 |
+
"clip_ratio/low_mean": 0.0234375,
|
| 132 |
+
"clip_ratio/low_min": 0.0234375,
|
| 133 |
+
"clip_ratio/region_mean": 0.0322265625,
|
| 134 |
+
"entropy": 0.8035241365432739,
|
| 135 |
+
"epoch": 0.175,
|
| 136 |
+
"grad_norm": 3.497161865234375,
|
| 137 |
+
"kl": 0.03351783752441406,
|
| 138 |
+
"learning_rate": 2.85e-06,
|
| 139 |
+
"loss": -0.004166277591139078,
|
| 140 |
+
"step": 7,
|
| 141 |
+
"step_time": 0.1107469599999149
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"clip_ratio/high_max": 0.0263671875,
|
| 145 |
+
"clip_ratio/high_mean": 0.0263671875,
|
| 146 |
+
"clip_ratio/low_mean": 0.0068359375,
|
| 147 |
+
"clip_ratio/low_min": 0.0068359375,
|
| 148 |
+
"clip_ratio/region_mean": 0.033203125,
|
| 149 |
+
"entropy": 0.5180478096008301,
|
| 150 |
+
"epoch": 0.2,
|
| 151 |
+
"grad_norm": 1.883070945739746,
|
| 152 |
+
"kl": 0.04320289194583893,
|
| 153 |
+
"learning_rate": 2.825e-06,
|
| 154 |
+
"loss": -0.053033724427223206,
|
| 155 |
+
"step": 8,
|
| 156 |
+
"step_time": 0.11128661100019599
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"clip_ratio/high_max": 0.0009765625,
|
| 160 |
+
"clip_ratio/high_mean": 0.0009765625,
|
| 161 |
+
"clip_ratio/low_mean": 0.0,
|
| 162 |
+
"clip_ratio/low_min": 0.0,
|
| 163 |
+
"clip_ratio/region_mean": 0.0009765625,
|
| 164 |
+
"completions/clipped_ratio": 1.0,
|
| 165 |
+
"completions/max_length": 256.0,
|
| 166 |
+
"completions/max_terminated_length": 0.0,
|
| 167 |
+
"completions/mean_length": 256.0,
|
| 168 |
+
"completions/mean_terminated_length": 0.0,
|
| 169 |
+
"completions/min_length": 256.0,
|
| 170 |
+
"completions/min_terminated_length": 0.0,
|
| 171 |
+
"entropy": 0.6499772071838379,
|
| 172 |
+
"epoch": 0.225,
|
| 173 |
+
"frac_reward_zero_std": 0.0,
|
| 174 |
+
"grad_norm": 3.706423759460449,
|
| 175 |
+
"kl": 0.041926056146621704,
|
| 176 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 177 |
+
"loss": -0.12885785102844238,
|
| 178 |
+
"num_tokens": 29444.0,
|
| 179 |
+
"reward": 0.5444201231002808,
|
| 180 |
+
"reward_std": 0.1789802759885788,
|
| 181 |
+
"rewards/AnswererRewardFunction/mean": 0.5444201231002808,
|
| 182 |
+
"rewards/AnswererRewardFunction/std": 0.1789802610874176,
|
| 183 |
+
"step": 9,
|
| 184 |
+
"step_time": 8.014759111000103
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"clip_ratio/high_max": 0.009765625,
|
| 188 |
+
"clip_ratio/high_mean": 0.009765625,
|
| 189 |
+
"clip_ratio/low_mean": 0.0048828125,
|
| 190 |
+
"clip_ratio/low_min": 0.0048828125,
|
| 191 |
+
"clip_ratio/region_mean": 0.0146484375,
|
| 192 |
+
"entropy": 0.743914008140564,
|
| 193 |
+
"epoch": 0.25,
|
| 194 |
+
"grad_norm": 2.893528699874878,
|
| 195 |
+
"kl": 0.08459636569023132,
|
| 196 |
+
"learning_rate": 2.775e-06,
|
| 197 |
+
"loss": -0.013386903330683708,
|
| 198 |
+
"step": 10,
|
| 199 |
+
"step_time": 0.11163025199994081
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"clip_ratio/high_max": 0.0029296875,
|
| 203 |
+
"clip_ratio/high_mean": 0.0029296875,
|
| 204 |
+
"clip_ratio/low_mean": 0.021484375,
|
| 205 |
+
"clip_ratio/low_min": 0.021484375,
|
| 206 |
+
"clip_ratio/region_mean": 0.0244140625,
|
| 207 |
+
"entropy": 0.697109580039978,
|
| 208 |
+
"epoch": 0.275,
|
| 209 |
+
"grad_norm": 2.882779359817505,
|
| 210 |
+
"kl": 0.04983264207839966,
|
| 211 |
+
"learning_rate": 2.75e-06,
|
| 212 |
+
"loss": 0.006847226992249489,
|
| 213 |
+
"step": 11,
|
| 214 |
+
"step_time": 0.11213077200000043
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"clip_ratio/high_max": 0.0,
|
| 218 |
+
"clip_ratio/high_mean": 0.0,
|
| 219 |
+
"clip_ratio/low_mean": 0.041015625,
|
| 220 |
+
"clip_ratio/low_min": 0.041015625,
|
| 221 |
+
"clip_ratio/region_mean": 0.041015625,
|
| 222 |
+
"entropy": 1.1653341054916382,
|
| 223 |
+
"epoch": 0.3,
|
| 224 |
+
"grad_norm": 3.08870267868042,
|
| 225 |
+
"kl": 0.08870606869459152,
|
| 226 |
+
"learning_rate": 2.725e-06,
|
| 227 |
+
"loss": 0.13892900943756104,
|
| 228 |
+
"step": 12,
|
| 229 |
+
"step_time": 0.11271204500008025
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"clip_ratio/high_max": 0.0009765625,
|
| 233 |
+
"clip_ratio/high_mean": 0.0009765625,
|
| 234 |
+
"clip_ratio/low_mean": 0.0,
|
| 235 |
+
"clip_ratio/low_min": 0.0,
|
| 236 |
+
"clip_ratio/region_mean": 0.0009765625,
|
| 237 |
+
"completions/clipped_ratio": 1.0,
|
| 238 |
+
"completions/max_length": 256.0,
|
| 239 |
+
"completions/max_terminated_length": 0.0,
|
| 240 |
+
"completions/mean_length": 256.0,
|
| 241 |
+
"completions/mean_terminated_length": 0.0,
|
| 242 |
+
"completions/min_length": 256.0,
|
| 243 |
+
"completions/min_terminated_length": 0.0,
|
| 244 |
+
"entropy": 0.63991379737854,
|
| 245 |
+
"epoch": 0.325,
|
| 246 |
+
"frac_reward_zero_std": 0.0,
|
| 247 |
+
"grad_norm": 3.3573482036590576,
|
| 248 |
+
"kl": 0.06412748992443085,
|
| 249 |
+
"learning_rate": 2.7e-06,
|
| 250 |
+
"loss": -0.16908901929855347,
|
| 251 |
+
"num_tokens": 39416.0,
|
| 252 |
+
"reward": 0.6549176573753357,
|
| 253 |
+
"reward_std": 0.09264716506004333,
|
| 254 |
+
"rewards/AnswererRewardFunction/mean": 0.6549176573753357,
|
| 255 |
+
"rewards/AnswererRewardFunction/std": 0.09264715760946274,
|
| 256 |
+
"step": 13,
|
| 257 |
+
"step_time": 7.928827267000088
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"clip_ratio/high_max": 0.001953125,
|
| 261 |
+
"clip_ratio/high_mean": 0.001953125,
|
| 262 |
+
"clip_ratio/low_mean": 0.001953125,
|
| 263 |
+
"clip_ratio/low_min": 0.001953125,
|
| 264 |
+
"clip_ratio/region_mean": 0.00390625,
|
| 265 |
+
"entropy": 0.3732947111129761,
|
| 266 |
+
"epoch": 0.35,
|
| 267 |
+
"grad_norm": 1.7340325117111206,
|
| 268 |
+
"kl": 0.043330118060112,
|
| 269 |
+
"learning_rate": 2.6750000000000002e-06,
|
| 270 |
+
"loss": 0.006798031274229288,
|
| 271 |
+
"step": 14,
|
| 272 |
+
"step_time": 0.11074772500001018
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"clip_ratio/high_max": 0.01171875,
|
| 276 |
+
"clip_ratio/high_mean": 0.01171875,
|
| 277 |
+
"clip_ratio/low_mean": 0.009765625,
|
| 278 |
+
"clip_ratio/low_min": 0.009765625,
|
| 279 |
+
"clip_ratio/region_mean": 0.021484375,
|
| 280 |
+
"entropy": 0.8068607449531555,
|
| 281 |
+
"epoch": 0.375,
|
| 282 |
+
"grad_norm": 3.2802722454071045,
|
| 283 |
+
"kl": 0.0741429328918457,
|
| 284 |
+
"learning_rate": 2.65e-06,
|
| 285 |
+
"loss": 0.05089528113603592,
|
| 286 |
+
"step": 15,
|
| 287 |
+
"step_time": 0.11073857499991391
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"clip_ratio/high_max": 0.015625,
|
| 291 |
+
"clip_ratio/high_mean": 0.015625,
|
| 292 |
+
"clip_ratio/low_mean": 0.0166015625,
|
| 293 |
+
"clip_ratio/low_min": 0.0166015625,
|
| 294 |
+
"clip_ratio/region_mean": 0.0322265625,
|
| 295 |
+
"entropy": 0.7192014455795288,
|
| 296 |
+
"epoch": 0.4,
|
| 297 |
+
"grad_norm": 4.314362049102783,
|
| 298 |
+
"kl": 0.07965575158596039,
|
| 299 |
+
"learning_rate": 2.6250000000000003e-06,
|
| 300 |
+
"loss": 0.11372166872024536,
|
| 301 |
+
"step": 16,
|
| 302 |
+
"step_time": 0.11177824599985797
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"clip_ratio/high_max": 0.0,
|
| 306 |
+
"clip_ratio/high_mean": 0.0,
|
| 307 |
+
"clip_ratio/low_mean": 0.0,
|
| 308 |
+
"clip_ratio/low_min": 0.0,
|
| 309 |
+
"clip_ratio/region_mean": 0.0,
|
| 310 |
+
"completions/clipped_ratio": 1.0,
|
| 311 |
+
"completions/max_length": 256.0,
|
| 312 |
+
"completions/max_terminated_length": 0.0,
|
| 313 |
+
"completions/mean_length": 256.0,
|
| 314 |
+
"completions/mean_terminated_length": 0.0,
|
| 315 |
+
"completions/min_length": 256.0,
|
| 316 |
+
"completions/min_terminated_length": 0.0,
|
| 317 |
+
"entropy": 0.6362192630767822,
|
| 318 |
+
"epoch": 0.425,
|
| 319 |
+
"frac_reward_zero_std": 0.0,
|
| 320 |
+
"grad_norm": 3.0532214641571045,
|
| 321 |
+
"kl": 0.0684322863817215,
|
| 322 |
+
"learning_rate": 2.6e-06,
|
| 323 |
+
"loss": 0.0310845747590065,
|
| 324 |
+
"num_tokens": 49824.0,
|
| 325 |
+
"reward": 0.8395085334777832,
|
| 326 |
+
"reward_std": 0.15289129316806793,
|
| 327 |
+
"rewards/AnswererRewardFunction/mean": 0.8395085334777832,
|
| 328 |
+
"rewards/AnswererRewardFunction/std": 0.15289129316806793,
|
| 329 |
+
"step": 17,
|
| 330 |
+
"step_time": 7.96237816200005
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"clip_ratio/high_max": 0.0068359375,
|
| 334 |
+
"clip_ratio/high_mean": 0.0068359375,
|
| 335 |
+
"clip_ratio/low_mean": 0.00390625,
|
| 336 |
+
"clip_ratio/low_min": 0.00390625,
|
| 337 |
+
"clip_ratio/region_mean": 0.0107421875,
|
| 338 |
+
"entropy": 0.6381477117538452,
|
| 339 |
+
"epoch": 0.45,
|
| 340 |
+
"grad_norm": 4.106065273284912,
|
| 341 |
+
"kl": 0.09328670799732208,
|
| 342 |
+
"learning_rate": 2.575e-06,
|
| 343 |
+
"loss": -0.07613930106163025,
|
| 344 |
+
"step": 18,
|
| 345 |
+
"step_time": 0.1118761339998855
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"clip_ratio/high_max": 0.005859375,
|
| 349 |
+
"clip_ratio/high_mean": 0.005859375,
|
| 350 |
+
"clip_ratio/low_mean": 0.013671875,
|
| 351 |
+
"clip_ratio/low_min": 0.013671875,
|
| 352 |
+
"clip_ratio/region_mean": 0.01953125,
|
| 353 |
+
"entropy": 0.5689660310745239,
|
| 354 |
+
"epoch": 0.475,
|
| 355 |
+
"grad_norm": 1.5947890281677246,
|
| 356 |
+
"kl": 0.08299237489700317,
|
| 357 |
+
"learning_rate": 2.55e-06,
|
| 358 |
+
"loss": 0.03764766454696655,
|
| 359 |
+
"step": 19,
|
| 360 |
+
"step_time": 0.1142826679999871
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"clip_ratio/high_max": 0.0126953125,
|
| 364 |
+
"clip_ratio/high_mean": 0.0126953125,
|
| 365 |
+
"clip_ratio/low_mean": 0.0107421875,
|
| 366 |
+
"clip_ratio/low_min": 0.0107421875,
|
| 367 |
+
"clip_ratio/region_mean": 0.0234375,
|
| 368 |
+
"entropy": 0.6226636171340942,
|
| 369 |
+
"epoch": 0.5,
|
| 370 |
+
"grad_norm": 1.9378103017807007,
|
| 371 |
+
"kl": 0.08963169902563095,
|
| 372 |
+
"learning_rate": 2.525e-06,
|
| 373 |
+
"loss": 0.008068917319178581,
|
| 374 |
+
"step": 20,
|
| 375 |
+
"step_time": 0.11152214599997023
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"clip_ratio/high_max": 0.0,
|
| 379 |
+
"clip_ratio/high_mean": 0.0,
|
| 380 |
+
"clip_ratio/low_mean": 0.0,
|
| 381 |
+
"clip_ratio/low_min": 0.0,
|
| 382 |
+
"clip_ratio/region_mean": 0.0,
|
| 383 |
+
"completions/clipped_ratio": 1.0,
|
| 384 |
+
"completions/max_length": 256.0,
|
| 385 |
+
"completions/max_terminated_length": 0.0,
|
| 386 |
+
"completions/mean_length": 256.0,
|
| 387 |
+
"completions/mean_terminated_length": 0.0,
|
| 388 |
+
"completions/min_length": 256.0,
|
| 389 |
+
"completions/min_terminated_length": 0.0,
|
| 390 |
+
"entropy": 0.7389768362045288,
|
| 391 |
+
"epoch": 0.525,
|
| 392 |
+
"frac_reward_zero_std": 0.0,
|
| 393 |
+
"grad_norm": 3.434631109237671,
|
| 394 |
+
"kl": 0.12693551182746887,
|
| 395 |
+
"learning_rate": 2.5e-06,
|
| 396 |
+
"loss": -0.023861240595579147,
|
| 397 |
+
"num_tokens": 59924.0,
|
| 398 |
+
"reward": 0.6655067205429077,
|
| 399 |
+
"reward_std": 0.17209242284297943,
|
| 400 |
+
"rewards/AnswererRewardFunction/mean": 0.6655067205429077,
|
| 401 |
+
"rewards/AnswererRewardFunction/std": 0.17209243774414062,
|
| 402 |
+
"step": 21,
|
| 403 |
+
"step_time": 7.950470066999969
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"clip_ratio/high_max": 0.0048828125,
|
| 407 |
+
"clip_ratio/high_mean": 0.0048828125,
|
| 408 |
+
"clip_ratio/low_mean": 0.0029296875,
|
| 409 |
+
"clip_ratio/low_min": 0.0029296875,
|
| 410 |
+
"clip_ratio/region_mean": 0.0078125,
|
| 411 |
+
"entropy": 0.6952782869338989,
|
| 412 |
+
"epoch": 0.55,
|
| 413 |
+
"grad_norm": 3.3466062545776367,
|
| 414 |
+
"kl": 0.09632558375597,
|
| 415 |
+
"learning_rate": 2.475e-06,
|
| 416 |
+
"loss": -0.015093192458152771,
|
| 417 |
+
"step": 22,
|
| 418 |
+
"step_time": 0.11464352500001951
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"clip_ratio/high_max": 0.0078125,
|
| 422 |
+
"clip_ratio/high_mean": 0.0078125,
|
| 423 |
+
"clip_ratio/low_mean": 0.015625,
|
| 424 |
+
"clip_ratio/low_min": 0.015625,
|
| 425 |
+
"clip_ratio/region_mean": 0.0234375,
|
| 426 |
+
"entropy": 0.4058366119861603,
|
| 427 |
+
"epoch": 0.575,
|
| 428 |
+
"grad_norm": 1.916627287864685,
|
| 429 |
+
"kl": 0.10934790968894958,
|
| 430 |
+
"learning_rate": 2.45e-06,
|
| 431 |
+
"loss": 0.038687966763973236,
|
| 432 |
+
"step": 23,
|
| 433 |
+
"step_time": 0.11371357500001977
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"clip_ratio/high_max": 0.0107421875,
|
| 437 |
+
"clip_ratio/high_mean": 0.0107421875,
|
| 438 |
+
"clip_ratio/low_mean": 0.005859375,
|
| 439 |
+
"clip_ratio/low_min": 0.005859375,
|
| 440 |
+
"clip_ratio/region_mean": 0.0166015625,
|
| 441 |
+
"entropy": 0.4684597849845886,
|
| 442 |
+
"epoch": 0.6,
|
| 443 |
+
"grad_norm": 2.56581449508667,
|
| 444 |
+
"kl": 0.09573253244161606,
|
| 445 |
+
"learning_rate": 2.425e-06,
|
| 446 |
+
"loss": 0.0021250469144433737,
|
| 447 |
+
"step": 24,
|
| 448 |
+
"step_time": 0.11208250999993652
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"clip_ratio/high_max": 0.0,
|
| 452 |
+
"clip_ratio/high_mean": 0.0,
|
| 453 |
+
"clip_ratio/low_mean": 0.0,
|
| 454 |
+
"clip_ratio/low_min": 0.0,
|
| 455 |
+
"clip_ratio/region_mean": 0.0,
|
| 456 |
+
"completions/clipped_ratio": 1.0,
|
| 457 |
+
"completions/max_length": 256.0,
|
| 458 |
+
"completions/max_terminated_length": 0.0,
|
| 459 |
+
"completions/mean_length": 256.0,
|
| 460 |
+
"completions/mean_terminated_length": 0.0,
|
| 461 |
+
"completions/min_length": 256.0,
|
| 462 |
+
"completions/min_terminated_length": 0.0,
|
| 463 |
+
"entropy": 0.5364047288894653,
|
| 464 |
+
"epoch": 0.625,
|
| 465 |
+
"frac_reward_zero_std": 0.0,
|
| 466 |
+
"grad_norm": 3.1820530891418457,
|
| 467 |
+
"kl": 0.11052386462688446,
|
| 468 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 469 |
+
"loss": -0.12420213222503662,
|
| 470 |
+
"num_tokens": 69996.0,
|
| 471 |
+
"reward": 0.706213653087616,
|
| 472 |
+
"reward_std": 0.22473102807998657,
|
| 473 |
+
"rewards/AnswererRewardFunction/mean": 0.706213653087616,
|
| 474 |
+
"rewards/AnswererRewardFunction/std": 0.22473104298114777,
|
| 475 |
+
"step": 25,
|
| 476 |
+
"step_time": 7.935186696000073
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"clip_ratio/high_max": 0.0068359375,
|
| 480 |
+
"clip_ratio/high_mean": 0.0068359375,
|
| 481 |
+
"clip_ratio/low_mean": 0.0,
|
| 482 |
+
"clip_ratio/low_min": 0.0,
|
| 483 |
+
"clip_ratio/region_mean": 0.0068359375,
|
| 484 |
+
"entropy": 0.5579304695129395,
|
| 485 |
+
"epoch": 0.65,
|
| 486 |
+
"grad_norm": 3.2145562171936035,
|
| 487 |
+
"kl": 0.09837664663791656,
|
| 488 |
+
"learning_rate": 2.375e-06,
|
| 489 |
+
"loss": -0.048176076263189316,
|
| 490 |
+
"step": 26,
|
| 491 |
+
"step_time": 0.11106458200015368
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"clip_ratio/high_max": 0.005859375,
|
| 495 |
+
"clip_ratio/high_mean": 0.005859375,
|
| 496 |
+
"clip_ratio/low_mean": 0.0048828125,
|
| 497 |
+
"clip_ratio/low_min": 0.0048828125,
|
| 498 |
+
"clip_ratio/region_mean": 0.0107421875,
|
| 499 |
+
"entropy": 0.5162898898124695,
|
| 500 |
+
"epoch": 0.675,
|
| 501 |
+
"grad_norm": 2.879662036895752,
|
| 502 |
+
"kl": 0.11825042963027954,
|
| 503 |
+
"learning_rate": 2.35e-06,
|
| 504 |
+
"loss": 0.06364448368549347,
|
| 505 |
+
"step": 27,
|
| 506 |
+
"step_time": 0.11082449600007749
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"clip_ratio/high_max": 0.0048828125,
|
| 510 |
+
"clip_ratio/high_mean": 0.0048828125,
|
| 511 |
+
"clip_ratio/low_mean": 0.0068359375,
|
| 512 |
+
"clip_ratio/low_min": 0.0068359375,
|
| 513 |
+
"clip_ratio/region_mean": 0.01171875,
|
| 514 |
+
"entropy": 0.4608369469642639,
|
| 515 |
+
"epoch": 0.7,
|
| 516 |
+
"grad_norm": 2.5901477336883545,
|
| 517 |
+
"kl": 0.09662161022424698,
|
| 518 |
+
"learning_rate": 2.325e-06,
|
| 519 |
+
"loss": 0.11211410164833069,
|
| 520 |
+
"step": 28,
|
| 521 |
+
"step_time": 0.11086178800019297
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"clip_ratio/high_max": 0.0,
|
| 525 |
+
"clip_ratio/high_mean": 0.0,
|
| 526 |
+
"clip_ratio/low_mean": 0.0,
|
| 527 |
+
"clip_ratio/low_min": 0.0,
|
| 528 |
+
"clip_ratio/region_mean": 0.0,
|
| 529 |
+
"completions/clipped_ratio": 1.0,
|
| 530 |
+
"completions/max_length": 256.0,
|
| 531 |
+
"completions/max_terminated_length": 0.0,
|
| 532 |
+
"completions/mean_length": 256.0,
|
| 533 |
+
"completions/mean_terminated_length": 0.0,
|
| 534 |
+
"completions/min_length": 256.0,
|
| 535 |
+
"completions/min_terminated_length": 0.0,
|
| 536 |
+
"entropy": 0.42433732748031616,
|
| 537 |
+
"epoch": 0.725,
|
| 538 |
+
"frac_reward_zero_std": 0.0,
|
| 539 |
+
"grad_norm": 2.176980495452881,
|
| 540 |
+
"kl": 0.10524210333824158,
|
| 541 |
+
"learning_rate": 2.3000000000000004e-06,
|
| 542 |
+
"loss": 0.008818887174129486,
|
| 543 |
+
"num_tokens": 80208.0,
|
| 544 |
+
"reward": 0.7267547845840454,
|
| 545 |
+
"reward_std": 0.179252490401268,
|
| 546 |
+
"rewards/AnswererRewardFunction/mean": 0.7267547845840454,
|
| 547 |
+
"rewards/AnswererRewardFunction/std": 0.1792525053024292,
|
| 548 |
+
"step": 29,
|
| 549 |
+
"step_time": 7.899287777999916
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"clip_ratio/high_max": 0.0,
|
| 553 |
+
"clip_ratio/high_mean": 0.0,
|
| 554 |
+
"clip_ratio/low_mean": 0.005859375,
|
| 555 |
+
"clip_ratio/low_min": 0.005859375,
|
| 556 |
+
"clip_ratio/region_mean": 0.005859375,
|
| 557 |
+
"entropy": 0.7436986565589905,
|
| 558 |
+
"epoch": 0.75,
|
| 559 |
+
"grad_norm": 5.134580612182617,
|
| 560 |
+
"kl": 0.12144140899181366,
|
| 561 |
+
"learning_rate": 2.275e-06,
|
| 562 |
+
"loss": 0.20647373795509338,
|
| 563 |
+
"step": 30,
|
| 564 |
+
"step_time": 0.11205311000003348
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"clip_ratio/high_max": 0.0068359375,
|
| 568 |
+
"clip_ratio/high_mean": 0.0068359375,
|
| 569 |
+
"clip_ratio/low_mean": 0.00390625,
|
| 570 |
+
"clip_ratio/low_min": 0.00390625,
|
| 571 |
+
"clip_ratio/region_mean": 0.0107421875,
|
| 572 |
+
"entropy": 0.5429961681365967,
|
| 573 |
+
"epoch": 0.775,
|
| 574 |
+
"grad_norm": 2.0209434032440186,
|
| 575 |
+
"kl": 0.12714454531669617,
|
| 576 |
+
"learning_rate": 2.25e-06,
|
| 577 |
+
"loss": -0.05760453641414642,
|
| 578 |
+
"step": 31,
|
| 579 |
+
"step_time": 0.11339481400000295
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"clip_ratio/high_max": 0.01953125,
|
| 583 |
+
"clip_ratio/high_mean": 0.01953125,
|
| 584 |
+
"clip_ratio/low_mean": 0.0,
|
| 585 |
+
"clip_ratio/low_min": 0.0,
|
| 586 |
+
"clip_ratio/region_mean": 0.01953125,
|
| 587 |
+
"entropy": 0.32000452280044556,
|
| 588 |
+
"epoch": 0.8,
|
| 589 |
+
"grad_norm": 1.7520897388458252,
|
| 590 |
+
"kl": 0.11870498955249786,
|
| 591 |
+
"learning_rate": 2.2250000000000003e-06,
|
| 592 |
+
"loss": -0.1561477929353714,
|
| 593 |
+
"step": 32,
|
| 594 |
+
"step_time": 0.11099254700002348
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"clip_ratio/high_max": 0.0,
|
| 598 |
+
"clip_ratio/high_mean": 0.0,
|
| 599 |
+
"clip_ratio/low_mean": 0.0,
|
| 600 |
+
"clip_ratio/low_min": 0.0,
|
| 601 |
+
"clip_ratio/region_mean": 0.0,
|
| 602 |
+
"completions/clipped_ratio": 1.0,
|
| 603 |
+
"completions/max_length": 256.0,
|
| 604 |
+
"completions/max_terminated_length": 0.0,
|
| 605 |
+
"completions/mean_length": 256.0,
|
| 606 |
+
"completions/mean_terminated_length": 0.0,
|
| 607 |
+
"completions/min_length": 256.0,
|
| 608 |
+
"completions/min_terminated_length": 0.0,
|
| 609 |
+
"entropy": 0.37099331617355347,
|
| 610 |
+
"epoch": 0.825,
|
| 611 |
+
"frac_reward_zero_std": 0.0,
|
| 612 |
+
"grad_norm": 2.467416763305664,
|
| 613 |
+
"kl": 0.11815842986106873,
|
| 614 |
+
"learning_rate": 2.1999999999999997e-06,
|
| 615 |
+
"loss": -0.02738424763083458,
|
| 616 |
+
"num_tokens": 90620.0,
|
| 617 |
+
"reward": 0.8727706074714661,
|
| 618 |
+
"reward_std": 0.1972254365682602,
|
| 619 |
+
"rewards/AnswererRewardFunction/mean": 0.8727706074714661,
|
| 620 |
+
"rewards/AnswererRewardFunction/std": 0.1972254365682602,
|
| 621 |
+
"step": 33,
|
| 622 |
+
"step_time": 7.91921779900008
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"clip_ratio/high_max": 0.0009765625,
|
| 626 |
+
"clip_ratio/high_mean": 0.0009765625,
|
| 627 |
+
"clip_ratio/low_mean": 0.00390625,
|
| 628 |
+
"clip_ratio/low_min": 0.00390625,
|
| 629 |
+
"clip_ratio/region_mean": 0.0048828125,
|
| 630 |
+
"entropy": 0.3960208296775818,
|
| 631 |
+
"epoch": 0.85,
|
| 632 |
+
"grad_norm": 2.1480352878570557,
|
| 633 |
+
"kl": 0.09754370152950287,
|
| 634 |
+
"learning_rate": 2.175e-06,
|
| 635 |
+
"loss": 0.060052644461393356,
|
| 636 |
+
"step": 34,
|
| 637 |
+
"step_time": 0.11552422499994464
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"clip_ratio/high_max": 0.0107421875,
|
| 641 |
+
"clip_ratio/high_mean": 0.0107421875,
|
| 642 |
+
"clip_ratio/low_mean": 0.0048828125,
|
| 643 |
+
"clip_ratio/low_min": 0.0048828125,
|
| 644 |
+
"clip_ratio/region_mean": 0.015625,
|
| 645 |
+
"entropy": 0.542590856552124,
|
| 646 |
+
"epoch": 0.875,
|
| 647 |
+
"grad_norm": 3.6754674911499023,
|
| 648 |
+
"kl": 0.10792969167232513,
|
| 649 |
+
"learning_rate": 2.15e-06,
|
| 650 |
+
"loss": -0.0720224380493164,
|
| 651 |
+
"step": 35,
|
| 652 |
+
"step_time": 0.11197718300013548
|
| 653 |
+
},
|
| 654 |
+
{
|
| 655 |
+
"clip_ratio/high_max": 0.013671875,
|
| 656 |
+
"clip_ratio/high_mean": 0.013671875,
|
| 657 |
+
"clip_ratio/low_mean": 0.0087890625,
|
| 658 |
+
"clip_ratio/low_min": 0.0087890625,
|
| 659 |
+
"clip_ratio/region_mean": 0.0224609375,
|
| 660 |
+
"entropy": 0.3774242401123047,
|
| 661 |
+
"epoch": 0.9,
|
| 662 |
+
"grad_norm": 1250253952.0,
|
| 663 |
+
"kl": 23265218.0,
|
| 664 |
+
"learning_rate": 2.125e-06,
|
| 665 |
+
"loss": 58163.08203125,
|
| 666 |
+
"step": 36,
|
| 667 |
+
"step_time": 0.12223055300000851
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"clip_ratio/high_max": 0.0009765625,
|
| 671 |
+
"clip_ratio/high_mean": 0.0009765625,
|
| 672 |
+
"clip_ratio/low_mean": 0.0,
|
| 673 |
+
"clip_ratio/low_min": 0.0,
|
| 674 |
+
"clip_ratio/region_mean": 0.0009765625,
|
| 675 |
+
"completions/clipped_ratio": 1.0,
|
| 676 |
+
"completions/max_length": 256.0,
|
| 677 |
+
"completions/max_terminated_length": 0.0,
|
| 678 |
+
"completions/mean_length": 256.0,
|
| 679 |
+
"completions/mean_terminated_length": 0.0,
|
| 680 |
+
"completions/min_length": 256.0,
|
| 681 |
+
"completions/min_terminated_length": 0.0,
|
| 682 |
+
"entropy": 0.5863927602767944,
|
| 683 |
+
"epoch": 0.925,
|
| 684 |
+
"frac_reward_zero_std": 0.25,
|
| 685 |
+
"grad_norm": 1.7186689376831055,
|
| 686 |
+
"kl": 0.12184222787618637,
|
| 687 |
+
"learning_rate": 2.1e-06,
|
| 688 |
+
"loss": 0.01318025216460228,
|
| 689 |
+
"num_tokens": 100744.0,
|
| 690 |
+
"reward": 0.8322785496711731,
|
| 691 |
+
"reward_std": 0.17467200756072998,
|
| 692 |
+
"rewards/AnswererRewardFunction/mean": 0.8322785496711731,
|
| 693 |
+
"rewards/AnswererRewardFunction/std": 0.17467200756072998,
|
| 694 |
+
"step": 37,
|
| 695 |
+
"step_time": 7.957934481999928
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"clip_ratio/high_max": 0.0078125,
|
| 699 |
+
"clip_ratio/high_mean": 0.0078125,
|
| 700 |
+
"clip_ratio/low_mean": 0.0,
|
| 701 |
+
"clip_ratio/low_min": 0.0,
|
| 702 |
+
"clip_ratio/region_mean": 0.0078125,
|
| 703 |
+
"entropy": 0.4571380019187927,
|
| 704 |
+
"epoch": 0.95,
|
| 705 |
+
"grad_norm": 2.5918664932250977,
|
| 706 |
+
"kl": 0.15713587403297424,
|
| 707 |
+
"learning_rate": 2.075e-06,
|
| 708 |
+
"loss": -0.15538358688354492,
|
| 709 |
+
"step": 38,
|
| 710 |
+
"step_time": 0.11134468399995967
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"clip_ratio/high_max": 0.009765625,
|
| 714 |
+
"clip_ratio/high_mean": 0.009765625,
|
| 715 |
+
"clip_ratio/low_mean": 0.001953125,
|
| 716 |
+
"clip_ratio/low_min": 0.001953125,
|
| 717 |
+
"clip_ratio/region_mean": 0.01171875,
|
| 718 |
+
"entropy": 0.5838406085968018,
|
| 719 |
+
"epoch": 0.975,
|
| 720 |
+
"grad_norm": 3.389342784881592,
|
| 721 |
+
"kl": 0.16417548060417175,
|
| 722 |
+
"learning_rate": 2.0500000000000003e-06,
|
| 723 |
+
"loss": 0.036985911428928375,
|
| 724 |
+
"step": 39,
|
| 725 |
+
"step_time": 0.11151985699984834
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"clip_ratio/high_max": 0.0068359375,
|
| 729 |
+
"clip_ratio/high_mean": 0.0068359375,
|
| 730 |
+
"clip_ratio/low_mean": 0.005859375,
|
| 731 |
+
"clip_ratio/low_min": 0.005859375,
|
| 732 |
+
"clip_ratio/region_mean": 0.0126953125,
|
| 733 |
+
"entropy": 0.626500129699707,
|
| 734 |
+
"epoch": 1.0,
|
| 735 |
+
"grad_norm": 3.443002939224243,
|
| 736 |
+
"kl": 0.19608259201049805,
|
| 737 |
+
"learning_rate": 2.025e-06,
|
| 738 |
+
"loss": 0.1089545264840126,
|
| 739 |
+
"step": 40,
|
| 740 |
+
"step_time": 0.11399926900003265
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"clip_ratio/high_max": 0.0,
|
| 744 |
+
"clip_ratio/high_mean": 0.0,
|
| 745 |
+
"clip_ratio/low_mean": 0.0009765625,
|
| 746 |
+
"clip_ratio/low_min": 0.0009765625,
|
| 747 |
+
"clip_ratio/region_mean": 0.0009765625,
|
| 748 |
+
"completions/clipped_ratio": 1.0,
|
| 749 |
+
"completions/max_length": 256.0,
|
| 750 |
+
"completions/max_terminated_length": 0.0,
|
| 751 |
+
"completions/mean_length": 256.0,
|
| 752 |
+
"completions/mean_terminated_length": 0.0,
|
| 753 |
+
"completions/min_length": 256.0,
|
| 754 |
+
"completions/min_terminated_length": 0.0,
|
| 755 |
+
"entropy": 0.5399507880210876,
|
| 756 |
+
"epoch": 1.025,
|
| 757 |
+
"frac_reward_zero_std": 0.0,
|
| 758 |
+
"grad_norm": 2.0896756649017334,
|
| 759 |
+
"kl": 0.16342155635356903,
|
| 760 |
+
"learning_rate": 2e-06,
|
| 761 |
+
"loss": 0.09698553383350372,
|
| 762 |
+
"num_tokens": 110360.0,
|
| 763 |
+
"reward": 0.6938474774360657,
|
| 764 |
+
"reward_std": 0.07210177928209305,
|
| 765 |
+
"rewards/AnswererRewardFunction/mean": 0.6938474774360657,
|
| 766 |
+
"rewards/AnswererRewardFunction/std": 0.07210177928209305,
|
| 767 |
+
"step": 41,
|
| 768 |
+
"step_time": 8.036780455000098
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"clip_ratio/high_max": 0.0048828125,
|
| 772 |
+
"clip_ratio/high_mean": 0.0048828125,
|
| 773 |
+
"clip_ratio/low_mean": 0.001953125,
|
| 774 |
+
"clip_ratio/low_min": 0.001953125,
|
| 775 |
+
"clip_ratio/region_mean": 0.0068359375,
|
| 776 |
+
"entropy": 0.5720961689949036,
|
| 777 |
+
"epoch": 1.05,
|
| 778 |
+
"grad_norm": 3.426593065261841,
|
| 779 |
+
"kl": 0.19985011219978333,
|
| 780 |
+
"learning_rate": 1.975e-06,
|
| 781 |
+
"loss": -0.014200590550899506,
|
| 782 |
+
"step": 42,
|
| 783 |
+
"step_time": 0.11001263299999664
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"clip_ratio/high_max": 0.009765625,
|
| 787 |
+
"clip_ratio/high_mean": 0.009765625,
|
| 788 |
+
"clip_ratio/low_mean": 0.0048828125,
|
| 789 |
+
"clip_ratio/low_min": 0.0048828125,
|
| 790 |
+
"clip_ratio/region_mean": 0.0146484375,
|
| 791 |
+
"entropy": 0.5380070209503174,
|
| 792 |
+
"epoch": 1.075,
|
| 793 |
+
"grad_norm": 1.9040859937667847,
|
| 794 |
+
"kl": 0.1798008680343628,
|
| 795 |
+
"learning_rate": 1.95e-06,
|
| 796 |
+
"loss": -0.0504855252802372,
|
| 797 |
+
"step": 43,
|
| 798 |
+
"step_time": 0.10936977100004697
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"clip_ratio/high_max": 0.01953125,
|
| 802 |
+
"clip_ratio/high_mean": 0.01953125,
|
| 803 |
+
"clip_ratio/low_mean": 0.0029296875,
|
| 804 |
+
"clip_ratio/low_min": 0.0029296875,
|
| 805 |
+
"clip_ratio/region_mean": 0.0224609375,
|
| 806 |
+
"entropy": 0.45030832290649414,
|
| 807 |
+
"epoch": 1.1,
|
| 808 |
+
"grad_norm": 2.7696876525878906,
|
| 809 |
+
"kl": 0.1864057183265686,
|
| 810 |
+
"learning_rate": 1.925e-06,
|
| 811 |
+
"loss": -0.029610613361001015,
|
| 812 |
+
"step": 44,
|
| 813 |
+
"step_time": 0.11253336699996908
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"clip_ratio/high_max": 0.0,
|
| 817 |
+
"clip_ratio/high_mean": 0.0,
|
| 818 |
+
"clip_ratio/low_mean": 0.0,
|
| 819 |
+
"clip_ratio/low_min": 0.0,
|
| 820 |
+
"clip_ratio/region_mean": 0.0,
|
| 821 |
+
"completions/clipped_ratio": 1.0,
|
| 822 |
+
"completions/max_length": 256.0,
|
| 823 |
+
"completions/max_terminated_length": 0.0,
|
| 824 |
+
"completions/mean_length": 256.0,
|
| 825 |
+
"completions/mean_terminated_length": 0.0,
|
| 826 |
+
"completions/min_length": 256.0,
|
| 827 |
+
"completions/min_terminated_length": 0.0,
|
| 828 |
+
"entropy": 0.48534196615219116,
|
| 829 |
+
"epoch": 1.125,
|
| 830 |
+
"frac_reward_zero_std": 0.0,
|
| 831 |
+
"grad_norm": 2.1930065155029297,
|
| 832 |
+
"kl": 0.21630029380321503,
|
| 833 |
+
"learning_rate": 1.9e-06,
|
| 834 |
+
"loss": -0.0531463548541069,
|
| 835 |
+
"num_tokens": 120316.0,
|
| 836 |
+
"reward": 0.8261469602584839,
|
| 837 |
+
"reward_std": 0.13625741004943848,
|
| 838 |
+
"rewards/AnswererRewardFunction/mean": 0.8261469602584839,
|
| 839 |
+
"rewards/AnswererRewardFunction/std": 0.13625742495059967,
|
| 840 |
+
"step": 45,
|
| 841 |
+
"step_time": 7.959708546999991
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"clip_ratio/high_max": 0.001953125,
|
| 845 |
+
"clip_ratio/high_mean": 0.001953125,
|
| 846 |
+
"clip_ratio/low_mean": 0.0009765625,
|
| 847 |
+
"clip_ratio/low_min": 0.0009765625,
|
| 848 |
+
"clip_ratio/region_mean": 0.0029296875,
|
| 849 |
+
"entropy": 0.5528072714805603,
|
| 850 |
+
"epoch": 1.15,
|
| 851 |
+
"grad_norm": 3.234687328338623,
|
| 852 |
+
"kl": 0.20044510066509247,
|
| 853 |
+
"learning_rate": 1.875e-06,
|
| 854 |
+
"loss": -0.14086312055587769,
|
| 855 |
+
"step": 46,
|
| 856 |
+
"step_time": 0.11803400899998451
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"clip_ratio/high_max": 0.0068359375,
|
| 860 |
+
"clip_ratio/high_mean": 0.0068359375,
|
| 861 |
+
"clip_ratio/low_mean": 0.0068359375,
|
| 862 |
+
"clip_ratio/low_min": 0.0068359375,
|
| 863 |
+
"clip_ratio/region_mean": 0.013671875,
|
| 864 |
+
"entropy": 0.5274936556816101,
|
| 865 |
+
"epoch": 1.175,
|
| 866 |
+
"grad_norm": 2.5730295181274414,
|
| 867 |
+
"kl": 0.16597405076026917,
|
| 868 |
+
"learning_rate": 1.85e-06,
|
| 869 |
+
"loss": 0.023598745465278625,
|
| 870 |
+
"step": 47,
|
| 871 |
+
"step_time": 0.12857143599990195
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"clip_ratio/high_max": 0.0,
|
| 875 |
+
"clip_ratio/high_mean": 0.0,
|
| 876 |
+
"clip_ratio/low_mean": 0.0166015625,
|
| 877 |
+
"clip_ratio/low_min": 0.0166015625,
|
| 878 |
+
"clip_ratio/region_mean": 0.0166015625,
|
| 879 |
+
"entropy": 0.5713120102882385,
|
| 880 |
+
"epoch": 1.2,
|
| 881 |
+
"grad_norm": 3.3215487003326416,
|
| 882 |
+
"kl": 0.15826421976089478,
|
| 883 |
+
"learning_rate": 1.8249999999999999e-06,
|
| 884 |
+
"loss": 0.1730981469154358,
|
| 885 |
+
"step": 48,
|
| 886 |
+
"step_time": 0.1108278810002048
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"clip_ratio/high_max": 0.0009765625,
|
| 890 |
+
"clip_ratio/high_mean": 0.0009765625,
|
| 891 |
+
"clip_ratio/low_mean": 0.0009765625,
|
| 892 |
+
"clip_ratio/low_min": 0.0009765625,
|
| 893 |
+
"clip_ratio/region_mean": 0.001953125,
|
| 894 |
+
"completions/clipped_ratio": 1.0,
|
| 895 |
+
"completions/max_length": 256.0,
|
| 896 |
+
"completions/max_terminated_length": 0.0,
|
| 897 |
+
"completions/mean_length": 256.0,
|
| 898 |
+
"completions/mean_terminated_length": 0.0,
|
| 899 |
+
"completions/min_length": 256.0,
|
| 900 |
+
"completions/min_terminated_length": 0.0,
|
| 901 |
+
"entropy": 0.40691593289375305,
|
| 902 |
+
"epoch": 1.225,
|
| 903 |
+
"frac_reward_zero_std": 0.0,
|
| 904 |
+
"grad_norm": 2.9191150665283203,
|
| 905 |
+
"kl": 0.16014470160007477,
|
| 906 |
+
"learning_rate": 1.8e-06,
|
| 907 |
+
"loss": 0.16662299633026123,
|
| 908 |
+
"num_tokens": 130272.0,
|
| 909 |
+
"reward": 0.8778884410858154,
|
| 910 |
+
"reward_std": 0.17971795797348022,
|
| 911 |
+
"rewards/AnswererRewardFunction/mean": 0.8778884410858154,
|
| 912 |
+
"rewards/AnswererRewardFunction/std": 0.17971795797348022,
|
| 913 |
+
"step": 49,
|
| 914 |
+
"step_time": 8.018302381000012
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"clip_ratio/high_max": 0.0,
|
| 918 |
+
"clip_ratio/high_mean": 0.0,
|
| 919 |
+
"clip_ratio/low_mean": 0.00390625,
|
| 920 |
+
"clip_ratio/low_min": 0.00390625,
|
| 921 |
+
"clip_ratio/region_mean": 0.00390625,
|
| 922 |
+
"entropy": 0.3267471194267273,
|
| 923 |
+
"epoch": 1.25,
|
| 924 |
+
"grad_norm": 1.9587223529815674,
|
| 925 |
+
"kl": 0.17650750279426575,
|
| 926 |
+
"learning_rate": 1.7750000000000002e-06,
|
| 927 |
+
"loss": -0.019061673432588577,
|
| 928 |
+
"step": 50,
|
| 929 |
+
"step_time": 0.11307515899989085
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"clip_ratio/high_max": 0.009765625,
|
| 933 |
+
"clip_ratio/high_mean": 0.009765625,
|
| 934 |
+
"clip_ratio/low_mean": 0.0068359375,
|
| 935 |
+
"clip_ratio/low_min": 0.0068359375,
|
| 936 |
+
"clip_ratio/region_mean": 0.0166015625,
|
| 937 |
+
"entropy": 0.47037309408187866,
|
| 938 |
+
"epoch": 1.275,
|
| 939 |
+
"grad_norm": 1.7802549600601196,
|
| 940 |
+
"kl": 0.15526388585567474,
|
| 941 |
+
"learning_rate": 1.7500000000000002e-06,
|
| 942 |
+
"loss": -0.04963531345129013,
|
| 943 |
+
"step": 51,
|
| 944 |
+
"step_time": 0.11180471299985584
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"clip_ratio/high_max": 0.0068359375,
|
| 948 |
+
"clip_ratio/high_mean": 0.0068359375,
|
| 949 |
+
"clip_ratio/low_mean": 0.013671875,
|
| 950 |
+
"clip_ratio/low_min": 0.013671875,
|
| 951 |
+
"clip_ratio/region_mean": 0.0205078125,
|
| 952 |
+
"entropy": 0.5774157047271729,
|
| 953 |
+
"epoch": 1.3,
|
| 954 |
+
"grad_norm": 2.833186388015747,
|
| 955 |
+
"kl": 0.1678266078233719,
|
| 956 |
+
"learning_rate": 1.725e-06,
|
| 957 |
+
"loss": -0.093873992562294,
|
| 958 |
+
"step": 52,
|
| 959 |
+
"step_time": 0.112085678999847
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"clip_ratio/high_max": 0.0,
|
| 963 |
+
"clip_ratio/high_mean": 0.0,
|
| 964 |
+
"clip_ratio/low_mean": 0.0,
|
| 965 |
+
"clip_ratio/low_min": 0.0,
|
| 966 |
+
"clip_ratio/region_mean": 0.0,
|
| 967 |
+
"completions/clipped_ratio": 1.0,
|
| 968 |
+
"completions/max_length": 256.0,
|
| 969 |
+
"completions/max_terminated_length": 0.0,
|
| 970 |
+
"completions/mean_length": 256.0,
|
| 971 |
+
"completions/mean_terminated_length": 0.0,
|
| 972 |
+
"completions/min_length": 256.0,
|
| 973 |
+
"completions/min_terminated_length": 0.0,
|
| 974 |
+
"entropy": 0.5357224941253662,
|
| 975 |
+
"epoch": 1.325,
|
| 976 |
+
"frac_reward_zero_std": 0.0,
|
| 977 |
+
"grad_norm": 3.081116199493408,
|
| 978 |
+
"kl": 0.18712827563285828,
|
| 979 |
+
"learning_rate": 1.7e-06,
|
| 980 |
+
"loss": 0.10657118260860443,
|
| 981 |
+
"num_tokens": 140436.0,
|
| 982 |
+
"reward": 0.7755283117294312,
|
| 983 |
+
"reward_std": 0.15775497257709503,
|
| 984 |
+
"rewards/AnswererRewardFunction/mean": 0.7755283117294312,
|
| 985 |
+
"rewards/AnswererRewardFunction/std": 0.15775495767593384,
|
| 986 |
+
"step": 53,
|
| 987 |
+
"step_time": 8.014508306999915
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"clip_ratio/high_max": 0.0029296875,
|
| 991 |
+
"clip_ratio/high_mean": 0.0029296875,
|
| 992 |
+
"clip_ratio/low_mean": 0.005859375,
|
| 993 |
+
"clip_ratio/low_min": 0.005859375,
|
| 994 |
+
"clip_ratio/region_mean": 0.0087890625,
|
| 995 |
+
"entropy": 0.4421442747116089,
|
| 996 |
+
"epoch": 1.35,
|
| 997 |
+
"grad_norm": 2.3898942470550537,
|
| 998 |
+
"kl": 0.1956406980752945,
|
| 999 |
+
"learning_rate": 1.675e-06,
|
| 1000 |
+
"loss": 0.011319718323647976,
|
| 1001 |
+
"step": 54,
|
| 1002 |
+
"step_time": 0.11459642999989228
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"clip_ratio/high_max": 0.001953125,
|
| 1006 |
+
"clip_ratio/high_mean": 0.001953125,
|
| 1007 |
+
"clip_ratio/low_mean": 0.009765625,
|
| 1008 |
+
"clip_ratio/low_min": 0.009765625,
|
| 1009 |
+
"clip_ratio/region_mean": 0.01171875,
|
| 1010 |
+
"entropy": 0.4516857862472534,
|
| 1011 |
+
"epoch": 1.375,
|
| 1012 |
+
"grad_norm": 1.689841389656067,
|
| 1013 |
+
"kl": 0.18309825658798218,
|
| 1014 |
+
"learning_rate": 1.65e-06,
|
| 1015 |
+
"loss": 0.06909539550542831,
|
| 1016 |
+
"step": 55,
|
| 1017 |
+
"step_time": 0.11464596999985588
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"clip_ratio/high_max": 0.009765625,
|
| 1021 |
+
"clip_ratio/high_mean": 0.009765625,
|
| 1022 |
+
"clip_ratio/low_mean": 0.0,
|
| 1023 |
+
"clip_ratio/low_min": 0.0,
|
| 1024 |
+
"clip_ratio/region_mean": 0.009765625,
|
| 1025 |
+
"entropy": 0.3544921875,
|
| 1026 |
+
"epoch": 1.4,
|
| 1027 |
+
"grad_norm": 2.365453004837036,
|
| 1028 |
+
"kl": 0.16238607466220856,
|
| 1029 |
+
"learning_rate": 1.625e-06,
|
| 1030 |
+
"loss": -0.18329277634620667,
|
| 1031 |
+
"step": 56,
|
| 1032 |
+
"step_time": 0.1123857229999885
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"clip_ratio/high_max": 0.0,
|
| 1036 |
+
"clip_ratio/high_mean": 0.0,
|
| 1037 |
+
"clip_ratio/low_mean": 0.0009765625,
|
| 1038 |
+
"clip_ratio/low_min": 0.0009765625,
|
| 1039 |
+
"clip_ratio/region_mean": 0.0009765625,
|
| 1040 |
+
"completions/clipped_ratio": 1.0,
|
| 1041 |
+
"completions/max_length": 256.0,
|
| 1042 |
+
"completions/max_terminated_length": 0.0,
|
| 1043 |
+
"completions/mean_length": 256.0,
|
| 1044 |
+
"completions/mean_terminated_length": 0.0,
|
| 1045 |
+
"completions/min_length": 256.0,
|
| 1046 |
+
"completions/min_terminated_length": 0.0,
|
| 1047 |
+
"entropy": 0.6127996444702148,
|
| 1048 |
+
"epoch": 1.425,
|
| 1049 |
+
"frac_reward_zero_std": 0.0,
|
| 1050 |
+
"grad_norm": 4.42158842086792,
|
| 1051 |
+
"kl": 0.12734977900981903,
|
| 1052 |
+
"learning_rate": 1.6e-06,
|
| 1053 |
+
"loss": 0.23196423053741455,
|
| 1054 |
+
"num_tokens": 150864.0,
|
| 1055 |
+
"reward": 0.8476194143295288,
|
| 1056 |
+
"reward_std": 0.16014426946640015,
|
| 1057 |
+
"rewards/AnswererRewardFunction/mean": 0.8476194143295288,
|
| 1058 |
+
"rewards/AnswererRewardFunction/std": 0.16014428436756134,
|
| 1059 |
+
"step": 57,
|
| 1060 |
+
"step_time": 7.9517445380001845
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"clip_ratio/high_max": 0.00390625,
|
| 1064 |
+
"clip_ratio/high_mean": 0.00390625,
|
| 1065 |
+
"clip_ratio/low_mean": 0.001953125,
|
| 1066 |
+
"clip_ratio/low_min": 0.001953125,
|
| 1067 |
+
"clip_ratio/region_mean": 0.005859375,
|
| 1068 |
+
"entropy": 0.4672216773033142,
|
| 1069 |
+
"epoch": 1.45,
|
| 1070 |
+
"grad_norm": 2.5147714614868164,
|
| 1071 |
+
"kl": 0.13779059052467346,
|
| 1072 |
+
"learning_rate": 1.5750000000000002e-06,
|
| 1073 |
+
"loss": -0.06359703838825226,
|
| 1074 |
+
"step": 58,
|
| 1075 |
+
"step_time": 0.11299467099979665
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"clip_ratio/high_max": 0.0068359375,
|
| 1079 |
+
"clip_ratio/high_mean": 0.0068359375,
|
| 1080 |
+
"clip_ratio/low_mean": 0.0087890625,
|
| 1081 |
+
"clip_ratio/low_min": 0.0087890625,
|
| 1082 |
+
"clip_ratio/region_mean": 0.015625,
|
| 1083 |
+
"entropy": 0.48225951194763184,
|
| 1084 |
+
"epoch": 1.475,
|
| 1085 |
+
"grad_norm": 2.714545249938965,
|
| 1086 |
+
"kl": 0.1360112726688385,
|
| 1087 |
+
"learning_rate": 1.5500000000000002e-06,
|
| 1088 |
+
"loss": -0.08431179821491241,
|
| 1089 |
+
"step": 59,
|
| 1090 |
+
"step_time": 0.11132271799988303
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"clip_ratio/high_max": 0.001953125,
|
| 1094 |
+
"clip_ratio/high_mean": 0.001953125,
|
| 1095 |
+
"clip_ratio/low_mean": 0.009765625,
|
| 1096 |
+
"clip_ratio/low_min": 0.009765625,
|
| 1097 |
+
"clip_ratio/region_mean": 0.01171875,
|
| 1098 |
+
"entropy": 0.5878890752792358,
|
| 1099 |
+
"epoch": 1.5,
|
| 1100 |
+
"grad_norm": 2.2165496349334717,
|
| 1101 |
+
"kl": 0.1507602035999298,
|
| 1102 |
+
"learning_rate": 1.525e-06,
|
| 1103 |
+
"loss": -0.08282318711280823,
|
| 1104 |
+
"step": 60,
|
| 1105 |
+
"step_time": 0.11144096099997114
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"clip_ratio/high_max": 0.0,
|
| 1109 |
+
"clip_ratio/high_mean": 0.0,
|
| 1110 |
+
"clip_ratio/low_mean": 0.001953125,
|
| 1111 |
+
"clip_ratio/low_min": 0.001953125,
|
| 1112 |
+
"clip_ratio/region_mean": 0.001953125,
|
| 1113 |
+
"completions/clipped_ratio": 1.0,
|
| 1114 |
+
"completions/max_length": 256.0,
|
| 1115 |
+
"completions/max_terminated_length": 0.0,
|
| 1116 |
+
"completions/mean_length": 256.0,
|
| 1117 |
+
"completions/mean_terminated_length": 0.0,
|
| 1118 |
+
"completions/min_length": 256.0,
|
| 1119 |
+
"completions/min_terminated_length": 0.0,
|
| 1120 |
+
"entropy": 0.5174557566642761,
|
| 1121 |
+
"epoch": 1.525,
|
| 1122 |
+
"frac_reward_zero_std": 0.25,
|
| 1123 |
+
"grad_norm": 2.4777371883392334,
|
| 1124 |
+
"kl": 0.16321629285812378,
|
| 1125 |
+
"learning_rate": 1.5e-06,
|
| 1126 |
+
"loss": -0.01113765686750412,
|
| 1127 |
+
"num_tokens": 161064.0,
|
| 1128 |
+
"reward": 0.7816817164421082,
|
| 1129 |
+
"reward_std": 0.12676921486854553,
|
| 1130 |
+
"rewards/AnswererRewardFunction/mean": 0.7816817164421082,
|
| 1131 |
+
"rewards/AnswererRewardFunction/std": 0.12676922976970673,
|
| 1132 |
+
"step": 61,
|
| 1133 |
+
"step_time": 7.992088394999882
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"clip_ratio/high_max": 0.00390625,
|
| 1137 |
+
"clip_ratio/high_mean": 0.00390625,
|
| 1138 |
+
"clip_ratio/low_mean": 0.0009765625,
|
| 1139 |
+
"clip_ratio/low_min": 0.0009765625,
|
| 1140 |
+
"clip_ratio/region_mean": 0.0048828125,
|
| 1141 |
+
"entropy": 0.4549635648727417,
|
| 1142 |
+
"epoch": 1.55,
|
| 1143 |
+
"grad_norm": 3.0455403327941895,
|
| 1144 |
+
"kl": 0.163794606924057,
|
| 1145 |
+
"learning_rate": 1.4749999999999999e-06,
|
| 1146 |
+
"loss": -0.15500083565711975,
|
| 1147 |
+
"step": 62,
|
| 1148 |
+
"step_time": 0.11149133300000358
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"clip_ratio/high_max": 0.0,
|
| 1152 |
+
"clip_ratio/high_mean": 0.0,
|
| 1153 |
+
"clip_ratio/low_mean": 0.0029296875,
|
| 1154 |
+
"clip_ratio/low_min": 0.0029296875,
|
| 1155 |
+
"clip_ratio/region_mean": 0.0029296875,
|
| 1156 |
+
"entropy": 0.5892786979675293,
|
| 1157 |
+
"epoch": 1.575,
|
| 1158 |
+
"grad_norm": 2.1281309127807617,
|
| 1159 |
+
"kl": 0.11631283909082413,
|
| 1160 |
+
"learning_rate": 1.45e-06,
|
| 1161 |
+
"loss": 0.10522277653217316,
|
| 1162 |
+
"step": 63,
|
| 1163 |
+
"step_time": 0.11075748799999019
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"clip_ratio/high_max": 0.0,
|
| 1167 |
+
"clip_ratio/high_mean": 0.0,
|
| 1168 |
+
"clip_ratio/low_mean": 0.005859375,
|
| 1169 |
+
"clip_ratio/low_min": 0.005859375,
|
| 1170 |
+
"clip_ratio/region_mean": 0.005859375,
|
| 1171 |
+
"entropy": 0.367717981338501,
|
| 1172 |
+
"epoch": 1.6,
|
| 1173 |
+
"grad_norm": 1.2092372179031372,
|
| 1174 |
+
"kl": 0.1191803440451622,
|
| 1175 |
+
"learning_rate": 1.425e-06,
|
| 1176 |
+
"loss": 0.0626659244298935,
|
| 1177 |
+
"step": 64,
|
| 1178 |
+
"step_time": 0.11023821699996006
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"clip_ratio/high_max": 0.0,
|
| 1182 |
+
"clip_ratio/high_mean": 0.0,
|
| 1183 |
+
"clip_ratio/low_mean": 0.0009765625,
|
| 1184 |
+
"clip_ratio/low_min": 0.0009765625,
|
| 1185 |
+
"clip_ratio/region_mean": 0.0009765625,
|
| 1186 |
+
"completions/clipped_ratio": 0.9375,
|
| 1187 |
+
"completions/max_length": 256.0,
|
| 1188 |
+
"completions/max_terminated_length": 256.0,
|
| 1189 |
+
"completions/mean_length": 256.0,
|
| 1190 |
+
"completions/mean_terminated_length": 256.0,
|
| 1191 |
+
"completions/min_length": 256.0,
|
| 1192 |
+
"completions/min_terminated_length": 256.0,
|
| 1193 |
+
"entropy": 0.44592803716659546,
|
| 1194 |
+
"epoch": 1.625,
|
| 1195 |
+
"frac_reward_zero_std": 0.25,
|
| 1196 |
+
"grad_norm": 3.688664674758911,
|
| 1197 |
+
"kl": 0.18022018671035767,
|
| 1198 |
+
"learning_rate": 1.4000000000000001e-06,
|
| 1199 |
+
"loss": 0.177067369222641,
|
| 1200 |
+
"num_tokens": 171260.0,
|
| 1201 |
+
"reward": 0.8617823123931885,
|
| 1202 |
+
"reward_std": 0.13122481107711792,
|
| 1203 |
+
"rewards/AnswererRewardFunction/mean": 0.8617823123931885,
|
| 1204 |
+
"rewards/AnswererRewardFunction/std": 0.1312248259782791,
|
| 1205 |
+
"step": 65,
|
| 1206 |
+
"step_time": 8.018060397999989
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"clip_ratio/high_max": 0.0,
|
| 1210 |
+
"clip_ratio/high_mean": 0.0,
|
| 1211 |
+
"clip_ratio/low_mean": 0.0,
|
| 1212 |
+
"clip_ratio/low_min": 0.0,
|
| 1213 |
+
"clip_ratio/region_mean": 0.0,
|
| 1214 |
+
"entropy": 0.3555079400539398,
|
| 1215 |
+
"epoch": 1.65,
|
| 1216 |
+
"grad_norm": 1.4989449977874756,
|
| 1217 |
+
"kl": 0.18531376123428345,
|
| 1218 |
+
"learning_rate": 1.375e-06,
|
| 1219 |
+
"loss": 0.010452479124069214,
|
| 1220 |
+
"step": 66,
|
| 1221 |
+
"step_time": 0.1116537860000335
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"clip_ratio/high_max": 0.0048828125,
|
| 1225 |
+
"clip_ratio/high_mean": 0.0048828125,
|
| 1226 |
+
"clip_ratio/low_mean": 0.001953125,
|
| 1227 |
+
"clip_ratio/low_min": 0.001953125,
|
| 1228 |
+
"clip_ratio/region_mean": 0.0068359375,
|
| 1229 |
+
"entropy": 0.4429858326911926,
|
| 1230 |
+
"epoch": 1.675,
|
| 1231 |
+
"grad_norm": 2.744241237640381,
|
| 1232 |
+
"kl": 0.14088644087314606,
|
| 1233 |
+
"learning_rate": 1.35e-06,
|
| 1234 |
+
"loss": -0.1291317492723465,
|
| 1235 |
+
"step": 67,
|
| 1236 |
+
"step_time": 0.11084942999991654
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"clip_ratio/high_max": 0.0078125,
|
| 1240 |
+
"clip_ratio/high_mean": 0.0078125,
|
| 1241 |
+
"clip_ratio/low_mean": 0.0,
|
| 1242 |
+
"clip_ratio/low_min": 0.0,
|
| 1243 |
+
"clip_ratio/region_mean": 0.0078125,
|
| 1244 |
+
"entropy": 0.644950270652771,
|
| 1245 |
+
"epoch": 1.7,
|
| 1246 |
+
"grad_norm": 1.3252309560775757,
|
| 1247 |
+
"kl": 0.13620926439762115,
|
| 1248 |
+
"learning_rate": 1.325e-06,
|
| 1249 |
+
"loss": -0.05575542896986008,
|
| 1250 |
+
"step": 68,
|
| 1251 |
+
"step_time": 0.11180194899998241
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"clip_ratio/high_max": 0.0,
|
| 1255 |
+
"clip_ratio/high_mean": 0.0,
|
| 1256 |
+
"clip_ratio/low_mean": 0.0,
|
| 1257 |
+
"clip_ratio/low_min": 0.0,
|
| 1258 |
+
"clip_ratio/region_mean": 0.0,
|
| 1259 |
+
"completions/clipped_ratio": 1.0,
|
| 1260 |
+
"completions/max_length": 256.0,
|
| 1261 |
+
"completions/max_terminated_length": 0.0,
|
| 1262 |
+
"completions/mean_length": 256.0,
|
| 1263 |
+
"completions/mean_terminated_length": 0.0,
|
| 1264 |
+
"completions/min_length": 256.0,
|
| 1265 |
+
"completions/min_terminated_length": 0.0,
|
| 1266 |
+
"entropy": 0.572523295879364,
|
| 1267 |
+
"epoch": 1.725,
|
| 1268 |
+
"frac_reward_zero_std": 0.0,
|
| 1269 |
+
"grad_norm": 3.0132229328155518,
|
| 1270 |
+
"kl": 0.2324308454990387,
|
| 1271 |
+
"learning_rate": 1.3e-06,
|
| 1272 |
+
"loss": 0.0005797058693133295,
|
| 1273 |
+
"num_tokens": 181272.0,
|
| 1274 |
+
"reward": 0.8270344734191895,
|
| 1275 |
+
"reward_std": 0.1737375259399414,
|
| 1276 |
+
"rewards/AnswererRewardFunction/mean": 0.8270344734191895,
|
| 1277 |
+
"rewards/AnswererRewardFunction/std": 0.1737375408411026,
|
| 1278 |
+
"step": 69,
|
| 1279 |
+
"step_time": 8.038297763999935
|
| 1280 |
+
},
|
| 1281 |
+
{
|
| 1282 |
+
"clip_ratio/high_max": 0.0029296875,
|
| 1283 |
+
"clip_ratio/high_mean": 0.0029296875,
|
| 1284 |
+
"clip_ratio/low_mean": 0.001953125,
|
| 1285 |
+
"clip_ratio/low_min": 0.001953125,
|
| 1286 |
+
"clip_ratio/region_mean": 0.0048828125,
|
| 1287 |
+
"entropy": 0.4360680878162384,
|
| 1288 |
+
"epoch": 1.75,
|
| 1289 |
+
"grad_norm": 2.0692310333251953,
|
| 1290 |
+
"kl": 0.1413719207048416,
|
| 1291 |
+
"learning_rate": 1.275e-06,
|
| 1292 |
+
"loss": -0.09568831324577332,
|
| 1293 |
+
"step": 70,
|
| 1294 |
+
"step_time": 0.1141534529999717
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"clip_ratio/high_max": 0.0009765625,
|
| 1298 |
+
"clip_ratio/high_mean": 0.0009765625,
|
| 1299 |
+
"clip_ratio/low_mean": 0.0009765625,
|
| 1300 |
+
"clip_ratio/low_min": 0.0009765625,
|
| 1301 |
+
"clip_ratio/region_mean": 0.001953125,
|
| 1302 |
+
"entropy": 0.3167879283428192,
|
| 1303 |
+
"epoch": 1.775,
|
| 1304 |
+
"grad_norm": 2.3254780769348145,
|
| 1305 |
+
"kl": 0.12818901240825653,
|
| 1306 |
+
"learning_rate": 1.25e-06,
|
| 1307 |
+
"loss": 0.01562567427754402,
|
| 1308 |
+
"step": 71,
|
| 1309 |
+
"step_time": 0.11141504400006852
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"clip_ratio/high_max": 0.0029296875,
|
| 1313 |
+
"clip_ratio/high_mean": 0.0029296875,
|
| 1314 |
+
"clip_ratio/low_mean": 0.005859375,
|
| 1315 |
+
"clip_ratio/low_min": 0.005859375,
|
| 1316 |
+
"clip_ratio/region_mean": 0.0087890625,
|
| 1317 |
+
"entropy": 0.5078291893005371,
|
| 1318 |
+
"epoch": 1.8,
|
| 1319 |
+
"grad_norm": 2.3528542518615723,
|
| 1320 |
+
"kl": 0.12712544202804565,
|
| 1321 |
+
"learning_rate": 1.225e-06,
|
| 1322 |
+
"loss": 0.08115437626838684,
|
| 1323 |
+
"step": 72,
|
| 1324 |
+
"step_time": 0.11149820699984048
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"clip_ratio/high_max": 0.0,
|
| 1328 |
+
"clip_ratio/high_mean": 0.0,
|
| 1329 |
+
"clip_ratio/low_mean": 0.0,
|
| 1330 |
+
"clip_ratio/low_min": 0.0,
|
| 1331 |
+
"clip_ratio/region_mean": 0.0,
|
| 1332 |
+
"completions/clipped_ratio": 1.0,
|
| 1333 |
+
"completions/max_length": 256.0,
|
| 1334 |
+
"completions/max_terminated_length": 0.0,
|
| 1335 |
+
"completions/mean_length": 256.0,
|
| 1336 |
+
"completions/mean_terminated_length": 0.0,
|
| 1337 |
+
"completions/min_length": 256.0,
|
| 1338 |
+
"completions/min_terminated_length": 0.0,
|
| 1339 |
+
"entropy": 0.4892379641532898,
|
| 1340 |
+
"epoch": 1.825,
|
| 1341 |
+
"frac_reward_zero_std": 0.0,
|
| 1342 |
+
"grad_norm": 2.7544267177581787,
|
| 1343 |
+
"kl": 0.16546988487243652,
|
| 1344 |
+
"learning_rate": 1.2000000000000002e-06,
|
| 1345 |
+
"loss": 0.00041364727076143026,
|
| 1346 |
+
"num_tokens": 191280.0,
|
| 1347 |
+
"reward": 0.8263288736343384,
|
| 1348 |
+
"reward_std": 0.15687303245067596,
|
| 1349 |
+
"rewards/AnswererRewardFunction/mean": 0.8263288736343384,
|
| 1350 |
+
"rewards/AnswererRewardFunction/std": 0.15687303245067596,
|
| 1351 |
+
"step": 73,
|
| 1352 |
+
"step_time": 8.055825157000072
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"clip_ratio/high_max": 0.0048828125,
|
| 1356 |
+
"clip_ratio/high_mean": 0.0048828125,
|
| 1357 |
+
"clip_ratio/low_mean": 0.0009765625,
|
| 1358 |
+
"clip_ratio/low_min": 0.0009765625,
|
| 1359 |
+
"clip_ratio/region_mean": 0.005859375,
|
| 1360 |
+
"entropy": 0.4051170349121094,
|
| 1361 |
+
"epoch": 1.85,
|
| 1362 |
+
"grad_norm": 2.8695521354675293,
|
| 1363 |
+
"kl": 0.16272857785224915,
|
| 1364 |
+
"learning_rate": 1.175e-06,
|
| 1365 |
+
"loss": -0.06398412585258484,
|
| 1366 |
+
"step": 74,
|
| 1367 |
+
"step_time": 0.11129722400005448
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"clip_ratio/high_max": 0.0078125,
|
| 1371 |
+
"clip_ratio/high_mean": 0.0078125,
|
| 1372 |
+
"clip_ratio/low_mean": 0.0,
|
| 1373 |
+
"clip_ratio/low_min": 0.0,
|
| 1374 |
+
"clip_ratio/region_mean": 0.0078125,
|
| 1375 |
+
"entropy": 0.5494378805160522,
|
| 1376 |
+
"epoch": 1.875,
|
| 1377 |
+
"grad_norm": 1.5364563465118408,
|
| 1378 |
+
"kl": 0.18589141964912415,
|
| 1379 |
+
"learning_rate": 1.1500000000000002e-06,
|
| 1380 |
+
"loss": -0.10700538754463196,
|
| 1381 |
+
"step": 75,
|
| 1382 |
+
"step_time": 0.11156244700009665
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"clip_ratio/high_max": 0.0048828125,
|
| 1386 |
+
"clip_ratio/high_mean": 0.0048828125,
|
| 1387 |
+
"clip_ratio/low_mean": 0.00390625,
|
| 1388 |
+
"clip_ratio/low_min": 0.00390625,
|
| 1389 |
+
"clip_ratio/region_mean": 0.0087890625,
|
| 1390 |
+
"entropy": 0.5933808088302612,
|
| 1391 |
+
"epoch": 1.9,
|
| 1392 |
+
"grad_norm": 3.9189910888671875,
|
| 1393 |
+
"kl": 0.16836795210838318,
|
| 1394 |
+
"learning_rate": 1.125e-06,
|
| 1395 |
+
"loss": 0.17264717817306519,
|
| 1396 |
+
"step": 76,
|
| 1397 |
+
"step_time": 0.11082776300008845
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"clip_ratio/high_max": 0.0,
|
| 1401 |
+
"clip_ratio/high_mean": 0.0,
|
| 1402 |
+
"clip_ratio/low_mean": 0.0,
|
| 1403 |
+
"clip_ratio/low_min": 0.0,
|
| 1404 |
+
"clip_ratio/region_mean": 0.0,
|
| 1405 |
+
"completions/clipped_ratio": 1.0,
|
| 1406 |
+
"completions/max_length": 256.0,
|
| 1407 |
+
"completions/max_terminated_length": 0.0,
|
| 1408 |
+
"completions/mean_length": 256.0,
|
| 1409 |
+
"completions/mean_terminated_length": 0.0,
|
| 1410 |
+
"completions/min_length": 256.0,
|
| 1411 |
+
"completions/min_terminated_length": 0.0,
|
| 1412 |
+
"entropy": 0.641096830368042,
|
| 1413 |
+
"epoch": 1.925,
|
| 1414 |
+
"frac_reward_zero_std": 0.0,
|
| 1415 |
+
"grad_norm": 3.785203456878662,
|
| 1416 |
+
"kl": 0.2009715437889099,
|
| 1417 |
+
"learning_rate": 1.0999999999999998e-06,
|
| 1418 |
+
"loss": -0.14720585942268372,
|
| 1419 |
+
"num_tokens": 201488.0,
|
| 1420 |
+
"reward": 0.8573338985443115,
|
| 1421 |
+
"reward_std": 0.13074268400669098,
|
| 1422 |
+
"rewards/AnswererRewardFunction/mean": 0.8573338985443115,
|
| 1423 |
+
"rewards/AnswererRewardFunction/std": 0.13074268400669098,
|
| 1424 |
+
"step": 77,
|
| 1425 |
+
"step_time": 7.9694944929999565
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"clip_ratio/high_max": 0.0,
|
| 1429 |
+
"clip_ratio/high_mean": 0.0,
|
| 1430 |
+
"clip_ratio/low_mean": 0.0,
|
| 1431 |
+
"clip_ratio/low_min": 0.0,
|
| 1432 |
+
"clip_ratio/region_mean": 0.0,
|
| 1433 |
+
"entropy": 0.4241795539855957,
|
| 1434 |
+
"epoch": 1.95,
|
| 1435 |
+
"grad_norm": 2.6870434284210205,
|
| 1436 |
+
"kl": 0.15703612565994263,
|
| 1437 |
+
"learning_rate": 1.075e-06,
|
| 1438 |
+
"loss": 0.16954365372657776,
|
| 1439 |
+
"step": 78,
|
| 1440 |
+
"step_time": 0.11322125600008803
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"clip_ratio/high_max": 0.0,
|
| 1444 |
+
"clip_ratio/high_mean": 0.0,
|
| 1445 |
+
"clip_ratio/low_mean": 0.0,
|
| 1446 |
+
"clip_ratio/low_min": 0.0,
|
| 1447 |
+
"clip_ratio/region_mean": 0.0,
|
| 1448 |
+
"entropy": 0.6060743927955627,
|
| 1449 |
+
"epoch": 1.975,
|
| 1450 |
+
"grad_norm": 2.581792116165161,
|
| 1451 |
+
"kl": 0.21035408973693848,
|
| 1452 |
+
"learning_rate": 1.05e-06,
|
| 1453 |
+
"loss": -0.02019679918885231,
|
| 1454 |
+
"step": 79,
|
| 1455 |
+
"step_time": 0.11112454099998104
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"clip_ratio/high_max": 0.0029296875,
|
| 1459 |
+
"clip_ratio/high_mean": 0.0029296875,
|
| 1460 |
+
"clip_ratio/low_mean": 0.005859375,
|
| 1461 |
+
"clip_ratio/low_min": 0.005859375,
|
| 1462 |
+
"clip_ratio/region_mean": 0.0087890625,
|
| 1463 |
+
"entropy": 0.5972496271133423,
|
| 1464 |
+
"epoch": 2.0,
|
| 1465 |
+
"grad_norm": 2.94528865814209,
|
| 1466 |
+
"kl": 0.21574267745018005,
|
| 1467 |
+
"learning_rate": 1.0250000000000001e-06,
|
| 1468 |
+
"loss": 0.0022968649864196777,
|
| 1469 |
+
"step": 80,
|
| 1470 |
+
"step_time": 0.11270847400010098
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"clip_ratio/high_max": 0.001953125,
|
| 1474 |
+
"clip_ratio/high_mean": 0.001953125,
|
| 1475 |
+
"clip_ratio/low_mean": 0.0,
|
| 1476 |
+
"clip_ratio/low_min": 0.0,
|
| 1477 |
+
"clip_ratio/region_mean": 0.001953125,
|
| 1478 |
+
"completions/clipped_ratio": 1.0,
|
| 1479 |
+
"completions/max_length": 256.0,
|
| 1480 |
+
"completions/max_terminated_length": 0.0,
|
| 1481 |
+
"completions/mean_length": 256.0,
|
| 1482 |
+
"completions/mean_terminated_length": 0.0,
|
| 1483 |
+
"completions/min_length": 256.0,
|
| 1484 |
+
"completions/min_terminated_length": 0.0,
|
| 1485 |
+
"entropy": 0.7111639380455017,
|
| 1486 |
+
"epoch": 2.025,
|
| 1487 |
+
"frac_reward_zero_std": 0.0,
|
| 1488 |
+
"grad_norm": 2.9343693256378174,
|
| 1489 |
+
"kl": 0.16421130299568176,
|
| 1490 |
+
"learning_rate": 1e-06,
|
| 1491 |
+
"loss": 0.01902194321155548,
|
| 1492 |
+
"num_tokens": 211768.0,
|
| 1493 |
+
"reward": 0.8376604914665222,
|
| 1494 |
+
"reward_std": 0.17479881644248962,
|
| 1495 |
+
"rewards/AnswererRewardFunction/mean": 0.8376604914665222,
|
| 1496 |
+
"rewards/AnswererRewardFunction/std": 0.17479883134365082,
|
| 1497 |
+
"step": 81,
|
| 1498 |
+
"step_time": 7.98263659600002
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"clip_ratio/high_max": 0.001953125,
|
| 1502 |
+
"clip_ratio/high_mean": 0.001953125,
|
| 1503 |
+
"clip_ratio/low_mean": 0.0,
|
| 1504 |
+
"clip_ratio/low_min": 0.0,
|
| 1505 |
+
"clip_ratio/region_mean": 0.001953125,
|
| 1506 |
+
"entropy": 0.46728429198265076,
|
| 1507 |
+
"epoch": 2.05,
|
| 1508 |
+
"grad_norm": 2.8638997077941895,
|
| 1509 |
+
"kl": 0.1915239840745926,
|
| 1510 |
+
"learning_rate": 9.75e-07,
|
| 1511 |
+
"loss": -0.08795377612113953,
|
| 1512 |
+
"step": 82,
|
| 1513 |
+
"step_time": 0.1123695169999337
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"clip_ratio/high_max": 0.0,
|
| 1517 |
+
"clip_ratio/high_mean": 0.0,
|
| 1518 |
+
"clip_ratio/low_mean": 0.001953125,
|
| 1519 |
+
"clip_ratio/low_min": 0.001953125,
|
| 1520 |
+
"clip_ratio/region_mean": 0.001953125,
|
| 1521 |
+
"entropy": 0.43050703406333923,
|
| 1522 |
+
"epoch": 2.075,
|
| 1523 |
+
"grad_norm": 2.538461685180664,
|
| 1524 |
+
"kl": 0.15114010870456696,
|
| 1525 |
+
"learning_rate": 9.5e-07,
|
| 1526 |
+
"loss": 0.07807829231023788,
|
| 1527 |
+
"step": 83,
|
| 1528 |
+
"step_time": 0.11493326299978435
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"clip_ratio/high_max": 0.0048828125,
|
| 1532 |
+
"clip_ratio/high_mean": 0.0048828125,
|
| 1533 |
+
"clip_ratio/low_mean": 0.001953125,
|
| 1534 |
+
"clip_ratio/low_min": 0.001953125,
|
| 1535 |
+
"clip_ratio/region_mean": 0.0068359375,
|
| 1536 |
+
"entropy": 0.3443547785282135,
|
| 1537 |
+
"epoch": 2.1,
|
| 1538 |
+
"grad_norm": 1.918908953666687,
|
| 1539 |
+
"kl": 0.17275559902191162,
|
| 1540 |
+
"learning_rate": 9.25e-07,
|
| 1541 |
+
"loss": -0.006664649583399296,
|
| 1542 |
+
"step": 84,
|
| 1543 |
+
"step_time": 0.11405003400000169
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"clip_ratio/high_max": 0.0,
|
| 1547 |
+
"clip_ratio/high_mean": 0.0,
|
| 1548 |
+
"clip_ratio/low_mean": 0.0,
|
| 1549 |
+
"clip_ratio/low_min": 0.0,
|
| 1550 |
+
"clip_ratio/region_mean": 0.0,
|
| 1551 |
+
"completions/clipped_ratio": 1.0,
|
| 1552 |
+
"completions/max_length": 256.0,
|
| 1553 |
+
"completions/max_terminated_length": 0.0,
|
| 1554 |
+
"completions/mean_length": 256.0,
|
| 1555 |
+
"completions/mean_terminated_length": 0.0,
|
| 1556 |
+
"completions/min_length": 256.0,
|
| 1557 |
+
"completions/min_terminated_length": 0.0,
|
| 1558 |
+
"entropy": 0.6337037682533264,
|
| 1559 |
+
"epoch": 2.125,
|
| 1560 |
+
"frac_reward_zero_std": 0.25,
|
| 1561 |
+
"grad_norm": 3.071202039718628,
|
| 1562 |
+
"kl": 0.2190721482038498,
|
| 1563 |
+
"learning_rate": 9e-07,
|
| 1564 |
+
"loss": -0.08100530505180359,
|
| 1565 |
+
"num_tokens": 221552.0,
|
| 1566 |
+
"reward": 0.8077311515808105,
|
| 1567 |
+
"reward_std": 0.17956602573394775,
|
| 1568 |
+
"rewards/AnswererRewardFunction/mean": 0.8077311515808105,
|
| 1569 |
+
"rewards/AnswererRewardFunction/std": 0.17956602573394775,
|
| 1570 |
+
"step": 85,
|
| 1571 |
+
"step_time": 7.966403669000101
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"clip_ratio/high_max": 0.0029296875,
|
| 1575 |
+
"clip_ratio/high_mean": 0.0029296875,
|
| 1576 |
+
"clip_ratio/low_mean": 0.0009765625,
|
| 1577 |
+
"clip_ratio/low_min": 0.0009765625,
|
| 1578 |
+
"clip_ratio/region_mean": 0.00390625,
|
| 1579 |
+
"entropy": 0.6190234422683716,
|
| 1580 |
+
"epoch": 2.15,
|
| 1581 |
+
"grad_norm": 1.9067888259887695,
|
| 1582 |
+
"kl": 0.16955198347568512,
|
| 1583 |
+
"learning_rate": 8.750000000000001e-07,
|
| 1584 |
+
"loss": -0.004910931922495365,
|
| 1585 |
+
"step": 86,
|
| 1586 |
+
"step_time": 0.11190421799983596
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"clip_ratio/high_max": 0.0009765625,
|
| 1590 |
+
"clip_ratio/high_mean": 0.0009765625,
|
| 1591 |
+
"clip_ratio/low_mean": 0.001953125,
|
| 1592 |
+
"clip_ratio/low_min": 0.001953125,
|
| 1593 |
+
"clip_ratio/region_mean": 0.0029296875,
|
| 1594 |
+
"entropy": 0.4382447600364685,
|
| 1595 |
+
"epoch": 2.175,
|
| 1596 |
+
"grad_norm": 3.9881021976470947,
|
| 1597 |
+
"kl": 0.23151037096977234,
|
| 1598 |
+
"learning_rate": 8.5e-07,
|
| 1599 |
+
"loss": 0.02522715926170349,
|
| 1600 |
+
"step": 87,
|
| 1601 |
+
"step_time": 0.11299947899988183
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"clip_ratio/high_max": 0.0048828125,
|
| 1605 |
+
"clip_ratio/high_mean": 0.0048828125,
|
| 1606 |
+
"clip_ratio/low_mean": 0.00390625,
|
| 1607 |
+
"clip_ratio/low_min": 0.00390625,
|
| 1608 |
+
"clip_ratio/region_mean": 0.0087890625,
|
| 1609 |
+
"entropy": 0.7477205395698547,
|
| 1610 |
+
"epoch": 2.2,
|
| 1611 |
+
"grad_norm": 1.7891472578048706,
|
| 1612 |
+
"kl": 0.20888686180114746,
|
| 1613 |
+
"learning_rate": 8.25e-07,
|
| 1614 |
+
"loss": 0.0635417029261589,
|
| 1615 |
+
"step": 88,
|
| 1616 |
+
"step_time": 0.11057215900018491
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"clip_ratio/high_max": 0.0,
|
| 1620 |
+
"clip_ratio/high_mean": 0.0,
|
| 1621 |
+
"clip_ratio/low_mean": 0.0,
|
| 1622 |
+
"clip_ratio/low_min": 0.0,
|
| 1623 |
+
"clip_ratio/region_mean": 0.0,
|
| 1624 |
+
"completions/clipped_ratio": 1.0,
|
| 1625 |
+
"completions/max_length": 256.0,
|
| 1626 |
+
"completions/max_terminated_length": 0.0,
|
| 1627 |
+
"completions/mean_length": 256.0,
|
| 1628 |
+
"completions/mean_terminated_length": 0.0,
|
| 1629 |
+
"completions/min_length": 256.0,
|
| 1630 |
+
"completions/min_terminated_length": 0.0,
|
| 1631 |
+
"entropy": 0.46236395835876465,
|
| 1632 |
+
"epoch": 2.225,
|
| 1633 |
+
"frac_reward_zero_std": 0.0,
|
| 1634 |
+
"grad_norm": 2.330125093460083,
|
| 1635 |
+
"kl": 0.25636792182922363,
|
| 1636 |
+
"learning_rate": 8e-07,
|
| 1637 |
+
"loss": 0.09635935723781586,
|
| 1638 |
+
"num_tokens": 231308.0,
|
| 1639 |
+
"reward": 0.7483423948287964,
|
| 1640 |
+
"reward_std": 0.11571943759918213,
|
| 1641 |
+
"rewards/AnswererRewardFunction/mean": 0.7483423948287964,
|
| 1642 |
+
"rewards/AnswererRewardFunction/std": 0.11571943759918213,
|
| 1643 |
+
"step": 89,
|
| 1644 |
+
"step_time": 7.988306075999844
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"clip_ratio/high_max": 0.0029296875,
|
| 1648 |
+
"clip_ratio/high_mean": 0.0029296875,
|
| 1649 |
+
"clip_ratio/low_mean": 0.0009765625,
|
| 1650 |
+
"clip_ratio/low_min": 0.0009765625,
|
| 1651 |
+
"clip_ratio/region_mean": 0.00390625,
|
| 1652 |
+
"entropy": 0.495961993932724,
|
| 1653 |
+
"epoch": 2.25,
|
| 1654 |
+
"grad_norm": 2.54791522026062,
|
| 1655 |
+
"kl": 0.17650391161441803,
|
| 1656 |
+
"learning_rate": 7.750000000000001e-07,
|
| 1657 |
+
"loss": -0.021507274359464645,
|
| 1658 |
+
"step": 90,
|
| 1659 |
+
"step_time": 0.11082631299996137
|
| 1660 |
+
}
|
| 1661 |
+
],
|
| 1662 |
+
"logging_steps": 1,
|
| 1663 |
+
"max_steps": 120,
|
| 1664 |
+
"num_input_tokens_seen": 231308,
|
| 1665 |
+
"num_train_epochs": 3,
|
| 1666 |
+
"save_steps": 30,
|
| 1667 |
+
"stateful_callbacks": {
|
| 1668 |
+
"TrainerControl": {
|
| 1669 |
+
"args": {
|
| 1670 |
+
"should_epoch_stop": false,
|
| 1671 |
+
"should_evaluate": false,
|
| 1672 |
+
"should_log": false,
|
| 1673 |
+
"should_save": true,
|
| 1674 |
+
"should_training_stop": false
|
| 1675 |
+
},
|
| 1676 |
+
"attributes": {}
|
| 1677 |
+
}
|
| 1678 |
+
},
|
| 1679 |
+
"total_flos": 0.0,
|
| 1680 |
+
"train_batch_size": 4,
|
| 1681 |
+
"trial_name": null,
|
| 1682 |
+
"trial_params": null
|
| 1683 |
+
}
|
self_play_hf_l40s_full/round_001/answerer_train/checkpoint-90/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a9aa1741e484209ae738c1b939926d2de381bea548a0b79516d821f11e36862f
|
| 3 |
+
size 7249
|