Hugging Face's logo

Siddeshwar1625
/
osint-checkpoints

Safetensors
Model card Files
xet
 Community
Siddeshwar1625 commited on
Commit
6f76252
·
verified ·
1 Parent(s): 500744c

Intermediate checkpoint upload step=90 (generator_train)

Browse files
Files changed (12) hide show
  1. .gitattributes +1 -0
  2. self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/chat_template.jinja +54 -0
  3. self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/config.json +57 -0
  4. self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/generation_config.json +13 -0
  5. self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/model.safetensors +3 -0
  6. self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/optimizer.pt +3 -0
  7. self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/rng_state.pth +3 -0
  8. self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/scheduler.pt +3 -0
  9. self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/tokenizer.json +3 -0
  10. self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/tokenizer_config.json +32 -0
  11. self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/trainer_state.json +1683 -0
  12. self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/training_args.bin +3 -0
.gitattributes CHANGED
@@ -44,3 +44,4 @@ self_play_hf_a10g_train/round_002/answerer_train/checkpoint-50/tokenizer.json fi
44
  self_play_hf_a10g_train/round_002/answerer_train/final_model/tokenizer.json filter=lfs diff=lfs merge=lfs -text
45
  self_play_hf_l40s_full/round_001/generator_train/checkpoint-30/tokenizer.json filter=lfs diff=lfs merge=lfs -text
46
  self_play_hf_l40s_full/round_001/generator_train/checkpoint-60/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 
 
44
  self_play_hf_a10g_train/round_002/answerer_train/final_model/tokenizer.json filter=lfs diff=lfs merge=lfs -text
45
  self_play_hf_l40s_full/round_001/generator_train/checkpoint-30/tokenizer.json filter=lfs diff=lfs merge=lfs -text
46
  self_play_hf_l40s_full/round_001/generator_train/checkpoint-60/tokenizer.json filter=lfs diff=lfs merge=lfs -text
47
+ self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/tokenizer.json filter=lfs diff=lfs merge=lfs -text
self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/chat_template.jinja ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- if tools %}
2
+ {{- '<|im_start|>system\n' }}
3
+ {%- if messages[0]['role'] == 'system' %}
4
+ {{- messages[0]['content'] }}
5
+ {%- else %}
6
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
7
+ {%- endif %}
8
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
9
+ {%- for tool in tools %}
10
+ {{- "\n" }}
11
+ {{- tool | tojson }}
12
+ {%- endfor %}
13
+ {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
14
+ {%- else %}
15
+ {%- if messages[0]['role'] == 'system' %}
16
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
17
+ {%- else %}
18
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
19
+ {%- endif %}
20
+ {%- endif %}
21
+ {%- for message in messages %}
22
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
23
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
24
+ {%- elif message.role == "assistant" %}
25
+ {{- '<|im_start|>' + message.role }}
26
+ {%- if message.content %}
27
+ {{- '\n' + message.content }}
28
+ {%- endif %}
29
+ {%- for tool_call in message.tool_calls %}
30
+ {%- if tool_call.function is defined %}
31
+ {%- set tool_call = tool_call.function %}
32
+ {%- endif %}
33
+ {{- '\n<tool_call>\n{"name": "' }}
34
+ {{- tool_call.name }}
35
+ {{- '", "arguments": ' }}
36
+ {{- tool_call.arguments | tojson }}
37
+ {{- '}\n</tool_call>' }}
38
+ {%- endfor %}
39
+ {{- '<|im_end|>\n' }}
40
+ {%- elif message.role == "tool" %}
41
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
42
+ {{- '<|im_start|>user' }}
43
+ {%- endif %}
44
+ {{- '\n<tool_response>\n' }}
45
+ {{- message.content }}
46
+ {{- '\n</tool_response>' }}
47
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
48
+ {{- '<|im_end|>\n' }}
49
+ {%- endif %}
50
+ {%- endif %}
51
+ {%- endfor %}
52
+ {%- if add_generation_prompt %}
53
+ {{- '<|im_start|>assistant\n' }}
54
+ {%- endif %}
self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/config.json ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen2ForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": null,
7
+ "dtype": "float32",
8
+ "eos_token_id": 151645,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 896,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 4864,
13
+ "layer_types": [
14
+ "full_attention",
15
+ "full_attention",
16
+ "full_attention",
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention"
38
+ ],
39
+ "max_position_embeddings": 32768,
40
+ "max_window_layers": 21,
41
+ "model_type": "qwen2",
42
+ "num_attention_heads": 14,
43
+ "num_hidden_layers": 24,
44
+ "num_key_value_heads": 2,
45
+ "pad_token_id": 151643,
46
+ "rms_norm_eps": 1e-06,
47
+ "rope_parameters": {
48
+ "rope_theta": 1000000.0,
49
+ "rope_type": "default"
50
+ },
51
+ "sliding_window": null,
52
+ "tie_word_embeddings": true,
53
+ "transformers_version": "5.6.2",
54
+ "use_cache": false,
55
+ "use_sliding_window": false,
56
+ "vocab_size": 151936
57
+ }
self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/generation_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_sample": true,
3
+ "eos_token_id": [
4
+ 151645,
5
+ 151643
6
+ ],
7
+ "pad_token_id": 151643,
8
+ "repetition_penalty": 1.1,
9
+ "temperature": 0.7,
10
+ "top_k": 20,
11
+ "top_p": 0.8,
12
+ "transformers_version": "5.6.2"
13
+ }
self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:42ae053c5381392363c250ad3fb91be9feb240278a40228079a8490e0e0620ed
3
+ size 1976163472
self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:303fb039a5ea9a08957bcddce4c319522d651c6eff3b245111f5ce792dbcb0f9
3
+ size 3952509771
self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b5001a1413e890182ec459c1edbf745017684c89f471fbb0cafcaadbc1070a9f
3
+ size 14645
self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9466ecaa07ea9f7aa3b70a452615802059d7a87e6220a892a7f18c7703f928c0
3
+ size 1465
self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
3
+ size 11421892
self_play_hf_l40s_full/round_001/generator_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": false,
24
+ "local_files_only": false,
25
+ "model_max_length": 131072,
26
+ "pad_token": "<|endoftext|>",
27
+ "padding_side": "left",
28
+ "split_special_tokens": false,
29
+ "tokenizer_class": "Qwen2Tokenizer",
30
+ "truncation_side": "left",
31
+ "unk_token": null
32
+ }
self_play_hf_l40s_full/round_001/generator_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": 3.75,
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": 0.001953125,
14
+ "clip_ratio/high_mean": 0.001953125,
15
+ "clip_ratio/low_mean": 0.0,
16
+ "clip_ratio/low_min": 0.0,
17
+ "clip_ratio/region_mean": 0.001953125,
18
+ "completions/clipped_ratio": 1.0,
19
+ "completions/max_length": 384.0,
20
+ "completions/max_terminated_length": 0.0,
21
+ "completions/mean_length": 384.0,
22
+ "completions/mean_terminated_length": 0.0,
23
+ "completions/min_length": 384.0,
24
+ "completions/min_terminated_length": 0.0,
25
+ "entropy": 1.9362258911132812,
26
+ "epoch": 0.041666666666666664,
27
+ "frac_reward_zero_std": 0.0,
28
+ "grad_norm": 3.343637704849243,
29
+ "kl": 0.0005498419050127268,
30
+ "learning_rate": 5e-06,
31
+ "loss": 0.13995857536792755,
32
+ "num_tokens": 25244.0,
33
+ "reward": -0.4352343678474426,
34
+ "reward_std": 0.306624174118042,
35
+ "rewards/GeneratorRewardFunction/mean": -0.4352343678474426,
36
+ "rewards/GeneratorRewardFunction/std": 0.306624174118042,
37
+ "step": 1,
38
+ "step_time": 12.856334962999995
39
+ },
40
+ {
41
+ "clip_ratio/high_max": 0.00390625,
42
+ "clip_ratio/high_mean": 0.00390625,
43
+ "clip_ratio/low_mean": 0.0026041667442768812,
44
+ "clip_ratio/low_min": 0.0026041667442768812,
45
+ "clip_ratio/region_mean": 0.0065104165114462376,
46
+ "entropy": 1.2692097425460815,
47
+ "epoch": 0.08333333333333333,
48
+ "grad_norm": 2.8688511848449707,
49
+ "kl": 0.0015357633819803596,
50
+ "learning_rate": 4.958333333333334e-06,
51
+ "loss": -0.06688034534454346,
52
+ "step": 2,
53
+ "step_time": 0.21943785100000923
54
+ },
55
+ {
56
+ "clip_ratio/high_max": 0.011067708022892475,
57
+ "clip_ratio/high_mean": 0.011067708022892475,
58
+ "clip_ratio/low_mean": 0.009765625,
59
+ "clip_ratio/low_min": 0.009765625,
60
+ "clip_ratio/region_mean": 0.02083333395421505,
61
+ "entropy": 1.8704347610473633,
62
+ "epoch": 0.125,
63
+ "grad_norm": 2.4751532077789307,
64
+ "kl": 0.005275897681713104,
65
+ "learning_rate": 4.9166666666666665e-06,
66
+ "loss": 0.010688100941479206,
67
+ "step": 3,
68
+ "step_time": 0.21936836800000492
69
+ },
70
+ {
71
+ "clip_ratio/high_max": 0.01692708395421505,
72
+ "clip_ratio/high_mean": 0.01692708395421505,
73
+ "clip_ratio/low_mean": 0.01171875,
74
+ "clip_ratio/low_min": 0.01171875,
75
+ "clip_ratio/region_mean": 0.02864583395421505,
76
+ "entropy": 1.1848320960998535,
77
+ "epoch": 0.16666666666666666,
78
+ "grad_norm": 1.5594402551651,
79
+ "kl": 0.006554471794515848,
80
+ "learning_rate": 4.875e-06,
81
+ "loss": -0.08190122246742249,
82
+ "step": 4,
83
+ "step_time": 0.21970520599998622
84
+ },
85
+ {
86
+ "clip_ratio/high_max": 0.0006510416860692203,
87
+ "clip_ratio/high_mean": 0.0006510416860692203,
88
+ "clip_ratio/low_mean": 0.0006510416860692203,
89
+ "clip_ratio/low_min": 0.0006510416860692203,
90
+ "clip_ratio/region_mean": 0.0013020833721384406,
91
+ "completions/clipped_ratio": 1.0,
92
+ "completions/max_length": 384.0,
93
+ "completions/max_terminated_length": 0.0,
94
+ "completions/mean_length": 384.0,
95
+ "completions/mean_terminated_length": 0.0,
96
+ "completions/min_length": 384.0,
97
+ "completions/min_terminated_length": 0.0,
98
+ "entropy": 1.2801975011825562,
99
+ "epoch": 0.20833333333333334,
100
+ "frac_reward_zero_std": 0.0,
101
+ "grad_norm": 3.291785717010498,
102
+ "kl": 0.02072032354772091,
103
+ "learning_rate": 4.833333333333333e-06,
104
+ "loss": 0.0352378748357296,
105
+ "num_tokens": 50440.0,
106
+ "reward": -0.32179686427116394,
107
+ "reward_std": 0.29624873399734497,
108
+ "rewards/GeneratorRewardFunction/mean": -0.32179686427116394,
109
+ "rewards/GeneratorRewardFunction/std": 0.29624876379966736,
110
+ "step": 5,
111
+ "step_time": 12.408729128999994
112
+ },
113
+ {
114
+ "clip_ratio/high_max": 0.005859375,
115
+ "clip_ratio/high_mean": 0.005859375,
116
+ "clip_ratio/low_mean": 0.0032552082557231188,
117
+ "clip_ratio/low_min": 0.0032552082557231188,
118
+ "clip_ratio/region_mean": 0.009114583022892475,
119
+ "entropy": 1.0433989763259888,
120
+ "epoch": 0.25,
121
+ "grad_norm": 2.3185477256774902,
122
+ "kl": 0.012637416832149029,
123
+ "learning_rate": 4.791666666666668e-06,
124
+ "loss": 0.07621120661497116,
125
+ "step": 6,
126
+ "step_time": 0.2254061419999971
127
+ },
128
+ {
129
+ "clip_ratio/high_max": 0.014322916977107525,
130
+ "clip_ratio/high_mean": 0.014322916977107525,
131
+ "clip_ratio/low_mean": 0.0052083334885537624,
132
+ "clip_ratio/low_min": 0.0052083334885537624,
133
+ "clip_ratio/region_mean": 0.01953125,
134
+ "entropy": 1.1590553522109985,
135
+ "epoch": 0.2916666666666667,
136
+ "grad_norm": 2.0043411254882812,
137
+ "kl": 0.01664155162870884,
138
+ "learning_rate": 4.75e-06,
139
+ "loss": -0.08277323096990585,
140
+ "step": 7,
141
+ "step_time": 0.22571980800000802
142
+ },
143
+ {
144
+ "clip_ratio/high_max": 0.011067708022892475,
145
+ "clip_ratio/high_mean": 0.011067708022892475,
146
+ "clip_ratio/low_mean": 0.02018229104578495,
147
+ "clip_ratio/low_min": 0.02018229104578495,
148
+ "clip_ratio/region_mean": 0.03125,
149
+ "entropy": 1.6304351091384888,
150
+ "epoch": 0.3333333333333333,
151
+ "grad_norm": 2.157682418823242,
152
+ "kl": 0.025665713474154472,
153
+ "learning_rate": 4.708333333333334e-06,
154
+ "loss": -0.02718915045261383,
155
+ "step": 8,
156
+ "step_time": 0.22516641699999695
157
+ },
158
+ {
159
+ "clip_ratio/high_max": 0.0,
160
+ "clip_ratio/high_mean": 0.0,
161
+ "clip_ratio/low_mean": 0.0006510416860692203,
162
+ "clip_ratio/low_min": 0.0006510416860692203,
163
+ "clip_ratio/region_mean": 0.0006510416860692203,
164
+ "completions/clipped_ratio": 1.0,
165
+ "completions/max_length": 384.0,
166
+ "completions/max_terminated_length": 0.0,
167
+ "completions/mean_length": 384.0,
168
+ "completions/mean_terminated_length": 0.0,
169
+ "completions/min_length": 384.0,
170
+ "completions/min_terminated_length": 0.0,
171
+ "entropy": 1.5122060775756836,
172
+ "epoch": 0.375,
173
+ "frac_reward_zero_std": 0.0,
174
+ "grad_norm": 2.3004961013793945,
175
+ "kl": 0.02992842346429825,
176
+ "learning_rate": 4.666666666666667e-06,
177
+ "loss": -0.017293695360422134,
178
+ "num_tokens": 75884.0,
179
+ "reward": -0.28343749046325684,
180
+ "reward_std": 0.27557042241096497,
181
+ "rewards/GeneratorRewardFunction/mean": -0.28343749046325684,
182
+ "rewards/GeneratorRewardFunction/std": 0.27557045221328735,
183
+ "step": 9,
184
+ "step_time": 12.346420661999986
185
+ },
186
+ {
187
+ "clip_ratio/high_max": 0.02018229104578495,
188
+ "clip_ratio/high_mean": 0.02018229104578495,
189
+ "clip_ratio/low_mean": 0.0006510416860692203,
190
+ "clip_ratio/low_min": 0.0006510416860692203,
191
+ "clip_ratio/region_mean": 0.02083333395421505,
192
+ "entropy": 1.4475125074386597,
193
+ "epoch": 0.4166666666666667,
194
+ "grad_norm": 1.7099461555480957,
195
+ "kl": 0.04340241476893425,
196
+ "learning_rate": 4.625000000000001e-06,
197
+ "loss": -0.04388632997870445,
198
+ "step": 10,
199
+ "step_time": 0.22484948900000745
200
+ },
201
+ {
202
+ "clip_ratio/high_max": 0.01888020895421505,
203
+ "clip_ratio/high_mean": 0.01888020895421505,
204
+ "clip_ratio/low_mean": 0.0065104165114462376,
205
+ "clip_ratio/low_min": 0.0065104165114462376,
206
+ "clip_ratio/region_mean": 0.025390625,
207
+ "entropy": 1.113129734992981,
208
+ "epoch": 0.4583333333333333,
209
+ "grad_norm": 2.5379157066345215,
210
+ "kl": 0.0384308397769928,
211
+ "learning_rate": 4.583333333333333e-06,
212
+ "loss": 0.06734556704759598,
213
+ "step": 11,
214
+ "step_time": 0.2249934319999909
215
+ },
216
+ {
217
+ "clip_ratio/high_max": 0.0364583320915699,
218
+ "clip_ratio/high_mean": 0.0364583320915699,
219
+ "clip_ratio/low_mean": 0.013671875,
220
+ "clip_ratio/low_min": 0.013671875,
221
+ "clip_ratio/region_mean": 0.0501302070915699,
222
+ "entropy": 1.2381106615066528,
223
+ "epoch": 0.5,
224
+ "grad_norm": 2.9557695388793945,
225
+ "kl": 0.05566346272826195,
226
+ "learning_rate": 4.541666666666667e-06,
227
+ "loss": -0.005549523513764143,
228
+ "step": 12,
229
+ "step_time": 0.22474803699998347
230
+ },
231
+ {
232
+ "clip_ratio/high_max": 0.0006510416860692203,
233
+ "clip_ratio/high_mean": 0.0006510416860692203,
234
+ "clip_ratio/low_mean": 0.0006510416860692203,
235
+ "clip_ratio/low_min": 0.0006510416860692203,
236
+ "clip_ratio/region_mean": 0.0013020833721384406,
237
+ "completions/clipped_ratio": 1.0,
238
+ "completions/max_length": 384.0,
239
+ "completions/max_terminated_length": 0.0,
240
+ "completions/mean_length": 384.0,
241
+ "completions/mean_terminated_length": 0.0,
242
+ "completions/min_length": 384.0,
243
+ "completions/min_terminated_length": 0.0,
244
+ "entropy": 1.487646222114563,
245
+ "epoch": 0.5416666666666666,
246
+ "frac_reward_zero_std": 0.0,
247
+ "grad_norm": 3.007843494415283,
248
+ "kl": 0.057869553565979004,
249
+ "learning_rate": 4.5e-06,
250
+ "loss": 0.007636225316673517,
251
+ "num_tokens": 101112.0,
252
+ "reward": -0.13499999046325684,
253
+ "reward_std": 0.20259153842926025,
254
+ "rewards/GeneratorRewardFunction/mean": -0.13499999046325684,
255
+ "rewards/GeneratorRewardFunction/std": 0.20259153842926025,
256
+ "step": 13,
257
+ "step_time": 12.321908172000008
258
+ },
259
+ {
260
+ "clip_ratio/high_max": 0.0065104165114462376,
261
+ "clip_ratio/high_mean": 0.0065104165114462376,
262
+ "clip_ratio/low_mean": 0.0,
263
+ "clip_ratio/low_min": 0.0,
264
+ "clip_ratio/region_mean": 0.0065104165114462376,
265
+ "entropy": 1.9252228736877441,
266
+ "epoch": 0.5833333333333334,
267
+ "grad_norm": 2.5925612449645996,
268
+ "kl": 0.12386713176965714,
269
+ "learning_rate": 4.4583333333333336e-06,
270
+ "loss": -0.14144371449947357,
271
+ "step": 14,
272
+ "step_time": 0.2183703009999931
273
+ },
274
+ {
275
+ "clip_ratio/high_max": 0.01953125,
276
+ "clip_ratio/high_mean": 0.01953125,
277
+ "clip_ratio/low_mean": 0.0078125,
278
+ "clip_ratio/low_min": 0.0078125,
279
+ "clip_ratio/region_mean": 0.02734375,
280
+ "entropy": 1.9338630437850952,
281
+ "epoch": 0.625,
282
+ "grad_norm": 2.6728527545928955,
283
+ "kl": 0.06178775057196617,
284
+ "learning_rate": 4.416666666666667e-06,
285
+ "loss": 0.029460331425070763,
286
+ "step": 15,
287
+ "step_time": 0.21820302699998706
288
+ },
289
+ {
290
+ "clip_ratio/high_max": 0.02083333395421505,
291
+ "clip_ratio/high_mean": 0.02083333395421505,
292
+ "clip_ratio/low_mean": 0.015625,
293
+ "clip_ratio/low_min": 0.015625,
294
+ "clip_ratio/region_mean": 0.0364583320915699,
295
+ "entropy": 1.4223586320877075,
296
+ "epoch": 0.6666666666666666,
297
+ "grad_norm": 2.478961944580078,
298
+ "kl": 0.07023113965988159,
299
+ "learning_rate": 4.3750000000000005e-06,
300
+ "loss": 0.10646841675043106,
301
+ "step": 16,
302
+ "step_time": 0.21740093799999727
303
+ },
304
+ {
305
+ "clip_ratio/high_max": 0.0,
306
+ "clip_ratio/high_mean": 0.0,
307
+ "clip_ratio/low_mean": 0.0,
308
+ "clip_ratio/low_min": 0.0,
309
+ "clip_ratio/region_mean": 0.0,
310
+ "completions/clipped_ratio": 0.9375,
311
+ "completions/max_length": 384.0,
312
+ "completions/max_terminated_length": 384.0,
313
+ "completions/mean_length": 384.0,
314
+ "completions/mean_terminated_length": 384.0,
315
+ "completions/min_length": 384.0,
316
+ "completions/min_terminated_length": 384.0,
317
+ "entropy": 1.4837397336959839,
318
+ "epoch": 0.7083333333333334,
319
+ "frac_reward_zero_std": 0.25,
320
+ "grad_norm": 1.4059587717056274,
321
+ "kl": 0.06152656301856041,
322
+ "learning_rate": 4.333333333333334e-06,
323
+ "loss": -0.09148877114057541,
324
+ "num_tokens": 126436.0,
325
+ "reward": -0.1550000011920929,
326
+ "reward_std": 0.23754999041557312,
327
+ "rewards/GeneratorRewardFunction/mean": -0.1550000011920929,
328
+ "rewards/GeneratorRewardFunction/std": 0.23754999041557312,
329
+ "step": 17,
330
+ "step_time": 12.288484535999999
331
+ },
332
+ {
333
+ "clip_ratio/high_max": 0.0078125,
334
+ "clip_ratio/high_mean": 0.0078125,
335
+ "clip_ratio/low_mean": 0.0013020833721384406,
336
+ "clip_ratio/low_min": 0.0013020833721384406,
337
+ "clip_ratio/region_mean": 0.009114583022892475,
338
+ "entropy": 1.9985771179199219,
339
+ "epoch": 0.75,
340
+ "grad_norm": 2.826540470123291,
341
+ "kl": 0.07520538568496704,
342
+ "learning_rate": 4.2916666666666665e-06,
343
+ "loss": 0.040274348109960556,
344
+ "step": 18,
345
+ "step_time": 0.22446858000000702
346
+ },
347
+ {
348
+ "clip_ratio/high_max": 0.0234375,
349
+ "clip_ratio/high_mean": 0.0234375,
350
+ "clip_ratio/low_mean": 0.0,
351
+ "clip_ratio/low_min": 0.0,
352
+ "clip_ratio/region_mean": 0.0234375,
353
+ "entropy": 1.5372291803359985,
354
+ "epoch": 0.7916666666666666,
355
+ "grad_norm": 1.147870659828186,
356
+ "kl": 0.08851856738328934,
357
+ "learning_rate": 4.25e-06,
358
+ "loss": -0.0608501173555851,
359
+ "step": 19,
360
+ "step_time": 0.22480693100001758
361
+ },
362
+ {
363
+ "clip_ratio/high_max": 0.01627604104578495,
364
+ "clip_ratio/high_mean": 0.01627604104578495,
365
+ "clip_ratio/low_mean": 0.01171875,
366
+ "clip_ratio/low_min": 0.01171875,
367
+ "clip_ratio/region_mean": 0.02799479104578495,
368
+ "entropy": 1.5571645498275757,
369
+ "epoch": 0.8333333333333334,
370
+ "grad_norm": 4.398966312408447,
371
+ "kl": 0.06785543262958527,
372
+ "learning_rate": 4.208333333333333e-06,
373
+ "loss": 0.11559239774942398,
374
+ "step": 20,
375
+ "step_time": 0.2247580810000045
376
+ },
377
+ {
378
+ "clip_ratio/high_max": 0.0006510416860692203,
379
+ "clip_ratio/high_mean": 0.0006510416860692203,
380
+ "clip_ratio/low_mean": 0.0,
381
+ "clip_ratio/low_min": 0.0,
382
+ "clip_ratio/region_mean": 0.0006510416860692203,
383
+ "completions/clipped_ratio": 1.0,
384
+ "completions/max_length": 384.0,
385
+ "completions/max_terminated_length": 0.0,
386
+ "completions/mean_length": 384.0,
387
+ "completions/mean_terminated_length": 0.0,
388
+ "completions/min_length": 384.0,
389
+ "completions/min_terminated_length": 0.0,
390
+ "entropy": 1.9454313516616821,
391
+ "epoch": 0.875,
392
+ "frac_reward_zero_std": 0.0,
393
+ "grad_norm": 3.373609781265259,
394
+ "kl": 0.09310229867696762,
395
+ "learning_rate": 4.166666666666667e-06,
396
+ "loss": 0.05512508749961853,
397
+ "num_tokens": 151596.0,
398
+ "reward": -0.31937500834465027,
399
+ "reward_std": 0.2807067036628723,
400
+ "rewards/GeneratorRewardFunction/mean": -0.31937500834465027,
401
+ "rewards/GeneratorRewardFunction/std": 0.2807067036628723,
402
+ "step": 21,
403
+ "step_time": 12.236714247999998
404
+ },
405
+ {
406
+ "clip_ratio/high_max": 0.0045572915114462376,
407
+ "clip_ratio/high_mean": 0.0045572915114462376,
408
+ "clip_ratio/low_mean": 0.0,
409
+ "clip_ratio/low_min": 0.0,
410
+ "clip_ratio/region_mean": 0.0045572915114462376,
411
+ "entropy": 1.5520371198654175,
412
+ "epoch": 0.9166666666666666,
413
+ "grad_norm": 2.0294437408447266,
414
+ "kl": 0.07163921743631363,
415
+ "learning_rate": 4.125e-06,
416
+ "loss": -0.12930750846862793,
417
+ "step": 22,
418
+ "step_time": 0.22085711700000843
419
+ },
420
+ {
421
+ "clip_ratio/high_max": 0.0071614584885537624,
422
+ "clip_ratio/high_mean": 0.0071614584885537624,
423
+ "clip_ratio/low_mean": 0.0,
424
+ "clip_ratio/low_min": 0.0,
425
+ "clip_ratio/region_mean": 0.0071614584885537624,
426
+ "entropy": 2.460364580154419,
427
+ "epoch": 0.9583333333333334,
428
+ "grad_norm": 3.18788743019104,
429
+ "kl": 0.08949334174394608,
430
+ "learning_rate": 4.083333333333334e-06,
431
+ "loss": -0.10965806245803833,
432
+ "step": 23,
433
+ "step_time": 0.21995270400000777
434
+ },
435
+ {
436
+ "clip_ratio/high_max": 0.0,
437
+ "clip_ratio/high_mean": 0.0,
438
+ "clip_ratio/low_mean": 0.011067708022892475,
439
+ "clip_ratio/low_min": 0.011067708022892475,
440
+ "clip_ratio/region_mean": 0.011067708022892475,
441
+ "entropy": 1.6922578811645508,
442
+ "epoch": 1.0,
443
+ "grad_norm": 3.1108548641204834,
444
+ "kl": 0.09264727681875229,
445
+ "learning_rate": 4.041666666666667e-06,
446
+ "loss": 0.18657775223255157,
447
+ "step": 24,
448
+ "step_time": 0.22040663499998914
449
+ },
450
+ {
451
+ "clip_ratio/high_max": 0.0006510416860692203,
452
+ "clip_ratio/high_mean": 0.0006510416860692203,
453
+ "clip_ratio/low_mean": 0.0,
454
+ "clip_ratio/low_min": 0.0,
455
+ "clip_ratio/region_mean": 0.0006510416860692203,
456
+ "completions/clipped_ratio": 1.0,
457
+ "completions/max_length": 384.0,
458
+ "completions/max_terminated_length": 0.0,
459
+ "completions/mean_length": 384.0,
460
+ "completions/mean_terminated_length": 0.0,
461
+ "completions/min_length": 384.0,
462
+ "completions/min_terminated_length": 0.0,
463
+ "entropy": 1.0486950874328613,
464
+ "epoch": 1.0416666666666667,
465
+ "frac_reward_zero_std": 0.25,
466
+ "grad_norm": 2.6160550117492676,
467
+ "kl": 0.11301010102033615,
468
+ "learning_rate": 4.000000000000001e-06,
469
+ "loss": -0.11105048656463623,
470
+ "num_tokens": 177212.0,
471
+ "reward": -0.11343750357627869,
472
+ "reward_std": 0.16382630169391632,
473
+ "rewards/GeneratorRewardFunction/mean": -0.11343750357627869,
474
+ "rewards/GeneratorRewardFunction/std": 0.16382630169391632,
475
+ "step": 25,
476
+ "step_time": 12.298926971000014
477
+ },
478
+ {
479
+ "clip_ratio/high_max": 0.001953125,
480
+ "clip_ratio/high_mean": 0.001953125,
481
+ "clip_ratio/low_mean": 0.0,
482
+ "clip_ratio/low_min": 0.0,
483
+ "clip_ratio/region_mean": 0.001953125,
484
+ "entropy": 2.317030191421509,
485
+ "epoch": 1.0833333333333333,
486
+ "grad_norm": 1.2411166429519653,
487
+ "kl": 0.1501876264810562,
488
+ "learning_rate": 3.958333333333333e-06,
489
+ "loss": -0.021454373374581337,
490
+ "step": 26,
491
+ "step_time": 0.22782803399999807
492
+ },
493
+ {
494
+ "clip_ratio/high_max": 0.013671875,
495
+ "clip_ratio/high_mean": 0.013671875,
496
+ "clip_ratio/low_mean": 0.0026041667442768812,
497
+ "clip_ratio/low_min": 0.0026041667442768812,
498
+ "clip_ratio/region_mean": 0.01627604104578495,
499
+ "entropy": 1.8395847082138062,
500
+ "epoch": 1.125,
501
+ "grad_norm": 2.227196216583252,
502
+ "kl": 0.10138175636529922,
503
+ "learning_rate": 3.916666666666667e-06,
504
+ "loss": -0.044178307056427,
505
+ "step": 27,
506
+ "step_time": 0.22749500599999806
507
+ },
508
+ {
509
+ "clip_ratio/high_max": 0.01953125,
510
+ "clip_ratio/high_mean": 0.01953125,
511
+ "clip_ratio/low_mean": 0.008463541977107525,
512
+ "clip_ratio/low_min": 0.008463541977107525,
513
+ "clip_ratio/region_mean": 0.02799479104578495,
514
+ "entropy": 2.2681424617767334,
515
+ "epoch": 1.1666666666666667,
516
+ "grad_norm": 3.8757026195526123,
517
+ "kl": 0.12476732581853867,
518
+ "learning_rate": 3.875e-06,
519
+ "loss": 0.1814102679491043,
520
+ "step": 28,
521
+ "step_time": 0.22761540100000843
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": 384.0,
531
+ "completions/max_terminated_length": 0.0,
532
+ "completions/mean_length": 384.0,
533
+ "completions/mean_terminated_length": 0.0,
534
+ "completions/min_length": 384.0,
535
+ "completions/min_terminated_length": 0.0,
536
+ "entropy": 2.2231695652008057,
537
+ "epoch": 1.2083333333333333,
538
+ "frac_reward_zero_std": 0.0,
539
+ "grad_norm": 1.559815764427185,
540
+ "kl": 0.1458280235528946,
541
+ "learning_rate": 3.833333333333334e-06,
542
+ "loss": -0.037189874798059464,
543
+ "num_tokens": 202200.0,
544
+ "reward": -0.13124999403953552,
545
+ "reward_std": 0.1798378825187683,
546
+ "rewards/GeneratorRewardFunction/mean": -0.13124999403953552,
547
+ "rewards/GeneratorRewardFunction/std": 0.1798378825187683,
548
+ "step": 29,
549
+ "step_time": 12.206748850999986
550
+ },
551
+ {
552
+ "clip_ratio/high_max": 0.00390625,
553
+ "clip_ratio/high_mean": 0.00390625,
554
+ "clip_ratio/low_mean": 0.0006510416860692203,
555
+ "clip_ratio/low_min": 0.0006510416860692203,
556
+ "clip_ratio/region_mean": 0.0045572915114462376,
557
+ "entropy": 1.855025291442871,
558
+ "epoch": 1.25,
559
+ "grad_norm": 2.0162570476531982,
560
+ "kl": 0.1043163314461708,
561
+ "learning_rate": 3.7916666666666666e-06,
562
+ "loss": -0.09465664625167847,
563
+ "step": 30,
564
+ "step_time": 0.2218639939999889
565
+ },
566
+ {
567
+ "clip_ratio/high_max": 0.014322916977107525,
568
+ "clip_ratio/high_mean": 0.014322916977107525,
569
+ "clip_ratio/low_mean": 0.0,
570
+ "clip_ratio/low_min": 0.0,
571
+ "clip_ratio/region_mean": 0.014322916977107525,
572
+ "entropy": 1.8362489938735962,
573
+ "epoch": 1.2916666666666667,
574
+ "grad_norm": 2.5384345054626465,
575
+ "kl": 0.13157621026039124,
576
+ "learning_rate": 3.7500000000000005e-06,
577
+ "loss": 0.014540579169988632,
578
+ "step": 31,
579
+ "step_time": 0.2185786840000219
580
+ },
581
+ {
582
+ "clip_ratio/high_max": 0.0078125,
583
+ "clip_ratio/high_mean": 0.0078125,
584
+ "clip_ratio/low_mean": 0.0032552082557231188,
585
+ "clip_ratio/low_min": 0.0032552082557231188,
586
+ "clip_ratio/region_mean": 0.011067708022892475,
587
+ "entropy": 2.758424758911133,
588
+ "epoch": 1.3333333333333333,
589
+ "grad_norm": 4.852933883666992,
590
+ "kl": 0.12440217286348343,
591
+ "learning_rate": 3.708333333333334e-06,
592
+ "loss": 0.12054908275604248,
593
+ "step": 32,
594
+ "step_time": 0.21956204600002138
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": 384.0,
604
+ "completions/max_terminated_length": 0.0,
605
+ "completions/mean_length": 384.0,
606
+ "completions/mean_terminated_length": 0.0,
607
+ "completions/min_length": 384.0,
608
+ "completions/min_terminated_length": 0.0,
609
+ "entropy": 1.9736067056655884,
610
+ "epoch": 1.375,
611
+ "frac_reward_zero_std": 0.0,
612
+ "grad_norm": 2.6751697063446045,
613
+ "kl": 0.13457757234573364,
614
+ "learning_rate": 3.6666666666666666e-06,
615
+ "loss": -0.033180128782987595,
616
+ "num_tokens": 227556.0,
617
+ "reward": -0.16562500596046448,
618
+ "reward_std": 0.21113088726997375,
619
+ "rewards/GeneratorRewardFunction/mean": -0.16562500596046448,
620
+ "rewards/GeneratorRewardFunction/std": 0.21113088726997375,
621
+ "step": 33,
622
+ "step_time": 12.259839492000026
623
+ },
624
+ {
625
+ "clip_ratio/high_max": 0.00390625,
626
+ "clip_ratio/high_mean": 0.00390625,
627
+ "clip_ratio/low_mean": 0.0013020833721384406,
628
+ "clip_ratio/low_min": 0.0013020833721384406,
629
+ "clip_ratio/region_mean": 0.0052083334885537624,
630
+ "entropy": 2.609300374984741,
631
+ "epoch": 1.4166666666666667,
632
+ "grad_norm": 2.6082329750061035,
633
+ "kl": 0.14972178637981415,
634
+ "learning_rate": 3.625e-06,
635
+ "loss": 0.043182551860809326,
636
+ "step": 34,
637
+ "step_time": 0.22564089899998407
638
+ },
639
+ {
640
+ "clip_ratio/high_max": 0.0032552082557231188,
641
+ "clip_ratio/high_mean": 0.0032552082557231188,
642
+ "clip_ratio/low_mean": 0.0052083334885537624,
643
+ "clip_ratio/low_min": 0.0052083334885537624,
644
+ "clip_ratio/region_mean": 0.008463541977107525,
645
+ "entropy": 1.7806358337402344,
646
+ "epoch": 1.4583333333333333,
647
+ "grad_norm": 2.4095988273620605,
648
+ "kl": 0.16750717163085938,
649
+ "learning_rate": 3.5833333333333335e-06,
650
+ "loss": 0.028078051283955574,
651
+ "step": 35,
652
+ "step_time": 0.22601868800001057
653
+ },
654
+ {
655
+ "clip_ratio/high_max": 0.009765625,
656
+ "clip_ratio/high_mean": 0.009765625,
657
+ "clip_ratio/low_mean": 0.0026041667442768812,
658
+ "clip_ratio/low_min": 0.0026041667442768812,
659
+ "clip_ratio/region_mean": 0.012369791977107525,
660
+ "entropy": 2.1615936756134033,
661
+ "epoch": 1.5,
662
+ "grad_norm": 2.4958155155181885,
663
+ "kl": 0.14145778119564056,
664
+ "learning_rate": 3.5416666666666673e-06,
665
+ "loss": -0.03617622330784798,
666
+ "step": 36,
667
+ "step_time": 0.22432236999998167
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": 384.0,
677
+ "completions/max_terminated_length": 0.0,
678
+ "completions/mean_length": 384.0,
679
+ "completions/mean_terminated_length": 0.0,
680
+ "completions/min_length": 384.0,
681
+ "completions/min_terminated_length": 0.0,
682
+ "entropy": 1.9265685081481934,
683
+ "epoch": 1.5416666666666665,
684
+ "frac_reward_zero_std": 0.0,
685
+ "grad_norm": 2.3400332927703857,
686
+ "kl": 0.14584331214427948,
687
+ "learning_rate": 3.5e-06,
688
+ "loss": -0.06899306923151016,
689
+ "num_tokens": 252640.0,
690
+ "reward": -0.16875000298023224,
691
+ "reward_std": 0.22686631977558136,
692
+ "rewards/GeneratorRewardFunction/mean": -0.16875000298023224,
693
+ "rewards/GeneratorRewardFunction/std": 0.22686634957790375,
694
+ "step": 37,
695
+ "step_time": 12.545449219000034
696
+ },
697
+ {
698
+ "clip_ratio/high_max": 0.001953125,
699
+ "clip_ratio/high_mean": 0.001953125,
700
+ "clip_ratio/low_mean": 0.0006510416860692203,
701
+ "clip_ratio/low_min": 0.0006510416860692203,
702
+ "clip_ratio/region_mean": 0.0026041667442768812,
703
+ "entropy": 2.562986373901367,
704
+ "epoch": 1.5833333333333335,
705
+ "grad_norm": 2.3512234687805176,
706
+ "kl": 0.16302359104156494,
707
+ "learning_rate": 3.4583333333333334e-06,
708
+ "loss": -0.06315269321203232,
709
+ "step": 38,
710
+ "step_time": 0.2186768419999794
711
+ },
712
+ {
713
+ "clip_ratio/high_max": 0.00390625,
714
+ "clip_ratio/high_mean": 0.00390625,
715
+ "clip_ratio/low_mean": 0.0026041667442768812,
716
+ "clip_ratio/low_min": 0.0026041667442768812,
717
+ "clip_ratio/region_mean": 0.0065104165114462376,
718
+ "entropy": 1.8095722198486328,
719
+ "epoch": 1.625,
720
+ "grad_norm": 4.144832611083984,
721
+ "kl": 0.17072321474552155,
722
+ "learning_rate": 3.416666666666667e-06,
723
+ "loss": 0.20085881650447845,
724
+ "step": 39,
725
+ "step_time": 0.2193114469999955
726
+ },
727
+ {
728
+ "clip_ratio/high_max": 0.017578125,
729
+ "clip_ratio/high_mean": 0.017578125,
730
+ "clip_ratio/low_mean": 0.001953125,
731
+ "clip_ratio/low_min": 0.001953125,
732
+ "clip_ratio/region_mean": 0.01953125,
733
+ "entropy": 1.678245186805725,
734
+ "epoch": 1.6666666666666665,
735
+ "grad_norm": 2.2242610454559326,
736
+ "kl": 0.1867058277130127,
737
+ "learning_rate": 3.3750000000000003e-06,
738
+ "loss": -0.06573846191167831,
739
+ "step": 40,
740
+ "step_time": 0.21886433700001362
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": 384.0,
750
+ "completions/max_terminated_length": 0.0,
751
+ "completions/mean_length": 384.0,
752
+ "completions/mean_terminated_length": 0.0,
753
+ "completions/min_length": 384.0,
754
+ "completions/min_terminated_length": 0.0,
755
+ "entropy": 2.451829195022583,
756
+ "epoch": 1.7083333333333335,
757
+ "frac_reward_zero_std": 0.0,
758
+ "grad_norm": 3.871298313140869,
759
+ "kl": 0.19436387717723846,
760
+ "learning_rate": 3.3333333333333333e-06,
761
+ "loss": -0.22527764737606049,
762
+ "num_tokens": 277896.0,
763
+ "reward": -0.1756249964237213,
764
+ "reward_std": 0.22031700611114502,
765
+ "rewards/GeneratorRewardFunction/mean": -0.1756249964237213,
766
+ "rewards/GeneratorRewardFunction/std": 0.22031700611114502,
767
+ "step": 41,
768
+ "step_time": 12.257713568000042
769
+ },
770
+ {
771
+ "clip_ratio/high_max": 0.0006510416860692203,
772
+ "clip_ratio/high_mean": 0.0006510416860692203,
773
+ "clip_ratio/low_mean": 0.0026041667442768812,
774
+ "clip_ratio/low_min": 0.0026041667442768812,
775
+ "clip_ratio/region_mean": 0.0032552082557231188,
776
+ "entropy": 2.2009453773498535,
777
+ "epoch": 1.75,
778
+ "grad_norm": 3.4078774452209473,
779
+ "kl": 0.1595371961593628,
780
+ "learning_rate": 3.2916666666666668e-06,
781
+ "loss": 0.18696677684783936,
782
+ "step": 42,
783
+ "step_time": 0.2246266409999862
784
+ },
785
+ {
786
+ "clip_ratio/high_max": 0.0032552082557231188,
787
+ "clip_ratio/high_mean": 0.0032552082557231188,
788
+ "clip_ratio/low_mean": 0.0045572915114462376,
789
+ "clip_ratio/low_min": 0.0045572915114462376,
790
+ "clip_ratio/region_mean": 0.0078125,
791
+ "entropy": 2.0628411769866943,
792
+ "epoch": 1.7916666666666665,
793
+ "grad_norm": 2.7552006244659424,
794
+ "kl": 0.16367904841899872,
795
+ "learning_rate": 3.2500000000000002e-06,
796
+ "loss": 0.08732300996780396,
797
+ "step": 43,
798
+ "step_time": 0.22506770999996206
799
+ },
800
+ {
801
+ "clip_ratio/high_max": 0.02018229104578495,
802
+ "clip_ratio/high_mean": 0.02018229104578495,
803
+ "clip_ratio/low_mean": 0.0026041667442768812,
804
+ "clip_ratio/low_min": 0.0026041667442768812,
805
+ "clip_ratio/region_mean": 0.02278645895421505,
806
+ "entropy": 2.0429086685180664,
807
+ "epoch": 1.8333333333333335,
808
+ "grad_norm": 2.219151735305786,
809
+ "kl": 0.18261682987213135,
810
+ "learning_rate": 3.2083333333333337e-06,
811
+ "loss": -0.04571157321333885,
812
+ "step": 44,
813
+ "step_time": 0.22442952299996932
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": 384.0,
823
+ "completions/max_terminated_length": 0.0,
824
+ "completions/mean_length": 384.0,
825
+ "completions/mean_terminated_length": 0.0,
826
+ "completions/min_length": 384.0,
827
+ "completions/min_terminated_length": 0.0,
828
+ "entropy": 2.9552838802337646,
829
+ "epoch": 1.875,
830
+ "frac_reward_zero_std": 0.0,
831
+ "grad_norm": 4.345765113830566,
832
+ "kl": 0.15903545916080475,
833
+ "learning_rate": 3.1666666666666667e-06,
834
+ "loss": 0.16413524746894836,
835
+ "num_tokens": 303192.0,
836
+ "reward": -0.11562499403953552,
837
+ "reward_std": 0.14958691596984863,
838
+ "rewards/GeneratorRewardFunction/mean": -0.11562499403953552,
839
+ "rewards/GeneratorRewardFunction/std": 0.14958691596984863,
840
+ "step": 45,
841
+ "step_time": 12.18034159199999
842
+ },
843
+ {
844
+ "clip_ratio/high_max": 0.0032552082557231188,
845
+ "clip_ratio/high_mean": 0.0032552082557231188,
846
+ "clip_ratio/low_mean": 0.0,
847
+ "clip_ratio/low_min": 0.0,
848
+ "clip_ratio/region_mean": 0.0032552082557231188,
849
+ "entropy": 2.230048418045044,
850
+ "epoch": 1.9166666666666665,
851
+ "grad_norm": 3.0355405807495117,
852
+ "kl": 0.16114504635334015,
853
+ "learning_rate": 3.125e-06,
854
+ "loss": -0.029725370928645134,
855
+ "step": 46,
856
+ "step_time": 0.2170054759999971
857
+ },
858
+ {
859
+ "clip_ratio/high_max": 0.010416666977107525,
860
+ "clip_ratio/high_mean": 0.010416666977107525,
861
+ "clip_ratio/low_mean": 0.0,
862
+ "clip_ratio/low_min": 0.0,
863
+ "clip_ratio/region_mean": 0.010416666977107525,
864
+ "entropy": 2.3869032859802246,
865
+ "epoch": 1.9583333333333335,
866
+ "grad_norm": 2.259568929672241,
867
+ "kl": 0.18078042566776276,
868
+ "learning_rate": 3.0833333333333336e-06,
869
+ "loss": -0.12276550382375717,
870
+ "step": 47,
871
+ "step_time": 0.21701960600000803
872
+ },
873
+ {
874
+ "clip_ratio/high_max": 0.010416666977107525,
875
+ "clip_ratio/high_mean": 0.010416666977107525,
876
+ "clip_ratio/low_mean": 0.0071614584885537624,
877
+ "clip_ratio/low_min": 0.0071614584885537624,
878
+ "clip_ratio/region_mean": 0.017578125,
879
+ "entropy": 1.6151641607284546,
880
+ "epoch": 2.0,
881
+ "grad_norm": 2.439406156539917,
882
+ "kl": 0.14620588719844818,
883
+ "learning_rate": 3.0416666666666666e-06,
884
+ "loss": -0.01039380207657814,
885
+ "step": 48,
886
+ "step_time": 0.21717840000002298
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": 384.0,
896
+ "completions/max_terminated_length": 0.0,
897
+ "completions/mean_length": 384.0,
898
+ "completions/mean_terminated_length": 0.0,
899
+ "completions/min_length": 384.0,
900
+ "completions/min_terminated_length": 0.0,
901
+ "entropy": 2.299091100692749,
902
+ "epoch": 2.0416666666666665,
903
+ "frac_reward_zero_std": 0.0,
904
+ "grad_norm": 2.469008207321167,
905
+ "kl": 0.17820477485656738,
906
+ "learning_rate": 3e-06,
907
+ "loss": 0.03437386080622673,
908
+ "num_tokens": 328804.0,
909
+ "reward": -0.11281250417232513,
910
+ "reward_std": 0.14768172800540924,
911
+ "rewards/GeneratorRewardFunction/mean": -0.11281250417232513,
912
+ "rewards/GeneratorRewardFunction/std": 0.14768172800540924,
913
+ "step": 49,
914
+ "step_time": 12.214060969000002
915
+ },
916
+ {
917
+ "clip_ratio/high_max": 0.0045572915114462376,
918
+ "clip_ratio/high_mean": 0.0045572915114462376,
919
+ "clip_ratio/low_mean": 0.0,
920
+ "clip_ratio/low_min": 0.0,
921
+ "clip_ratio/region_mean": 0.0045572915114462376,
922
+ "entropy": 1.7260881662368774,
923
+ "epoch": 2.0833333333333335,
924
+ "grad_norm": 2.3388266563415527,
925
+ "kl": 0.14910505712032318,
926
+ "learning_rate": 2.9583333333333335e-06,
927
+ "loss": -0.1556338518857956,
928
+ "step": 50,
929
+ "step_time": 0.2256633300000317
930
+ },
931
+ {
932
+ "clip_ratio/high_max": 0.011067708022892475,
933
+ "clip_ratio/high_mean": 0.011067708022892475,
934
+ "clip_ratio/low_mean": 0.0026041667442768812,
935
+ "clip_ratio/low_min": 0.0026041667442768812,
936
+ "clip_ratio/region_mean": 0.013671875,
937
+ "entropy": 1.9709595441818237,
938
+ "epoch": 2.125,
939
+ "grad_norm": 3.018829107284546,
940
+ "kl": 0.20297934114933014,
941
+ "learning_rate": 2.916666666666667e-06,
942
+ "loss": -0.013471716083586216,
943
+ "step": 51,
944
+ "step_time": 0.22631718399998135
945
+ },
946
+ {
947
+ "clip_ratio/high_max": 0.008463541977107525,
948
+ "clip_ratio/high_mean": 0.008463541977107525,
949
+ "clip_ratio/low_mean": 0.001953125,
950
+ "clip_ratio/low_min": 0.001953125,
951
+ "clip_ratio/region_mean": 0.010416666977107525,
952
+ "entropy": 2.6199281215667725,
953
+ "epoch": 2.1666666666666665,
954
+ "grad_norm": 4.550353527069092,
955
+ "kl": 0.14221708476543427,
956
+ "learning_rate": 2.875e-06,
957
+ "loss": 0.13723115622997284,
958
+ "step": 52,
959
+ "step_time": 0.2253769019999936
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": 384.0,
969
+ "completions/max_terminated_length": 0.0,
970
+ "completions/mean_length": 384.0,
971
+ "completions/mean_terminated_length": 0.0,
972
+ "completions/min_length": 384.0,
973
+ "completions/min_terminated_length": 0.0,
974
+ "entropy": 2.2031538486480713,
975
+ "epoch": 2.2083333333333335,
976
+ "frac_reward_zero_std": 0.0,
977
+ "grad_norm": 2.5543603897094727,
978
+ "kl": 0.18505902588367462,
979
+ "learning_rate": 2.8333333333333335e-06,
980
+ "loss": 0.0006821660790592432,
981
+ "num_tokens": 354128.0,
982
+ "reward": -0.15625,
983
+ "reward_std": 0.17485707998275757,
984
+ "rewards/GeneratorRewardFunction/mean": -0.15625,
985
+ "rewards/GeneratorRewardFunction/std": 0.17485707998275757,
986
+ "step": 53,
987
+ "step_time": 12.302670394000017
988
+ },
989
+ {
990
+ "clip_ratio/high_max": 0.001953125,
991
+ "clip_ratio/high_mean": 0.001953125,
992
+ "clip_ratio/low_mean": 0.0006510416860692203,
993
+ "clip_ratio/low_min": 0.0006510416860692203,
994
+ "clip_ratio/region_mean": 0.0026041667442768812,
995
+ "entropy": 2.024317979812622,
996
+ "epoch": 2.25,
997
+ "grad_norm": 2.0381131172180176,
998
+ "kl": 0.18907330930233002,
999
+ "learning_rate": 2.791666666666667e-06,
1000
+ "loss": 0.049581658095121384,
1001
+ "step": 54,
1002
+ "step_time": 0.21795685400002185
1003
+ },
1004
+ {
1005
+ "clip_ratio/high_max": 0.009765625,
1006
+ "clip_ratio/high_mean": 0.009765625,
1007
+ "clip_ratio/low_mean": 0.0,
1008
+ "clip_ratio/low_min": 0.0,
1009
+ "clip_ratio/region_mean": 0.009765625,
1010
+ "entropy": 2.274016857147217,
1011
+ "epoch": 2.2916666666666665,
1012
+ "grad_norm": 3.3998916149139404,
1013
+ "kl": 0.19920788705348969,
1014
+ "learning_rate": 2.7500000000000004e-06,
1015
+ "loss": -0.19976168870925903,
1016
+ "step": 55,
1017
+ "step_time": 0.21774610800002847
1018
+ },
1019
+ {
1020
+ "clip_ratio/high_max": 0.0071614584885537624,
1021
+ "clip_ratio/high_mean": 0.0071614584885537624,
1022
+ "clip_ratio/low_mean": 0.0006510416860692203,
1023
+ "clip_ratio/low_min": 0.0006510416860692203,
1024
+ "clip_ratio/region_mean": 0.0078125,
1025
+ "entropy": 2.4295196533203125,
1026
+ "epoch": 2.3333333333333335,
1027
+ "grad_norm": 3.8777236938476562,
1028
+ "kl": 0.1597600132226944,
1029
+ "learning_rate": 2.7083333333333334e-06,
1030
+ "loss": 0.15159599483013153,
1031
+ "step": 56,
1032
+ "step_time": 0.21687880799998993
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": 384.0,
1042
+ "completions/max_terminated_length": 0.0,
1043
+ "completions/mean_length": 384.0,
1044
+ "completions/mean_terminated_length": 0.0,
1045
+ "completions/min_length": 384.0,
1046
+ "completions/min_terminated_length": 0.0,
1047
+ "entropy": 2.244577169418335,
1048
+ "epoch": 2.375,
1049
+ "frac_reward_zero_std": 0.0,
1050
+ "grad_norm": 2.2973275184631348,
1051
+ "kl": 0.1547890454530716,
1052
+ "learning_rate": 2.666666666666667e-06,
1053
+ "loss": -0.13385160267353058,
1054
+ "num_tokens": 379284.0,
1055
+ "reward": -0.07656250149011612,
1056
+ "reward_std": 0.0422578901052475,
1057
+ "rewards/GeneratorRewardFunction/mean": -0.07656250149011612,
1058
+ "rewards/GeneratorRewardFunction/std": 0.0422578901052475,
1059
+ "step": 57,
1060
+ "step_time": 12.35130471399998
1061
+ },
1062
+ {
1063
+ "clip_ratio/high_max": 0.0032552082557231188,
1064
+ "clip_ratio/high_mean": 0.0032552082557231188,
1065
+ "clip_ratio/low_mean": 0.0,
1066
+ "clip_ratio/low_min": 0.0,
1067
+ "clip_ratio/region_mean": 0.0032552082557231188,
1068
+ "entropy": 1.7890634536743164,
1069
+ "epoch": 2.4166666666666665,
1070
+ "grad_norm": 2.7311065196990967,
1071
+ "kl": 0.16195693612098694,
1072
+ "learning_rate": 2.6250000000000003e-06,
1073
+ "loss": -0.141176238656044,
1074
+ "step": 58,
1075
+ "step_time": 0.22516193699999576
1076
+ },
1077
+ {
1078
+ "clip_ratio/high_max": 0.0,
1079
+ "clip_ratio/high_mean": 0.0,
1080
+ "clip_ratio/low_mean": 0.0026041667442768812,
1081
+ "clip_ratio/low_min": 0.0026041667442768812,
1082
+ "clip_ratio/region_mean": 0.0026041667442768812,
1083
+ "entropy": 2.727632522583008,
1084
+ "epoch": 2.4583333333333335,
1085
+ "grad_norm": 6.948690414428711,
1086
+ "kl": 0.19930227100849152,
1087
+ "learning_rate": 2.5833333333333337e-06,
1088
+ "loss": 0.32165396213531494,
1089
+ "step": 59,
1090
+ "step_time": 0.22437341799997057
1091
+ },
1092
+ {
1093
+ "clip_ratio/high_max": 0.011067708022892475,
1094
+ "clip_ratio/high_mean": 0.011067708022892475,
1095
+ "clip_ratio/low_mean": 0.0006510416860692203,
1096
+ "clip_ratio/low_min": 0.0006510416860692203,
1097
+ "clip_ratio/region_mean": 0.01171875,
1098
+ "entropy": 2.4251506328582764,
1099
+ "epoch": 2.5,
1100
+ "grad_norm": 2.1018707752227783,
1101
+ "kl": 0.1818857640028,
1102
+ "learning_rate": 2.5416666666666668e-06,
1103
+ "loss": -0.04400597885251045,
1104
+ "step": 60,
1105
+ "step_time": 0.2244098689999987
1106
+ },
1107
+ {
1108
+ "clip_ratio/high_max": 0.0006510416860692203,
1109
+ "clip_ratio/high_mean": 0.0006510416860692203,
1110
+ "clip_ratio/low_mean": 0.0,
1111
+ "clip_ratio/low_min": 0.0,
1112
+ "clip_ratio/region_mean": 0.0006510416860692203,
1113
+ "completions/clipped_ratio": 1.0,
1114
+ "completions/max_length": 384.0,
1115
+ "completions/max_terminated_length": 0.0,
1116
+ "completions/mean_length": 384.0,
1117
+ "completions/mean_terminated_length": 0.0,
1118
+ "completions/min_length": 384.0,
1119
+ "completions/min_terminated_length": 0.0,
1120
+ "entropy": 2.3666765689849854,
1121
+ "epoch": 2.5416666666666665,
1122
+ "frac_reward_zero_std": 0.0,
1123
+ "grad_norm": 3.457057476043701,
1124
+ "kl": 0.20910978317260742,
1125
+ "learning_rate": 2.5e-06,
1126
+ "loss": 0.08574869483709335,
1127
+ "num_tokens": 404348.0,
1128
+ "reward": -0.22625000774860382,
1129
+ "reward_std": 0.2726322412490845,
1130
+ "rewards/GeneratorRewardFunction/mean": -0.22625000774860382,
1131
+ "rewards/GeneratorRewardFunction/std": 0.2726322412490845,
1132
+ "step": 61,
1133
+ "step_time": 12.438367398999958
1134
+ },
1135
+ {
1136
+ "clip_ratio/high_max": 0.0006510416860692203,
1137
+ "clip_ratio/high_mean": 0.0006510416860692203,
1138
+ "clip_ratio/low_mean": 0.0,
1139
+ "clip_ratio/low_min": 0.0,
1140
+ "clip_ratio/region_mean": 0.0006510416860692203,
1141
+ "entropy": 2.5774478912353516,
1142
+ "epoch": 2.5833333333333335,
1143
+ "grad_norm": 1.9898955821990967,
1144
+ "kl": 0.18942372500896454,
1145
+ "learning_rate": 2.4583333333333332e-06,
1146
+ "loss": -0.07343629747629166,
1147
+ "step": 62,
1148
+ "step_time": 0.22581257799998866
1149
+ },
1150
+ {
1151
+ "clip_ratio/high_max": 0.0032552082557231188,
1152
+ "clip_ratio/high_mean": 0.0032552082557231188,
1153
+ "clip_ratio/low_mean": 0.0,
1154
+ "clip_ratio/low_min": 0.0,
1155
+ "clip_ratio/region_mean": 0.0032552082557231188,
1156
+ "entropy": 2.429924964904785,
1157
+ "epoch": 2.625,
1158
+ "grad_norm": 3.2511487007141113,
1159
+ "kl": 0.18745101988315582,
1160
+ "learning_rate": 2.4166666666666667e-06,
1161
+ "loss": -0.17632712423801422,
1162
+ "step": 63,
1163
+ "step_time": 0.2255372609999995
1164
+ },
1165
+ {
1166
+ "clip_ratio/high_max": 0.0032552082557231188,
1167
+ "clip_ratio/high_mean": 0.0032552082557231188,
1168
+ "clip_ratio/low_mean": 0.0026041667442768812,
1169
+ "clip_ratio/low_min": 0.0026041667442768812,
1170
+ "clip_ratio/region_mean": 0.005859375,
1171
+ "entropy": 2.37870717048645,
1172
+ "epoch": 2.6666666666666665,
1173
+ "grad_norm": 3.8804032802581787,
1174
+ "kl": 0.152898907661438,
1175
+ "learning_rate": 2.375e-06,
1176
+ "loss": 0.16565446555614471,
1177
+ "step": 64,
1178
+ "step_time": 0.22526705500001754
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": 384.0,
1188
+ "completions/max_terminated_length": 0.0,
1189
+ "completions/mean_length": 384.0,
1190
+ "completions/mean_terminated_length": 0.0,
1191
+ "completions/min_length": 384.0,
1192
+ "completions/min_terminated_length": 0.0,
1193
+ "entropy": 2.2809789180755615,
1194
+ "epoch": 2.7083333333333335,
1195
+ "frac_reward_zero_std": 0.25,
1196
+ "grad_norm": 2.90450119972229,
1197
+ "kl": 0.1772325038909912,
1198
+ "learning_rate": 2.3333333333333336e-06,
1199
+ "loss": -0.05227809026837349,
1200
+ "num_tokens": 429432.0,
1201
+ "reward": -0.08624999970197678,
1202
+ "reward_std": 0.09694501012563705,
1203
+ "rewards/GeneratorRewardFunction/mean": -0.08624999970197678,
1204
+ "rewards/GeneratorRewardFunction/std": 0.09694500267505646,
1205
+ "step": 65,
1206
+ "step_time": 12.340170748000048
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": 2.3075809478759766,
1215
+ "epoch": 2.75,
1216
+ "grad_norm": 2.807197332382202,
1217
+ "kl": 0.2230893224477768,
1218
+ "learning_rate": 2.2916666666666666e-06,
1219
+ "loss": 0.14831461012363434,
1220
+ "step": 66,
1221
+ "step_time": 0.2178340820000244
1222
+ },
1223
+ {
1224
+ "clip_ratio/high_max": 0.0013020833721384406,
1225
+ "clip_ratio/high_mean": 0.0013020833721384406,
1226
+ "clip_ratio/low_mean": 0.0,
1227
+ "clip_ratio/low_min": 0.0,
1228
+ "clip_ratio/region_mean": 0.0013020833721384406,
1229
+ "entropy": 2.4482500553131104,
1230
+ "epoch": 2.7916666666666665,
1231
+ "grad_norm": 0.43737831711769104,
1232
+ "kl": 0.172013521194458,
1233
+ "learning_rate": 2.25e-06,
1234
+ "loss": -0.019011570140719414,
1235
+ "step": 67,
1236
+ "step_time": 0.21846536599997535
1237
+ },
1238
+ {
1239
+ "clip_ratio/high_max": 0.0052083334885537624,
1240
+ "clip_ratio/high_mean": 0.0052083334885537624,
1241
+ "clip_ratio/low_mean": 0.001953125,
1242
+ "clip_ratio/low_min": 0.001953125,
1243
+ "clip_ratio/region_mean": 0.0071614584885537624,
1244
+ "entropy": 2.5060949325561523,
1245
+ "epoch": 2.8333333333333335,
1246
+ "grad_norm": 2.4608078002929688,
1247
+ "kl": 0.20921878516674042,
1248
+ "learning_rate": 2.2083333333333335e-06,
1249
+ "loss": -0.07550495862960815,
1250
+ "step": 68,
1251
+ "step_time": 0.21823405100002446
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": 384.0,
1261
+ "completions/max_terminated_length": 0.0,
1262
+ "completions/mean_length": 384.0,
1263
+ "completions/mean_terminated_length": 0.0,
1264
+ "completions/min_length": 384.0,
1265
+ "completions/min_terminated_length": 0.0,
1266
+ "entropy": 2.1956374645233154,
1267
+ "epoch": 2.875,
1268
+ "frac_reward_zero_std": 0.0,
1269
+ "grad_norm": 2.7649459838867188,
1270
+ "kl": 0.20148341357707977,
1271
+ "learning_rate": 2.166666666666667e-06,
1272
+ "loss": 0.0005037125083617866,
1273
+ "num_tokens": 454788.0,
1274
+ "reward": -0.15281249582767487,
1275
+ "reward_std": 0.2081102877855301,
1276
+ "rewards/GeneratorRewardFunction/mean": -0.15281249582767487,
1277
+ "rewards/GeneratorRewardFunction/std": 0.20811030268669128,
1278
+ "step": 69,
1279
+ "step_time": 12.46666924799996
1280
+ },
1281
+ {
1282
+ "clip_ratio/high_max": 0.001953125,
1283
+ "clip_ratio/high_mean": 0.001953125,
1284
+ "clip_ratio/low_mean": 0.0,
1285
+ "clip_ratio/low_min": 0.0,
1286
+ "clip_ratio/region_mean": 0.001953125,
1287
+ "entropy": 2.1872847080230713,
1288
+ "epoch": 2.9166666666666665,
1289
+ "grad_norm": 2.1137373447418213,
1290
+ "kl": 0.2070050984621048,
1291
+ "learning_rate": 2.125e-06,
1292
+ "loss": -0.0964542105793953,
1293
+ "step": 70,
1294
+ "step_time": 0.21906004099997745
1295
+ },
1296
+ {
1297
+ "clip_ratio/high_max": 0.0026041667442768812,
1298
+ "clip_ratio/high_mean": 0.0026041667442768812,
1299
+ "clip_ratio/low_mean": 0.0006510416860692203,
1300
+ "clip_ratio/low_min": 0.0006510416860692203,
1301
+ "clip_ratio/region_mean": 0.0032552082557231188,
1302
+ "entropy": 2.914250135421753,
1303
+ "epoch": 2.9583333333333335,
1304
+ "grad_norm": 4.2155232429504395,
1305
+ "kl": 0.19951386749744415,
1306
+ "learning_rate": 2.0833333333333334e-06,
1307
+ "loss": 0.09215161204338074,
1308
+ "step": 71,
1309
+ "step_time": 0.21903813999995236
1310
+ },
1311
+ {
1312
+ "clip_ratio/high_max": 0.0078125,
1313
+ "clip_ratio/high_mean": 0.0078125,
1314
+ "clip_ratio/low_mean": 0.0006510416860692203,
1315
+ "clip_ratio/low_min": 0.0006510416860692203,
1316
+ "clip_ratio/region_mean": 0.008463541977107525,
1317
+ "entropy": 2.5522615909576416,
1318
+ "epoch": 3.0,
1319
+ "grad_norm": 2.679938793182373,
1320
+ "kl": 0.17947690188884735,
1321
+ "learning_rate": 2.041666666666667e-06,
1322
+ "loss": 0.0045168730430305,
1323
+ "step": 72,
1324
+ "step_time": 0.2201463600000011
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": 384.0,
1334
+ "completions/max_terminated_length": 0.0,
1335
+ "completions/mean_length": 384.0,
1336
+ "completions/mean_terminated_length": 0.0,
1337
+ "completions/min_length": 384.0,
1338
+ "completions/min_terminated_length": 0.0,
1339
+ "entropy": 2.0034985542297363,
1340
+ "epoch": 3.0416666666666665,
1341
+ "frac_reward_zero_std": 0.25,
1342
+ "grad_norm": 2.8730416297912598,
1343
+ "kl": 0.18952839076519012,
1344
+ "learning_rate": 2.0000000000000003e-06,
1345
+ "loss": 0.0004738100979011506,
1346
+ "num_tokens": 480176.0,
1347
+ "reward": -0.10687500238418579,
1348
+ "reward_std": 0.15892215073108673,
1349
+ "rewards/GeneratorRewardFunction/mean": -0.10687500238418579,
1350
+ "rewards/GeneratorRewardFunction/std": 0.15892215073108673,
1351
+ "step": 73,
1352
+ "step_time": 12.335805735999998
1353
+ },
1354
+ {
1355
+ "clip_ratio/high_max": 0.0013020833721384406,
1356
+ "clip_ratio/high_mean": 0.0013020833721384406,
1357
+ "clip_ratio/low_mean": 0.0,
1358
+ "clip_ratio/low_min": 0.0,
1359
+ "clip_ratio/region_mean": 0.0013020833721384406,
1360
+ "entropy": 2.350316286087036,
1361
+ "epoch": 3.0833333333333335,
1362
+ "grad_norm": 3.1329355239868164,
1363
+ "kl": 0.20395760238170624,
1364
+ "learning_rate": 1.9583333333333334e-06,
1365
+ "loss": -0.013183367438614368,
1366
+ "step": 74,
1367
+ "step_time": 0.2247725990000049
1368
+ },
1369
+ {
1370
+ "clip_ratio/high_max": 0.001953125,
1371
+ "clip_ratio/high_mean": 0.001953125,
1372
+ "clip_ratio/low_mean": 0.0006510416860692203,
1373
+ "clip_ratio/low_min": 0.0006510416860692203,
1374
+ "clip_ratio/region_mean": 0.0026041667442768812,
1375
+ "entropy": 2.3521158695220947,
1376
+ "epoch": 3.125,
1377
+ "grad_norm": 1.9546054601669312,
1378
+ "kl": 0.20290331542491913,
1379
+ "learning_rate": 1.916666666666667e-06,
1380
+ "loss": 0.0031558917835354805,
1381
+ "step": 75,
1382
+ "step_time": 0.22477466999998796
1383
+ },
1384
+ {
1385
+ "clip_ratio/high_max": 0.001953125,
1386
+ "clip_ratio/high_mean": 0.001953125,
1387
+ "clip_ratio/low_mean": 0.0,
1388
+ "clip_ratio/low_min": 0.0,
1389
+ "clip_ratio/region_mean": 0.001953125,
1390
+ "entropy": 2.429577112197876,
1391
+ "epoch": 3.1666666666666665,
1392
+ "grad_norm": 1.2733182907104492,
1393
+ "kl": 0.28720995783805847,
1394
+ "learning_rate": 1.8750000000000003e-06,
1395
+ "loss": 0.011465795338153839,
1396
+ "step": 76,
1397
+ "step_time": 0.2244293130000301
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": 384.0,
1407
+ "completions/max_terminated_length": 0.0,
1408
+ "completions/mean_length": 384.0,
1409
+ "completions/mean_terminated_length": 0.0,
1410
+ "completions/min_length": 384.0,
1411
+ "completions/min_terminated_length": 0.0,
1412
+ "entropy": 2.655349016189575,
1413
+ "epoch": 3.2083333333333335,
1414
+ "frac_reward_zero_std": 0.0,
1415
+ "grad_norm": 2.0284245014190674,
1416
+ "kl": 0.2677163779735565,
1417
+ "learning_rate": 1.8333333333333333e-06,
1418
+ "loss": -0.046080395579338074,
1419
+ "num_tokens": 505692.0,
1420
+ "reward": -0.27125000953674316,
1421
+ "reward_std": 0.26958298683166504,
1422
+ "rewards/GeneratorRewardFunction/mean": -0.27125000953674316,
1423
+ "rewards/GeneratorRewardFunction/std": 0.2695830166339874,
1424
+ "step": 77,
1425
+ "step_time": 12.360189741999989
1426
+ },
1427
+ {
1428
+ "clip_ratio/high_max": 0.001953125,
1429
+ "clip_ratio/high_mean": 0.001953125,
1430
+ "clip_ratio/low_mean": 0.0006510416860692203,
1431
+ "clip_ratio/low_min": 0.0006510416860692203,
1432
+ "clip_ratio/region_mean": 0.0026041667442768812,
1433
+ "entropy": 3.2132160663604736,
1434
+ "epoch": 3.25,
1435
+ "grad_norm": 3.93788743019104,
1436
+ "kl": 0.18754476308822632,
1437
+ "learning_rate": 1.7916666666666667e-06,
1438
+ "loss": 0.08930965512990952,
1439
+ "step": 78,
1440
+ "step_time": 0.22499302199997828
1441
+ },
1442
+ {
1443
+ "clip_ratio/high_max": 0.0032552082557231188,
1444
+ "clip_ratio/high_mean": 0.0032552082557231188,
1445
+ "clip_ratio/low_mean": 0.0,
1446
+ "clip_ratio/low_min": 0.0,
1447
+ "clip_ratio/region_mean": 0.0032552082557231188,
1448
+ "entropy": 2.5116894245147705,
1449
+ "epoch": 3.2916666666666665,
1450
+ "grad_norm": 2.4447059631347656,
1451
+ "kl": 0.2354326993227005,
1452
+ "learning_rate": 1.75e-06,
1453
+ "loss": -0.07100113481283188,
1454
+ "step": 79,
1455
+ "step_time": 0.22562992499996426
1456
+ },
1457
+ {
1458
+ "clip_ratio/high_max": 0.001953125,
1459
+ "clip_ratio/high_mean": 0.001953125,
1460
+ "clip_ratio/low_mean": 0.0013020833721384406,
1461
+ "clip_ratio/low_min": 0.0013020833721384406,
1462
+ "clip_ratio/region_mean": 0.0032552082557231188,
1463
+ "entropy": 2.6606433391571045,
1464
+ "epoch": 3.3333333333333335,
1465
+ "grad_norm": 2.5851616859436035,
1466
+ "kl": 0.23066651821136475,
1467
+ "learning_rate": 1.7083333333333334e-06,
1468
+ "loss": 0.030483582988381386,
1469
+ "step": 80,
1470
+ "step_time": 0.22511524399999416
1471
+ },
1472
+ {
1473
+ "clip_ratio/high_max": 0.0013020833721384406,
1474
+ "clip_ratio/high_mean": 0.0013020833721384406,
1475
+ "clip_ratio/low_mean": 0.0,
1476
+ "clip_ratio/low_min": 0.0,
1477
+ "clip_ratio/region_mean": 0.0013020833721384406,
1478
+ "completions/clipped_ratio": 1.0,
1479
+ "completions/max_length": 384.0,
1480
+ "completions/max_terminated_length": 0.0,
1481
+ "completions/mean_length": 384.0,
1482
+ "completions/mean_terminated_length": 0.0,
1483
+ "completions/min_length": 384.0,
1484
+ "completions/min_terminated_length": 0.0,
1485
+ "entropy": 2.848414421081543,
1486
+ "epoch": 3.375,
1487
+ "frac_reward_zero_std": 0.0,
1488
+ "grad_norm": 2.853872776031494,
1489
+ "kl": 0.21739749610424042,
1490
+ "learning_rate": 1.6666666666666667e-06,
1491
+ "loss": -0.01184266060590744,
1492
+ "num_tokens": 530384.0,
1493
+ "reward": -0.1821874976158142,
1494
+ "reward_std": 0.23522308468818665,
1495
+ "rewards/GeneratorRewardFunction/mean": -0.1821874976158142,
1496
+ "rewards/GeneratorRewardFunction/std": 0.23522308468818665,
1497
+ "step": 81,
1498
+ "step_time": 12.474556322000012
1499
+ },
1500
+ {
1501
+ "clip_ratio/high_max": 0.0006510416860692203,
1502
+ "clip_ratio/high_mean": 0.0006510416860692203,
1503
+ "clip_ratio/low_mean": 0.0,
1504
+ "clip_ratio/low_min": 0.0,
1505
+ "clip_ratio/region_mean": 0.0006510416860692203,
1506
+ "entropy": 2.7031612396240234,
1507
+ "epoch": 3.4166666666666665,
1508
+ "grad_norm": 1.9774682521820068,
1509
+ "kl": 0.17283748090267181,
1510
+ "learning_rate": 1.6250000000000001e-06,
1511
+ "loss": -0.11227142065763474,
1512
+ "step": 82,
1513
+ "step_time": 0.2153717829999664
1514
+ },
1515
+ {
1516
+ "clip_ratio/high_max": 0.0032552082557231188,
1517
+ "clip_ratio/high_mean": 0.0032552082557231188,
1518
+ "clip_ratio/low_mean": 0.0,
1519
+ "clip_ratio/low_min": 0.0,
1520
+ "clip_ratio/region_mean": 0.0032552082557231188,
1521
+ "entropy": 2.0835349559783936,
1522
+ "epoch": 3.4583333333333335,
1523
+ "grad_norm": 2.6850008964538574,
1524
+ "kl": 0.194760262966156,
1525
+ "learning_rate": 1.5833333333333333e-06,
1526
+ "loss": -0.0002996302500832826,
1527
+ "step": 83,
1528
+ "step_time": 0.21602736900001673
1529
+ },
1530
+ {
1531
+ "clip_ratio/high_max": 0.001953125,
1532
+ "clip_ratio/high_mean": 0.001953125,
1533
+ "clip_ratio/low_mean": 0.0013020833721384406,
1534
+ "clip_ratio/low_min": 0.0013020833721384406,
1535
+ "clip_ratio/region_mean": 0.0032552082557231188,
1536
+ "entropy": 1.9008747339248657,
1537
+ "epoch": 3.5,
1538
+ "grad_norm": 3.73012638092041,
1539
+ "kl": 0.19497458636760712,
1540
+ "learning_rate": 1.5416666666666668e-06,
1541
+ "loss": 0.12751369178295135,
1542
+ "step": 84,
1543
+ "step_time": 0.21625546300003862
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": 384.0,
1553
+ "completions/max_terminated_length": 0.0,
1554
+ "completions/mean_length": 384.0,
1555
+ "completions/mean_terminated_length": 0.0,
1556
+ "completions/min_length": 384.0,
1557
+ "completions/min_terminated_length": 0.0,
1558
+ "entropy": 2.5717437267303467,
1559
+ "epoch": 3.5416666666666665,
1560
+ "frac_reward_zero_std": 0.0,
1561
+ "grad_norm": 3.2951369285583496,
1562
+ "kl": 0.1804773062467575,
1563
+ "learning_rate": 1.5e-06,
1564
+ "loss": 0.13690312206745148,
1565
+ "num_tokens": 555616.0,
1566
+ "reward": -0.20656248927116394,
1567
+ "reward_std": 0.224593847990036,
1568
+ "rewards/GeneratorRewardFunction/mean": -0.20656248927116394,
1569
+ "rewards/GeneratorRewardFunction/std": 0.224593847990036,
1570
+ "step": 85,
1571
+ "step_time": 12.407897509999998
1572
+ },
1573
+ {
1574
+ "clip_ratio/high_max": 0.0,
1575
+ "clip_ratio/high_mean": 0.0,
1576
+ "clip_ratio/low_mean": 0.0006510416860692203,
1577
+ "clip_ratio/low_min": 0.0006510416860692203,
1578
+ "clip_ratio/region_mean": 0.0006510416860692203,
1579
+ "entropy": 1.9566816091537476,
1580
+ "epoch": 3.5833333333333335,
1581
+ "grad_norm": 2.5742263793945312,
1582
+ "kl": 0.2944556176662445,
1583
+ "learning_rate": 1.4583333333333335e-06,
1584
+ "loss": 0.00038116463110782206,
1585
+ "step": 86,
1586
+ "step_time": 0.2183175630000278
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": 2.4693102836608887,
1595
+ "epoch": 3.625,
1596
+ "grad_norm": 2.8581724166870117,
1597
+ "kl": 0.24227438867092133,
1598
+ "learning_rate": 1.4166666666666667e-06,
1599
+ "loss": -0.009446345269680023,
1600
+ "step": 87,
1601
+ "step_time": 0.21834270400000833
1602
+ },
1603
+ {
1604
+ "clip_ratio/high_max": 0.0006510416860692203,
1605
+ "clip_ratio/high_mean": 0.0006510416860692203,
1606
+ "clip_ratio/low_mean": 0.0,
1607
+ "clip_ratio/low_min": 0.0,
1608
+ "clip_ratio/region_mean": 0.0006510416860692203,
1609
+ "entropy": 2.551859140396118,
1610
+ "epoch": 3.6666666666666665,
1611
+ "grad_norm": 2.1804516315460205,
1612
+ "kl": 0.20367814600467682,
1613
+ "learning_rate": 1.3750000000000002e-06,
1614
+ "loss": -0.1257125288248062,
1615
+ "step": 88,
1616
+ "step_time": 0.2176733800000079
1617
+ },
1618
+ {
1619
+ "clip_ratio/high_max": 0.0006510416860692203,
1620
+ "clip_ratio/high_mean": 0.0006510416860692203,
1621
+ "clip_ratio/low_mean": 0.0,
1622
+ "clip_ratio/low_min": 0.0,
1623
+ "clip_ratio/region_mean": 0.0006510416860692203,
1624
+ "completions/clipped_ratio": 1.0,
1625
+ "completions/max_length": 384.0,
1626
+ "completions/max_terminated_length": 0.0,
1627
+ "completions/mean_length": 384.0,
1628
+ "completions/mean_terminated_length": 0.0,
1629
+ "completions/min_length": 384.0,
1630
+ "completions/min_terminated_length": 0.0,
1631
+ "entropy": 2.4765048027038574,
1632
+ "epoch": 3.7083333333333335,
1633
+ "frac_reward_zero_std": 0.0,
1634
+ "grad_norm": 3.6547114849090576,
1635
+ "kl": 0.23193800449371338,
1636
+ "learning_rate": 1.3333333333333334e-06,
1637
+ "loss": -0.0005866275751031935,
1638
+ "num_tokens": 581116.0,
1639
+ "reward": -0.26734375953674316,
1640
+ "reward_std": 0.28873497247695923,
1641
+ "rewards/GeneratorRewardFunction/mean": -0.26734375953674316,
1642
+ "rewards/GeneratorRewardFunction/std": 0.28873497247695923,
1643
+ "step": 89,
1644
+ "step_time": 12.265255932999992
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": 2.2775588035583496,
1653
+ "epoch": 3.75,
1654
+ "grad_norm": 3.115778923034668,
1655
+ "kl": 0.24746482074260712,
1656
+ "learning_rate": 1.2916666666666669e-06,
1657
+ "loss": 0.11805403232574463,
1658
+ "step": 90,
1659
+ "step_time": 0.22551655099999834
1660
+ }
1661
+ ],
1662
+ "logging_steps": 1,
1663
+ "max_steps": 120,
1664
+ "num_input_tokens_seen": 581116,
1665
+ "num_train_epochs": 5,
1666
+ "save_steps": 30,
1667
+ "stateful_callbacks": {
1668
+ "TrainerControl": {
1669
+ "args": {
1670
+ "should_epoch_stop": false,
1671
+ "should_evaluate": false,
1672
+ "should_log": false,
1673
+ "should_save": true,
1674
+ "should_training_stop": false
1675
+ },
1676
+ "attributes": {}
1677
+ }
1678
+ },
1679
+ "total_flos": 0.0,
1680
+ "train_batch_size": 4,
1681
+ "trial_name": null,
1682
+ "trial_params": null
1683
+ }
self_play_hf_l40s_full/round_001/generator_train/checkpoint-90/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31a29342dccc7e8741482029f38c21ec23009b84a898a5318c35a0ac0c919d25
3
+ size 7249