Siddeshwar1625 commited on
Commit
b5cf27c
·
verified ·
1 Parent(s): 0c523d6

Intermediate checkpoint upload step=90 (answerer_train)

Browse files
.gitattributes CHANGED
@@ -69,3 +69,4 @@ self_play_hf_l40s_full/round_003/generator_train/checkpoint-120/tokenizer.json f
69
  self_play_hf_l40s_full/round_003/generator_train/final_model/tokenizer.json filter=lfs diff=lfs merge=lfs -text
70
  self_play_hf_l40s_full/round_003/answerer_train/checkpoint-30/tokenizer.json filter=lfs diff=lfs merge=lfs -text
71
  self_play_hf_l40s_full/round_003/answerer_train/checkpoint-60/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 
 
69
  self_play_hf_l40s_full/round_003/generator_train/final_model/tokenizer.json filter=lfs diff=lfs merge=lfs -text
70
  self_play_hf_l40s_full/round_003/answerer_train/checkpoint-30/tokenizer.json filter=lfs diff=lfs merge=lfs -text
71
  self_play_hf_l40s_full/round_003/answerer_train/checkpoint-60/tokenizer.json filter=lfs diff=lfs merge=lfs -text
72
+ self_play_hf_l40s_full/round_003/answerer_train/checkpoint-90/tokenizer.json filter=lfs diff=lfs merge=lfs -text
self_play_hf_l40s_full/round_003/answerer_train/checkpoint-90/chat_template.jinja ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- if tools %}
2
+ {{- '<|im_start|>system\n' }}
3
+ {%- if messages[0]['role'] == 'system' %}
4
+ {{- messages[0]['content'] }}
5
+ {%- else %}
6
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
7
+ {%- endif %}
8
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
9
+ {%- for tool in tools %}
10
+ {{- "\n" }}
11
+ {{- tool | tojson }}
12
+ {%- endfor %}
13
+ {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
14
+ {%- else %}
15
+ {%- if messages[0]['role'] == 'system' %}
16
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
17
+ {%- else %}
18
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
19
+ {%- endif %}
20
+ {%- endif %}
21
+ {%- for message in messages %}
22
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
23
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
24
+ {%- elif message.role == "assistant" %}
25
+ {{- '<|im_start|>' + message.role }}
26
+ {%- if message.content %}
27
+ {{- '\n' + message.content }}
28
+ {%- endif %}
29
+ {%- for tool_call in message.tool_calls %}
30
+ {%- if tool_call.function is defined %}
31
+ {%- set tool_call = tool_call.function %}
32
+ {%- endif %}
33
+ {{- '\n<tool_call>\n{"name": "' }}
34
+ {{- tool_call.name }}
35
+ {{- '", "arguments": ' }}
36
+ {{- tool_call.arguments | tojson }}
37
+ {{- '}\n</tool_call>' }}
38
+ {%- endfor %}
39
+ {{- '<|im_end|>\n' }}
40
+ {%- elif message.role == "tool" %}
41
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
42
+ {{- '<|im_start|>user' }}
43
+ {%- endif %}
44
+ {{- '\n<tool_response>\n' }}
45
+ {{- message.content }}
46
+ {{- '\n</tool_response>' }}
47
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
48
+ {{- '<|im_end|>\n' }}
49
+ {%- endif %}
50
+ {%- endif %}
51
+ {%- endfor %}
52
+ {%- if add_generation_prompt %}
53
+ {{- '<|im_start|>assistant\n' }}
54
+ {%- endif %}
self_play_hf_l40s_full/round_003/answerer_train/checkpoint-90/config.json ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen2ForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": null,
7
+ "dtype": "float32",
8
+ "eos_token_id": 151645,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 896,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 4864,
13
+ "layer_types": [
14
+ "full_attention",
15
+ "full_attention",
16
+ "full_attention",
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention"
38
+ ],
39
+ "max_position_embeddings": 32768,
40
+ "max_window_layers": 21,
41
+ "model_type": "qwen2",
42
+ "num_attention_heads": 14,
43
+ "num_hidden_layers": 24,
44
+ "num_key_value_heads": 2,
45
+ "pad_token_id": 151643,
46
+ "rms_norm_eps": 1e-06,
47
+ "rope_parameters": {
48
+ "rope_theta": 1000000.0,
49
+ "rope_type": "default"
50
+ },
51
+ "sliding_window": null,
52
+ "tie_word_embeddings": true,
53
+ "transformers_version": "5.6.2",
54
+ "use_cache": false,
55
+ "use_sliding_window": false,
56
+ "vocab_size": 151936
57
+ }
self_play_hf_l40s_full/round_003/answerer_train/checkpoint-90/generation_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_sample": true,
3
+ "eos_token_id": [
4
+ 151645,
5
+ 151643
6
+ ],
7
+ "pad_token_id": 151643,
8
+ "repetition_penalty": 1.1,
9
+ "temperature": 0.7,
10
+ "top_k": 20,
11
+ "top_p": 0.8,
12
+ "transformers_version": "5.6.2"
13
+ }
self_play_hf_l40s_full/round_003/answerer_train/checkpoint-90/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6c47c72630cffe1ede635f3968bcffec120b4bc8fd73f256c32117b6d31ae3d9
3
+ size 1976163472
self_play_hf_l40s_full/round_003/answerer_train/checkpoint-90/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:081d6abc88ff62f494387200ebce6988ae5469cb89be77db1e8d3f05e9128115
3
+ size 3952509771
self_play_hf_l40s_full/round_003/answerer_train/checkpoint-90/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26f6240389ad2111a606d4aef5c6832d0136df7debe9f496847f25e20ecf886b
3
+ size 14645
self_play_hf_l40s_full/round_003/answerer_train/checkpoint-90/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:716b80a1d03fa94e9f2454097632e924f71019b15d068a8837fcbbbce1910893
3
+ size 1465
self_play_hf_l40s_full/round_003/answerer_train/checkpoint-90/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
3
+ size 11421892
self_play_hf_l40s_full/round_003/answerer_train/checkpoint-90/tokenizer_config.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "backend": "tokenizers",
4
+ "bos_token": null,
5
+ "clean_up_tokenization_spaces": false,
6
+ "eos_token": "<|im_end|>",
7
+ "errors": "replace",
8
+ "extra_special_tokens": [
9
+ "<|im_start|>",
10
+ "<|im_end|>",
11
+ "<|object_ref_start|>",
12
+ "<|object_ref_end|>",
13
+ "<|box_start|>",
14
+ "<|box_end|>",
15
+ "<|quad_start|>",
16
+ "<|quad_end|>",
17
+ "<|vision_start|>",
18
+ "<|vision_end|>",
19
+ "<|vision_pad|>",
20
+ "<|image_pad|>",
21
+ "<|video_pad|>"
22
+ ],
23
+ "is_local": true,
24
+ "local_files_only": false,
25
+ "model_max_length": 131072,
26
+ "pad_token": "<|endoftext|>",
27
+ "padding_side": "left",
28
+ "split_special_tokens": false,
29
+ "tokenizer_class": "Qwen2Tokenizer",
30
+ "truncation_side": "left",
31
+ "unk_token": null
32
+ }
self_play_hf_l40s_full/round_003/answerer_train/checkpoint-90/trainer_state.json ADDED
@@ -0,0 +1,1683 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 2.25,
6
+ "eval_steps": 500,
7
+ "global_step": 90,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "clip_ratio/high_max": 0.0,
14
+ "clip_ratio/high_mean": 0.0,
15
+ "clip_ratio/low_mean": 0.0,
16
+ "clip_ratio/low_min": 0.0,
17
+ "clip_ratio/region_mean": 0.0,
18
+ "completions/clipped_ratio": 1.0,
19
+ "completions/max_length": 256.0,
20
+ "completions/max_terminated_length": 0.0,
21
+ "completions/mean_length": 256.0,
22
+ "completions/mean_terminated_length": 0.0,
23
+ "completions/min_length": 256.0,
24
+ "completions/min_terminated_length": 0.0,
25
+ "entropy": 1.508577823638916,
26
+ "epoch": 0.025,
27
+ "frac_reward_zero_std": 0.0,
28
+ "grad_norm": 3.7061917781829834,
29
+ "kl": 0.0,
30
+ "learning_rate": 3e-06,
31
+ "loss": 0.06245460361242294,
32
+ "num_tokens": 9932.0,
33
+ "reward": 0.7367546558380127,
34
+ "reward_std": 0.18893879652023315,
35
+ "rewards/AnswererRewardFunction/mean": 0.7367546558380127,
36
+ "rewards/AnswererRewardFunction/std": 0.18893879652023315,
37
+ "step": 1,
38
+ "step_time": 7.948499319999883
39
+ },
40
+ {
41
+ "clip_ratio/high_max": 0.0302734375,
42
+ "clip_ratio/high_mean": 0.0302734375,
43
+ "clip_ratio/low_mean": 0.01171875,
44
+ "clip_ratio/low_min": 0.01171875,
45
+ "clip_ratio/region_mean": 0.0419921875,
46
+ "entropy": 1.748507022857666,
47
+ "epoch": 0.05,
48
+ "grad_norm": 3.0338211059570312,
49
+ "kl": 0.0139366639778018,
50
+ "learning_rate": 2.9750000000000003e-06,
51
+ "loss": 0.0011969313491135836,
52
+ "step": 2,
53
+ "step_time": 0.11838497100006862
54
+ },
55
+ {
56
+ "clip_ratio/high_max": 0.060546875,
57
+ "clip_ratio/high_mean": 0.060546875,
58
+ "clip_ratio/low_mean": 0.0322265625,
59
+ "clip_ratio/low_min": 0.0322265625,
60
+ "clip_ratio/region_mean": 0.0927734375,
61
+ "entropy": 1.7642402648925781,
62
+ "epoch": 0.075,
63
+ "grad_norm": 13.311189651489258,
64
+ "kl": 1.1347599029541016,
65
+ "learning_rate": 2.9499999999999997e-06,
66
+ "loss": -0.035761456936597824,
67
+ "step": 3,
68
+ "step_time": 0.11420342099972913
69
+ },
70
+ {
71
+ "clip_ratio/high_max": 0.0927734375,
72
+ "clip_ratio/high_mean": 0.0927734375,
73
+ "clip_ratio/low_mean": 0.0400390625,
74
+ "clip_ratio/low_min": 0.0400390625,
75
+ "clip_ratio/region_mean": 0.1328125,
76
+ "entropy": 1.7131476402282715,
77
+ "epoch": 0.1,
78
+ "grad_norm": 3.361721992492676,
79
+ "kl": 0.14338737726211548,
80
+ "learning_rate": 2.925e-06,
81
+ "loss": -0.01821894198656082,
82
+ "step": 4,
83
+ "step_time": 0.1116166990000238
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": 1.7969765663146973,
99
+ "epoch": 0.125,
100
+ "frac_reward_zero_std": 0.5,
101
+ "grad_norm": 0.037479810416698456,
102
+ "kl": 0.03041483834385872,
103
+ "learning_rate": 2.9e-06,
104
+ "loss": -0.0008414739277213812,
105
+ "num_tokens": 19740.0,
106
+ "reward": 0.8083143830299377,
107
+ "reward_std": 0.1783086657524109,
108
+ "rewards/AnswererRewardFunction/mean": 0.8083143830299377,
109
+ "rewards/AnswererRewardFunction/std": 0.1783086657524109,
110
+ "step": 5,
111
+ "step_time": 7.940309183999943
112
+ },
113
+ {
114
+ "clip_ratio/high_max": 0.0068359375,
115
+ "clip_ratio/high_mean": 0.0068359375,
116
+ "clip_ratio/low_mean": 0.0,
117
+ "clip_ratio/low_min": 0.0,
118
+ "clip_ratio/region_mean": 0.0068359375,
119
+ "entropy": 1.9499748945236206,
120
+ "epoch": 0.15,
121
+ "grad_norm": 1.9231107234954834,
122
+ "kl": 0.03690164536237717,
123
+ "learning_rate": 2.875e-06,
124
+ "loss": -0.09412746876478195,
125
+ "step": 6,
126
+ "step_time": 0.10904968900013046
127
+ },
128
+ {
129
+ "clip_ratio/high_max": 0.0,
130
+ "clip_ratio/high_mean": 0.0,
131
+ "clip_ratio/low_mean": 0.0087890625,
132
+ "clip_ratio/low_min": 0.0087890625,
133
+ "clip_ratio/region_mean": 0.0087890625,
134
+ "entropy": 1.7189953327178955,
135
+ "epoch": 0.175,
136
+ "grad_norm": 2.766663074493408,
137
+ "kl": 0.03734806925058365,
138
+ "learning_rate": 2.85e-06,
139
+ "loss": 0.09430856257677078,
140
+ "step": 7,
141
+ "step_time": 0.10698909699976866
142
+ },
143
+ {
144
+ "clip_ratio/high_max": 0.0556640625,
145
+ "clip_ratio/high_mean": 0.0556640625,
146
+ "clip_ratio/low_mean": 0.021484375,
147
+ "clip_ratio/low_min": 0.021484375,
148
+ "clip_ratio/region_mean": 0.0771484375,
149
+ "entropy": 1.8247244358062744,
150
+ "epoch": 0.2,
151
+ "grad_norm": 2.4841396808624268,
152
+ "kl": 0.06804034113883972,
153
+ "learning_rate": 2.825e-06,
154
+ "loss": 0.0009648638078942895,
155
+ "step": 8,
156
+ "step_time": 0.10721996099982789
157
+ },
158
+ {
159
+ "clip_ratio/high_max": 0.0,
160
+ "clip_ratio/high_mean": 0.0,
161
+ "clip_ratio/low_mean": 0.0,
162
+ "clip_ratio/low_min": 0.0,
163
+ "clip_ratio/region_mean": 0.0,
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": 1.9720534086227417,
172
+ "epoch": 0.225,
173
+ "frac_reward_zero_std": 0.25,
174
+ "grad_norm": 3.5232784748077393,
175
+ "kl": 0.06137782335281372,
176
+ "learning_rate": 2.8000000000000003e-06,
177
+ "loss": 0.02912595309317112,
178
+ "num_tokens": 29508.0,
179
+ "reward": 0.7273093461990356,
180
+ "reward_std": 0.18703272938728333,
181
+ "rewards/AnswererRewardFunction/mean": 0.7273093461990356,
182
+ "rewards/AnswererRewardFunction/std": 0.18703272938728333,
183
+ "step": 9,
184
+ "step_time": 7.9921620860000075
185
+ },
186
+ {
187
+ "clip_ratio/high_max": 0.0078125,
188
+ "clip_ratio/high_mean": 0.0078125,
189
+ "clip_ratio/low_mean": 0.0,
190
+ "clip_ratio/low_min": 0.0,
191
+ "clip_ratio/region_mean": 0.0078125,
192
+ "entropy": 1.8374459743499756,
193
+ "epoch": 0.25,
194
+ "grad_norm": 1.6457006931304932,
195
+ "kl": 0.06881466507911682,
196
+ "learning_rate": 2.775e-06,
197
+ "loss": -0.09420805424451828,
198
+ "step": 10,
199
+ "step_time": 0.11236686899974302
200
+ },
201
+ {
202
+ "clip_ratio/high_max": 0.005859375,
203
+ "clip_ratio/high_mean": 0.005859375,
204
+ "clip_ratio/low_mean": 0.0302734375,
205
+ "clip_ratio/low_min": 0.0302734375,
206
+ "clip_ratio/region_mean": 0.0361328125,
207
+ "entropy": 1.7282634973526,
208
+ "epoch": 0.275,
209
+ "grad_norm": 2.8160879611968994,
210
+ "kl": 0.15302874147891998,
211
+ "learning_rate": 2.75e-06,
212
+ "loss": 0.03125021979212761,
213
+ "step": 11,
214
+ "step_time": 0.11106374300015887
215
+ },
216
+ {
217
+ "clip_ratio/high_max": 0.044921875,
218
+ "clip_ratio/high_mean": 0.044921875,
219
+ "clip_ratio/low_mean": 0.01953125,
220
+ "clip_ratio/low_min": 0.01953125,
221
+ "clip_ratio/region_mean": 0.064453125,
222
+ "entropy": 1.6645623445510864,
223
+ "epoch": 0.3,
224
+ "grad_norm": 3.6318187713623047,
225
+ "kl": 0.06911536306142807,
226
+ "learning_rate": 2.725e-06,
227
+ "loss": 0.028275620192289352,
228
+ "step": 12,
229
+ "step_time": 0.11270152599990979
230
+ },
231
+ {
232
+ "clip_ratio/high_max": 0.001953125,
233
+ "clip_ratio/high_mean": 0.001953125,
234
+ "clip_ratio/low_mean": 0.0,
235
+ "clip_ratio/low_min": 0.0,
236
+ "clip_ratio/region_mean": 0.001953125,
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": 1.7028954029083252,
245
+ "epoch": 0.325,
246
+ "frac_reward_zero_std": 0.5,
247
+ "grad_norm": 1.2632755041122437,
248
+ "kl": 0.07647612690925598,
249
+ "learning_rate": 2.7e-06,
250
+ "loss": -0.05399465560913086,
251
+ "num_tokens": 39508.0,
252
+ "reward": 0.7219144701957703,
253
+ "reward_std": 0.1761765331029892,
254
+ "rewards/AnswererRewardFunction/mean": 0.7219144701957703,
255
+ "rewards/AnswererRewardFunction/std": 0.1761765331029892,
256
+ "step": 13,
257
+ "step_time": 8.053660665999814
258
+ },
259
+ {
260
+ "clip_ratio/high_max": 0.0048828125,
261
+ "clip_ratio/high_mean": 0.0048828125,
262
+ "clip_ratio/low_mean": 0.0009765625,
263
+ "clip_ratio/low_min": 0.0009765625,
264
+ "clip_ratio/region_mean": 0.005859375,
265
+ "entropy": 1.4734899997711182,
266
+ "epoch": 0.35,
267
+ "grad_norm": 1.8239141702651978,
268
+ "kl": 0.0759836882352829,
269
+ "learning_rate": 2.6750000000000002e-06,
270
+ "loss": -0.007084880489856005,
271
+ "step": 14,
272
+ "step_time": 0.1127117660003023
273
+ },
274
+ {
275
+ "clip_ratio/high_max": 0.0166015625,
276
+ "clip_ratio/high_mean": 0.0166015625,
277
+ "clip_ratio/low_mean": 0.0048828125,
278
+ "clip_ratio/low_min": 0.0048828125,
279
+ "clip_ratio/region_mean": 0.021484375,
280
+ "entropy": 1.5525939464569092,
281
+ "epoch": 0.375,
282
+ "grad_norm": 1.4309831857681274,
283
+ "kl": 0.10118035972118378,
284
+ "learning_rate": 2.65e-06,
285
+ "loss": 0.03022492118179798,
286
+ "step": 15,
287
+ "step_time": 0.14820013999997173
288
+ },
289
+ {
290
+ "clip_ratio/high_max": 0.046875,
291
+ "clip_ratio/high_mean": 0.046875,
292
+ "clip_ratio/low_mean": 0.013671875,
293
+ "clip_ratio/low_min": 0.013671875,
294
+ "clip_ratio/region_mean": 0.060546875,
295
+ "entropy": 1.9967842102050781,
296
+ "epoch": 0.4,
297
+ "grad_norm": 5.337355136871338,
298
+ "kl": 0.1319834589958191,
299
+ "learning_rate": 2.6250000000000003e-06,
300
+ "loss": 0.03607366606593132,
301
+ "step": 16,
302
+ "step_time": 0.11176063600032649
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": 2.0281615257263184,
318
+ "epoch": 0.425,
319
+ "frac_reward_zero_std": 0.25,
320
+ "grad_norm": 2.718438148498535,
321
+ "kl": 0.14551398158073425,
322
+ "learning_rate": 2.6e-06,
323
+ "loss": -0.17034515738487244,
324
+ "num_tokens": 49932.0,
325
+ "reward": 0.8668582439422607,
326
+ "reward_std": 0.1170896589756012,
327
+ "rewards/AnswererRewardFunction/mean": 0.8668582439422607,
328
+ "rewards/AnswererRewardFunction/std": 0.1170896664261818,
329
+ "step": 17,
330
+ "step_time": 7.967793787000119
331
+ },
332
+ {
333
+ "clip_ratio/high_max": 0.0,
334
+ "clip_ratio/high_mean": 0.0,
335
+ "clip_ratio/low_mean": 0.0048828125,
336
+ "clip_ratio/low_min": 0.0048828125,
337
+ "clip_ratio/region_mean": 0.0048828125,
338
+ "entropy": 2.224254846572876,
339
+ "epoch": 0.45,
340
+ "grad_norm": 4.844032287597656,
341
+ "kl": 0.15413105487823486,
342
+ "learning_rate": 2.575e-06,
343
+ "loss": 0.18968576192855835,
344
+ "step": 18,
345
+ "step_time": 0.11300960199969268
346
+ },
347
+ {
348
+ "clip_ratio/high_max": 0.0361328125,
349
+ "clip_ratio/high_mean": 0.0361328125,
350
+ "clip_ratio/low_mean": 0.00390625,
351
+ "clip_ratio/low_min": 0.00390625,
352
+ "clip_ratio/region_mean": 0.0400390625,
353
+ "entropy": 2.0003585815429688,
354
+ "epoch": 0.475,
355
+ "grad_norm": 2.2668349742889404,
356
+ "kl": 0.15107640624046326,
357
+ "learning_rate": 2.55e-06,
358
+ "loss": -0.039400599896907806,
359
+ "step": 19,
360
+ "step_time": 0.11324099700004808
361
+ },
362
+ {
363
+ "clip_ratio/high_max": 0.0185546875,
364
+ "clip_ratio/high_mean": 0.0185546875,
365
+ "clip_ratio/low_mean": 0.01171875,
366
+ "clip_ratio/low_min": 0.01171875,
367
+ "clip_ratio/region_mean": 0.0302734375,
368
+ "entropy": 2.0104799270629883,
369
+ "epoch": 0.5,
370
+ "grad_norm": 1.9012264013290405,
371
+ "kl": 0.17855022847652435,
372
+ "learning_rate": 2.525e-06,
373
+ "loss": 0.023593107238411903,
374
+ "step": 20,
375
+ "step_time": 0.11757079700009854
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": 1.998047947883606,
391
+ "epoch": 0.525,
392
+ "frac_reward_zero_std": 1.0,
393
+ "grad_norm": 0.05554308369755745,
394
+ "kl": 0.17033995687961578,
395
+ "learning_rate": 2.5e-06,
396
+ "loss": 0.00042584986658766866,
397
+ "num_tokens": 60100.0,
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": 8.04426064099971
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": 2.163236141204834,
412
+ "epoch": 0.55,
413
+ "grad_norm": 0.06196916475892067,
414
+ "kl": 0.19698408246040344,
415
+ "learning_rate": 2.475e-06,
416
+ "loss": 0.0004924602108076215,
417
+ "step": 22,
418
+ "step_time": 0.11172266499988837
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": 2.032647132873535,
427
+ "epoch": 0.575,
428
+ "grad_norm": 0.06370746344327927,
429
+ "kl": 0.20239636301994324,
430
+ "learning_rate": 2.45e-06,
431
+ "loss": 0.0005059909308329225,
432
+ "step": 23,
433
+ "step_time": 0.11149290999992445
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": 2.058699607849121,
442
+ "epoch": 0.6,
443
+ "grad_norm": 0.06226831674575806,
444
+ "kl": 0.174312561750412,
445
+ "learning_rate": 2.425e-06,
446
+ "loss": 0.00043578140321187675,
447
+ "step": 24,
448
+ "step_time": 0.11008746099969358
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": 1.8655509948730469,
464
+ "epoch": 0.625,
465
+ "frac_reward_zero_std": 1.0,
466
+ "grad_norm": 0.048708297312259674,
467
+ "kl": 0.16897301375865936,
468
+ "learning_rate": 2.4000000000000003e-06,
469
+ "loss": 0.00042243252391926944,
470
+ "num_tokens": 70088.0,
471
+ "reward": 0.8790740966796875,
472
+ "reward_std": 0.10388742387294769,
473
+ "rewards/AnswererRewardFunction/mean": 0.8790740966796875,
474
+ "rewards/AnswererRewardFunction/std": 0.1038874164223671,
475
+ "step": 25,
476
+ "step_time": 7.980882594999912
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": 1.8213839530944824,
485
+ "epoch": 0.65,
486
+ "grad_norm": 0.09992338716983795,
487
+ "kl": 0.20379483699798584,
488
+ "learning_rate": 2.375e-06,
489
+ "loss": 0.000509487115778029,
490
+ "step": 26,
491
+ "step_time": 0.11193690600021
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": 1.7698655128479004,
500
+ "epoch": 0.675,
501
+ "grad_norm": 0.14092643558979034,
502
+ "kl": 0.25070053339004517,
503
+ "learning_rate": 2.35e-06,
504
+ "loss": 0.0006267513381317258,
505
+ "step": 27,
506
+ "step_time": 0.11155138000003717
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": 2.0493922233581543,
515
+ "epoch": 0.7,
516
+ "grad_norm": 0.08153603225946426,
517
+ "kl": 0.2104545682668686,
518
+ "learning_rate": 2.325e-06,
519
+ "loss": 0.0005261364276520908,
520
+ "step": 28,
521
+ "step_time": 0.11111651900000652
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": 2.0664830207824707,
537
+ "epoch": 0.725,
538
+ "frac_reward_zero_std": 1.0,
539
+ "grad_norm": 0.0705447718501091,
540
+ "kl": 0.23184415698051453,
541
+ "learning_rate": 2.3000000000000004e-06,
542
+ "loss": 0.0005796104087494314,
543
+ "num_tokens": 80300.0,
544
+ "reward": 0.8702797293663025,
545
+ "reward_std": 0.11096905916929245,
546
+ "rewards/AnswererRewardFunction/mean": 0.8702797293663025,
547
+ "rewards/AnswererRewardFunction/std": 0.11096906661987305,
548
+ "step": 29,
549
+ "step_time": 7.967512303000149
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": 2.122178316116333,
558
+ "epoch": 0.75,
559
+ "grad_norm": 0.05567191541194916,
560
+ "kl": 0.2171575129032135,
561
+ "learning_rate": 2.275e-06,
562
+ "loss": 0.000542893772944808,
563
+ "step": 30,
564
+ "step_time": 0.11231415000020206
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": 1.939367413520813,
573
+ "epoch": 0.775,
574
+ "grad_norm": 0.05278945714235306,
575
+ "kl": 0.17951712012290955,
576
+ "learning_rate": 2.25e-06,
577
+ "loss": 0.0004487927653826773,
578
+ "step": 31,
579
+ "step_time": 0.11434970699974656
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": 1.7900925874710083,
588
+ "epoch": 0.8,
589
+ "grad_norm": 0.11904135346412659,
590
+ "kl": 0.22381603717803955,
591
+ "learning_rate": 2.2250000000000003e-06,
592
+ "loss": 0.0005595399998128414,
593
+ "step": 32,
594
+ "step_time": 0.11125125099988509
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": 2.1402230262756348,
610
+ "epoch": 0.825,
611
+ "frac_reward_zero_std": 0.75,
612
+ "grad_norm": 0.059307511895895004,
613
+ "kl": 0.19554322957992554,
614
+ "learning_rate": 2.1999999999999997e-06,
615
+ "loss": 0.0004888580879196525,
616
+ "num_tokens": 90780.0,
617
+ "reward": 0.9159495830535889,
618
+ "reward_std": 0.06545510143041611,
619
+ "rewards/AnswererRewardFunction/mean": 0.9159495830535889,
620
+ "rewards/AnswererRewardFunction/std": 0.0654551088809967,
621
+ "step": 33,
622
+ "step_time": 7.981194668000171
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": 1.8783634901046753,
631
+ "epoch": 0.85,
632
+ "grad_norm": 0.04647519066929817,
633
+ "kl": 0.17279550433158875,
634
+ "learning_rate": 2.175e-06,
635
+ "loss": 0.00043198870844207704,
636
+ "step": 34,
637
+ "step_time": 0.12349177899977803
638
+ },
639
+ {
640
+ "clip_ratio/high_max": 0.0009765625,
641
+ "clip_ratio/high_mean": 0.0009765625,
642
+ "clip_ratio/low_mean": 0.001953125,
643
+ "clip_ratio/low_min": 0.001953125,
644
+ "clip_ratio/region_mean": 0.0029296875,
645
+ "entropy": 2.043071746826172,
646
+ "epoch": 0.875,
647
+ "grad_norm": 3.108652353286743,
648
+ "kl": 0.17774607241153717,
649
+ "learning_rate": 2.15e-06,
650
+ "loss": 0.03142169862985611,
651
+ "step": 35,
652
+ "step_time": 0.11168615199994747
653
+ },
654
+ {
655
+ "clip_ratio/high_max": 0.001953125,
656
+ "clip_ratio/high_mean": 0.001953125,
657
+ "clip_ratio/low_mean": 0.0,
658
+ "clip_ratio/low_min": 0.0,
659
+ "clip_ratio/region_mean": 0.001953125,
660
+ "entropy": 2.1080150604248047,
661
+ "epoch": 0.9,
662
+ "grad_norm": 1.021546721458435,
663
+ "kl": 0.17375531792640686,
664
+ "learning_rate": 2.125e-06,
665
+ "loss": -0.03066263720393181,
666
+ "step": 36,
667
+ "step_time": 0.11152969900012977
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": 2.118429660797119,
683
+ "epoch": 0.925,
684
+ "frac_reward_zero_std": 1.0,
685
+ "grad_norm": 0.04497942700982094,
686
+ "kl": 0.18413573503494263,
687
+ "learning_rate": 2.1e-06,
688
+ "loss": 0.0004603393317665905,
689
+ "num_tokens": 100732.0,
690
+ "reward": 0.8883568644523621,
691
+ "reward_std": 0.08731486648321152,
692
+ "rewards/AnswererRewardFunction/mean": 0.8883568644523621,
693
+ "rewards/AnswererRewardFunction/std": 0.08731484413146973,
694
+ "step": 37,
695
+ "step_time": 7.960166341000331
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": 1.9833812713623047,
704
+ "epoch": 0.95,
705
+ "grad_norm": 0.04675867408514023,
706
+ "kl": 0.17259088158607483,
707
+ "learning_rate": 2.075e-06,
708
+ "loss": 0.00043147720862179995,
709
+ "step": 38,
710
+ "step_time": 0.10994235699990895
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": 1.8975489139556885,
719
+ "epoch": 0.975,
720
+ "grad_norm": 0.039946023374795914,
721
+ "kl": 0.16186058521270752,
722
+ "learning_rate": 2.0500000000000003e-06,
723
+ "loss": 0.0004046514513902366,
724
+ "step": 39,
725
+ "step_time": 0.10970632199996544
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": 1.8897861242294312,
734
+ "epoch": 1.0,
735
+ "grad_norm": 0.0415203720331192,
736
+ "kl": 0.15467771887779236,
737
+ "learning_rate": 2.025e-06,
738
+ "loss": 0.0003866942715831101,
739
+ "step": 40,
740
+ "step_time": 0.10993808700004593
741
+ },
742
+ {
743
+ "clip_ratio/high_max": 0.0009765625,
744
+ "clip_ratio/high_mean": 0.0009765625,
745
+ "clip_ratio/low_mean": 0.0,
746
+ "clip_ratio/low_min": 0.0,
747
+ "clip_ratio/region_mean": 0.0009765625,
748
+ "completions/clipped_ratio": 1.0,
749
+ "completions/max_length": 256.0,
750
+ "completions/max_terminated_length": 0.0,
751
+ "completions/mean_length": 256.0,
752
+ "completions/mean_terminated_length": 0.0,
753
+ "completions/min_length": 256.0,
754
+ "completions/min_terminated_length": 0.0,
755
+ "entropy": 1.8479175567626953,
756
+ "epoch": 1.025,
757
+ "frac_reward_zero_std": 0.75,
758
+ "grad_norm": 2.5839734077453613,
759
+ "kl": 0.17626172304153442,
760
+ "learning_rate": 2e-06,
761
+ "loss": -0.10722197592258453,
762
+ "num_tokens": 110236.0,
763
+ "reward": 0.741972804069519,
764
+ "reward_std": 0.0906248465180397,
765
+ "rewards/AnswererRewardFunction/mean": 0.741972804069519,
766
+ "rewards/AnswererRewardFunction/std": 0.0906248390674591,
767
+ "step": 41,
768
+ "step_time": 7.969326156000079
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": 1.7285118103027344,
777
+ "epoch": 1.05,
778
+ "grad_norm": 2.326667070388794,
779
+ "kl": 0.17114567756652832,
780
+ "learning_rate": 1.975e-06,
781
+ "loss": 0.10866044461727142,
782
+ "step": 42,
783
+ "step_time": 0.10928293899996788
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": 1.8972716331481934,
792
+ "epoch": 1.075,
793
+ "grad_norm": 0.04685656353831291,
794
+ "kl": 0.18068528175354004,
795
+ "learning_rate": 1.95e-06,
796
+ "loss": 0.0004517132183536887,
797
+ "step": 43,
798
+ "step_time": 0.10901798299983056
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": 1.759249210357666,
807
+ "epoch": 1.1,
808
+ "grad_norm": 0.0477583147585392,
809
+ "kl": 0.19557110965251923,
810
+ "learning_rate": 1.925e-06,
811
+ "loss": 0.0004889278206974268,
812
+ "step": 44,
813
+ "step_time": 0.10952626300013435
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": 1.993303656578064,
829
+ "epoch": 1.125,
830
+ "frac_reward_zero_std": 1.0,
831
+ "grad_norm": 0.041841983795166016,
832
+ "kl": 0.1679525524377823,
833
+ "learning_rate": 1.9e-06,
834
+ "loss": 0.00041988136945292354,
835
+ "num_tokens": 120228.0,
836
+ "reward": 0.8790740966796875,
837
+ "reward_std": 0.10388742387294769,
838
+ "rewards/AnswererRewardFunction/mean": 0.8790740966796875,
839
+ "rewards/AnswererRewardFunction/std": 0.1038874164223671,
840
+ "step": 45,
841
+ "step_time": 7.970393666000291
842
+ },
843
+ {
844
+ "clip_ratio/high_max": 0.0,
845
+ "clip_ratio/high_mean": 0.0,
846
+ "clip_ratio/low_mean": 0.0,
847
+ "clip_ratio/low_min": 0.0,
848
+ "clip_ratio/region_mean": 0.0,
849
+ "entropy": 1.984572410583496,
850
+ "epoch": 1.15,
851
+ "grad_norm": 0.05101794749498367,
852
+ "kl": 0.17519888281822205,
853
+ "learning_rate": 1.875e-06,
854
+ "loss": 0.00043799722334370017,
855
+ "step": 46,
856
+ "step_time": 0.1115572050002811
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": 1.7677072286605835,
865
+ "epoch": 1.175,
866
+ "grad_norm": 0.04441745951771736,
867
+ "kl": 0.1697729527950287,
868
+ "learning_rate": 1.85e-06,
869
+ "loss": 0.0004244323936291039,
870
+ "step": 47,
871
+ "step_time": 0.11457784000003812
872
+ },
873
+ {
874
+ "clip_ratio/high_max": 0.0,
875
+ "clip_ratio/high_mean": 0.0,
876
+ "clip_ratio/low_mean": 0.0,
877
+ "clip_ratio/low_min": 0.0,
878
+ "clip_ratio/region_mean": 0.0,
879
+ "entropy": 1.6262599229812622,
880
+ "epoch": 1.2,
881
+ "grad_norm": 0.03933388739824295,
882
+ "kl": 0.1536058932542801,
883
+ "learning_rate": 1.8249999999999999e-06,
884
+ "loss": 0.00038401474012061954,
885
+ "step": 48,
886
+ "step_time": 0.11322292199974981
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": 1.9592680931091309,
902
+ "epoch": 1.225,
903
+ "frac_reward_zero_std": 1.0,
904
+ "grad_norm": 0.050161559134721756,
905
+ "kl": 0.17297984659671783,
906
+ "learning_rate": 1.8e-06,
907
+ "loss": 0.00043244962580502033,
908
+ "num_tokens": 130268.0,
909
+ "reward": 0.9394216537475586,
910
+ "reward_std": 0.00734142167493701,
911
+ "rewards/AnswererRewardFunction/mean": 0.9394216537475586,
912
+ "rewards/AnswererRewardFunction/std": 0.00734142167493701,
913
+ "step": 49,
914
+ "step_time": 7.934003324000059
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": 1.9569470882415771,
923
+ "epoch": 1.25,
924
+ "grad_norm": 0.046731311827898026,
925
+ "kl": 0.1752794086933136,
926
+ "learning_rate": 1.7750000000000002e-06,
927
+ "loss": 0.00043819850543513894,
928
+ "step": 50,
929
+ "step_time": 0.11127928799987785
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": 1.9778821468353271,
938
+ "epoch": 1.275,
939
+ "grad_norm": 0.053713273257017136,
940
+ "kl": 0.16941609978675842,
941
+ "learning_rate": 1.7500000000000002e-06,
942
+ "loss": 0.0004235402448102832,
943
+ "step": 51,
944
+ "step_time": 0.11058060300001671
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": 1.870413064956665,
953
+ "epoch": 1.3,
954
+ "grad_norm": 0.04283243790268898,
955
+ "kl": 0.16937875747680664,
956
+ "learning_rate": 1.725e-06,
957
+ "loss": 0.00042344690882600844,
958
+ "step": 52,
959
+ "step_time": 0.11374437099993884
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": 1.931403636932373,
975
+ "epoch": 1.325,
976
+ "frac_reward_zero_std": 1.0,
977
+ "grad_norm": 0.059217408299446106,
978
+ "kl": 0.16964863240718842,
979
+ "learning_rate": 1.7e-06,
980
+ "loss": 0.0004241215356159955,
981
+ "num_tokens": 140452.0,
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": 7.961514715999783
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": 1.9520056247711182,
996
+ "epoch": 1.35,
997
+ "grad_norm": 0.04660027474164963,
998
+ "kl": 0.15617473423480988,
999
+ "learning_rate": 1.675e-06,
1000
+ "loss": 0.00039043682045303285,
1001
+ "step": 54,
1002
+ "step_time": 0.11305551600025865
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": 2.0161426067352295,
1011
+ "epoch": 1.375,
1012
+ "grad_norm": 0.05104615166783333,
1013
+ "kl": 0.1674293428659439,
1014
+ "learning_rate": 1.65e-06,
1015
+ "loss": 0.00041857335600070655,
1016
+ "step": 55,
1017
+ "step_time": 0.11170585599984406
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": 1.9064700603485107,
1026
+ "epoch": 1.4,
1027
+ "grad_norm": 0.03587495535612106,
1028
+ "kl": 0.14189420640468597,
1029
+ "learning_rate": 1.625e-06,
1030
+ "loss": 0.0003547355008777231,
1031
+ "step": 56,
1032
+ "step_time": 0.11207809500001531
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": 1.8830595016479492,
1048
+ "epoch": 1.425,
1049
+ "frac_reward_zero_std": 0.5,
1050
+ "grad_norm": 4.119225025177002,
1051
+ "kl": 0.15411227941513062,
1052
+ "learning_rate": 1.6e-06,
1053
+ "loss": 0.03162853792309761,
1054
+ "num_tokens": 150760.0,
1055
+ "reward": 0.9279924631118774,
1056
+ "reward_std": 0.08381706476211548,
1057
+ "rewards/AnswererRewardFunction/mean": 0.9279924631118774,
1058
+ "rewards/AnswererRewardFunction/std": 0.08381707966327667,
1059
+ "step": 57,
1060
+ "step_time": 7.956761510999968
1061
+ },
1062
+ {
1063
+ "clip_ratio/high_max": 0.0,
1064
+ "clip_ratio/high_mean": 0.0,
1065
+ "clip_ratio/low_mean": 0.0,
1066
+ "clip_ratio/low_min": 0.0,
1067
+ "clip_ratio/region_mean": 0.0,
1068
+ "entropy": 1.990787148475647,
1069
+ "epoch": 1.45,
1070
+ "grad_norm": 0.9625141620635986,
1071
+ "kl": 0.1472472846508026,
1072
+ "learning_rate": 1.5750000000000002e-06,
1073
+ "loss": -0.03082755208015442,
1074
+ "step": 58,
1075
+ "step_time": 0.11239712299993698
1076
+ },
1077
+ {
1078
+ "clip_ratio/high_max": 0.001953125,
1079
+ "clip_ratio/high_mean": 0.001953125,
1080
+ "clip_ratio/low_mean": 0.0009765625,
1081
+ "clip_ratio/low_min": 0.0009765625,
1082
+ "clip_ratio/region_mean": 0.0029296875,
1083
+ "entropy": 2.0103137493133545,
1084
+ "epoch": 1.475,
1085
+ "grad_norm": 1.3257628679275513,
1086
+ "kl": 0.13910706341266632,
1087
+ "learning_rate": 1.5500000000000002e-06,
1088
+ "loss": 0.0002091672213282436,
1089
+ "step": 59,
1090
+ "step_time": 0.12151307400017686
1091
+ },
1092
+ {
1093
+ "clip_ratio/high_max": 0.001953125,
1094
+ "clip_ratio/high_mean": 0.001953125,
1095
+ "clip_ratio/low_mean": 0.0068359375,
1096
+ "clip_ratio/low_min": 0.0068359375,
1097
+ "clip_ratio/region_mean": 0.0087890625,
1098
+ "entropy": 1.9193438291549683,
1099
+ "epoch": 1.5,
1100
+ "grad_norm": 1.3438688516616821,
1101
+ "kl": 0.14766687154769897,
1102
+ "learning_rate": 1.525e-06,
1103
+ "loss": -6.373357609845698e-05,
1104
+ "step": 60,
1105
+ "step_time": 0.11405404700008148
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": 1.8955051898956299,
1121
+ "epoch": 1.525,
1122
+ "frac_reward_zero_std": 1.0,
1123
+ "grad_norm": 0.0531984344124794,
1124
+ "kl": 0.1584109514951706,
1125
+ "learning_rate": 1.5e-06,
1126
+ "loss": 0.0003960274043492973,
1127
+ "num_tokens": 160980.0,
1128
+ "reward": 0.8702797293663025,
1129
+ "reward_std": 0.11096905916929245,
1130
+ "rewards/AnswererRewardFunction/mean": 0.8702797293663025,
1131
+ "rewards/AnswererRewardFunction/std": 0.11096906661987305,
1132
+ "step": 61,
1133
+ "step_time": 8.018725697999798
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": 1.7957987785339355,
1142
+ "epoch": 1.55,
1143
+ "grad_norm": 0.0425194650888443,
1144
+ "kl": 0.1376654952764511,
1145
+ "learning_rate": 1.4749999999999999e-06,
1146
+ "loss": 0.0003441637381911278,
1147
+ "step": 62,
1148
+ "step_time": 0.11205720399993879
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": 2.031898021697998,
1157
+ "epoch": 1.575,
1158
+ "grad_norm": 0.03276832774281502,
1159
+ "kl": 0.1293315291404724,
1160
+ "learning_rate": 1.45e-06,
1161
+ "loss": 0.0003233288007322699,
1162
+ "step": 63,
1163
+ "step_time": 0.11122407899983955
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": 1.7628273963928223,
1172
+ "epoch": 1.6,
1173
+ "grad_norm": 0.03524264320731163,
1174
+ "kl": 0.1332603096961975,
1175
+ "learning_rate": 1.425e-06,
1176
+ "loss": 0.0003331507614348084,
1177
+ "step": 64,
1178
+ "step_time": 0.11078836900014721
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": 1.7855920791625977,
1194
+ "epoch": 1.625,
1195
+ "frac_reward_zero_std": 1.0,
1196
+ "grad_norm": 0.03824622556567192,
1197
+ "kl": 0.1409425437450409,
1198
+ "learning_rate": 1.4000000000000001e-06,
1199
+ "loss": 0.00035235637915320694,
1200
+ "num_tokens": 171220.0,
1201
+ "reward": 0.8702797293663025,
1202
+ "reward_std": 0.11096905916929245,
1203
+ "rewards/AnswererRewardFunction/mean": 0.8702797293663025,
1204
+ "rewards/AnswererRewardFunction/std": 0.11096906661987305,
1205
+ "step": 65,
1206
+ "step_time": 7.966622833000201
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": 1.9455888271331787,
1215
+ "epoch": 1.65,
1216
+ "grad_norm": 0.03869345411658287,
1217
+ "kl": 0.12960346043109894,
1218
+ "learning_rate": 1.375e-06,
1219
+ "loss": 0.0003240086371079087,
1220
+ "step": 66,
1221
+ "step_time": 0.11310339100009514
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": 1.871221899986267,
1230
+ "epoch": 1.675,
1231
+ "grad_norm": 0.030493447557091713,
1232
+ "kl": 0.12974175810813904,
1233
+ "learning_rate": 1.35e-06,
1234
+ "loss": 0.0003243543906137347,
1235
+ "step": 67,
1236
+ "step_time": 0.11169741200001226
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": 1.7456235885620117,
1245
+ "epoch": 1.7,
1246
+ "grad_norm": 0.038182202726602554,
1247
+ "kl": 0.16313272714614868,
1248
+ "learning_rate": 1.325e-06,
1249
+ "loss": 0.00040783180156722665,
1250
+ "step": 68,
1251
+ "step_time": 0.11167954099983035
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": 1.9022241830825806,
1267
+ "epoch": 1.725,
1268
+ "frac_reward_zero_std": 1.0,
1269
+ "grad_norm": 0.03612837567925453,
1270
+ "kl": 0.1507774144411087,
1271
+ "learning_rate": 1.3e-06,
1272
+ "loss": 0.0003769434988498688,
1273
+ "num_tokens": 181128.0,
1274
+ "reward": 0.8372920751571655,
1275
+ "reward_std": 0.09813749045133591,
1276
+ "rewards/AnswererRewardFunction/mean": 0.8372920751571655,
1277
+ "rewards/AnswererRewardFunction/std": 0.0981374979019165,
1278
+ "step": 69,
1279
+ "step_time": 8.082920007000212
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": 1.7322332859039307,
1288
+ "epoch": 1.75,
1289
+ "grad_norm": 0.04425577074289322,
1290
+ "kl": 0.13741829991340637,
1291
+ "learning_rate": 1.275e-06,
1292
+ "loss": 0.0003435457474552095,
1293
+ "step": 70,
1294
+ "step_time": 0.11250203699955819
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": 1.7408225536346436,
1303
+ "epoch": 1.775,
1304
+ "grad_norm": 0.034772347658872604,
1305
+ "kl": 0.11738508939743042,
1306
+ "learning_rate": 1.25e-06,
1307
+ "loss": 0.00029346271185204387,
1308
+ "step": 71,
1309
+ "step_time": 0.11231103299996903
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": 2.035499095916748,
1318
+ "epoch": 1.8,
1319
+ "grad_norm": 0.032121725380420685,
1320
+ "kl": 0.11111746728420258,
1321
+ "learning_rate": 1.225e-06,
1322
+ "loss": 0.00027779367519542575,
1323
+ "step": 72,
1324
+ "step_time": 0.11149108700010402
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": 1.8566285371780396,
1340
+ "epoch": 1.825,
1341
+ "frac_reward_zero_std": 1.0,
1342
+ "grad_norm": 0.04663561284542084,
1343
+ "kl": 0.12325030565261841,
1344
+ "learning_rate": 1.2000000000000002e-06,
1345
+ "loss": 0.00030812574550509453,
1346
+ "num_tokens": 191192.0,
1347
+ "reward": 0.9358674883842468,
1348
+ "reward_std": 0.0063578663393855095,
1349
+ "rewards/AnswererRewardFunction/mean": 0.9358674883842468,
1350
+ "rewards/AnswererRewardFunction/std": 0.006357857491821051,
1351
+ "step": 73,
1352
+ "step_time": 8.056543536999925
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": 1.8830863237380981,
1361
+ "epoch": 1.85,
1362
+ "grad_norm": 0.030907513573765755,
1363
+ "kl": 0.1182084009051323,
1364
+ "learning_rate": 1.175e-06,
1365
+ "loss": 0.000295520992949605,
1366
+ "step": 74,
1367
+ "step_time": 0.11116390200004389
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": 1.971235990524292,
1376
+ "epoch": 1.875,
1377
+ "grad_norm": 0.03279021754860878,
1378
+ "kl": 0.11992711573839188,
1379
+ "learning_rate": 1.1500000000000002e-06,
1380
+ "loss": 0.0002998177951667458,
1381
+ "step": 75,
1382
+ "step_time": 0.11309119199995621
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": 1.9007563591003418,
1391
+ "epoch": 1.9,
1392
+ "grad_norm": 0.06583976000547409,
1393
+ "kl": 0.16469089686870575,
1394
+ "learning_rate": 1.125e-06,
1395
+ "loss": 0.00041172723285853863,
1396
+ "step": 76,
1397
+ "step_time": 0.12020159899975624
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": 1.791405200958252,
1413
+ "epoch": 1.925,
1414
+ "frac_reward_zero_std": 0.75,
1415
+ "grad_norm": 1.5868710279464722,
1416
+ "kl": 0.12453828006982803,
1417
+ "learning_rate": 1.0999999999999998e-06,
1418
+ "loss": -0.05364468693733215,
1419
+ "num_tokens": 201464.0,
1420
+ "reward": 0.8520751595497131,
1421
+ "reward_std": 0.10910354554653168,
1422
+ "rewards/AnswererRewardFunction/mean": 0.8520751595497131,
1423
+ "rewards/AnswererRewardFunction/std": 0.10910355299711227,
1424
+ "step": 77,
1425
+ "step_time": 7.989794482999969
1426
+ },
1427
+ {
1428
+ "clip_ratio/high_max": 0.0009765625,
1429
+ "clip_ratio/high_mean": 0.0009765625,
1430
+ "clip_ratio/low_mean": 0.0,
1431
+ "clip_ratio/low_min": 0.0,
1432
+ "clip_ratio/region_mean": 0.0009765625,
1433
+ "entropy": 1.7989006042480469,
1434
+ "epoch": 1.95,
1435
+ "grad_norm": 1.6091150045394897,
1436
+ "kl": 0.12970173358917236,
1437
+ "learning_rate": 1.075e-06,
1438
+ "loss": -0.05402720347046852,
1439
+ "step": 78,
1440
+ "step_time": 0.11137092700028006
1441
+ },
1442
+ {
1443
+ "clip_ratio/high_max": 0.0,
1444
+ "clip_ratio/high_mean": 0.0,
1445
+ "clip_ratio/low_mean": 0.0009765625,
1446
+ "clip_ratio/low_min": 0.0009765625,
1447
+ "clip_ratio/region_mean": 0.0009765625,
1448
+ "entropy": 1.7526617050170898,
1449
+ "epoch": 1.975,
1450
+ "grad_norm": 1.5686577558517456,
1451
+ "kl": 0.13482609391212463,
1452
+ "learning_rate": 1.05e-06,
1453
+ "loss": 0.05471678823232651,
1454
+ "step": 79,
1455
+ "step_time": 0.11093630699997448
1456
+ },
1457
+ {
1458
+ "clip_ratio/high_max": 0.0,
1459
+ "clip_ratio/high_mean": 0.0,
1460
+ "clip_ratio/low_mean": 0.001953125,
1461
+ "clip_ratio/low_min": 0.001953125,
1462
+ "clip_ratio/region_mean": 0.001953125,
1463
+ "entropy": 1.8272747993469238,
1464
+ "epoch": 2.0,
1465
+ "grad_norm": 1.6918859481811523,
1466
+ "kl": 0.10058464854955673,
1467
+ "learning_rate": 1.0250000000000001e-06,
1468
+ "loss": 0.05438198894262314,
1469
+ "step": 80,
1470
+ "step_time": 0.11253189099988958
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": 1.6942585706710815,
1486
+ "epoch": 2.025,
1487
+ "frac_reward_zero_std": 1.0,
1488
+ "grad_norm": 0.6853057146072388,
1489
+ "kl": 0.17735232412815094,
1490
+ "learning_rate": 1e-06,
1491
+ "loss": 0.00044338079169392586,
1492
+ "num_tokens": 211624.0,
1493
+ "reward": 0.9411008358001709,
1494
+ "reward_std": 0.015719598159193993,
1495
+ "rewards/AnswererRewardFunction/mean": 0.9411008358001709,
1496
+ "rewards/AnswererRewardFunction/std": 0.01571960747241974,
1497
+ "step": 81,
1498
+ "step_time": 8.016376127000058
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": 1.7056351900100708,
1507
+ "epoch": 2.05,
1508
+ "grad_norm": 0.034336455166339874,
1509
+ "kl": 0.12249764800071716,
1510
+ "learning_rate": 9.75e-07,
1511
+ "loss": 0.0003062441246584058,
1512
+ "step": 82,
1513
+ "step_time": 0.11230414100009511
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": 1.831735372543335,
1522
+ "epoch": 2.075,
1523
+ "grad_norm": 0.060906197875738144,
1524
+ "kl": 0.13377204537391663,
1525
+ "learning_rate": 9.5e-07,
1526
+ "loss": 0.00033443013671785593,
1527
+ "step": 83,
1528
+ "step_time": 0.11227076900013344
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": 1.661720871925354,
1537
+ "epoch": 2.1,
1538
+ "grad_norm": 0.03820506110787392,
1539
+ "kl": 0.11871674656867981,
1540
+ "learning_rate": 9.25e-07,
1541
+ "loss": 0.0002967918699141592,
1542
+ "step": 84,
1543
+ "step_time": 0.11260722900033215
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": 1.9599846601486206,
1559
+ "epoch": 2.125,
1560
+ "frac_reward_zero_std": 1.0,
1561
+ "grad_norm": 0.0332174077630043,
1562
+ "kl": 0.11171815544366837,
1563
+ "learning_rate": 9e-07,
1564
+ "loss": 0.0002792953746393323,
1565
+ "num_tokens": 221472.0,
1566
+ "reward": 0.877388060092926,
1567
+ "reward_std": 0.1153634712100029,
1568
+ "rewards/AnswererRewardFunction/mean": 0.877388060092926,
1569
+ "rewards/AnswererRewardFunction/std": 0.1153634786605835,
1570
+ "step": 85,
1571
+ "step_time": 7.952262335999876
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": 1.651448130607605,
1580
+ "epoch": 2.15,
1581
+ "grad_norm": 0.03528081998229027,
1582
+ "kl": 0.10771322250366211,
1583
+ "learning_rate": 8.750000000000001e-07,
1584
+ "loss": 0.00026928307488560677,
1585
+ "step": 86,
1586
+ "step_time": 0.11112980300003983
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": 1.8455110788345337,
1595
+ "epoch": 2.175,
1596
+ "grad_norm": 0.03294501453638077,
1597
+ "kl": 0.11150027811527252,
1598
+ "learning_rate": 8.5e-07,
1599
+ "loss": 0.0002787506964523345,
1600
+ "step": 87,
1601
+ "step_time": 0.11136706800016327
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": 1.8199677467346191,
1610
+ "epoch": 2.2,
1611
+ "grad_norm": 0.039475854486227036,
1612
+ "kl": 0.11303894221782684,
1613
+ "learning_rate": 8.25e-07,
1614
+ "loss": 0.000282597349723801,
1615
+ "step": 88,
1616
+ "step_time": 0.11021387400023741
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": 1.7036609649658203,
1632
+ "epoch": 2.225,
1633
+ "frac_reward_zero_std": 1.0,
1634
+ "grad_norm": 0.04740661010146141,
1635
+ "kl": 0.12941786646842957,
1636
+ "learning_rate": 8e-07,
1637
+ "loss": 0.000323544634738937,
1638
+ "num_tokens": 231280.0,
1639
+ "reward": 0.8227691054344177,
1640
+ "reward_std": 0.11510813981294632,
1641
+ "rewards/AnswererRewardFunction/mean": 0.8227691054344177,
1642
+ "rewards/AnswererRewardFunction/std": 0.11510813981294632,
1643
+ "step": 89,
1644
+ "step_time": 8.003638340999714
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": 1.7878626585006714,
1653
+ "epoch": 2.25,
1654
+ "grad_norm": 0.03674144670367241,
1655
+ "kl": 0.11576652526855469,
1656
+ "learning_rate": 7.750000000000001e-07,
1657
+ "loss": 0.00028941628988832235,
1658
+ "step": 90,
1659
+ "step_time": 0.11103760499963755
1660
+ }
1661
+ ],
1662
+ "logging_steps": 1,
1663
+ "max_steps": 120,
1664
+ "num_input_tokens_seen": 231280,
1665
+ "num_train_epochs": 3,
1666
+ "save_steps": 30,
1667
+ "stateful_callbacks": {
1668
+ "TrainerControl": {
1669
+ "args": {
1670
+ "should_epoch_stop": false,
1671
+ "should_evaluate": false,
1672
+ "should_log": false,
1673
+ "should_save": true,
1674
+ "should_training_stop": false
1675
+ },
1676
+ "attributes": {}
1677
+ }
1678
+ },
1679
+ "total_flos": 0.0,
1680
+ "train_batch_size": 4,
1681
+ "trial_name": null,
1682
+ "trial_params": null
1683
+ }
self_play_hf_l40s_full/round_003/answerer_train/checkpoint-90/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e14d48f850dc10b8e5263f794946cc41733613b5eaf53ccba69ef8a3d5687b3e
3
+ size 7249