Intermediate checkpoint upload step=120 (answerer_train)
Browse files- .gitattributes +1 -0
- self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/chat_template.jinja +54 -0
- self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/config.json +57 -0
- self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/generation_config.json +13 -0
- self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/model.safetensors +3 -0
- self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/optimizer.pt +3 -0
- self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/rng_state.pth +3 -0
- self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/scheduler.pt +3 -0
- self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/tokenizer.json +3 -0
- self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/tokenizer_config.json +32 -0
- self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/trainer_state.json +2224 -0
- self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/training_args.bin +3 -0
.gitattributes
CHANGED
|
@@ -100,3 +100,4 @@ self_play_hf_l40s_full/round_006/generator_train/final_model/tokenizer.json filt
|
|
| 100 |
self_play_hf_l40s_full/round_006/answerer_train/checkpoint-30/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 101 |
self_play_hf_l40s_full/round_006/answerer_train/checkpoint-60/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 102 |
self_play_hf_l40s_full/round_006/answerer_train/checkpoint-90/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 100 |
self_play_hf_l40s_full/round_006/answerer_train/checkpoint-30/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 101 |
self_play_hf_l40s_full/round_006/answerer_train/checkpoint-60/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 102 |
self_play_hf_l40s_full/round_006/answerer_train/checkpoint-90/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 103 |
+
self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/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_006/answerer_train/checkpoint-120/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_006/answerer_train/checkpoint-120/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_006/answerer_train/checkpoint-120/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:de315cc60d389929358a9b10a4939096f0b84a4c9824dd2ff53456f326606fc5
|
| 3 |
+
size 1976163472
|
self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0ed8c2705d58361d774cbc5dd32dd77dbbe7162046c2a7a5c4238b461b10a334
|
| 3 |
+
size 3952509771
|
self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:09d387fab291cf55ab5d51fbc665b5b617f39b612148cbfb2c8bec740f61e07b
|
| 3 |
+
size 14645
|
self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7278566d3e343c7c551e937aa2102c1aa1c1b0a72103eeaeaa73c012e20ea6df
|
| 3 |
+
size 1465
|
self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/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_006/answerer_train/checkpoint-120/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_006/answerer_train/checkpoint-120/trainer_state.json
ADDED
|
@@ -0,0 +1,2224 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 3.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 120,
|
| 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": 2.1889684200286865,
|
| 26 |
+
"epoch": 0.025,
|
| 27 |
+
"frac_reward_zero_std": 0.25,
|
| 28 |
+
"grad_norm": 1.8947820663452148,
|
| 29 |
+
"kl": 0.0,
|
| 30 |
+
"learning_rate": 3e-06,
|
| 31 |
+
"loss": -0.12490925192832947,
|
| 32 |
+
"num_tokens": 9876.0,
|
| 33 |
+
"reward": 0.6893415451049805,
|
| 34 |
+
"reward_std": 0.16361689567565918,
|
| 35 |
+
"rewards/AnswererRewardFunction/mean": 0.6893415451049805,
|
| 36 |
+
"rewards/AnswererRewardFunction/std": 0.16361691057682037,
|
| 37 |
+
"step": 1,
|
| 38 |
+
"step_time": 7.946243045999836
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"clip_ratio/high_max": 0.021484375,
|
| 42 |
+
"clip_ratio/high_mean": 0.021484375,
|
| 43 |
+
"clip_ratio/low_mean": 0.0009765625,
|
| 44 |
+
"clip_ratio/low_min": 0.0009765625,
|
| 45 |
+
"clip_ratio/region_mean": 0.0224609375,
|
| 46 |
+
"entropy": 1.8370153903961182,
|
| 47 |
+
"epoch": 0.05,
|
| 48 |
+
"grad_norm": 4.110031604766846,
|
| 49 |
+
"kl": 0.6599398851394653,
|
| 50 |
+
"learning_rate": 2.9750000000000003e-06,
|
| 51 |
+
"loss": 0.0643470287322998,
|
| 52 |
+
"step": 2,
|
| 53 |
+
"step_time": 0.11823628799993458
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"clip_ratio/high_max": 0.0595703125,
|
| 57 |
+
"clip_ratio/high_mean": 0.0595703125,
|
| 58 |
+
"clip_ratio/low_mean": 0.0263671875,
|
| 59 |
+
"clip_ratio/low_min": 0.0263671875,
|
| 60 |
+
"clip_ratio/region_mean": 0.0859375,
|
| 61 |
+
"entropy": 2.297727584838867,
|
| 62 |
+
"epoch": 0.075,
|
| 63 |
+
"grad_norm": 296.2247314453125,
|
| 64 |
+
"kl": 79.00199890136719,
|
| 65 |
+
"learning_rate": 2.9499999999999997e-06,
|
| 66 |
+
"loss": 0.23638710379600525,
|
| 67 |
+
"step": 3,
|
| 68 |
+
"step_time": 0.11235964000024978
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"clip_ratio/high_max": 0.05859375,
|
| 72 |
+
"clip_ratio/high_mean": 0.05859375,
|
| 73 |
+
"clip_ratio/low_mean": 0.087890625,
|
| 74 |
+
"clip_ratio/low_min": 0.087890625,
|
| 75 |
+
"clip_ratio/region_mean": 0.146484375,
|
| 76 |
+
"entropy": 1.4839229583740234,
|
| 77 |
+
"epoch": 0.1,
|
| 78 |
+
"grad_norm": 10.855915069580078,
|
| 79 |
+
"kl": 3.744659900665283,
|
| 80 |
+
"learning_rate": 2.925e-06,
|
| 81 |
+
"loss": 0.0730188637971878,
|
| 82 |
+
"step": 4,
|
| 83 |
+
"step_time": 0.11265637799988326
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"clip_ratio/high_max": 0.0,
|
| 87 |
+
"clip_ratio/high_mean": 0.0,
|
| 88 |
+
"clip_ratio/low_mean": 0.0,
|
| 89 |
+
"clip_ratio/low_min": 0.0,
|
| 90 |
+
"clip_ratio/region_mean": 0.0,
|
| 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": 0.6120401620864868,
|
| 99 |
+
"epoch": 0.125,
|
| 100 |
+
"frac_reward_zero_std": 0.75,
|
| 101 |
+
"grad_norm": 0.009019882418215275,
|
| 102 |
+
"kl": 0.01311470940709114,
|
| 103 |
+
"learning_rate": 2.9e-06,
|
| 104 |
+
"loss": 3.278677468188107e-05,
|
| 105 |
+
"num_tokens": 19608.0,
|
| 106 |
+
"reward": 0.457000195980072,
|
| 107 |
+
"reward_std": 0.07295458763837814,
|
| 108 |
+
"rewards/AnswererRewardFunction/mean": 0.457000195980072,
|
| 109 |
+
"rewards/AnswererRewardFunction/std": 0.07295458763837814,
|
| 110 |
+
"step": 5,
|
| 111 |
+
"step_time": 7.926103121999404
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"clip_ratio/high_max": 0.0,
|
| 115 |
+
"clip_ratio/high_mean": 0.0,
|
| 116 |
+
"clip_ratio/low_mean": 0.0,
|
| 117 |
+
"clip_ratio/low_min": 0.0,
|
| 118 |
+
"clip_ratio/region_mean": 0.0,
|
| 119 |
+
"entropy": 1.1489129066467285,
|
| 120 |
+
"epoch": 0.15,
|
| 121 |
+
"grad_norm": 0.015938131138682365,
|
| 122 |
+
"kl": 0.030105752870440483,
|
| 123 |
+
"learning_rate": 2.875e-06,
|
| 124 |
+
"loss": 7.526438275817782e-05,
|
| 125 |
+
"step": 6,
|
| 126 |
+
"step_time": 0.11689895999916189
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"clip_ratio/high_max": 0.005859375,
|
| 130 |
+
"clip_ratio/high_mean": 0.005859375,
|
| 131 |
+
"clip_ratio/low_mean": 0.0048828125,
|
| 132 |
+
"clip_ratio/low_min": 0.0048828125,
|
| 133 |
+
"clip_ratio/region_mean": 0.0107421875,
|
| 134 |
+
"entropy": 0.6303037405014038,
|
| 135 |
+
"epoch": 0.175,
|
| 136 |
+
"grad_norm": 3.0208628177642822,
|
| 137 |
+
"kl": 0.02866116538643837,
|
| 138 |
+
"learning_rate": 2.85e-06,
|
| 139 |
+
"loss": -0.028981417417526245,
|
| 140 |
+
"step": 7,
|
| 141 |
+
"step_time": 0.10871063499962474
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"clip_ratio/high_max": 0.0,
|
| 145 |
+
"clip_ratio/high_mean": 0.0,
|
| 146 |
+
"clip_ratio/low_mean": 0.005859375,
|
| 147 |
+
"clip_ratio/low_min": 0.005859375,
|
| 148 |
+
"clip_ratio/region_mean": 0.005859375,
|
| 149 |
+
"entropy": 2.0413095951080322,
|
| 150 |
+
"epoch": 0.2,
|
| 151 |
+
"grad_norm": 1.0012094974517822,
|
| 152 |
+
"kl": 0.05237983167171478,
|
| 153 |
+
"learning_rate": 2.825e-06,
|
| 154 |
+
"loss": 0.03189055621623993,
|
| 155 |
+
"step": 8,
|
| 156 |
+
"step_time": 0.10796216900052968
|
| 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": 2.2237331867218018,
|
| 172 |
+
"epoch": 0.225,
|
| 173 |
+
"frac_reward_zero_std": 0.5,
|
| 174 |
+
"grad_norm": 1.620787501335144,
|
| 175 |
+
"kl": 0.0605711005628109,
|
| 176 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 177 |
+
"loss": 0.02867729589343071,
|
| 178 |
+
"num_tokens": 29436.0,
|
| 179 |
+
"reward": 0.7435240745544434,
|
| 180 |
+
"reward_std": 0.23860794305801392,
|
| 181 |
+
"rewards/AnswererRewardFunction/mean": 0.7435240745544434,
|
| 182 |
+
"rewards/AnswererRewardFunction/std": 0.2386079579591751,
|
| 183 |
+
"step": 9,
|
| 184 |
+
"step_time": 7.88594858200031
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"clip_ratio/high_max": 0.001953125,
|
| 188 |
+
"clip_ratio/high_mean": 0.001953125,
|
| 189 |
+
"clip_ratio/low_mean": 0.0009765625,
|
| 190 |
+
"clip_ratio/low_min": 0.0009765625,
|
| 191 |
+
"clip_ratio/region_mean": 0.0029296875,
|
| 192 |
+
"entropy": 2.133094310760498,
|
| 193 |
+
"epoch": 0.25,
|
| 194 |
+
"grad_norm": 3.653198003768921,
|
| 195 |
+
"kl": 0.07492633908987045,
|
| 196 |
+
"learning_rate": 2.775e-06,
|
| 197 |
+
"loss": 0.06285852193832397,
|
| 198 |
+
"step": 10,
|
| 199 |
+
"step_time": 0.11135919299977104
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"clip_ratio/high_max": 0.0078125,
|
| 203 |
+
"clip_ratio/high_mean": 0.0078125,
|
| 204 |
+
"clip_ratio/low_mean": 0.0087890625,
|
| 205 |
+
"clip_ratio/low_min": 0.0087890625,
|
| 206 |
+
"clip_ratio/region_mean": 0.0166015625,
|
| 207 |
+
"entropy": 1.841313362121582,
|
| 208 |
+
"epoch": 0.275,
|
| 209 |
+
"grad_norm": 2.8710782527923584,
|
| 210 |
+
"kl": 0.05763581395149231,
|
| 211 |
+
"learning_rate": 2.75e-06,
|
| 212 |
+
"loss": -0.06052658334374428,
|
| 213 |
+
"step": 11,
|
| 214 |
+
"step_time": 0.11127804100033245
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"clip_ratio/high_max": 0.0146484375,
|
| 218 |
+
"clip_ratio/high_mean": 0.0146484375,
|
| 219 |
+
"clip_ratio/low_mean": 0.0,
|
| 220 |
+
"clip_ratio/low_min": 0.0,
|
| 221 |
+
"clip_ratio/region_mean": 0.0146484375,
|
| 222 |
+
"entropy": 1.7089946269989014,
|
| 223 |
+
"epoch": 0.3,
|
| 224 |
+
"grad_norm": 0.9450397491455078,
|
| 225 |
+
"kl": 0.25390055775642395,
|
| 226 |
+
"learning_rate": 2.725e-06,
|
| 227 |
+
"loss": -0.02993064746260643,
|
| 228 |
+
"step": 12,
|
| 229 |
+
"step_time": 0.11213281900018046
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"clip_ratio/high_max": 0.0,
|
| 233 |
+
"clip_ratio/high_mean": 0.0,
|
| 234 |
+
"clip_ratio/low_mean": 0.0,
|
| 235 |
+
"clip_ratio/low_min": 0.0,
|
| 236 |
+
"clip_ratio/region_mean": 0.0,
|
| 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": 2.0214595794677734,
|
| 245 |
+
"epoch": 0.325,
|
| 246 |
+
"frac_reward_zero_std": 0.5,
|
| 247 |
+
"grad_norm": 0.8579655289649963,
|
| 248 |
+
"kl": 0.08672940731048584,
|
| 249 |
+
"learning_rate": 2.7e-06,
|
| 250 |
+
"loss": -0.031010489910840988,
|
| 251 |
+
"num_tokens": 39540.0,
|
| 252 |
+
"reward": 0.8283045291900635,
|
| 253 |
+
"reward_std": 0.18679988384246826,
|
| 254 |
+
"rewards/AnswererRewardFunction/mean": 0.8283045291900635,
|
| 255 |
+
"rewards/AnswererRewardFunction/std": 0.18679988384246826,
|
| 256 |
+
"step": 13,
|
| 257 |
+
"step_time": 7.924993149999864
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"clip_ratio/high_max": 0.0009765625,
|
| 261 |
+
"clip_ratio/high_mean": 0.0009765625,
|
| 262 |
+
"clip_ratio/low_mean": 0.0,
|
| 263 |
+
"clip_ratio/low_min": 0.0,
|
| 264 |
+
"clip_ratio/region_mean": 0.0009765625,
|
| 265 |
+
"entropy": 2.297172784805298,
|
| 266 |
+
"epoch": 0.35,
|
| 267 |
+
"grad_norm": 1.3188775777816772,
|
| 268 |
+
"kl": 0.13232484459877014,
|
| 269 |
+
"learning_rate": 2.6750000000000002e-06,
|
| 270 |
+
"loss": -0.06186162680387497,
|
| 271 |
+
"step": 14,
|
| 272 |
+
"step_time": 0.11182027799986827
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"clip_ratio/high_max": 0.0068359375,
|
| 276 |
+
"clip_ratio/high_mean": 0.0068359375,
|
| 277 |
+
"clip_ratio/low_mean": 0.0341796875,
|
| 278 |
+
"clip_ratio/low_min": 0.0341796875,
|
| 279 |
+
"clip_ratio/region_mean": 0.041015625,
|
| 280 |
+
"entropy": 1.809241771697998,
|
| 281 |
+
"epoch": 0.375,
|
| 282 |
+
"grad_norm": 2.606775999069214,
|
| 283 |
+
"kl": 0.1521003544330597,
|
| 284 |
+
"learning_rate": 2.65e-06,
|
| 285 |
+
"loss": 0.06266739964485168,
|
| 286 |
+
"step": 15,
|
| 287 |
+
"step_time": 0.11220332800075994
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"clip_ratio/high_max": 0.021484375,
|
| 291 |
+
"clip_ratio/high_mean": 0.021484375,
|
| 292 |
+
"clip_ratio/low_mean": 0.0,
|
| 293 |
+
"clip_ratio/low_min": 0.0,
|
| 294 |
+
"clip_ratio/region_mean": 0.021484375,
|
| 295 |
+
"entropy": 1.8247143030166626,
|
| 296 |
+
"epoch": 0.4,
|
| 297 |
+
"grad_norm": 3.210848808288574,
|
| 298 |
+
"kl": 0.1755138635635376,
|
| 299 |
+
"learning_rate": 2.6250000000000003e-06,
|
| 300 |
+
"loss": 0.03220568597316742,
|
| 301 |
+
"step": 16,
|
| 302 |
+
"step_time": 0.11238660199978767
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"clip_ratio/high_max": 0.0009765625,
|
| 306 |
+
"clip_ratio/high_mean": 0.0009765625,
|
| 307 |
+
"clip_ratio/low_mean": 0.0,
|
| 308 |
+
"clip_ratio/low_min": 0.0,
|
| 309 |
+
"clip_ratio/region_mean": 0.0009765625,
|
| 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": 2.2773776054382324,
|
| 318 |
+
"epoch": 0.425,
|
| 319 |
+
"frac_reward_zero_std": 0.5,
|
| 320 |
+
"grad_norm": 0.971729040145874,
|
| 321 |
+
"kl": 0.12777887284755707,
|
| 322 |
+
"learning_rate": 2.6e-06,
|
| 323 |
+
"loss": -0.030927438288927078,
|
| 324 |
+
"num_tokens": 49856.0,
|
| 325 |
+
"reward": 0.8496553301811218,
|
| 326 |
+
"reward_std": 0.16016095876693726,
|
| 327 |
+
"rewards/AnswererRewardFunction/mean": 0.8496553301811218,
|
| 328 |
+
"rewards/AnswererRewardFunction/std": 0.16016097366809845,
|
| 329 |
+
"step": 17,
|
| 330 |
+
"step_time": 7.993629739000426
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"clip_ratio/high_max": 0.0,
|
| 334 |
+
"clip_ratio/high_mean": 0.0,
|
| 335 |
+
"clip_ratio/low_mean": 0.001953125,
|
| 336 |
+
"clip_ratio/low_min": 0.001953125,
|
| 337 |
+
"clip_ratio/region_mean": 0.001953125,
|
| 338 |
+
"entropy": 2.0068607330322266,
|
| 339 |
+
"epoch": 0.45,
|
| 340 |
+
"grad_norm": 2.8351244926452637,
|
| 341 |
+
"kl": 0.15364348888397217,
|
| 342 |
+
"learning_rate": 2.575e-06,
|
| 343 |
+
"loss": 0.15774129331111908,
|
| 344 |
+
"step": 18,
|
| 345 |
+
"step_time": 0.11176471900034812
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"clip_ratio/high_max": 0.0029296875,
|
| 349 |
+
"clip_ratio/high_mean": 0.0029296875,
|
| 350 |
+
"clip_ratio/low_mean": 0.0,
|
| 351 |
+
"clip_ratio/low_min": 0.0,
|
| 352 |
+
"clip_ratio/region_mean": 0.0029296875,
|
| 353 |
+
"entropy": 2.258239269256592,
|
| 354 |
+
"epoch": 0.475,
|
| 355 |
+
"grad_norm": 1.3264741897583008,
|
| 356 |
+
"kl": 0.15728481113910675,
|
| 357 |
+
"learning_rate": 2.55e-06,
|
| 358 |
+
"loss": -0.061835240572690964,
|
| 359 |
+
"step": 19,
|
| 360 |
+
"step_time": 0.11123546799990436
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"clip_ratio/high_max": 0.0029296875,
|
| 364 |
+
"clip_ratio/high_mean": 0.0029296875,
|
| 365 |
+
"clip_ratio/low_mean": 0.0068359375,
|
| 366 |
+
"clip_ratio/low_min": 0.0068359375,
|
| 367 |
+
"clip_ratio/region_mean": 0.009765625,
|
| 368 |
+
"entropy": 2.259558916091919,
|
| 369 |
+
"epoch": 0.5,
|
| 370 |
+
"grad_norm": 2.8228580951690674,
|
| 371 |
+
"kl": 0.17330552637577057,
|
| 372 |
+
"learning_rate": 2.525e-06,
|
| 373 |
+
"loss": -0.06122318655252457,
|
| 374 |
+
"step": 20,
|
| 375 |
+
"step_time": 0.1108333300007871
|
| 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": 2.179208755493164,
|
| 391 |
+
"epoch": 0.525,
|
| 392 |
+
"frac_reward_zero_std": 1.0,
|
| 393 |
+
"grad_norm": 0.08803759515285492,
|
| 394 |
+
"kl": 0.12282326817512512,
|
| 395 |
+
"learning_rate": 2.5e-06,
|
| 396 |
+
"loss": 0.0003070581587962806,
|
| 397 |
+
"num_tokens": 59948.0,
|
| 398 |
+
"reward": 0.8848026990890503,
|
| 399 |
+
"reward_std": 0.08498957008123398,
|
| 400 |
+
"rewards/AnswererRewardFunction/mean": 0.8848026990890503,
|
| 401 |
+
"rewards/AnswererRewardFunction/std": 0.08498956263065338,
|
| 402 |
+
"step": 21,
|
| 403 |
+
"step_time": 7.9203589380003905
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"clip_ratio/high_max": 0.0,
|
| 407 |
+
"clip_ratio/high_mean": 0.0,
|
| 408 |
+
"clip_ratio/low_mean": 0.0,
|
| 409 |
+
"clip_ratio/low_min": 0.0,
|
| 410 |
+
"clip_ratio/region_mean": 0.0,
|
| 411 |
+
"entropy": 2.306586980819702,
|
| 412 |
+
"epoch": 0.55,
|
| 413 |
+
"grad_norm": 0.052290912717580795,
|
| 414 |
+
"kl": 0.16986945271492004,
|
| 415 |
+
"learning_rate": 2.475e-06,
|
| 416 |
+
"loss": 0.0004246736061759293,
|
| 417 |
+
"step": 22,
|
| 418 |
+
"step_time": 0.11205570599940984
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"clip_ratio/high_max": 0.0,
|
| 422 |
+
"clip_ratio/high_mean": 0.0,
|
| 423 |
+
"clip_ratio/low_mean": 0.0,
|
| 424 |
+
"clip_ratio/low_min": 0.0,
|
| 425 |
+
"clip_ratio/region_mean": 0.0,
|
| 426 |
+
"entropy": 2.19669508934021,
|
| 427 |
+
"epoch": 0.575,
|
| 428 |
+
"grad_norm": 0.05250624194741249,
|
| 429 |
+
"kl": 0.17336246371269226,
|
| 430 |
+
"learning_rate": 2.45e-06,
|
| 431 |
+
"loss": 0.00043340615229681134,
|
| 432 |
+
"step": 23,
|
| 433 |
+
"step_time": 0.11391655599982187
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"clip_ratio/high_max": 0.0,
|
| 437 |
+
"clip_ratio/high_mean": 0.0,
|
| 438 |
+
"clip_ratio/low_mean": 0.0,
|
| 439 |
+
"clip_ratio/low_min": 0.0,
|
| 440 |
+
"clip_ratio/region_mean": 0.0,
|
| 441 |
+
"entropy": 2.0577125549316406,
|
| 442 |
+
"epoch": 0.6,
|
| 443 |
+
"grad_norm": 0.04398365691304207,
|
| 444 |
+
"kl": 0.0968913584947586,
|
| 445 |
+
"learning_rate": 2.425e-06,
|
| 446 |
+
"loss": 0.00024222838692367077,
|
| 447 |
+
"step": 24,
|
| 448 |
+
"step_time": 0.11333441400074662
|
| 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": 2.0512969493865967,
|
| 464 |
+
"epoch": 0.625,
|
| 465 |
+
"frac_reward_zero_std": 1.0,
|
| 466 |
+
"grad_norm": 0.051768120378255844,
|
| 467 |
+
"kl": 0.16503679752349854,
|
| 468 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 469 |
+
"loss": 0.00041259199497289956,
|
| 470 |
+
"num_tokens": 69888.0,
|
| 471 |
+
"reward": 0.8755199909210205,
|
| 472 |
+
"reward_std": 0.10159505903720856,
|
| 473 |
+
"rewards/AnswererRewardFunction/mean": 0.8755199909210205,
|
| 474 |
+
"rewards/AnswererRewardFunction/std": 0.10159505903720856,
|
| 475 |
+
"step": 25,
|
| 476 |
+
"step_time": 7.957063270000617
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"clip_ratio/high_max": 0.0,
|
| 480 |
+
"clip_ratio/high_mean": 0.0,
|
| 481 |
+
"clip_ratio/low_mean": 0.0,
|
| 482 |
+
"clip_ratio/low_min": 0.0,
|
| 483 |
+
"clip_ratio/region_mean": 0.0,
|
| 484 |
+
"entropy": 2.2467222213745117,
|
| 485 |
+
"epoch": 0.65,
|
| 486 |
+
"grad_norm": 0.04632311314344406,
|
| 487 |
+
"kl": 0.13820040225982666,
|
| 488 |
+
"learning_rate": 2.375e-06,
|
| 489 |
+
"loss": 0.0003455010009929538,
|
| 490 |
+
"step": 26,
|
| 491 |
+
"step_time": 0.11183958200035704
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"clip_ratio/high_max": 0.0,
|
| 495 |
+
"clip_ratio/high_mean": 0.0,
|
| 496 |
+
"clip_ratio/low_mean": 0.0,
|
| 497 |
+
"clip_ratio/low_min": 0.0,
|
| 498 |
+
"clip_ratio/region_mean": 0.0,
|
| 499 |
+
"entropy": 2.199760913848877,
|
| 500 |
+
"epoch": 0.675,
|
| 501 |
+
"grad_norm": 0.07714560627937317,
|
| 502 |
+
"kl": 0.16415069997310638,
|
| 503 |
+
"learning_rate": 2.35e-06,
|
| 504 |
+
"loss": 0.00041037675691768527,
|
| 505 |
+
"step": 27,
|
| 506 |
+
"step_time": 0.11124441800075147
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"clip_ratio/high_max": 0.0,
|
| 510 |
+
"clip_ratio/high_mean": 0.0,
|
| 511 |
+
"clip_ratio/low_mean": 0.0,
|
| 512 |
+
"clip_ratio/low_min": 0.0,
|
| 513 |
+
"clip_ratio/region_mean": 0.0,
|
| 514 |
+
"entropy": 2.108006238937378,
|
| 515 |
+
"epoch": 0.7,
|
| 516 |
+
"grad_norm": 0.049742791801691055,
|
| 517 |
+
"kl": 0.16107158362865448,
|
| 518 |
+
"learning_rate": 2.325e-06,
|
| 519 |
+
"loss": 0.0004026789392810315,
|
| 520 |
+
"step": 28,
|
| 521 |
+
"step_time": 0.11063443700004427
|
| 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": 2.3680968284606934,
|
| 537 |
+
"epoch": 0.725,
|
| 538 |
+
"frac_reward_zero_std": 1.0,
|
| 539 |
+
"grad_norm": 0.03891065716743469,
|
| 540 |
+
"kl": 0.1308700293302536,
|
| 541 |
+
"learning_rate": 2.3000000000000004e-06,
|
| 542 |
+
"loss": 0.00032717504655011,
|
| 543 |
+
"num_tokens": 80032.0,
|
| 544 |
+
"reward": 0.9358674883842468,
|
| 545 |
+
"reward_std": 0.0063578663393855095,
|
| 546 |
+
"rewards/AnswererRewardFunction/mean": 0.9358674883842468,
|
| 547 |
+
"rewards/AnswererRewardFunction/std": 0.006357857491821051,
|
| 548 |
+
"step": 29,
|
| 549 |
+
"step_time": 7.913411731999986
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"clip_ratio/high_max": 0.0,
|
| 553 |
+
"clip_ratio/high_mean": 0.0,
|
| 554 |
+
"clip_ratio/low_mean": 0.0,
|
| 555 |
+
"clip_ratio/low_min": 0.0,
|
| 556 |
+
"clip_ratio/region_mean": 0.0,
|
| 557 |
+
"entropy": 2.180490016937256,
|
| 558 |
+
"epoch": 0.75,
|
| 559 |
+
"grad_norm": 0.039776675403118134,
|
| 560 |
+
"kl": 0.14568868279457092,
|
| 561 |
+
"learning_rate": 2.275e-06,
|
| 562 |
+
"loss": 0.00036422169068828225,
|
| 563 |
+
"step": 30,
|
| 564 |
+
"step_time": 0.11067452799943567
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"clip_ratio/high_max": 0.0,
|
| 568 |
+
"clip_ratio/high_mean": 0.0,
|
| 569 |
+
"clip_ratio/low_mean": 0.0,
|
| 570 |
+
"clip_ratio/low_min": 0.0,
|
| 571 |
+
"clip_ratio/region_mean": 0.0,
|
| 572 |
+
"entropy": 2.4302010536193848,
|
| 573 |
+
"epoch": 0.775,
|
| 574 |
+
"grad_norm": 0.051406338810920715,
|
| 575 |
+
"kl": 0.16150793433189392,
|
| 576 |
+
"learning_rate": 2.25e-06,
|
| 577 |
+
"loss": 0.0004037698090542108,
|
| 578 |
+
"step": 31,
|
| 579 |
+
"step_time": 0.11291773700031627
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"clip_ratio/high_max": 0.0,
|
| 583 |
+
"clip_ratio/high_mean": 0.0,
|
| 584 |
+
"clip_ratio/low_mean": 0.0,
|
| 585 |
+
"clip_ratio/low_min": 0.0,
|
| 586 |
+
"clip_ratio/region_mean": 0.0,
|
| 587 |
+
"entropy": 2.173776626586914,
|
| 588 |
+
"epoch": 0.8,
|
| 589 |
+
"grad_norm": 0.03287765383720398,
|
| 590 |
+
"kl": 0.131052628159523,
|
| 591 |
+
"learning_rate": 2.2250000000000003e-06,
|
| 592 |
+
"loss": 0.0003276315692346543,
|
| 593 |
+
"step": 32,
|
| 594 |
+
"step_time": 0.11095032400044147
|
| 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": 2.1301045417785645,
|
| 610 |
+
"epoch": 0.825,
|
| 611 |
+
"frac_reward_zero_std": 1.0,
|
| 612 |
+
"grad_norm": 0.07125719636678696,
|
| 613 |
+
"kl": 0.14635741710662842,
|
| 614 |
+
"learning_rate": 2.1999999999999997e-06,
|
| 615 |
+
"loss": 0.00036589353112503886,
|
| 616 |
+
"num_tokens": 90360.0,
|
| 617 |
+
"reward": 0.9323133230209351,
|
| 618 |
+
"reward_std": 0.0,
|
| 619 |
+
"rewards/AnswererRewardFunction/mean": 0.9323133230209351,
|
| 620 |
+
"rewards/AnswererRewardFunction/std": 0.0,
|
| 621 |
+
"step": 33,
|
| 622 |
+
"step_time": 7.960296397000093
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"clip_ratio/high_max": 0.0,
|
| 626 |
+
"clip_ratio/high_mean": 0.0,
|
| 627 |
+
"clip_ratio/low_mean": 0.0,
|
| 628 |
+
"clip_ratio/low_min": 0.0,
|
| 629 |
+
"clip_ratio/region_mean": 0.0,
|
| 630 |
+
"entropy": 2.0239813327789307,
|
| 631 |
+
"epoch": 0.85,
|
| 632 |
+
"grad_norm": 0.0397072434425354,
|
| 633 |
+
"kl": 0.13195659220218658,
|
| 634 |
+
"learning_rate": 2.175e-06,
|
| 635 |
+
"loss": 0.0003298914816696197,
|
| 636 |
+
"step": 34,
|
| 637 |
+
"step_time": 0.1131003039999996
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"clip_ratio/high_max": 0.0,
|
| 641 |
+
"clip_ratio/high_mean": 0.0,
|
| 642 |
+
"clip_ratio/low_mean": 0.0,
|
| 643 |
+
"clip_ratio/low_min": 0.0,
|
| 644 |
+
"clip_ratio/region_mean": 0.0,
|
| 645 |
+
"entropy": 2.308377504348755,
|
| 646 |
+
"epoch": 0.875,
|
| 647 |
+
"grad_norm": 0.040582410991191864,
|
| 648 |
+
"kl": 0.13440832495689392,
|
| 649 |
+
"learning_rate": 2.15e-06,
|
| 650 |
+
"loss": 0.00033602083567529917,
|
| 651 |
+
"step": 35,
|
| 652 |
+
"step_time": 0.15085302500028774
|
| 653 |
+
},
|
| 654 |
+
{
|
| 655 |
+
"clip_ratio/high_max": 0.0,
|
| 656 |
+
"clip_ratio/high_mean": 0.0,
|
| 657 |
+
"clip_ratio/low_mean": 0.0,
|
| 658 |
+
"clip_ratio/low_min": 0.0,
|
| 659 |
+
"clip_ratio/region_mean": 0.0,
|
| 660 |
+
"entropy": 2.1328980922698975,
|
| 661 |
+
"epoch": 0.9,
|
| 662 |
+
"grad_norm": 0.034442681819200516,
|
| 663 |
+
"kl": 0.1075764000415802,
|
| 664 |
+
"learning_rate": 2.125e-06,
|
| 665 |
+
"loss": 0.0002689409884624183,
|
| 666 |
+
"step": 36,
|
| 667 |
+
"step_time": 0.1115491030004705
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"clip_ratio/high_max": 0.0,
|
| 671 |
+
"clip_ratio/high_mean": 0.0,
|
| 672 |
+
"clip_ratio/low_mean": 0.0,
|
| 673 |
+
"clip_ratio/low_min": 0.0,
|
| 674 |
+
"clip_ratio/region_mean": 0.0,
|
| 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": 2.149538516998291,
|
| 683 |
+
"epoch": 0.925,
|
| 684 |
+
"frac_reward_zero_std": 1.0,
|
| 685 |
+
"grad_norm": 0.03783974051475525,
|
| 686 |
+
"kl": 0.1285218596458435,
|
| 687 |
+
"learning_rate": 2.1e-06,
|
| 688 |
+
"loss": 0.00032130462932400405,
|
| 689 |
+
"num_tokens": 100360.0,
|
| 690 |
+
"reward": 0.8372920751571655,
|
| 691 |
+
"reward_std": 0.09813749045133591,
|
| 692 |
+
"rewards/AnswererRewardFunction/mean": 0.8372920751571655,
|
| 693 |
+
"rewards/AnswererRewardFunction/std": 0.0981374979019165,
|
| 694 |
+
"step": 37,
|
| 695 |
+
"step_time": 7.983764245999737
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"clip_ratio/high_max": 0.0,
|
| 699 |
+
"clip_ratio/high_mean": 0.0,
|
| 700 |
+
"clip_ratio/low_mean": 0.0,
|
| 701 |
+
"clip_ratio/low_min": 0.0,
|
| 702 |
+
"clip_ratio/region_mean": 0.0,
|
| 703 |
+
"entropy": 2.095836639404297,
|
| 704 |
+
"epoch": 0.95,
|
| 705 |
+
"grad_norm": 0.03456847742199898,
|
| 706 |
+
"kl": 0.11794041097164154,
|
| 707 |
+
"learning_rate": 2.075e-06,
|
| 708 |
+
"loss": 0.000294851022772491,
|
| 709 |
+
"step": 38,
|
| 710 |
+
"step_time": 0.11436837399924116
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"clip_ratio/high_max": 0.0,
|
| 714 |
+
"clip_ratio/high_mean": 0.0,
|
| 715 |
+
"clip_ratio/low_mean": 0.0,
|
| 716 |
+
"clip_ratio/low_min": 0.0,
|
| 717 |
+
"clip_ratio/region_mean": 0.0,
|
| 718 |
+
"entropy": 2.0270957946777344,
|
| 719 |
+
"epoch": 0.975,
|
| 720 |
+
"grad_norm": 0.44049227237701416,
|
| 721 |
+
"kl": 0.21028028428554535,
|
| 722 |
+
"learning_rate": 2.0500000000000003e-06,
|
| 723 |
+
"loss": 0.0005257007433101535,
|
| 724 |
+
"step": 39,
|
| 725 |
+
"step_time": 0.11228540899992367
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"clip_ratio/high_max": 0.0,
|
| 729 |
+
"clip_ratio/high_mean": 0.0,
|
| 730 |
+
"clip_ratio/low_mean": 0.0,
|
| 731 |
+
"clip_ratio/low_min": 0.0,
|
| 732 |
+
"clip_ratio/region_mean": 0.0,
|
| 733 |
+
"entropy": 2.257455587387085,
|
| 734 |
+
"epoch": 1.0,
|
| 735 |
+
"grad_norm": 0.0388871431350708,
|
| 736 |
+
"kl": 0.13575704395771027,
|
| 737 |
+
"learning_rate": 2.025e-06,
|
| 738 |
+
"loss": 0.00033939260174520314,
|
| 739 |
+
"step": 40,
|
| 740 |
+
"step_time": 0.11100837200046954
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"clip_ratio/high_max": 0.0,
|
| 744 |
+
"clip_ratio/high_mean": 0.0,
|
| 745 |
+
"clip_ratio/low_mean": 0.0,
|
| 746 |
+
"clip_ratio/low_min": 0.0,
|
| 747 |
+
"clip_ratio/region_mean": 0.0,
|
| 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": 2.2085485458374023,
|
| 756 |
+
"epoch": 1.025,
|
| 757 |
+
"frac_reward_zero_std": 1.0,
|
| 758 |
+
"grad_norm": 0.03739944472908974,
|
| 759 |
+
"kl": 0.1360509842634201,
|
| 760 |
+
"learning_rate": 2e-06,
|
| 761 |
+
"loss": 0.0003401274443604052,
|
| 762 |
+
"num_tokens": 110016.0,
|
| 763 |
+
"reward": 0.8263232707977295,
|
| 764 |
+
"reward_std": 0.11883138865232468,
|
| 765 |
+
"rewards/AnswererRewardFunction/mean": 0.8263232707977295,
|
| 766 |
+
"rewards/AnswererRewardFunction/std": 0.11883135885000229,
|
| 767 |
+
"step": 41,
|
| 768 |
+
"step_time": 7.922437872999581
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"clip_ratio/high_max": 0.0,
|
| 772 |
+
"clip_ratio/high_mean": 0.0,
|
| 773 |
+
"clip_ratio/low_mean": 0.0,
|
| 774 |
+
"clip_ratio/low_min": 0.0,
|
| 775 |
+
"clip_ratio/region_mean": 0.0,
|
| 776 |
+
"entropy": 2.4597482681274414,
|
| 777 |
+
"epoch": 1.05,
|
| 778 |
+
"grad_norm": 0.038012683391571045,
|
| 779 |
+
"kl": 0.13071982562541962,
|
| 780 |
+
"learning_rate": 1.975e-06,
|
| 781 |
+
"loss": 0.00032679951982572675,
|
| 782 |
+
"step": 42,
|
| 783 |
+
"step_time": 0.10853852999935043
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"clip_ratio/high_max": 0.0,
|
| 787 |
+
"clip_ratio/high_mean": 0.0,
|
| 788 |
+
"clip_ratio/low_mean": 0.0,
|
| 789 |
+
"clip_ratio/low_min": 0.0,
|
| 790 |
+
"clip_ratio/region_mean": 0.0,
|
| 791 |
+
"entropy": 2.238304376602173,
|
| 792 |
+
"epoch": 1.075,
|
| 793 |
+
"grad_norm": 0.03296751156449318,
|
| 794 |
+
"kl": 0.11736348271369934,
|
| 795 |
+
"learning_rate": 1.95e-06,
|
| 796 |
+
"loss": 0.00029340869514271617,
|
| 797 |
+
"step": 43,
|
| 798 |
+
"step_time": 0.10875790699992649
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"clip_ratio/high_max": 0.0,
|
| 802 |
+
"clip_ratio/high_mean": 0.0,
|
| 803 |
+
"clip_ratio/low_mean": 0.0,
|
| 804 |
+
"clip_ratio/low_min": 0.0,
|
| 805 |
+
"clip_ratio/region_mean": 0.0,
|
| 806 |
+
"entropy": 2.2012887001037598,
|
| 807 |
+
"epoch": 1.1,
|
| 808 |
+
"grad_norm": 0.03213638812303543,
|
| 809 |
+
"kl": 0.11174218356609344,
|
| 810 |
+
"learning_rate": 1.925e-06,
|
| 811 |
+
"loss": 0.000279355444945395,
|
| 812 |
+
"step": 44,
|
| 813 |
+
"step_time": 0.10765942200032441
|
| 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": 2.246687889099121,
|
| 829 |
+
"epoch": 1.125,
|
| 830 |
+
"frac_reward_zero_std": 0.75,
|
| 831 |
+
"grad_norm": 3.2677783966064453,
|
| 832 |
+
"kl": 0.0954747349023819,
|
| 833 |
+
"learning_rate": 1.9e-06,
|
| 834 |
+
"loss": 0.06269100308418274,
|
| 835 |
+
"num_tokens": 119912.0,
|
| 836 |
+
"reward": 0.8023627996444702,
|
| 837 |
+
"reward_std": 0.11866423487663269,
|
| 838 |
+
"rewards/AnswererRewardFunction/mean": 0.8023627996444702,
|
| 839 |
+
"rewards/AnswererRewardFunction/std": 0.11866423487663269,
|
| 840 |
+
"step": 45,
|
| 841 |
+
"step_time": 7.962604862999797
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"clip_ratio/high_max": 0.001953125,
|
| 845 |
+
"clip_ratio/high_mean": 0.001953125,
|
| 846 |
+
"clip_ratio/low_mean": 0.0,
|
| 847 |
+
"clip_ratio/low_min": 0.0,
|
| 848 |
+
"clip_ratio/region_mean": 0.001953125,
|
| 849 |
+
"entropy": 2.204995632171631,
|
| 850 |
+
"epoch": 1.15,
|
| 851 |
+
"grad_norm": 0.9533382058143616,
|
| 852 |
+
"kl": 0.11488857865333557,
|
| 853 |
+
"learning_rate": 1.875e-06,
|
| 854 |
+
"loss": -0.03102012351155281,
|
| 855 |
+
"step": 46,
|
| 856 |
+
"step_time": 0.11413234800056671
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"clip_ratio/high_max": 0.0,
|
| 860 |
+
"clip_ratio/high_mean": 0.0,
|
| 861 |
+
"clip_ratio/low_mean": 0.0,
|
| 862 |
+
"clip_ratio/low_min": 0.0,
|
| 863 |
+
"clip_ratio/region_mean": 0.0,
|
| 864 |
+
"entropy": 2.193535804748535,
|
| 865 |
+
"epoch": 1.175,
|
| 866 |
+
"grad_norm": 0.028080500662326813,
|
| 867 |
+
"kl": 0.11596246063709259,
|
| 868 |
+
"learning_rate": 1.85e-06,
|
| 869 |
+
"loss": 0.0002899061073549092,
|
| 870 |
+
"step": 47,
|
| 871 |
+
"step_time": 0.1131547060003868
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"clip_ratio/high_max": 0.00390625,
|
| 875 |
+
"clip_ratio/high_mean": 0.00390625,
|
| 876 |
+
"clip_ratio/low_mean": 0.0,
|
| 877 |
+
"clip_ratio/low_min": 0.0,
|
| 878 |
+
"clip_ratio/region_mean": 0.00390625,
|
| 879 |
+
"entropy": 2.2996320724487305,
|
| 880 |
+
"epoch": 1.2,
|
| 881 |
+
"grad_norm": 0.9624528884887695,
|
| 882 |
+
"kl": 0.09764513373374939,
|
| 883 |
+
"learning_rate": 1.8249999999999999e-06,
|
| 884 |
+
"loss": -0.030984658747911453,
|
| 885 |
+
"step": 48,
|
| 886 |
+
"step_time": 0.11348181499943166
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"clip_ratio/high_max": 0.0,
|
| 890 |
+
"clip_ratio/high_mean": 0.0,
|
| 891 |
+
"clip_ratio/low_mean": 0.0,
|
| 892 |
+
"clip_ratio/low_min": 0.0,
|
| 893 |
+
"clip_ratio/region_mean": 0.0,
|
| 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": 2.1679866313934326,
|
| 902 |
+
"epoch": 1.225,
|
| 903 |
+
"frac_reward_zero_std": 1.0,
|
| 904 |
+
"grad_norm": 0.027085669338703156,
|
| 905 |
+
"kl": 0.10030621290206909,
|
| 906 |
+
"learning_rate": 1.8e-06,
|
| 907 |
+
"loss": 0.0002507655299268663,
|
| 908 |
+
"num_tokens": 129820.0,
|
| 909 |
+
"reward": 0.8280093669891357,
|
| 910 |
+
"reward_std": 0.10857454687356949,
|
| 911 |
+
"rewards/AnswererRewardFunction/mean": 0.8280093669891357,
|
| 912 |
+
"rewards/AnswererRewardFunction/std": 0.10857456177473068,
|
| 913 |
+
"step": 49,
|
| 914 |
+
"step_time": 7.983247337000648
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"clip_ratio/high_max": 0.0,
|
| 918 |
+
"clip_ratio/high_mean": 0.0,
|
| 919 |
+
"clip_ratio/low_mean": 0.0,
|
| 920 |
+
"clip_ratio/low_min": 0.0,
|
| 921 |
+
"clip_ratio/region_mean": 0.0,
|
| 922 |
+
"entropy": 2.363748073577881,
|
| 923 |
+
"epoch": 1.25,
|
| 924 |
+
"grad_norm": 0.042082529515028,
|
| 925 |
+
"kl": 0.12735603749752045,
|
| 926 |
+
"learning_rate": 1.7750000000000002e-06,
|
| 927 |
+
"loss": 0.00031839008443057537,
|
| 928 |
+
"step": 50,
|
| 929 |
+
"step_time": 0.11199663499974122
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"clip_ratio/high_max": 0.0,
|
| 933 |
+
"clip_ratio/high_mean": 0.0,
|
| 934 |
+
"clip_ratio/low_mean": 0.0,
|
| 935 |
+
"clip_ratio/low_min": 0.0,
|
| 936 |
+
"clip_ratio/region_mean": 0.0,
|
| 937 |
+
"entropy": 2.1276743412017822,
|
| 938 |
+
"epoch": 1.275,
|
| 939 |
+
"grad_norm": 0.03107348270714283,
|
| 940 |
+
"kl": 0.11797476559877396,
|
| 941 |
+
"learning_rate": 1.7500000000000002e-06,
|
| 942 |
+
"loss": 0.0002949369081761688,
|
| 943 |
+
"step": 51,
|
| 944 |
+
"step_time": 0.11148821400001907
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"clip_ratio/high_max": 0.0,
|
| 948 |
+
"clip_ratio/high_mean": 0.0,
|
| 949 |
+
"clip_ratio/low_mean": 0.0,
|
| 950 |
+
"clip_ratio/low_min": 0.0,
|
| 951 |
+
"clip_ratio/region_mean": 0.0,
|
| 952 |
+
"entropy": 2.328831672668457,
|
| 953 |
+
"epoch": 1.3,
|
| 954 |
+
"grad_norm": 0.038491472601890564,
|
| 955 |
+
"kl": 0.13311246037483215,
|
| 956 |
+
"learning_rate": 1.725e-06,
|
| 957 |
+
"loss": 0.0003327811718918383,
|
| 958 |
+
"step": 52,
|
| 959 |
+
"step_time": 0.11236872300014511
|
| 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": 2.4427859783172607,
|
| 975 |
+
"epoch": 1.325,
|
| 976 |
+
"frac_reward_zero_std": 1.0,
|
| 977 |
+
"grad_norm": 0.02960454300045967,
|
| 978 |
+
"kl": 0.10422872751951218,
|
| 979 |
+
"learning_rate": 1.7e-06,
|
| 980 |
+
"loss": 0.00026057183276861906,
|
| 981 |
+
"num_tokens": 139968.0,
|
| 982 |
+
"reward": 0.8848026990890503,
|
| 983 |
+
"reward_std": 0.08498957008123398,
|
| 984 |
+
"rewards/AnswererRewardFunction/mean": 0.8848026990890503,
|
| 985 |
+
"rewards/AnswererRewardFunction/std": 0.08498956263065338,
|
| 986 |
+
"step": 53,
|
| 987 |
+
"step_time": 7.788033633999476
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"clip_ratio/high_max": 0.0,
|
| 991 |
+
"clip_ratio/high_mean": 0.0,
|
| 992 |
+
"clip_ratio/low_mean": 0.0,
|
| 993 |
+
"clip_ratio/low_min": 0.0,
|
| 994 |
+
"clip_ratio/region_mean": 0.0,
|
| 995 |
+
"entropy": 2.5127248764038086,
|
| 996 |
+
"epoch": 1.35,
|
| 997 |
+
"grad_norm": 0.04145759716629982,
|
| 998 |
+
"kl": 0.1229250580072403,
|
| 999 |
+
"learning_rate": 1.675e-06,
|
| 1000 |
+
"loss": 0.0003073126426897943,
|
| 1001 |
+
"step": 54,
|
| 1002 |
+
"step_time": 0.11092337199988833
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"clip_ratio/high_max": 0.0,
|
| 1006 |
+
"clip_ratio/high_mean": 0.0,
|
| 1007 |
+
"clip_ratio/low_mean": 0.0,
|
| 1008 |
+
"clip_ratio/low_min": 0.0,
|
| 1009 |
+
"clip_ratio/region_mean": 0.0,
|
| 1010 |
+
"entropy": 2.483670234680176,
|
| 1011 |
+
"epoch": 1.375,
|
| 1012 |
+
"grad_norm": 0.03221055120229721,
|
| 1013 |
+
"kl": 0.11193286627531052,
|
| 1014 |
+
"learning_rate": 1.65e-06,
|
| 1015 |
+
"loss": 0.0002798321656882763,
|
| 1016 |
+
"step": 55,
|
| 1017 |
+
"step_time": 0.11110094600007869
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"clip_ratio/high_max": 0.0,
|
| 1021 |
+
"clip_ratio/high_mean": 0.0,
|
| 1022 |
+
"clip_ratio/low_mean": 0.0,
|
| 1023 |
+
"clip_ratio/low_min": 0.0,
|
| 1024 |
+
"clip_ratio/region_mean": 0.0,
|
| 1025 |
+
"entropy": 2.4600026607513428,
|
| 1026 |
+
"epoch": 1.4,
|
| 1027 |
+
"grad_norm": 0.032304368913173676,
|
| 1028 |
+
"kl": 0.1237458735704422,
|
| 1029 |
+
"learning_rate": 1.625e-06,
|
| 1030 |
+
"loss": 0.0003093646664638072,
|
| 1031 |
+
"step": 56,
|
| 1032 |
+
"step_time": 0.1109393220003767
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"clip_ratio/high_max": 0.0,
|
| 1036 |
+
"clip_ratio/high_mean": 0.0,
|
| 1037 |
+
"clip_ratio/low_mean": 0.0,
|
| 1038 |
+
"clip_ratio/low_min": 0.0,
|
| 1039 |
+
"clip_ratio/region_mean": 0.0,
|
| 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": 1.7456945180892944,
|
| 1048 |
+
"epoch": 1.425,
|
| 1049 |
+
"frac_reward_zero_std": 0.5,
|
| 1050 |
+
"grad_norm": 1.4578479528427124,
|
| 1051 |
+
"kl": 0.09936054795980453,
|
| 1052 |
+
"learning_rate": 1.6e-06,
|
| 1053 |
+
"loss": 0.031486742198467255,
|
| 1054 |
+
"num_tokens": 150328.0,
|
| 1055 |
+
"reward": 0.8651799559593201,
|
| 1056 |
+
"reward_std": 0.18344323337078094,
|
| 1057 |
+
"rewards/AnswererRewardFunction/mean": 0.8651799559593201,
|
| 1058 |
+
"rewards/AnswererRewardFunction/std": 0.18344323337078094,
|
| 1059 |
+
"step": 57,
|
| 1060 |
+
"step_time": 7.922721363000164
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"clip_ratio/high_max": 0.0,
|
| 1064 |
+
"clip_ratio/high_mean": 0.0,
|
| 1065 |
+
"clip_ratio/low_mean": 0.0009765625,
|
| 1066 |
+
"clip_ratio/low_min": 0.0009765625,
|
| 1067 |
+
"clip_ratio/region_mean": 0.0009765625,
|
| 1068 |
+
"entropy": 1.9535131454467773,
|
| 1069 |
+
"epoch": 1.45,
|
| 1070 |
+
"grad_norm": 1.1535450220108032,
|
| 1071 |
+
"kl": 0.12368227541446686,
|
| 1072 |
+
"learning_rate": 1.5750000000000002e-06,
|
| 1073 |
+
"loss": 0.06276790797710419,
|
| 1074 |
+
"step": 58,
|
| 1075 |
+
"step_time": 0.11167901599947072
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"clip_ratio/high_max": 0.005859375,
|
| 1079 |
+
"clip_ratio/high_mean": 0.005859375,
|
| 1080 |
+
"clip_ratio/low_mean": 0.0,
|
| 1081 |
+
"clip_ratio/low_min": 0.0,
|
| 1082 |
+
"clip_ratio/region_mean": 0.005859375,
|
| 1083 |
+
"entropy": 2.162888526916504,
|
| 1084 |
+
"epoch": 1.475,
|
| 1085 |
+
"grad_norm": 1.3076512813568115,
|
| 1086 |
+
"kl": 0.10642383247613907,
|
| 1087 |
+
"learning_rate": 1.5500000000000002e-06,
|
| 1088 |
+
"loss": -0.062260571867227554,
|
| 1089 |
+
"step": 59,
|
| 1090 |
+
"step_time": 0.1131628670000282
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"clip_ratio/high_max": 0.001953125,
|
| 1094 |
+
"clip_ratio/high_mean": 0.001953125,
|
| 1095 |
+
"clip_ratio/low_mean": 0.0,
|
| 1096 |
+
"clip_ratio/low_min": 0.0,
|
| 1097 |
+
"clip_ratio/region_mean": 0.001953125,
|
| 1098 |
+
"entropy": 2.2302868366241455,
|
| 1099 |
+
"epoch": 1.5,
|
| 1100 |
+
"grad_norm": 1.0681862831115723,
|
| 1101 |
+
"kl": 0.1137072965502739,
|
| 1102 |
+
"learning_rate": 1.525e-06,
|
| 1103 |
+
"loss": -0.030728042125701904,
|
| 1104 |
+
"step": 60,
|
| 1105 |
+
"step_time": 0.11138374100028159
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"clip_ratio/high_max": 0.0,
|
| 1109 |
+
"clip_ratio/high_mean": 0.0,
|
| 1110 |
+
"clip_ratio/low_mean": 0.0,
|
| 1111 |
+
"clip_ratio/low_min": 0.0,
|
| 1112 |
+
"clip_ratio/region_mean": 0.0,
|
| 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": 2.427663803100586,
|
| 1121 |
+
"epoch": 1.525,
|
| 1122 |
+
"frac_reward_zero_std": 1.0,
|
| 1123 |
+
"grad_norm": 0.03646298125386238,
|
| 1124 |
+
"kl": 0.1284782737493515,
|
| 1125 |
+
"learning_rate": 1.5e-06,
|
| 1126 |
+
"loss": 0.0003211956936866045,
|
| 1127 |
+
"num_tokens": 160432.0,
|
| 1128 |
+
"reward": 0.9358674883842468,
|
| 1129 |
+
"reward_std": 0.0063578663393855095,
|
| 1130 |
+
"rewards/AnswererRewardFunction/mean": 0.9358674883842468,
|
| 1131 |
+
"rewards/AnswererRewardFunction/std": 0.006357857491821051,
|
| 1132 |
+
"step": 61,
|
| 1133 |
+
"step_time": 7.902743857000132
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"clip_ratio/high_max": 0.0,
|
| 1137 |
+
"clip_ratio/high_mean": 0.0,
|
| 1138 |
+
"clip_ratio/low_mean": 0.0,
|
| 1139 |
+
"clip_ratio/low_min": 0.0,
|
| 1140 |
+
"clip_ratio/region_mean": 0.0,
|
| 1141 |
+
"entropy": 2.569056272506714,
|
| 1142 |
+
"epoch": 1.55,
|
| 1143 |
+
"grad_norm": 0.037735678255558014,
|
| 1144 |
+
"kl": 0.11821131408214569,
|
| 1145 |
+
"learning_rate": 1.4749999999999999e-06,
|
| 1146 |
+
"loss": 0.00029552826890721917,
|
| 1147 |
+
"step": 62,
|
| 1148 |
+
"step_time": 0.11205038600019179
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"clip_ratio/high_max": 0.0,
|
| 1152 |
+
"clip_ratio/high_mean": 0.0,
|
| 1153 |
+
"clip_ratio/low_mean": 0.0,
|
| 1154 |
+
"clip_ratio/low_min": 0.0,
|
| 1155 |
+
"clip_ratio/region_mean": 0.0,
|
| 1156 |
+
"entropy": 2.2847204208374023,
|
| 1157 |
+
"epoch": 1.575,
|
| 1158 |
+
"grad_norm": 0.04250877723097801,
|
| 1159 |
+
"kl": 0.12299974262714386,
|
| 1160 |
+
"learning_rate": 1.45e-06,
|
| 1161 |
+
"loss": 0.00030749934376217425,
|
| 1162 |
+
"step": 63,
|
| 1163 |
+
"step_time": 0.1111710889999813
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"clip_ratio/high_max": 0.0,
|
| 1167 |
+
"clip_ratio/high_mean": 0.0,
|
| 1168 |
+
"clip_ratio/low_mean": 0.0,
|
| 1169 |
+
"clip_ratio/low_min": 0.0,
|
| 1170 |
+
"clip_ratio/region_mean": 0.0,
|
| 1171 |
+
"entropy": 2.354663848876953,
|
| 1172 |
+
"epoch": 1.6,
|
| 1173 |
+
"grad_norm": 0.05598018690943718,
|
| 1174 |
+
"kl": 0.1540047526359558,
|
| 1175 |
+
"learning_rate": 1.425e-06,
|
| 1176 |
+
"loss": 0.00038501189555972815,
|
| 1177 |
+
"step": 64,
|
| 1178 |
+
"step_time": 0.11028651000015088
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"clip_ratio/high_max": 0.0,
|
| 1182 |
+
"clip_ratio/high_mean": 0.0,
|
| 1183 |
+
"clip_ratio/low_mean": 0.0,
|
| 1184 |
+
"clip_ratio/low_min": 0.0,
|
| 1185 |
+
"clip_ratio/region_mean": 0.0,
|
| 1186 |
+
"completions/clipped_ratio": 1.0,
|
| 1187 |
+
"completions/max_length": 256.0,
|
| 1188 |
+
"completions/max_terminated_length": 0.0,
|
| 1189 |
+
"completions/mean_length": 256.0,
|
| 1190 |
+
"completions/mean_terminated_length": 0.0,
|
| 1191 |
+
"completions/min_length": 256.0,
|
| 1192 |
+
"completions/min_terminated_length": 0.0,
|
| 1193 |
+
"entropy": 2.4809746742248535,
|
| 1194 |
+
"epoch": 1.625,
|
| 1195 |
+
"frac_reward_zero_std": 1.0,
|
| 1196 |
+
"grad_norm": 0.03586937487125397,
|
| 1197 |
+
"kl": 0.13512319326400757,
|
| 1198 |
+
"learning_rate": 1.4000000000000001e-06,
|
| 1199 |
+
"loss": 0.00033780798548832536,
|
| 1200 |
+
"num_tokens": 170616.0,
|
| 1201 |
+
"reward": 0.9358674883842468,
|
| 1202 |
+
"reward_std": 0.0063578663393855095,
|
| 1203 |
+
"rewards/AnswererRewardFunction/mean": 0.9358674883842468,
|
| 1204 |
+
"rewards/AnswererRewardFunction/std": 0.006357857491821051,
|
| 1205 |
+
"step": 65,
|
| 1206 |
+
"step_time": 7.88926543100024
|
| 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": 2.3049070835113525,
|
| 1215 |
+
"epoch": 1.65,
|
| 1216 |
+
"grad_norm": 0.03147316351532936,
|
| 1217 |
+
"kl": 0.12661941349506378,
|
| 1218 |
+
"learning_rate": 1.375e-06,
|
| 1219 |
+
"loss": 0.000316548568662256,
|
| 1220 |
+
"step": 66,
|
| 1221 |
+
"step_time": 0.11236034299963649
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"clip_ratio/high_max": 0.0,
|
| 1225 |
+
"clip_ratio/high_mean": 0.0,
|
| 1226 |
+
"clip_ratio/low_mean": 0.0,
|
| 1227 |
+
"clip_ratio/low_min": 0.0,
|
| 1228 |
+
"clip_ratio/region_mean": 0.0,
|
| 1229 |
+
"entropy": 2.4119033813476562,
|
| 1230 |
+
"epoch": 1.675,
|
| 1231 |
+
"grad_norm": 0.03738119453191757,
|
| 1232 |
+
"kl": 0.12834127247333527,
|
| 1233 |
+
"learning_rate": 1.35e-06,
|
| 1234 |
+
"loss": 0.00032085314160212874,
|
| 1235 |
+
"step": 67,
|
| 1236 |
+
"step_time": 0.11753063199921598
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"clip_ratio/high_max": 0.0,
|
| 1240 |
+
"clip_ratio/high_mean": 0.0,
|
| 1241 |
+
"clip_ratio/low_mean": 0.0,
|
| 1242 |
+
"clip_ratio/low_min": 0.0,
|
| 1243 |
+
"clip_ratio/region_mean": 0.0,
|
| 1244 |
+
"entropy": 2.1333839893341064,
|
| 1245 |
+
"epoch": 1.7,
|
| 1246 |
+
"grad_norm": 0.035668984055519104,
|
| 1247 |
+
"kl": 0.12106461077928543,
|
| 1248 |
+
"learning_rate": 1.325e-06,
|
| 1249 |
+
"loss": 0.0003026615595445037,
|
| 1250 |
+
"step": 68,
|
| 1251 |
+
"step_time": 0.1119794260002891
|
| 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": 2.2524616718292236,
|
| 1267 |
+
"epoch": 1.725,
|
| 1268 |
+
"frac_reward_zero_std": 1.0,
|
| 1269 |
+
"grad_norm": 0.03496238961815834,
|
| 1270 |
+
"kl": 0.122513048350811,
|
| 1271 |
+
"learning_rate": 1.3e-06,
|
| 1272 |
+
"loss": 0.0003062825999222696,
|
| 1273 |
+
"num_tokens": 180516.0,
|
| 1274 |
+
"reward": 0.7625542283058167,
|
| 1275 |
+
"reward_std": 0.10456417500972748,
|
| 1276 |
+
"rewards/AnswererRewardFunction/mean": 0.7625542283058167,
|
| 1277 |
+
"rewards/AnswererRewardFunction/std": 0.10456419736146927,
|
| 1278 |
+
"step": 69,
|
| 1279 |
+
"step_time": 7.958654460999242
|
| 1280 |
+
},
|
| 1281 |
+
{
|
| 1282 |
+
"clip_ratio/high_max": 0.0,
|
| 1283 |
+
"clip_ratio/high_mean": 0.0,
|
| 1284 |
+
"clip_ratio/low_mean": 0.0,
|
| 1285 |
+
"clip_ratio/low_min": 0.0,
|
| 1286 |
+
"clip_ratio/region_mean": 0.0,
|
| 1287 |
+
"entropy": 2.4326138496398926,
|
| 1288 |
+
"epoch": 1.75,
|
| 1289 |
+
"grad_norm": 0.035222653299570084,
|
| 1290 |
+
"kl": 0.14358815550804138,
|
| 1291 |
+
"learning_rate": 1.275e-06,
|
| 1292 |
+
"loss": 0.00035897037014365196,
|
| 1293 |
+
"step": 70,
|
| 1294 |
+
"step_time": 0.11120955699971091
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"clip_ratio/high_max": 0.0,
|
| 1298 |
+
"clip_ratio/high_mean": 0.0,
|
| 1299 |
+
"clip_ratio/low_mean": 0.0,
|
| 1300 |
+
"clip_ratio/low_min": 0.0,
|
| 1301 |
+
"clip_ratio/region_mean": 0.0,
|
| 1302 |
+
"entropy": 2.4045815467834473,
|
| 1303 |
+
"epoch": 1.775,
|
| 1304 |
+
"grad_norm": 0.033389367163181305,
|
| 1305 |
+
"kl": 0.1303025484085083,
|
| 1306 |
+
"learning_rate": 1.25e-06,
|
| 1307 |
+
"loss": 0.00032575635123066604,
|
| 1308 |
+
"step": 71,
|
| 1309 |
+
"step_time": 0.11156805400059966
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"clip_ratio/high_max": 0.0,
|
| 1313 |
+
"clip_ratio/high_mean": 0.0,
|
| 1314 |
+
"clip_ratio/low_mean": 0.0,
|
| 1315 |
+
"clip_ratio/low_min": 0.0,
|
| 1316 |
+
"clip_ratio/region_mean": 0.0,
|
| 1317 |
+
"entropy": 2.31404972076416,
|
| 1318 |
+
"epoch": 1.8,
|
| 1319 |
+
"grad_norm": 0.037959031760692596,
|
| 1320 |
+
"kl": 0.12655681371688843,
|
| 1321 |
+
"learning_rate": 1.225e-06,
|
| 1322 |
+
"loss": 0.0003163920482620597,
|
| 1323 |
+
"step": 72,
|
| 1324 |
+
"step_time": 0.11096716199972434
|
| 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": 2.3193564414978027,
|
| 1340 |
+
"epoch": 1.825,
|
| 1341 |
+
"frac_reward_zero_std": 1.0,
|
| 1342 |
+
"grad_norm": 0.031556062400341034,
|
| 1343 |
+
"kl": 0.13829217851161957,
|
| 1344 |
+
"learning_rate": 1.2000000000000002e-06,
|
| 1345 |
+
"loss": 0.00034573045559227467,
|
| 1346 |
+
"num_tokens": 190524.0,
|
| 1347 |
+
"reward": 0.8368297815322876,
|
| 1348 |
+
"reward_std": 0.1396068036556244,
|
| 1349 |
+
"rewards/AnswererRewardFunction/mean": 0.8368297815322876,
|
| 1350 |
+
"rewards/AnswererRewardFunction/std": 0.1396068036556244,
|
| 1351 |
+
"step": 73,
|
| 1352 |
+
"step_time": 8.002297633000126
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"clip_ratio/high_max": 0.0,
|
| 1356 |
+
"clip_ratio/high_mean": 0.0,
|
| 1357 |
+
"clip_ratio/low_mean": 0.0,
|
| 1358 |
+
"clip_ratio/low_min": 0.0,
|
| 1359 |
+
"clip_ratio/region_mean": 0.0,
|
| 1360 |
+
"entropy": 2.3310608863830566,
|
| 1361 |
+
"epoch": 1.85,
|
| 1362 |
+
"grad_norm": 0.03229216858744621,
|
| 1363 |
+
"kl": 0.11433228850364685,
|
| 1364 |
+
"learning_rate": 1.175e-06,
|
| 1365 |
+
"loss": 0.00028583069797605276,
|
| 1366 |
+
"step": 74,
|
| 1367 |
+
"step_time": 0.11197342300056334
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"clip_ratio/high_max": 0.0,
|
| 1371 |
+
"clip_ratio/high_mean": 0.0,
|
| 1372 |
+
"clip_ratio/low_mean": 0.0,
|
| 1373 |
+
"clip_ratio/low_min": 0.0,
|
| 1374 |
+
"clip_ratio/region_mean": 0.0,
|
| 1375 |
+
"entropy": 2.260051727294922,
|
| 1376 |
+
"epoch": 1.875,
|
| 1377 |
+
"grad_norm": 0.05061931163072586,
|
| 1378 |
+
"kl": 0.13256871700286865,
|
| 1379 |
+
"learning_rate": 1.1500000000000002e-06,
|
| 1380 |
+
"loss": 0.0003314217901788652,
|
| 1381 |
+
"step": 75,
|
| 1382 |
+
"step_time": 0.11072676599997067
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"clip_ratio/high_max": 0.0,
|
| 1386 |
+
"clip_ratio/high_mean": 0.0,
|
| 1387 |
+
"clip_ratio/low_mean": 0.0,
|
| 1388 |
+
"clip_ratio/low_min": 0.0,
|
| 1389 |
+
"clip_ratio/region_mean": 0.0,
|
| 1390 |
+
"entropy": 2.226618766784668,
|
| 1391 |
+
"epoch": 1.9,
|
| 1392 |
+
"grad_norm": 0.058189865201711655,
|
| 1393 |
+
"kl": 0.14827311038970947,
|
| 1394 |
+
"learning_rate": 1.125e-06,
|
| 1395 |
+
"loss": 0.00037068274104967713,
|
| 1396 |
+
"step": 76,
|
| 1397 |
+
"step_time": 0.11129284000071493
|
| 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": 2.2741575241088867,
|
| 1413 |
+
"epoch": 1.925,
|
| 1414 |
+
"frac_reward_zero_std": 1.0,
|
| 1415 |
+
"grad_norm": 0.03143996000289917,
|
| 1416 |
+
"kl": 0.12221985310316086,
|
| 1417 |
+
"learning_rate": 1.0999999999999998e-06,
|
| 1418 |
+
"loss": 0.0003055496490560472,
|
| 1419 |
+
"num_tokens": 200720.0,
|
| 1420 |
+
"reward": 0.8848026990890503,
|
| 1421 |
+
"reward_std": 0.08498957008123398,
|
| 1422 |
+
"rewards/AnswererRewardFunction/mean": 0.8848026990890503,
|
| 1423 |
+
"rewards/AnswererRewardFunction/std": 0.08498956263065338,
|
| 1424 |
+
"step": 77,
|
| 1425 |
+
"step_time": 7.931931125999654
|
| 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": 2.2842938899993896,
|
| 1434 |
+
"epoch": 1.95,
|
| 1435 |
+
"grad_norm": 0.03443538770079613,
|
| 1436 |
+
"kl": 0.10899876803159714,
|
| 1437 |
+
"learning_rate": 1.075e-06,
|
| 1438 |
+
"loss": 0.00027249689446762204,
|
| 1439 |
+
"step": 78,
|
| 1440 |
+
"step_time": 0.11264212100013538
|
| 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": 2.268710136413574,
|
| 1449 |
+
"epoch": 1.975,
|
| 1450 |
+
"grad_norm": 0.05908680334687233,
|
| 1451 |
+
"kl": 0.15448565781116486,
|
| 1452 |
+
"learning_rate": 1.05e-06,
|
| 1453 |
+
"loss": 0.0003862141165882349,
|
| 1454 |
+
"step": 79,
|
| 1455 |
+
"step_time": 0.11114387899942813
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"clip_ratio/high_max": 0.0,
|
| 1459 |
+
"clip_ratio/high_mean": 0.0,
|
| 1460 |
+
"clip_ratio/low_mean": 0.0,
|
| 1461 |
+
"clip_ratio/low_min": 0.0,
|
| 1462 |
+
"clip_ratio/region_mean": 0.0,
|
| 1463 |
+
"entropy": 2.3151421546936035,
|
| 1464 |
+
"epoch": 2.0,
|
| 1465 |
+
"grad_norm": 0.03168993815779686,
|
| 1466 |
+
"kl": 0.13113561272621155,
|
| 1467 |
+
"learning_rate": 1.0250000000000001e-06,
|
| 1468 |
+
"loss": 0.00032783905044198036,
|
| 1469 |
+
"step": 80,
|
| 1470 |
+
"step_time": 0.11127813099938066
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"clip_ratio/high_max": 0.0,
|
| 1474 |
+
"clip_ratio/high_mean": 0.0,
|
| 1475 |
+
"clip_ratio/low_mean": 0.0,
|
| 1476 |
+
"clip_ratio/low_min": 0.0,
|
| 1477 |
+
"clip_ratio/region_mean": 0.0,
|
| 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": 2.1372506618499756,
|
| 1486 |
+
"epoch": 2.025,
|
| 1487 |
+
"frac_reward_zero_std": 1.0,
|
| 1488 |
+
"grad_norm": 0.03289029374718666,
|
| 1489 |
+
"kl": 0.121255062520504,
|
| 1490 |
+
"learning_rate": 1e-06,
|
| 1491 |
+
"loss": 0.00030313764000311494,
|
| 1492 |
+
"num_tokens": 210928.0,
|
| 1493 |
+
"reward": 0.8014031052589417,
|
| 1494 |
+
"reward_std": 0.13520348072052002,
|
| 1495 |
+
"rewards/AnswererRewardFunction/mean": 0.8014031052589417,
|
| 1496 |
+
"rewards/AnswererRewardFunction/std": 0.1352034956216812,
|
| 1497 |
+
"step": 81,
|
| 1498 |
+
"step_time": 7.988964082999701
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"clip_ratio/high_max": 0.0,
|
| 1502 |
+
"clip_ratio/high_mean": 0.0,
|
| 1503 |
+
"clip_ratio/low_mean": 0.0,
|
| 1504 |
+
"clip_ratio/low_min": 0.0,
|
| 1505 |
+
"clip_ratio/region_mean": 0.0,
|
| 1506 |
+
"entropy": 2.3523125648498535,
|
| 1507 |
+
"epoch": 2.05,
|
| 1508 |
+
"grad_norm": 0.03441542014479637,
|
| 1509 |
+
"kl": 0.13307708501815796,
|
| 1510 |
+
"learning_rate": 9.75e-07,
|
| 1511 |
+
"loss": 0.00033269269624724984,
|
| 1512 |
+
"step": 82,
|
| 1513 |
+
"step_time": 0.11172376699960296
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"clip_ratio/high_max": 0.0,
|
| 1517 |
+
"clip_ratio/high_mean": 0.0,
|
| 1518 |
+
"clip_ratio/low_mean": 0.0,
|
| 1519 |
+
"clip_ratio/low_min": 0.0,
|
| 1520 |
+
"clip_ratio/region_mean": 0.0,
|
| 1521 |
+
"entropy": 2.42449951171875,
|
| 1522 |
+
"epoch": 2.075,
|
| 1523 |
+
"grad_norm": 0.03447401151061058,
|
| 1524 |
+
"kl": 0.1456916630268097,
|
| 1525 |
+
"learning_rate": 9.5e-07,
|
| 1526 |
+
"loss": 0.0003642291412688792,
|
| 1527 |
+
"step": 83,
|
| 1528 |
+
"step_time": 0.11197693299982348
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"clip_ratio/high_max": 0.0,
|
| 1532 |
+
"clip_ratio/high_mean": 0.0,
|
| 1533 |
+
"clip_ratio/low_mean": 0.0,
|
| 1534 |
+
"clip_ratio/low_min": 0.0,
|
| 1535 |
+
"clip_ratio/region_mean": 0.0,
|
| 1536 |
+
"entropy": 2.2901861667633057,
|
| 1537 |
+
"epoch": 2.1,
|
| 1538 |
+
"grad_norm": 0.03624199330806732,
|
| 1539 |
+
"kl": 0.12441237270832062,
|
| 1540 |
+
"learning_rate": 9.25e-07,
|
| 1541 |
+
"loss": 0.0003110308898612857,
|
| 1542 |
+
"step": 84,
|
| 1543 |
+
"step_time": 0.11186316000021179
|
| 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": 2.2687549591064453,
|
| 1559 |
+
"epoch": 2.125,
|
| 1560 |
+
"frac_reward_zero_std": 1.0,
|
| 1561 |
+
"grad_norm": 0.028875086456537247,
|
| 1562 |
+
"kl": 0.11365413665771484,
|
| 1563 |
+
"learning_rate": 9e-07,
|
| 1564 |
+
"loss": 0.0002841353416442871,
|
| 1565 |
+
"num_tokens": 220696.0,
|
| 1566 |
+
"reward": 0.8315634727478027,
|
| 1567 |
+
"reward_std": 0.11233747750520706,
|
| 1568 |
+
"rewards/AnswererRewardFunction/mean": 0.8315634727478027,
|
| 1569 |
+
"rewards/AnswererRewardFunction/std": 0.11233748495578766,
|
| 1570 |
+
"step": 85,
|
| 1571 |
+
"step_time": 7.969419599000503
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"clip_ratio/high_max": 0.0,
|
| 1575 |
+
"clip_ratio/high_mean": 0.0,
|
| 1576 |
+
"clip_ratio/low_mean": 0.0,
|
| 1577 |
+
"clip_ratio/low_min": 0.0,
|
| 1578 |
+
"clip_ratio/region_mean": 0.0,
|
| 1579 |
+
"entropy": 2.319582939147949,
|
| 1580 |
+
"epoch": 2.15,
|
| 1581 |
+
"grad_norm": 0.042500272393226624,
|
| 1582 |
+
"kl": 0.13395462930202484,
|
| 1583 |
+
"learning_rate": 8.750000000000001e-07,
|
| 1584 |
+
"loss": 0.00033488660119473934,
|
| 1585 |
+
"step": 86,
|
| 1586 |
+
"step_time": 0.11122223699931055
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"clip_ratio/high_max": 0.0,
|
| 1590 |
+
"clip_ratio/high_mean": 0.0,
|
| 1591 |
+
"clip_ratio/low_mean": 0.0,
|
| 1592 |
+
"clip_ratio/low_min": 0.0,
|
| 1593 |
+
"clip_ratio/region_mean": 0.0,
|
| 1594 |
+
"entropy": 2.251894235610962,
|
| 1595 |
+
"epoch": 2.175,
|
| 1596 |
+
"grad_norm": 0.0328298918902874,
|
| 1597 |
+
"kl": 0.09911029785871506,
|
| 1598 |
+
"learning_rate": 8.5e-07,
|
| 1599 |
+
"loss": 0.00024777575163170695,
|
| 1600 |
+
"step": 87,
|
| 1601 |
+
"step_time": 0.11061579400029586
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"clip_ratio/high_max": 0.0,
|
| 1605 |
+
"clip_ratio/high_mean": 0.0,
|
| 1606 |
+
"clip_ratio/low_mean": 0.0,
|
| 1607 |
+
"clip_ratio/low_min": 0.0,
|
| 1608 |
+
"clip_ratio/region_mean": 0.0,
|
| 1609 |
+
"entropy": 2.2412831783294678,
|
| 1610 |
+
"epoch": 2.2,
|
| 1611 |
+
"grad_norm": 0.03572091460227966,
|
| 1612 |
+
"kl": 0.11483246088027954,
|
| 1613 |
+
"learning_rate": 8.25e-07,
|
| 1614 |
+
"loss": 0.0002870811440516263,
|
| 1615 |
+
"step": 88,
|
| 1616 |
+
"step_time": 0.11135016000025644
|
| 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": 2.2329115867614746,
|
| 1632 |
+
"epoch": 2.225,
|
| 1633 |
+
"frac_reward_zero_std": 1.0,
|
| 1634 |
+
"grad_norm": 0.03398412466049194,
|
| 1635 |
+
"kl": 0.10536332428455353,
|
| 1636 |
+
"learning_rate": 8e-07,
|
| 1637 |
+
"loss": 0.00026340829208493233,
|
| 1638 |
+
"num_tokens": 230476.0,
|
| 1639 |
+
"reward": 0.8402513265609741,
|
| 1640 |
+
"reward_std": 0.13532696664333344,
|
| 1641 |
+
"rewards/AnswererRewardFunction/mean": 0.8402513265609741,
|
| 1642 |
+
"rewards/AnswererRewardFunction/std": 0.13532696664333344,
|
| 1643 |
+
"step": 89,
|
| 1644 |
+
"step_time": 7.977572359999613
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"clip_ratio/high_max": 0.0,
|
| 1648 |
+
"clip_ratio/high_mean": 0.0,
|
| 1649 |
+
"clip_ratio/low_mean": 0.0,
|
| 1650 |
+
"clip_ratio/low_min": 0.0,
|
| 1651 |
+
"clip_ratio/region_mean": 0.0,
|
| 1652 |
+
"entropy": 2.220691680908203,
|
| 1653 |
+
"epoch": 2.25,
|
| 1654 |
+
"grad_norm": 0.029248526319861412,
|
| 1655 |
+
"kl": 0.11119011789560318,
|
| 1656 |
+
"learning_rate": 7.750000000000001e-07,
|
| 1657 |
+
"loss": 0.00027797528309747577,
|
| 1658 |
+
"step": 90,
|
| 1659 |
+
"step_time": 0.10857933900024364
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"clip_ratio/high_max": 0.0,
|
| 1663 |
+
"clip_ratio/high_mean": 0.0,
|
| 1664 |
+
"clip_ratio/low_mean": 0.0,
|
| 1665 |
+
"clip_ratio/low_min": 0.0,
|
| 1666 |
+
"clip_ratio/region_mean": 0.0,
|
| 1667 |
+
"entropy": 2.380763292312622,
|
| 1668 |
+
"epoch": 2.275,
|
| 1669 |
+
"grad_norm": 0.06103343889117241,
|
| 1670 |
+
"kl": 0.11687684059143066,
|
| 1671 |
+
"learning_rate": 7.5e-07,
|
| 1672 |
+
"loss": 0.0002921920968219638,
|
| 1673 |
+
"step": 91,
|
| 1674 |
+
"step_time": 0.11029780799981381
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"clip_ratio/high_max": 0.0,
|
| 1678 |
+
"clip_ratio/high_mean": 0.0,
|
| 1679 |
+
"clip_ratio/low_mean": 0.0,
|
| 1680 |
+
"clip_ratio/low_min": 0.0,
|
| 1681 |
+
"clip_ratio/region_mean": 0.0,
|
| 1682 |
+
"entropy": 2.324506998062134,
|
| 1683 |
+
"epoch": 2.3,
|
| 1684 |
+
"grad_norm": 0.028312548995018005,
|
| 1685 |
+
"kl": 0.1045643538236618,
|
| 1686 |
+
"learning_rate": 7.25e-07,
|
| 1687 |
+
"loss": 0.0002614108962006867,
|
| 1688 |
+
"step": 92,
|
| 1689 |
+
"step_time": 0.10763527099970815
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"clip_ratio/high_max": 0.0,
|
| 1693 |
+
"clip_ratio/high_mean": 0.0,
|
| 1694 |
+
"clip_ratio/low_mean": 0.0,
|
| 1695 |
+
"clip_ratio/low_min": 0.0,
|
| 1696 |
+
"clip_ratio/region_mean": 0.0,
|
| 1697 |
+
"completions/clipped_ratio": 1.0,
|
| 1698 |
+
"completions/max_length": 256.0,
|
| 1699 |
+
"completions/max_terminated_length": 0.0,
|
| 1700 |
+
"completions/mean_length": 256.0,
|
| 1701 |
+
"completions/mean_terminated_length": 0.0,
|
| 1702 |
+
"completions/min_length": 256.0,
|
| 1703 |
+
"completions/min_terminated_length": 0.0,
|
| 1704 |
+
"entropy": 2.2441272735595703,
|
| 1705 |
+
"epoch": 2.325,
|
| 1706 |
+
"frac_reward_zero_std": 1.0,
|
| 1707 |
+
"grad_norm": 0.031592387706041336,
|
| 1708 |
+
"kl": 0.11078132688999176,
|
| 1709 |
+
"learning_rate": 7.000000000000001e-07,
|
| 1710 |
+
"loss": 0.000276953331194818,
|
| 1711 |
+
"num_tokens": 240712.0,
|
| 1712 |
+
"reward": 0.9323133230209351,
|
| 1713 |
+
"reward_std": 0.0,
|
| 1714 |
+
"rewards/AnswererRewardFunction/mean": 0.9323133230209351,
|
| 1715 |
+
"rewards/AnswererRewardFunction/std": 0.0,
|
| 1716 |
+
"step": 93,
|
| 1717 |
+
"step_time": 7.962796914999672
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"clip_ratio/high_max": 0.0,
|
| 1721 |
+
"clip_ratio/high_mean": 0.0,
|
| 1722 |
+
"clip_ratio/low_mean": 0.0,
|
| 1723 |
+
"clip_ratio/low_min": 0.0,
|
| 1724 |
+
"clip_ratio/region_mean": 0.0,
|
| 1725 |
+
"entropy": 2.2512917518615723,
|
| 1726 |
+
"epoch": 2.35,
|
| 1727 |
+
"grad_norm": 0.0284142903983593,
|
| 1728 |
+
"kl": 0.1118057370185852,
|
| 1729 |
+
"learning_rate": 6.75e-07,
|
| 1730 |
+
"loss": 0.00027951435185968876,
|
| 1731 |
+
"step": 94,
|
| 1732 |
+
"step_time": 0.1111687049997272
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"clip_ratio/high_max": 0.0,
|
| 1736 |
+
"clip_ratio/high_mean": 0.0,
|
| 1737 |
+
"clip_ratio/low_mean": 0.0,
|
| 1738 |
+
"clip_ratio/low_min": 0.0,
|
| 1739 |
+
"clip_ratio/region_mean": 0.0,
|
| 1740 |
+
"entropy": 2.063572406768799,
|
| 1741 |
+
"epoch": 2.375,
|
| 1742 |
+
"grad_norm": 0.03613530844449997,
|
| 1743 |
+
"kl": 0.09560176730155945,
|
| 1744 |
+
"learning_rate": 6.5e-07,
|
| 1745 |
+
"loss": 0.00023900439555291086,
|
| 1746 |
+
"step": 95,
|
| 1747 |
+
"step_time": 0.10962757299967052
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"clip_ratio/high_max": 0.0,
|
| 1751 |
+
"clip_ratio/high_mean": 0.0,
|
| 1752 |
+
"clip_ratio/low_mean": 0.0,
|
| 1753 |
+
"clip_ratio/low_min": 0.0,
|
| 1754 |
+
"clip_ratio/region_mean": 0.0,
|
| 1755 |
+
"entropy": 2.301564931869507,
|
| 1756 |
+
"epoch": 2.4,
|
| 1757 |
+
"grad_norm": 0.03563942015171051,
|
| 1758 |
+
"kl": 0.11534290015697479,
|
| 1759 |
+
"learning_rate": 6.25e-07,
|
| 1760 |
+
"loss": 0.00028835725970566273,
|
| 1761 |
+
"step": 96,
|
| 1762 |
+
"step_time": 0.11151011100082542
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"clip_ratio/high_max": 0.0,
|
| 1766 |
+
"clip_ratio/high_mean": 0.0,
|
| 1767 |
+
"clip_ratio/low_mean": 0.0,
|
| 1768 |
+
"clip_ratio/low_min": 0.0,
|
| 1769 |
+
"clip_ratio/region_mean": 0.0,
|
| 1770 |
+
"completions/clipped_ratio": 1.0,
|
| 1771 |
+
"completions/max_length": 256.0,
|
| 1772 |
+
"completions/max_terminated_length": 0.0,
|
| 1773 |
+
"completions/mean_length": 256.0,
|
| 1774 |
+
"completions/mean_terminated_length": 0.0,
|
| 1775 |
+
"completions/min_length": 256.0,
|
| 1776 |
+
"completions/min_terminated_length": 0.0,
|
| 1777 |
+
"entropy": 2.2916674613952637,
|
| 1778 |
+
"epoch": 2.425,
|
| 1779 |
+
"frac_reward_zero_std": 0.75,
|
| 1780 |
+
"grad_norm": 1.3385826349258423,
|
| 1781 |
+
"kl": 0.0960921049118042,
|
| 1782 |
+
"learning_rate": 6.000000000000001e-07,
|
| 1783 |
+
"loss": -0.06221439689397812,
|
| 1784 |
+
"num_tokens": 250640.0,
|
| 1785 |
+
"reward": 0.8200892210006714,
|
| 1786 |
+
"reward_std": 0.13361600041389465,
|
| 1787 |
+
"rewards/AnswererRewardFunction/mean": 0.8200892210006714,
|
| 1788 |
+
"rewards/AnswererRewardFunction/std": 0.13361601531505585,
|
| 1789 |
+
"step": 97,
|
| 1790 |
+
"step_time": 7.975149017999684
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"clip_ratio/high_max": 0.0,
|
| 1794 |
+
"clip_ratio/high_mean": 0.0,
|
| 1795 |
+
"clip_ratio/low_mean": 0.0,
|
| 1796 |
+
"clip_ratio/low_min": 0.0,
|
| 1797 |
+
"clip_ratio/region_mean": 0.0,
|
| 1798 |
+
"entropy": 2.4413206577301025,
|
| 1799 |
+
"epoch": 2.45,
|
| 1800 |
+
"grad_norm": 2.7124135494232178,
|
| 1801 |
+
"kl": 0.11070380359888077,
|
| 1802 |
+
"learning_rate": 5.750000000000001e-07,
|
| 1803 |
+
"loss": 0.09390329569578171,
|
| 1804 |
+
"step": 98,
|
| 1805 |
+
"step_time": 0.11212589400020079
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"clip_ratio/high_max": 0.0,
|
| 1809 |
+
"clip_ratio/high_mean": 0.0,
|
| 1810 |
+
"clip_ratio/low_mean": 0.0,
|
| 1811 |
+
"clip_ratio/low_min": 0.0,
|
| 1812 |
+
"clip_ratio/region_mean": 0.0,
|
| 1813 |
+
"entropy": 2.297349452972412,
|
| 1814 |
+
"epoch": 2.475,
|
| 1815 |
+
"grad_norm": 0.035885103046894073,
|
| 1816 |
+
"kl": 0.09237588196992874,
|
| 1817 |
+
"learning_rate": 5.499999999999999e-07,
|
| 1818 |
+
"loss": 0.00023093970958143473,
|
| 1819 |
+
"step": 99,
|
| 1820 |
+
"step_time": 0.11208251399966684
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"clip_ratio/high_max": 0.0,
|
| 1824 |
+
"clip_ratio/high_mean": 0.0,
|
| 1825 |
+
"clip_ratio/low_mean": 0.0,
|
| 1826 |
+
"clip_ratio/low_min": 0.0,
|
| 1827 |
+
"clip_ratio/region_mean": 0.0,
|
| 1828 |
+
"entropy": 2.258030652999878,
|
| 1829 |
+
"epoch": 2.5,
|
| 1830 |
+
"grad_norm": 0.9310756325721741,
|
| 1831 |
+
"kl": 0.10258521139621735,
|
| 1832 |
+
"learning_rate": 5.25e-07,
|
| 1833 |
+
"loss": -0.03099227324128151,
|
| 1834 |
+
"step": 100,
|
| 1835 |
+
"step_time": 0.11538284000016574
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"clip_ratio/high_max": 0.0,
|
| 1839 |
+
"clip_ratio/high_mean": 0.0,
|
| 1840 |
+
"clip_ratio/low_mean": 0.0,
|
| 1841 |
+
"clip_ratio/low_min": 0.0,
|
| 1842 |
+
"clip_ratio/region_mean": 0.0,
|
| 1843 |
+
"completions/clipped_ratio": 1.0,
|
| 1844 |
+
"completions/max_length": 256.0,
|
| 1845 |
+
"completions/max_terminated_length": 0.0,
|
| 1846 |
+
"completions/mean_length": 256.0,
|
| 1847 |
+
"completions/mean_terminated_length": 0.0,
|
| 1848 |
+
"completions/min_length": 256.0,
|
| 1849 |
+
"completions/min_terminated_length": 0.0,
|
| 1850 |
+
"entropy": 2.2856216430664062,
|
| 1851 |
+
"epoch": 2.525,
|
| 1852 |
+
"frac_reward_zero_std": 1.0,
|
| 1853 |
+
"grad_norm": 0.027649035677313805,
|
| 1854 |
+
"kl": 0.10456856340169907,
|
| 1855 |
+
"learning_rate": 5e-07,
|
| 1856 |
+
"loss": 0.0002614214026834816,
|
| 1857 |
+
"num_tokens": 260712.0,
|
| 1858 |
+
"reward": 0.8883568644523621,
|
| 1859 |
+
"reward_std": 0.08731484413146973,
|
| 1860 |
+
"rewards/AnswererRewardFunction/mean": 0.8883568644523621,
|
| 1861 |
+
"rewards/AnswererRewardFunction/std": 0.08731484413146973,
|
| 1862 |
+
"step": 101,
|
| 1863 |
+
"step_time": 7.974483150999731
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"clip_ratio/high_max": 0.0,
|
| 1867 |
+
"clip_ratio/high_mean": 0.0,
|
| 1868 |
+
"clip_ratio/low_mean": 0.0,
|
| 1869 |
+
"clip_ratio/low_min": 0.0,
|
| 1870 |
+
"clip_ratio/region_mean": 0.0,
|
| 1871 |
+
"entropy": 2.256802558898926,
|
| 1872 |
+
"epoch": 2.55,
|
| 1873 |
+
"grad_norm": 0.025713058188557625,
|
| 1874 |
+
"kl": 0.09895136952400208,
|
| 1875 |
+
"learning_rate": 4.75e-07,
|
| 1876 |
+
"loss": 0.0002473784261383116,
|
| 1877 |
+
"step": 102,
|
| 1878 |
+
"step_time": 0.1179694649999874
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"clip_ratio/high_max": 0.0,
|
| 1882 |
+
"clip_ratio/high_mean": 0.0,
|
| 1883 |
+
"clip_ratio/low_mean": 0.0,
|
| 1884 |
+
"clip_ratio/low_min": 0.0,
|
| 1885 |
+
"clip_ratio/region_mean": 0.0,
|
| 1886 |
+
"entropy": 2.2987618446350098,
|
| 1887 |
+
"epoch": 2.575,
|
| 1888 |
+
"grad_norm": 0.03702561929821968,
|
| 1889 |
+
"kl": 0.10545263439416885,
|
| 1890 |
+
"learning_rate": 4.5e-07,
|
| 1891 |
+
"loss": 0.0002636315766721964,
|
| 1892 |
+
"step": 103,
|
| 1893 |
+
"step_time": 0.11270309300016379
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"clip_ratio/high_max": 0.0,
|
| 1897 |
+
"clip_ratio/high_mean": 0.0,
|
| 1898 |
+
"clip_ratio/low_mean": 0.0,
|
| 1899 |
+
"clip_ratio/low_min": 0.0,
|
| 1900 |
+
"clip_ratio/region_mean": 0.0,
|
| 1901 |
+
"entropy": 2.2298688888549805,
|
| 1902 |
+
"epoch": 2.6,
|
| 1903 |
+
"grad_norm": 0.032740261405706406,
|
| 1904 |
+
"kl": 0.10465572774410248,
|
| 1905 |
+
"learning_rate": 4.25e-07,
|
| 1906 |
+
"loss": 0.00026163930306211114,
|
| 1907 |
+
"step": 104,
|
| 1908 |
+
"step_time": 0.11148421600046277
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"clip_ratio/high_max": 0.0,
|
| 1912 |
+
"clip_ratio/high_mean": 0.0,
|
| 1913 |
+
"clip_ratio/low_mean": 0.0,
|
| 1914 |
+
"clip_ratio/low_min": 0.0,
|
| 1915 |
+
"clip_ratio/region_mean": 0.0,
|
| 1916 |
+
"completions/clipped_ratio": 1.0,
|
| 1917 |
+
"completions/max_length": 256.0,
|
| 1918 |
+
"completions/max_terminated_length": 0.0,
|
| 1919 |
+
"completions/mean_length": 256.0,
|
| 1920 |
+
"completions/mean_terminated_length": 0.0,
|
| 1921 |
+
"completions/min_length": 256.0,
|
| 1922 |
+
"completions/min_terminated_length": 0.0,
|
| 1923 |
+
"entropy": 2.4025683403015137,
|
| 1924 |
+
"epoch": 2.625,
|
| 1925 |
+
"frac_reward_zero_std": 1.0,
|
| 1926 |
+
"grad_norm": 0.03896611928939819,
|
| 1927 |
+
"kl": 0.10709057748317719,
|
| 1928 |
+
"learning_rate": 4e-07,
|
| 1929 |
+
"loss": 0.0002677264274097979,
|
| 1930 |
+
"num_tokens": 270652.0,
|
| 1931 |
+
"reward": 0.8187265992164612,
|
| 1932 |
+
"reward_std": 0.11731186509132385,
|
| 1933 |
+
"rewards/AnswererRewardFunction/mean": 0.8187265992164612,
|
| 1934 |
+
"rewards/AnswererRewardFunction/std": 0.11731186509132385,
|
| 1935 |
+
"step": 105,
|
| 1936 |
+
"step_time": 7.961027192999609
|
| 1937 |
+
},
|
| 1938 |
+
{
|
| 1939 |
+
"clip_ratio/high_max": 0.0,
|
| 1940 |
+
"clip_ratio/high_mean": 0.0,
|
| 1941 |
+
"clip_ratio/low_mean": 0.0,
|
| 1942 |
+
"clip_ratio/low_min": 0.0,
|
| 1943 |
+
"clip_ratio/region_mean": 0.0,
|
| 1944 |
+
"entropy": 2.294612407684326,
|
| 1945 |
+
"epoch": 2.65,
|
| 1946 |
+
"grad_norm": 0.02885749377310276,
|
| 1947 |
+
"kl": 0.09969587624073029,
|
| 1948 |
+
"learning_rate": 3.75e-07,
|
| 1949 |
+
"loss": 0.00024923967430368066,
|
| 1950 |
+
"step": 106,
|
| 1951 |
+
"step_time": 0.11204869899938785
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"clip_ratio/high_max": 0.0,
|
| 1955 |
+
"clip_ratio/high_mean": 0.0,
|
| 1956 |
+
"clip_ratio/low_mean": 0.0,
|
| 1957 |
+
"clip_ratio/low_min": 0.0,
|
| 1958 |
+
"clip_ratio/region_mean": 0.0,
|
| 1959 |
+
"entropy": 2.2906203269958496,
|
| 1960 |
+
"epoch": 2.675,
|
| 1961 |
+
"grad_norm": 0.0278431698679924,
|
| 1962 |
+
"kl": 0.10312323272228241,
|
| 1963 |
+
"learning_rate": 3.5000000000000004e-07,
|
| 1964 |
+
"loss": 0.0002578080748207867,
|
| 1965 |
+
"step": 107,
|
| 1966 |
+
"step_time": 0.11121519100015576
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"clip_ratio/high_max": 0.0,
|
| 1970 |
+
"clip_ratio/high_mean": 0.0,
|
| 1971 |
+
"clip_ratio/low_mean": 0.0,
|
| 1972 |
+
"clip_ratio/low_min": 0.0,
|
| 1973 |
+
"clip_ratio/region_mean": 0.0,
|
| 1974 |
+
"entropy": 2.2572021484375,
|
| 1975 |
+
"epoch": 2.7,
|
| 1976 |
+
"grad_norm": 0.030486231669783592,
|
| 1977 |
+
"kl": 0.111005038022995,
|
| 1978 |
+
"learning_rate": 3.25e-07,
|
| 1979 |
+
"loss": 0.00027751256129704416,
|
| 1980 |
+
"step": 108,
|
| 1981 |
+
"step_time": 0.11487295599999925
|
| 1982 |
+
},
|
| 1983 |
+
{
|
| 1984 |
+
"clip_ratio/high_max": 0.0,
|
| 1985 |
+
"clip_ratio/high_mean": 0.0,
|
| 1986 |
+
"clip_ratio/low_mean": 0.0,
|
| 1987 |
+
"clip_ratio/low_min": 0.0,
|
| 1988 |
+
"clip_ratio/region_mean": 0.0,
|
| 1989 |
+
"completions/clipped_ratio": 1.0,
|
| 1990 |
+
"completions/max_length": 256.0,
|
| 1991 |
+
"completions/max_terminated_length": 0.0,
|
| 1992 |
+
"completions/mean_length": 256.0,
|
| 1993 |
+
"completions/mean_terminated_length": 0.0,
|
| 1994 |
+
"completions/min_length": 256.0,
|
| 1995 |
+
"completions/min_terminated_length": 0.0,
|
| 1996 |
+
"entropy": 2.2769908905029297,
|
| 1997 |
+
"epoch": 2.725,
|
| 1998 |
+
"frac_reward_zero_std": 1.0,
|
| 1999 |
+
"grad_norm": 0.032291825860738754,
|
| 2000 |
+
"kl": 0.09712264686822891,
|
| 2001 |
+
"learning_rate": 3.0000000000000004e-07,
|
| 2002 |
+
"loss": 0.00024280662182718515,
|
| 2003 |
+
"num_tokens": 280828.0,
|
| 2004 |
+
"reward": 0.9358674883842468,
|
| 2005 |
+
"reward_std": 0.0063578663393855095,
|
| 2006 |
+
"rewards/AnswererRewardFunction/mean": 0.9358674883842468,
|
| 2007 |
+
"rewards/AnswererRewardFunction/std": 0.006357857491821051,
|
| 2008 |
+
"step": 109,
|
| 2009 |
+
"step_time": 8.002215447999333
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"clip_ratio/high_max": 0.0,
|
| 2013 |
+
"clip_ratio/high_mean": 0.0,
|
| 2014 |
+
"clip_ratio/low_mean": 0.0,
|
| 2015 |
+
"clip_ratio/low_min": 0.0,
|
| 2016 |
+
"clip_ratio/region_mean": 0.0,
|
| 2017 |
+
"entropy": 2.209662437438965,
|
| 2018 |
+
"epoch": 2.75,
|
| 2019 |
+
"grad_norm": 0.026608958840370178,
|
| 2020 |
+
"kl": 0.10376521944999695,
|
| 2021 |
+
"learning_rate": 2.7499999999999996e-07,
|
| 2022 |
+
"loss": 0.00025941303465515375,
|
| 2023 |
+
"step": 110,
|
| 2024 |
+
"step_time": 0.11106319399914355
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"clip_ratio/high_max": 0.0,
|
| 2028 |
+
"clip_ratio/high_mean": 0.0,
|
| 2029 |
+
"clip_ratio/low_mean": 0.0,
|
| 2030 |
+
"clip_ratio/low_min": 0.0,
|
| 2031 |
+
"clip_ratio/region_mean": 0.0,
|
| 2032 |
+
"entropy": 2.2341318130493164,
|
| 2033 |
+
"epoch": 2.775,
|
| 2034 |
+
"grad_norm": 0.0365554504096508,
|
| 2035 |
+
"kl": 0.10377062857151031,
|
| 2036 |
+
"learning_rate": 2.5e-07,
|
| 2037 |
+
"loss": 0.00025942656793631613,
|
| 2038 |
+
"step": 111,
|
| 2039 |
+
"step_time": 0.11160198600009608
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"clip_ratio/high_max": 0.0,
|
| 2043 |
+
"clip_ratio/high_mean": 0.0,
|
| 2044 |
+
"clip_ratio/low_mean": 0.0,
|
| 2045 |
+
"clip_ratio/low_min": 0.0,
|
| 2046 |
+
"clip_ratio/region_mean": 0.0,
|
| 2047 |
+
"entropy": 2.2511730194091797,
|
| 2048 |
+
"epoch": 2.8,
|
| 2049 |
+
"grad_norm": 0.027873879298567772,
|
| 2050 |
+
"kl": 0.09454593807458878,
|
| 2051 |
+
"learning_rate": 2.25e-07,
|
| 2052 |
+
"loss": 0.00023636485275346786,
|
| 2053 |
+
"step": 112,
|
| 2054 |
+
"step_time": 0.11040955699991173
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"clip_ratio/high_max": 0.0,
|
| 2058 |
+
"clip_ratio/high_mean": 0.0,
|
| 2059 |
+
"clip_ratio/low_mean": 0.0,
|
| 2060 |
+
"clip_ratio/low_min": 0.0,
|
| 2061 |
+
"clip_ratio/region_mean": 0.0,
|
| 2062 |
+
"completions/clipped_ratio": 1.0,
|
| 2063 |
+
"completions/max_length": 256.0,
|
| 2064 |
+
"completions/max_terminated_length": 0.0,
|
| 2065 |
+
"completions/mean_length": 256.0,
|
| 2066 |
+
"completions/mean_terminated_length": 0.0,
|
| 2067 |
+
"completions/min_length": 256.0,
|
| 2068 |
+
"completions/min_terminated_length": 0.0,
|
| 2069 |
+
"entropy": 2.217411994934082,
|
| 2070 |
+
"epoch": 2.825,
|
| 2071 |
+
"frac_reward_zero_std": 1.0,
|
| 2072 |
+
"grad_norm": 0.025801831856369972,
|
| 2073 |
+
"kl": 0.1010793149471283,
|
| 2074 |
+
"learning_rate": 2e-07,
|
| 2075 |
+
"loss": 0.0002526982862036675,
|
| 2076 |
+
"num_tokens": 290956.0,
|
| 2077 |
+
"reward": 0.8848026990890503,
|
| 2078 |
+
"reward_std": 0.08498957008123398,
|
| 2079 |
+
"rewards/AnswererRewardFunction/mean": 0.8848026990890503,
|
| 2080 |
+
"rewards/AnswererRewardFunction/std": 0.08498956263065338,
|
| 2081 |
+
"step": 113,
|
| 2082 |
+
"step_time": 7.909033621999697
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"clip_ratio/high_max": 0.0,
|
| 2086 |
+
"clip_ratio/high_mean": 0.0,
|
| 2087 |
+
"clip_ratio/low_mean": 0.0,
|
| 2088 |
+
"clip_ratio/low_min": 0.0,
|
| 2089 |
+
"clip_ratio/region_mean": 0.0,
|
| 2090 |
+
"entropy": 2.4240846633911133,
|
| 2091 |
+
"epoch": 2.85,
|
| 2092 |
+
"grad_norm": 0.029559027403593063,
|
| 2093 |
+
"kl": 0.11018238961696625,
|
| 2094 |
+
"learning_rate": 1.7500000000000002e-07,
|
| 2095 |
+
"loss": 0.00027545596822164953,
|
| 2096 |
+
"step": 114,
|
| 2097 |
+
"step_time": 0.1100401760004388
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"clip_ratio/high_max": 0.0,
|
| 2101 |
+
"clip_ratio/high_mean": 0.0,
|
| 2102 |
+
"clip_ratio/low_mean": 0.0,
|
| 2103 |
+
"clip_ratio/low_min": 0.0,
|
| 2104 |
+
"clip_ratio/region_mean": 0.0,
|
| 2105 |
+
"entropy": 2.2784318923950195,
|
| 2106 |
+
"epoch": 2.875,
|
| 2107 |
+
"grad_norm": 0.029010634869337082,
|
| 2108 |
+
"kl": 0.09815891087055206,
|
| 2109 |
+
"learning_rate": 1.5000000000000002e-07,
|
| 2110 |
+
"loss": 0.0002453972992952913,
|
| 2111 |
+
"step": 115,
|
| 2112 |
+
"step_time": 0.11177435299941862
|
| 2113 |
+
},
|
| 2114 |
+
{
|
| 2115 |
+
"clip_ratio/high_max": 0.0,
|
| 2116 |
+
"clip_ratio/high_mean": 0.0,
|
| 2117 |
+
"clip_ratio/low_mean": 0.0,
|
| 2118 |
+
"clip_ratio/low_min": 0.0,
|
| 2119 |
+
"clip_ratio/region_mean": 0.0,
|
| 2120 |
+
"entropy": 2.2035021781921387,
|
| 2121 |
+
"epoch": 2.9,
|
| 2122 |
+
"grad_norm": 0.028797617182135582,
|
| 2123 |
+
"kl": 0.09739834070205688,
|
| 2124 |
+
"learning_rate": 1.25e-07,
|
| 2125 |
+
"loss": 0.0002434958441881463,
|
| 2126 |
+
"step": 116,
|
| 2127 |
+
"step_time": 0.11099860700051067
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"clip_ratio/high_max": 0.0,
|
| 2131 |
+
"clip_ratio/high_mean": 0.0,
|
| 2132 |
+
"clip_ratio/low_mean": 0.0,
|
| 2133 |
+
"clip_ratio/low_min": 0.0,
|
| 2134 |
+
"clip_ratio/region_mean": 0.0,
|
| 2135 |
+
"completions/clipped_ratio": 1.0,
|
| 2136 |
+
"completions/max_length": 256.0,
|
| 2137 |
+
"completions/max_terminated_length": 0.0,
|
| 2138 |
+
"completions/mean_length": 256.0,
|
| 2139 |
+
"completions/mean_terminated_length": 0.0,
|
| 2140 |
+
"completions/min_length": 256.0,
|
| 2141 |
+
"completions/min_terminated_length": 0.0,
|
| 2142 |
+
"entropy": 2.359414577484131,
|
| 2143 |
+
"epoch": 2.925,
|
| 2144 |
+
"frac_reward_zero_std": 1.0,
|
| 2145 |
+
"grad_norm": 0.02853987365961075,
|
| 2146 |
+
"kl": 0.11679545789957047,
|
| 2147 |
+
"learning_rate": 1e-07,
|
| 2148 |
+
"loss": 0.00029198863194324076,
|
| 2149 |
+
"num_tokens": 301080.0,
|
| 2150 |
+
"reward": 0.8755199909210205,
|
| 2151 |
+
"reward_std": 0.10159505903720856,
|
| 2152 |
+
"rewards/AnswererRewardFunction/mean": 0.8755199909210205,
|
| 2153 |
+
"rewards/AnswererRewardFunction/std": 0.10159505903720856,
|
| 2154 |
+
"step": 117,
|
| 2155 |
+
"step_time": 7.939703003000432
|
| 2156 |
+
},
|
| 2157 |
+
{
|
| 2158 |
+
"clip_ratio/high_max": 0.0,
|
| 2159 |
+
"clip_ratio/high_mean": 0.0,
|
| 2160 |
+
"clip_ratio/low_mean": 0.0,
|
| 2161 |
+
"clip_ratio/low_min": 0.0,
|
| 2162 |
+
"clip_ratio/region_mean": 0.0,
|
| 2163 |
+
"entropy": 2.1621551513671875,
|
| 2164 |
+
"epoch": 2.95,
|
| 2165 |
+
"grad_norm": 0.02482123300433159,
|
| 2166 |
+
"kl": 0.09331086277961731,
|
| 2167 |
+
"learning_rate": 7.500000000000001e-08,
|
| 2168 |
+
"loss": 0.00023327715462073684,
|
| 2169 |
+
"step": 118,
|
| 2170 |
+
"step_time": 0.1127649840000231
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"clip_ratio/high_max": 0.0,
|
| 2174 |
+
"clip_ratio/high_mean": 0.0,
|
| 2175 |
+
"clip_ratio/low_mean": 0.0,
|
| 2176 |
+
"clip_ratio/low_min": 0.0,
|
| 2177 |
+
"clip_ratio/region_mean": 0.0,
|
| 2178 |
+
"entropy": 2.151353359222412,
|
| 2179 |
+
"epoch": 2.975,
|
| 2180 |
+
"grad_norm": 0.027611907571554184,
|
| 2181 |
+
"kl": 0.09863051772117615,
|
| 2182 |
+
"learning_rate": 5e-08,
|
| 2183 |
+
"loss": 0.00024657626636326313,
|
| 2184 |
+
"step": 119,
|
| 2185 |
+
"step_time": 0.11157421799998701
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"clip_ratio/high_max": 0.0,
|
| 2189 |
+
"clip_ratio/high_mean": 0.0,
|
| 2190 |
+
"clip_ratio/low_mean": 0.0,
|
| 2191 |
+
"clip_ratio/low_min": 0.0,
|
| 2192 |
+
"clip_ratio/region_mean": 0.0,
|
| 2193 |
+
"entropy": 2.0556399822235107,
|
| 2194 |
+
"epoch": 3.0,
|
| 2195 |
+
"grad_norm": 0.03890826925635338,
|
| 2196 |
+
"kl": 0.11086487770080566,
|
| 2197 |
+
"learning_rate": 2.5e-08,
|
| 2198 |
+
"loss": 0.00027716218028217554,
|
| 2199 |
+
"step": 120,
|
| 2200 |
+
"step_time": 0.11192440499962686
|
| 2201 |
+
}
|
| 2202 |
+
],
|
| 2203 |
+
"logging_steps": 1,
|
| 2204 |
+
"max_steps": 120,
|
| 2205 |
+
"num_input_tokens_seen": 301080,
|
| 2206 |
+
"num_train_epochs": 3,
|
| 2207 |
+
"save_steps": 30,
|
| 2208 |
+
"stateful_callbacks": {
|
| 2209 |
+
"TrainerControl": {
|
| 2210 |
+
"args": {
|
| 2211 |
+
"should_epoch_stop": false,
|
| 2212 |
+
"should_evaluate": false,
|
| 2213 |
+
"should_log": false,
|
| 2214 |
+
"should_save": true,
|
| 2215 |
+
"should_training_stop": true
|
| 2216 |
+
},
|
| 2217 |
+
"attributes": {}
|
| 2218 |
+
}
|
| 2219 |
+
},
|
| 2220 |
+
"total_flos": 0.0,
|
| 2221 |
+
"train_batch_size": 4,
|
| 2222 |
+
"trial_name": null,
|
| 2223 |
+
"trial_params": null
|
| 2224 |
+
}
|
self_play_hf_l40s_full/round_006/answerer_train/checkpoint-120/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:99b43de3228fc906dd533a65034a9a07e1b0017bc5ef4fa7868a1a78d1cee45d
|
| 3 |
+
size 7249
|