Siddeshwar1625 commited on
Commit
0a4d6a5
·
verified ·
1 Parent(s): 1d1387b

Intermediate checkpoint upload step=120 (generator_train)

Browse files
.gitattributes CHANGED
@@ -85,3 +85,4 @@ self_play_hf_l40s_full/round_004/answerer_train/final_model/tokenizer.json filte
85
  self_play_hf_l40s_full/round_005/generator_train/checkpoint-30/tokenizer.json filter=lfs diff=lfs merge=lfs -text
86
  self_play_hf_l40s_full/round_005/generator_train/checkpoint-60/tokenizer.json filter=lfs diff=lfs merge=lfs -text
87
  self_play_hf_l40s_full/round_005/generator_train/checkpoint-90/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 
 
85
  self_play_hf_l40s_full/round_005/generator_train/checkpoint-30/tokenizer.json filter=lfs diff=lfs merge=lfs -text
86
  self_play_hf_l40s_full/round_005/generator_train/checkpoint-60/tokenizer.json filter=lfs diff=lfs merge=lfs -text
87
  self_play_hf_l40s_full/round_005/generator_train/checkpoint-90/tokenizer.json filter=lfs diff=lfs merge=lfs -text
88
+ self_play_hf_l40s_full/round_005/generator_train/checkpoint-120/tokenizer.json filter=lfs diff=lfs merge=lfs -text
self_play_hf_l40s_full/round_005/generator_train/checkpoint-120/chat_template.jinja ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- if tools %}
2
+ {{- '<|im_start|>system\n' }}
3
+ {%- if messages[0]['role'] == 'system' %}
4
+ {{- messages[0]['content'] }}
5
+ {%- else %}
6
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
7
+ {%- endif %}
8
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
9
+ {%- for tool in tools %}
10
+ {{- "\n" }}
11
+ {{- tool | tojson }}
12
+ {%- endfor %}
13
+ {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
14
+ {%- else %}
15
+ {%- if messages[0]['role'] == 'system' %}
16
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
17
+ {%- else %}
18
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
19
+ {%- endif %}
20
+ {%- endif %}
21
+ {%- for message in messages %}
22
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
23
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
24
+ {%- elif message.role == "assistant" %}
25
+ {{- '<|im_start|>' + message.role }}
26
+ {%- if message.content %}
27
+ {{- '\n' + message.content }}
28
+ {%- endif %}
29
+ {%- for tool_call in message.tool_calls %}
30
+ {%- if tool_call.function is defined %}
31
+ {%- set tool_call = tool_call.function %}
32
+ {%- endif %}
33
+ {{- '\n<tool_call>\n{"name": "' }}
34
+ {{- tool_call.name }}
35
+ {{- '", "arguments": ' }}
36
+ {{- tool_call.arguments | tojson }}
37
+ {{- '}\n</tool_call>' }}
38
+ {%- endfor %}
39
+ {{- '<|im_end|>\n' }}
40
+ {%- elif message.role == "tool" %}
41
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
42
+ {{- '<|im_start|>user' }}
43
+ {%- endif %}
44
+ {{- '\n<tool_response>\n' }}
45
+ {{- message.content }}
46
+ {{- '\n</tool_response>' }}
47
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
48
+ {{- '<|im_end|>\n' }}
49
+ {%- endif %}
50
+ {%- endif %}
51
+ {%- endfor %}
52
+ {%- if add_generation_prompt %}
53
+ {{- '<|im_start|>assistant\n' }}
54
+ {%- endif %}
self_play_hf_l40s_full/round_005/generator_train/checkpoint-120/config.json ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen2ForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": null,
7
+ "dtype": "float32",
8
+ "eos_token_id": 151645,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 896,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 4864,
13
+ "layer_types": [
14
+ "full_attention",
15
+ "full_attention",
16
+ "full_attention",
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention"
38
+ ],
39
+ "max_position_embeddings": 32768,
40
+ "max_window_layers": 21,
41
+ "model_type": "qwen2",
42
+ "num_attention_heads": 14,
43
+ "num_hidden_layers": 24,
44
+ "num_key_value_heads": 2,
45
+ "pad_token_id": 151643,
46
+ "rms_norm_eps": 1e-06,
47
+ "rope_parameters": {
48
+ "rope_theta": 1000000.0,
49
+ "rope_type": "default"
50
+ },
51
+ "sliding_window": null,
52
+ "tie_word_embeddings": true,
53
+ "transformers_version": "5.6.2",
54
+ "use_cache": false,
55
+ "use_sliding_window": false,
56
+ "vocab_size": 151936
57
+ }
self_play_hf_l40s_full/round_005/generator_train/checkpoint-120/generation_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_sample": true,
3
+ "eos_token_id": [
4
+ 151645,
5
+ 151643
6
+ ],
7
+ "pad_token_id": 151643,
8
+ "repetition_penalty": 1.1,
9
+ "temperature": 0.7,
10
+ "top_k": 20,
11
+ "top_p": 0.8,
12
+ "transformers_version": "5.6.2"
13
+ }
self_play_hf_l40s_full/round_005/generator_train/checkpoint-120/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3f2d531b41a4f6103c97f60295c77faba854dc8907a530ec20d48c472e5c9621
3
+ size 1976163472
self_play_hf_l40s_full/round_005/generator_train/checkpoint-120/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e7790d25b84db93c693c5ef59cf606f44af6b783a82fb135a00dbe3dd5553ee5
3
+ size 3952509771
self_play_hf_l40s_full/round_005/generator_train/checkpoint-120/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ddf233ab1c5767c7a618809cb16cda8677bbdb0ed7511223c707ea5be3318fab
3
+ size 14645
self_play_hf_l40s_full/round_005/generator_train/checkpoint-120/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:efad52b65f455156135da53c9927013d723b01aa46bbf7e57c3c8480f44fb55f
3
+ size 1465
self_play_hf_l40s_full/round_005/generator_train/checkpoint-120/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
3
+ size 11421892
self_play_hf_l40s_full/round_005/generator_train/checkpoint-120/tokenizer_config.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "backend": "tokenizers",
4
+ "bos_token": null,
5
+ "clean_up_tokenization_spaces": false,
6
+ "eos_token": "<|im_end|>",
7
+ "errors": "replace",
8
+ "extra_special_tokens": [
9
+ "<|im_start|>",
10
+ "<|im_end|>",
11
+ "<|object_ref_start|>",
12
+ "<|object_ref_end|>",
13
+ "<|box_start|>",
14
+ "<|box_end|>",
15
+ "<|quad_start|>",
16
+ "<|quad_end|>",
17
+ "<|vision_start|>",
18
+ "<|vision_end|>",
19
+ "<|vision_pad|>",
20
+ "<|image_pad|>",
21
+ "<|video_pad|>"
22
+ ],
23
+ "is_local": true,
24
+ "local_files_only": false,
25
+ "model_max_length": 131072,
26
+ "pad_token": "<|endoftext|>",
27
+ "padding_side": "left",
28
+ "split_special_tokens": false,
29
+ "tokenizer_class": "Qwen2Tokenizer",
30
+ "truncation_side": "left",
31
+ "unk_token": null
32
+ }
self_play_hf_l40s_full/round_005/generator_train/checkpoint-120/trainer_state.json ADDED
@@ -0,0 +1,2224 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 5.0,
6
+ "eval_steps": 500,
7
+ "global_step": 120,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "clip_ratio/high_max": 0.0,
14
+ "clip_ratio/high_mean": 0.0,
15
+ "clip_ratio/low_mean": 0.0,
16
+ "clip_ratio/low_min": 0.0,
17
+ "clip_ratio/region_mean": 0.0,
18
+ "completions/clipped_ratio": 1.0,
19
+ "completions/max_length": 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": 4.218832492828369,
26
+ "epoch": 0.041666666666666664,
27
+ "frac_reward_zero_std": 0.5,
28
+ "grad_norm": 2.3453547954559326,
29
+ "kl": 0.0003995061560999602,
30
+ "learning_rate": 5e-06,
31
+ "loss": 0.062480177730321884,
32
+ "num_tokens": 25296.0,
33
+ "reward": -0.16249999403953552,
34
+ "reward_std": 0.24186772108078003,
35
+ "rewards/GeneratorRewardFunction/mean": -0.16249999403953552,
36
+ "rewards/GeneratorRewardFunction/std": 0.24186773598194122,
37
+ "step": 1,
38
+ "step_time": 12.200799825999638
39
+ },
40
+ {
41
+ "clip_ratio/high_max": 0.0,
42
+ "clip_ratio/high_mean": 0.0,
43
+ "clip_ratio/low_mean": 0.0325520820915699,
44
+ "clip_ratio/low_min": 0.0325520820915699,
45
+ "clip_ratio/region_mean": 0.0325520820915699,
46
+ "entropy": 4.258089542388916,
47
+ "epoch": 0.08333333333333333,
48
+ "grad_norm": 1.9494712352752686,
49
+ "kl": 0.01056413259357214,
50
+ "learning_rate": 4.958333333333334e-06,
51
+ "loss": 0.10853039473295212,
52
+ "step": 2,
53
+ "step_time": 0.22676954300004581
54
+ },
55
+ {
56
+ "clip_ratio/high_max": 0.0618489570915699,
57
+ "clip_ratio/high_mean": 0.0618489570915699,
58
+ "clip_ratio/low_mean": 0.0,
59
+ "clip_ratio/low_min": 0.0,
60
+ "clip_ratio/region_mean": 0.0618489570915699,
61
+ "entropy": 3.910761594772339,
62
+ "epoch": 0.125,
63
+ "grad_norm": 1.5771934986114502,
64
+ "kl": 0.03891553357243538,
65
+ "learning_rate": 4.9166666666666665e-06,
66
+ "loss": -0.08321236819028854,
67
+ "step": 3,
68
+ "step_time": 0.2259805369994865
69
+ },
70
+ {
71
+ "clip_ratio/high_max": 0.0755208358168602,
72
+ "clip_ratio/high_mean": 0.0755208358168602,
73
+ "clip_ratio/low_mean": 0.0,
74
+ "clip_ratio/low_min": 0.0,
75
+ "clip_ratio/region_mean": 0.0755208358168602,
76
+ "entropy": 4.357999324798584,
77
+ "epoch": 0.16666666666666666,
78
+ "grad_norm": 1.4966953992843628,
79
+ "kl": 0.03107132576406002,
80
+ "learning_rate": 4.875e-06,
81
+ "loss": -0.08200164884328842,
82
+ "step": 4,
83
+ "step_time": 0.22625164300006873
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": 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": 3.2972638607025146,
99
+ "epoch": 0.20833333333333334,
100
+ "frac_reward_zero_std": 1.0,
101
+ "grad_norm": 0.012044131755828857,
102
+ "kl": 0.021904518827795982,
103
+ "learning_rate": 4.833333333333333e-06,
104
+ "loss": 5.476129081216641e-05,
105
+ "num_tokens": 50928.0,
106
+ "reward": -0.05000000074505806,
107
+ "reward_std": 0.0,
108
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
109
+ "rewards/GeneratorRewardFunction/std": 0.0,
110
+ "step": 5,
111
+ "step_time": 12.42558103400006
112
+ },
113
+ {
114
+ "clip_ratio/high_max": 0.0,
115
+ "clip_ratio/high_mean": 0.0,
116
+ "clip_ratio/low_mean": 0.0,
117
+ "clip_ratio/low_min": 0.0,
118
+ "clip_ratio/region_mean": 0.0,
119
+ "entropy": 4.389660358428955,
120
+ "epoch": 0.25,
121
+ "grad_norm": 0.01742626540362835,
122
+ "kl": 0.034423865377902985,
123
+ "learning_rate": 4.791666666666668e-06,
124
+ "loss": 8.605967741459608e-05,
125
+ "step": 6,
126
+ "step_time": 0.2294949209999686
127
+ },
128
+ {
129
+ "clip_ratio/high_max": 0.0,
130
+ "clip_ratio/high_mean": 0.0,
131
+ "clip_ratio/low_mean": 0.0,
132
+ "clip_ratio/low_min": 0.0,
133
+ "clip_ratio/region_mean": 0.0,
134
+ "entropy": 4.004185199737549,
135
+ "epoch": 0.2916666666666667,
136
+ "grad_norm": 0.018903816118836403,
137
+ "kl": 0.0423373244702816,
138
+ "learning_rate": 4.75e-06,
139
+ "loss": 0.000105843304481823,
140
+ "step": 7,
141
+ "step_time": 0.22864662399933877
142
+ },
143
+ {
144
+ "clip_ratio/high_max": 0.0,
145
+ "clip_ratio/high_mean": 0.0,
146
+ "clip_ratio/low_mean": 0.0,
147
+ "clip_ratio/low_min": 0.0,
148
+ "clip_ratio/region_mean": 0.0,
149
+ "entropy": 4.172105312347412,
150
+ "epoch": 0.3333333333333333,
151
+ "grad_norm": 0.02683727815747261,
152
+ "kl": 0.05335487797856331,
153
+ "learning_rate": 4.708333333333334e-06,
154
+ "loss": 0.00013338720600586385,
155
+ "step": 8,
156
+ "step_time": 0.22813066999970033
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": 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": 3.9359867572784424,
172
+ "epoch": 0.375,
173
+ "frac_reward_zero_std": 0.75,
174
+ "grad_norm": 0.7581679821014404,
175
+ "kl": 0.05953861400485039,
176
+ "learning_rate": 4.666666666666667e-06,
177
+ "loss": -0.031090745702385902,
178
+ "num_tokens": 76192.0,
179
+ "reward": -0.08749999850988388,
180
+ "reward_std": 0.14999999105930328,
181
+ "rewards/GeneratorRewardFunction/mean": -0.08749999850988388,
182
+ "rewards/GeneratorRewardFunction/std": 0.15000000596046448,
183
+ "step": 9,
184
+ "step_time": 12.354528401000607
185
+ },
186
+ {
187
+ "clip_ratio/high_max": 0.0006510416860692203,
188
+ "clip_ratio/high_mean": 0.0006510416860692203,
189
+ "clip_ratio/low_mean": 0.0,
190
+ "clip_ratio/low_min": 0.0,
191
+ "clip_ratio/region_mean": 0.0006510416860692203,
192
+ "entropy": 3.8035316467285156,
193
+ "epoch": 0.4166666666666667,
194
+ "grad_norm": 0.7192441821098328,
195
+ "kl": 0.05113422870635986,
196
+ "learning_rate": 4.625000000000001e-06,
197
+ "loss": -0.031105250120162964,
198
+ "step": 10,
199
+ "step_time": 0.22121003899974312
200
+ },
201
+ {
202
+ "clip_ratio/high_max": 0.0,
203
+ "clip_ratio/high_mean": 0.0,
204
+ "clip_ratio/low_mean": 0.0,
205
+ "clip_ratio/low_min": 0.0,
206
+ "clip_ratio/region_mean": 0.0,
207
+ "entropy": 4.490121841430664,
208
+ "epoch": 0.4583333333333333,
209
+ "grad_norm": 0.020974429324269295,
210
+ "kl": 0.06544887274503708,
211
+ "learning_rate": 4.583333333333333e-06,
212
+ "loss": 0.00016362218593712896,
213
+ "step": 11,
214
+ "step_time": 0.22179323100044712
215
+ },
216
+ {
217
+ "clip_ratio/high_max": 0.01692708395421505,
218
+ "clip_ratio/high_mean": 0.01692708395421505,
219
+ "clip_ratio/low_mean": 0.010416666977107525,
220
+ "clip_ratio/low_min": 0.010416666977107525,
221
+ "clip_ratio/region_mean": 0.02734375,
222
+ "entropy": 4.094717025756836,
223
+ "epoch": 0.5,
224
+ "grad_norm": 2.440980911254883,
225
+ "kl": 0.056731849908828735,
226
+ "learning_rate": 4.541666666666667e-06,
227
+ "loss": 0.06417856365442276,
228
+ "step": 12,
229
+ "step_time": 0.22233516199958103
230
+ },
231
+ {
232
+ "clip_ratio/high_max": 0.0,
233
+ "clip_ratio/high_mean": 0.0,
234
+ "clip_ratio/low_mean": 0.0,
235
+ "clip_ratio/low_min": 0.0,
236
+ "clip_ratio/region_mean": 0.0,
237
+ "completions/clipped_ratio": 1.0,
238
+ "completions/max_length": 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": 4.305630207061768,
245
+ "epoch": 0.5416666666666666,
246
+ "frac_reward_zero_std": 0.75,
247
+ "grad_norm": 0.7685584425926208,
248
+ "kl": 0.06827350705862045,
249
+ "learning_rate": 4.5e-06,
250
+ "loss": -0.03110312856733799,
251
+ "num_tokens": 101896.0,
252
+ "reward": -0.05656249821186066,
253
+ "reward_std": 0.026250001043081284,
254
+ "rewards/GeneratorRewardFunction/mean": -0.05656249821186066,
255
+ "rewards/GeneratorRewardFunction/std": 0.026250001043081284,
256
+ "step": 13,
257
+ "step_time": 12.351281250999818
258
+ },
259
+ {
260
+ "clip_ratio/high_max": 0.0006510416860692203,
261
+ "clip_ratio/high_mean": 0.0006510416860692203,
262
+ "clip_ratio/low_mean": 0.0,
263
+ "clip_ratio/low_min": 0.0,
264
+ "clip_ratio/region_mean": 0.0006510416860692203,
265
+ "entropy": 4.384155750274658,
266
+ "epoch": 0.5833333333333334,
267
+ "grad_norm": 1.074666976928711,
268
+ "kl": 0.09433463960886002,
269
+ "learning_rate": 4.4583333333333336e-06,
270
+ "loss": -0.06210761144757271,
271
+ "step": 14,
272
+ "step_time": 0.228220768999563
273
+ },
274
+ {
275
+ "clip_ratio/high_max": 0.0,
276
+ "clip_ratio/high_mean": 0.0,
277
+ "clip_ratio/low_mean": 0.0013020833721384406,
278
+ "clip_ratio/low_min": 0.0013020833721384406,
279
+ "clip_ratio/region_mean": 0.0013020833721384406,
280
+ "entropy": 4.467245578765869,
281
+ "epoch": 0.625,
282
+ "grad_norm": 2.4682021141052246,
283
+ "kl": 0.0888032540678978,
284
+ "learning_rate": 4.416666666666667e-06,
285
+ "loss": 0.09433958679437637,
286
+ "step": 15,
287
+ "step_time": 0.22796953400029452
288
+ },
289
+ {
290
+ "clip_ratio/high_max": 0.0,
291
+ "clip_ratio/high_mean": 0.0,
292
+ "clip_ratio/low_mean": 0.0,
293
+ "clip_ratio/low_min": 0.0,
294
+ "clip_ratio/region_mean": 0.0,
295
+ "entropy": 4.379356384277344,
296
+ "epoch": 0.6666666666666666,
297
+ "grad_norm": 0.05159206688404083,
298
+ "kl": 0.09723836183547974,
299
+ "learning_rate": 4.3750000000000005e-06,
300
+ "loss": 0.0002430959139019251,
301
+ "step": 16,
302
+ "step_time": 0.22781476000000112
303
+ },
304
+ {
305
+ "clip_ratio/high_max": 0.0,
306
+ "clip_ratio/high_mean": 0.0,
307
+ "clip_ratio/low_mean": 0.0,
308
+ "clip_ratio/low_min": 0.0,
309
+ "clip_ratio/region_mean": 0.0,
310
+ "completions/clipped_ratio": 1.0,
311
+ "completions/max_length": 384.0,
312
+ "completions/max_terminated_length": 0.0,
313
+ "completions/mean_length": 384.0,
314
+ "completions/mean_terminated_length": 0.0,
315
+ "completions/min_length": 384.0,
316
+ "completions/min_terminated_length": 0.0,
317
+ "entropy": 4.611701965332031,
318
+ "epoch": 0.7083333333333334,
319
+ "frac_reward_zero_std": 1.0,
320
+ "grad_norm": 0.027016103267669678,
321
+ "kl": 0.09803465753793716,
322
+ "learning_rate": 4.333333333333334e-06,
323
+ "loss": 0.00024508664500899613,
324
+ "num_tokens": 127124.0,
325
+ "reward": -0.05000000074505806,
326
+ "reward_std": 0.0,
327
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
328
+ "rewards/GeneratorRewardFunction/std": 0.0,
329
+ "step": 17,
330
+ "step_time": 12.34623024100074
331
+ },
332
+ {
333
+ "clip_ratio/high_max": 0.0,
334
+ "clip_ratio/high_mean": 0.0,
335
+ "clip_ratio/low_mean": 0.0,
336
+ "clip_ratio/low_min": 0.0,
337
+ "clip_ratio/region_mean": 0.0,
338
+ "entropy": 4.3698506355285645,
339
+ "epoch": 0.75,
340
+ "grad_norm": 0.027759015560150146,
341
+ "kl": 0.10794633626937866,
342
+ "learning_rate": 4.2916666666666665e-06,
343
+ "loss": 0.00026986582088284194,
344
+ "step": 18,
345
+ "step_time": 0.22221811200051889
346
+ },
347
+ {
348
+ "clip_ratio/high_max": 0.0,
349
+ "clip_ratio/high_mean": 0.0,
350
+ "clip_ratio/low_mean": 0.0,
351
+ "clip_ratio/low_min": 0.0,
352
+ "clip_ratio/region_mean": 0.0,
353
+ "entropy": 4.589417934417725,
354
+ "epoch": 0.7916666666666666,
355
+ "grad_norm": 0.033771730959415436,
356
+ "kl": 0.1163717731833458,
357
+ "learning_rate": 4.25e-06,
358
+ "loss": 0.0002909294271375984,
359
+ "step": 19,
360
+ "step_time": 0.22278399300012097
361
+ },
362
+ {
363
+ "clip_ratio/high_max": 0.0,
364
+ "clip_ratio/high_mean": 0.0,
365
+ "clip_ratio/low_mean": 0.0,
366
+ "clip_ratio/low_min": 0.0,
367
+ "clip_ratio/region_mean": 0.0,
368
+ "entropy": 4.381205081939697,
369
+ "epoch": 0.8333333333333334,
370
+ "grad_norm": 0.034088168293237686,
371
+ "kl": 0.12577062845230103,
372
+ "learning_rate": 4.208333333333333e-06,
373
+ "loss": 0.0003144265792798251,
374
+ "step": 20,
375
+ "step_time": 0.2227981229998477
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": 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": 4.56951904296875,
391
+ "epoch": 0.875,
392
+ "frac_reward_zero_std": 0.75,
393
+ "grad_norm": 0.030865877866744995,
394
+ "kl": 0.1258154958486557,
395
+ "learning_rate": 4.166666666666667e-06,
396
+ "loss": 0.00031453874544240534,
397
+ "num_tokens": 152420.0,
398
+ "reward": -0.06875000149011612,
399
+ "reward_std": 0.07500000298023224,
400
+ "rewards/GeneratorRewardFunction/mean": -0.06875000149011612,
401
+ "rewards/GeneratorRewardFunction/std": 0.07500000298023224,
402
+ "step": 21,
403
+ "step_time": 12.431868073999794
404
+ },
405
+ {
406
+ "clip_ratio/high_max": 0.0006510416860692203,
407
+ "clip_ratio/high_mean": 0.0006510416860692203,
408
+ "clip_ratio/low_mean": 0.0,
409
+ "clip_ratio/low_min": 0.0,
410
+ "clip_ratio/region_mean": 0.0006510416860692203,
411
+ "entropy": 4.256842136383057,
412
+ "epoch": 0.9166666666666666,
413
+ "grad_norm": 1.0559728145599365,
414
+ "kl": 0.11962169408798218,
415
+ "learning_rate": 4.125e-06,
416
+ "loss": -0.06200842559337616,
417
+ "step": 22,
418
+ "step_time": 0.22747174799951608
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": 4.441510200500488,
427
+ "epoch": 0.9583333333333334,
428
+ "grad_norm": 0.8300934433937073,
429
+ "kl": 0.12788282334804535,
430
+ "learning_rate": 4.083333333333334e-06,
431
+ "loss": -0.030932163819670677,
432
+ "step": 23,
433
+ "step_time": 0.2274885289998565
434
+ },
435
+ {
436
+ "clip_ratio/high_max": 0.0,
437
+ "clip_ratio/high_mean": 0.0,
438
+ "clip_ratio/low_mean": 0.0026041667442768812,
439
+ "clip_ratio/low_min": 0.0026041667442768812,
440
+ "clip_ratio/region_mean": 0.0026041667442768812,
441
+ "entropy": 4.461675643920898,
442
+ "epoch": 1.0,
443
+ "grad_norm": 2.326770067214966,
444
+ "kl": 0.1365349292755127,
445
+ "learning_rate": 4.041666666666667e-06,
446
+ "loss": 0.09416845440864563,
447
+ "step": 24,
448
+ "step_time": 0.22709047199987253
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": 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": 4.375850200653076,
464
+ "epoch": 1.0416666666666667,
465
+ "frac_reward_zero_std": 1.0,
466
+ "grad_norm": 0.031697701662778854,
467
+ "kl": 0.1358719766139984,
468
+ "learning_rate": 4.000000000000001e-06,
469
+ "loss": 0.0003396799729671329,
470
+ "num_tokens": 177592.0,
471
+ "reward": -0.05000000074505806,
472
+ "reward_std": 0.0,
473
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
474
+ "rewards/GeneratorRewardFunction/std": 0.0,
475
+ "step": 25,
476
+ "step_time": 12.350898434000555
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": 4.472981929779053,
485
+ "epoch": 1.0833333333333333,
486
+ "grad_norm": 0.033567290753126144,
487
+ "kl": 0.1449340432882309,
488
+ "learning_rate": 3.958333333333333e-06,
489
+ "loss": 0.0003623350930865854,
490
+ "step": 26,
491
+ "step_time": 0.22230296500038094
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": 4.471365451812744,
500
+ "epoch": 1.125,
501
+ "grad_norm": 0.033745091408491135,
502
+ "kl": 0.14426572620868683,
503
+ "learning_rate": 3.916666666666667e-06,
504
+ "loss": 0.0003606643294915557,
505
+ "step": 27,
506
+ "step_time": 0.22167697199984104
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": 4.190944671630859,
515
+ "epoch": 1.1666666666666667,
516
+ "grad_norm": 0.03441477566957474,
517
+ "kl": 0.14298784732818604,
518
+ "learning_rate": 3.875e-06,
519
+ "loss": 0.0003574696311261505,
520
+ "step": 28,
521
+ "step_time": 0.22162332900006732
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": 4.321583271026611,
537
+ "epoch": 1.2083333333333333,
538
+ "frac_reward_zero_std": 0.75,
539
+ "grad_norm": 2.1965482234954834,
540
+ "kl": 0.13321949541568756,
541
+ "learning_rate": 3.833333333333334e-06,
542
+ "loss": 0.09402058273553848,
543
+ "num_tokens": 203376.0,
544
+ "reward": -0.06875000149011612,
545
+ "reward_std": 0.07499999552965164,
546
+ "rewards/GeneratorRewardFunction/mean": -0.06875000149011612,
547
+ "rewards/GeneratorRewardFunction/std": 0.07500000298023224,
548
+ "step": 29,
549
+ "step_time": 12.401389794999886
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": 4.206346035003662,
558
+ "epoch": 1.25,
559
+ "grad_norm": 0.8052829504013062,
560
+ "kl": 0.14546459913253784,
561
+ "learning_rate": 3.7916666666666666e-06,
562
+ "loss": -0.03093709610402584,
563
+ "step": 30,
564
+ "step_time": 0.22923127000012755
565
+ },
566
+ {
567
+ "clip_ratio/high_max": 0.0032552082557231188,
568
+ "clip_ratio/high_mean": 0.0032552082557231188,
569
+ "clip_ratio/low_mean": 0.0,
570
+ "clip_ratio/low_min": 0.0,
571
+ "clip_ratio/region_mean": 0.0032552082557231188,
572
+ "entropy": 4.350301742553711,
573
+ "epoch": 1.2916666666666667,
574
+ "grad_norm": 1.0959073305130005,
575
+ "kl": 0.19109351933002472,
576
+ "learning_rate": 3.7500000000000005e-06,
577
+ "loss": -0.061980169266462326,
578
+ "step": 31,
579
+ "step_time": 0.22754814400013856
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": 4.322090148925781,
588
+ "epoch": 1.3333333333333333,
589
+ "grad_norm": 0.031234100461006165,
590
+ "kl": 0.12836213409900665,
591
+ "learning_rate": 3.708333333333334e-06,
592
+ "loss": 0.00032090532477013767,
593
+ "step": 32,
594
+ "step_time": 0.22721318900039478
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": 4.353842258453369,
610
+ "epoch": 1.375,
611
+ "frac_reward_zero_std": 1.0,
612
+ "grad_norm": 0.04642889276146889,
613
+ "kl": 0.15312014520168304,
614
+ "learning_rate": 3.6666666666666666e-06,
615
+ "loss": 0.00038280035369098186,
616
+ "num_tokens": 228780.0,
617
+ "reward": -0.05000000074505806,
618
+ "reward_std": 0.0,
619
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
620
+ "rewards/GeneratorRewardFunction/std": 0.0,
621
+ "step": 33,
622
+ "step_time": 12.362105676000283
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": 4.097261905670166,
631
+ "epoch": 1.4166666666666667,
632
+ "grad_norm": 0.047277797013521194,
633
+ "kl": 0.1807316094636917,
634
+ "learning_rate": 3.625e-06,
635
+ "loss": 0.00045182902249507606,
636
+ "step": 34,
637
+ "step_time": 0.2276034909991722
638
+ },
639
+ {
640
+ "clip_ratio/high_max": 0.0,
641
+ "clip_ratio/high_mean": 0.0,
642
+ "clip_ratio/low_mean": 0.0,
643
+ "clip_ratio/low_min": 0.0,
644
+ "clip_ratio/region_mean": 0.0,
645
+ "entropy": 4.232495307922363,
646
+ "epoch": 1.4583333333333333,
647
+ "grad_norm": 0.03665163367986679,
648
+ "kl": 0.16979527473449707,
649
+ "learning_rate": 3.5833333333333335e-06,
650
+ "loss": 0.0004244882147759199,
651
+ "step": 35,
652
+ "step_time": 0.22647975700056122
653
+ },
654
+ {
655
+ "clip_ratio/high_max": 0.0,
656
+ "clip_ratio/high_mean": 0.0,
657
+ "clip_ratio/low_mean": 0.0,
658
+ "clip_ratio/low_min": 0.0,
659
+ "clip_ratio/region_mean": 0.0,
660
+ "entropy": 4.373459339141846,
661
+ "epoch": 1.5,
662
+ "grad_norm": 0.03509606420993805,
663
+ "kl": 0.1546647995710373,
664
+ "learning_rate": 3.5416666666666673e-06,
665
+ "loss": 0.0003866619954351336,
666
+ "step": 36,
667
+ "step_time": 0.22700092600007338
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": 4.0846028327941895,
683
+ "epoch": 1.5416666666666665,
684
+ "frac_reward_zero_std": 0.75,
685
+ "grad_norm": 2.4683332443237305,
686
+ "kl": 0.16653424501419067,
687
+ "learning_rate": 3.5e-06,
688
+ "loss": 0.05828670784831047,
689
+ "num_tokens": 254116.0,
690
+ "reward": -0.05015625059604645,
691
+ "reward_std": 0.0006249994039535522,
692
+ "rewards/GeneratorRewardFunction/mean": -0.05015625059604645,
693
+ "rewards/GeneratorRewardFunction/std": 0.0006249994039535522,
694
+ "step": 37,
695
+ "step_time": 12.278016894999382
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": 4.041675567626953,
704
+ "epoch": 1.5833333333333335,
705
+ "grad_norm": 0.036362167447805405,
706
+ "kl": 0.17584657669067383,
707
+ "learning_rate": 3.4583333333333334e-06,
708
+ "loss": 0.00043961641495116055,
709
+ "step": 38,
710
+ "step_time": 0.22066016499957186
711
+ },
712
+ {
713
+ "clip_ratio/high_max": 0.0013020833721384406,
714
+ "clip_ratio/high_mean": 0.0013020833721384406,
715
+ "clip_ratio/low_mean": 0.0,
716
+ "clip_ratio/low_min": 0.0,
717
+ "clip_ratio/region_mean": 0.0013020833721384406,
718
+ "entropy": 4.3244171142578125,
719
+ "epoch": 1.625,
720
+ "grad_norm": 0.6458258032798767,
721
+ "kl": 0.1514502316713333,
722
+ "learning_rate": 3.416666666666667e-06,
723
+ "loss": -0.02857072651386261,
724
+ "step": 39,
725
+ "step_time": 0.2196550920007212
726
+ },
727
+ {
728
+ "clip_ratio/high_max": 0.0013020833721384406,
729
+ "clip_ratio/high_mean": 0.0013020833721384406,
730
+ "clip_ratio/low_mean": 0.0,
731
+ "clip_ratio/low_min": 0.0,
732
+ "clip_ratio/region_mean": 0.0013020833721384406,
733
+ "entropy": 4.209409236907959,
734
+ "epoch": 1.6666666666666665,
735
+ "grad_norm": 0.7317789196968079,
736
+ "kl": 0.13871996104717255,
737
+ "learning_rate": 3.3750000000000003e-06,
738
+ "loss": -0.028499560430645943,
739
+ "step": 40,
740
+ "step_time": 0.2200379099995189
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": 4.327061653137207,
756
+ "epoch": 1.7083333333333335,
757
+ "frac_reward_zero_std": 0.75,
758
+ "grad_norm": 0.7506322860717773,
759
+ "kl": 0.16559262573719025,
760
+ "learning_rate": 3.3333333333333333e-06,
761
+ "loss": -0.03082560934126377,
762
+ "num_tokens": 279536.0,
763
+ "reward": -0.08749999850988388,
764
+ "reward_std": 0.14999999105930328,
765
+ "rewards/GeneratorRewardFunction/mean": -0.08749999850988388,
766
+ "rewards/GeneratorRewardFunction/std": 0.15000000596046448,
767
+ "step": 41,
768
+ "step_time": 12.311622474000615
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": 4.28145170211792,
777
+ "epoch": 1.75,
778
+ "grad_norm": 0.037420228123664856,
779
+ "kl": 0.1730494350194931,
780
+ "learning_rate": 3.2916666666666668e-06,
781
+ "loss": 0.00043262357939966023,
782
+ "step": 42,
783
+ "step_time": 0.22799636899981124
784
+ },
785
+ {
786
+ "clip_ratio/high_max": 0.0026041667442768812,
787
+ "clip_ratio/high_mean": 0.0026041667442768812,
788
+ "clip_ratio/low_mean": 0.0026041667442768812,
789
+ "clip_ratio/low_min": 0.0026041667442768812,
790
+ "clip_ratio/region_mean": 0.0052083334885537624,
791
+ "entropy": 3.382018804550171,
792
+ "epoch": 1.7916666666666665,
793
+ "grad_norm": 2.7098262310028076,
794
+ "kl": 0.14611031115055084,
795
+ "learning_rate": 3.2500000000000002e-06,
796
+ "loss": 0.06306206434965134,
797
+ "step": 43,
798
+ "step_time": 0.22749470799953997
799
+ },
800
+ {
801
+ "clip_ratio/high_max": 0.008463541977107525,
802
+ "clip_ratio/high_mean": 0.008463541977107525,
803
+ "clip_ratio/low_mean": 0.0,
804
+ "clip_ratio/low_min": 0.0,
805
+ "clip_ratio/region_mean": 0.008463541977107525,
806
+ "entropy": 4.229821681976318,
807
+ "epoch": 1.8333333333333335,
808
+ "grad_norm": 0.6993237733840942,
809
+ "kl": 0.16464383900165558,
810
+ "learning_rate": 3.2083333333333337e-06,
811
+ "loss": -0.030900651589035988,
812
+ "step": 44,
813
+ "step_time": 0.2268984759994055
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": 4.33951997756958,
829
+ "epoch": 1.875,
830
+ "frac_reward_zero_std": 1.0,
831
+ "grad_norm": 0.032941851764917374,
832
+ "kl": 0.18063347041606903,
833
+ "learning_rate": 3.1666666666666667e-06,
834
+ "loss": 0.00045158364810049534,
835
+ "num_tokens": 304840.0,
836
+ "reward": -0.05000000074505806,
837
+ "reward_std": 0.0,
838
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
839
+ "rewards/GeneratorRewardFunction/std": 0.0,
840
+ "step": 45,
841
+ "step_time": 12.408172645000377
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": 4.255344867706299,
850
+ "epoch": 1.9166666666666665,
851
+ "grad_norm": 0.038530610501766205,
852
+ "kl": 0.1718256026506424,
853
+ "learning_rate": 3.125e-06,
854
+ "loss": 0.00042956401011906564,
855
+ "step": 46,
856
+ "step_time": 0.2266177340006834
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": 3.8836581707000732,
865
+ "epoch": 1.9583333333333335,
866
+ "grad_norm": 0.025322094559669495,
867
+ "kl": 0.13393373787403107,
868
+ "learning_rate": 3.0833333333333336e-06,
869
+ "loss": 0.0003348343598190695,
870
+ "step": 47,
871
+ "step_time": 0.22698829199998727
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": 4.329832077026367,
880
+ "epoch": 2.0,
881
+ "grad_norm": 0.1431460827589035,
882
+ "kl": 0.22837483882904053,
883
+ "learning_rate": 3.0416666666666666e-06,
884
+ "loss": 0.0005709370598196983,
885
+ "step": 48,
886
+ "step_time": 0.22729420899941033
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": 4.286876678466797,
902
+ "epoch": 2.0416666666666665,
903
+ "frac_reward_zero_std": 1.0,
904
+ "grad_norm": 0.031947534531354904,
905
+ "kl": 0.1614980697631836,
906
+ "learning_rate": 3e-06,
907
+ "loss": 0.0004037451872136444,
908
+ "num_tokens": 330064.0,
909
+ "reward": -0.05000000074505806,
910
+ "reward_std": 0.0,
911
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
912
+ "rewards/GeneratorRewardFunction/std": 0.0,
913
+ "step": 49,
914
+ "step_time": 12.335740126000019
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": 4.031215667724609,
923
+ "epoch": 2.0833333333333335,
924
+ "grad_norm": 0.030324924737215042,
925
+ "kl": 0.16199041903018951,
926
+ "learning_rate": 2.9583333333333335e-06,
927
+ "loss": 0.00040497604641132057,
928
+ "step": 50,
929
+ "step_time": 0.2283358570002747
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": 4.352718353271484,
938
+ "epoch": 2.125,
939
+ "grad_norm": 0.03301743045449257,
940
+ "kl": 0.17951743304729462,
941
+ "learning_rate": 2.916666666666667e-06,
942
+ "loss": 0.00044879360939376056,
943
+ "step": 51,
944
+ "step_time": 0.22797506900042208
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": 4.1786041259765625,
953
+ "epoch": 2.1666666666666665,
954
+ "grad_norm": 0.03034944273531437,
955
+ "kl": 0.16076649725437164,
956
+ "learning_rate": 2.875e-06,
957
+ "loss": 0.0004019162443000823,
958
+ "step": 52,
959
+ "step_time": 0.2283364969998729
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": 4.120919704437256,
975
+ "epoch": 2.2083333333333335,
976
+ "frac_reward_zero_std": 1.0,
977
+ "grad_norm": 0.030451813712716103,
978
+ "kl": 0.16769714653491974,
979
+ "learning_rate": 2.8333333333333335e-06,
980
+ "loss": 0.00041924286051653326,
981
+ "num_tokens": 355500.0,
982
+ "reward": -0.05000000074505806,
983
+ "reward_std": 0.0,
984
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
985
+ "rewards/GeneratorRewardFunction/std": 0.0,
986
+ "step": 53,
987
+ "step_time": 12.406069011
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": 4.143009185791016,
996
+ "epoch": 2.25,
997
+ "grad_norm": 0.04242139682173729,
998
+ "kl": 0.18013274669647217,
999
+ "learning_rate": 2.791666666666667e-06,
1000
+ "loss": 0.00045033186324872077,
1001
+ "step": 54,
1002
+ "step_time": 0.22930021999945893
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": 3.9695751667022705,
1011
+ "epoch": 2.2916666666666665,
1012
+ "grad_norm": 0.026388173922896385,
1013
+ "kl": 0.14907608926296234,
1014
+ "learning_rate": 2.7500000000000004e-06,
1015
+ "loss": 0.00037269017775543034,
1016
+ "step": 55,
1017
+ "step_time": 0.22874155800036533
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": 4.342067718505859,
1026
+ "epoch": 2.3333333333333335,
1027
+ "grad_norm": 0.03221583738923073,
1028
+ "kl": 0.17097871005535126,
1029
+ "learning_rate": 2.7083333333333334e-06,
1030
+ "loss": 0.00042744679376482964,
1031
+ "step": 56,
1032
+ "step_time": 0.22889051099991775
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": 4.179519176483154,
1048
+ "epoch": 2.375,
1049
+ "frac_reward_zero_std": 1.0,
1050
+ "grad_norm": 0.028109125792980194,
1051
+ "kl": 0.17036812007427216,
1052
+ "learning_rate": 2.666666666666667e-06,
1053
+ "loss": 0.0004259202687535435,
1054
+ "num_tokens": 381024.0,
1055
+ "reward": -0.05000000074505806,
1056
+ "reward_std": 0.0,
1057
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
1058
+ "rewards/GeneratorRewardFunction/std": 0.0,
1059
+ "step": 57,
1060
+ "step_time": 12.308869638000033
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": 4.261288166046143,
1069
+ "epoch": 2.4166666666666665,
1070
+ "grad_norm": 0.030122917145490646,
1071
+ "kl": 0.1687396764755249,
1072
+ "learning_rate": 2.6250000000000003e-06,
1073
+ "loss": 0.0004218492249492556,
1074
+ "step": 58,
1075
+ "step_time": 0.23021780399994896
1076
+ },
1077
+ {
1078
+ "clip_ratio/high_max": 0.0,
1079
+ "clip_ratio/high_mean": 0.0,
1080
+ "clip_ratio/low_mean": 0.0,
1081
+ "clip_ratio/low_min": 0.0,
1082
+ "clip_ratio/region_mean": 0.0,
1083
+ "entropy": 4.327264308929443,
1084
+ "epoch": 2.4583333333333335,
1085
+ "grad_norm": 0.030730031430721283,
1086
+ "kl": 0.17366604506969452,
1087
+ "learning_rate": 2.5833333333333337e-06,
1088
+ "loss": 0.0004341650346759707,
1089
+ "step": 59,
1090
+ "step_time": 0.23006772000007913
1091
+ },
1092
+ {
1093
+ "clip_ratio/high_max": 0.0,
1094
+ "clip_ratio/high_mean": 0.0,
1095
+ "clip_ratio/low_mean": 0.0,
1096
+ "clip_ratio/low_min": 0.0,
1097
+ "clip_ratio/region_mean": 0.0,
1098
+ "entropy": 4.266103267669678,
1099
+ "epoch": 2.5,
1100
+ "grad_norm": 0.03138311952352524,
1101
+ "kl": 0.16585773229599,
1102
+ "learning_rate": 2.5416666666666668e-06,
1103
+ "loss": 0.00041464436799287796,
1104
+ "step": 60,
1105
+ "step_time": 0.2302518249998684
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": 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": 4.149628162384033,
1121
+ "epoch": 2.5416666666666665,
1122
+ "frac_reward_zero_std": 0.75,
1123
+ "grad_norm": 0.031717631965875626,
1124
+ "kl": 0.1798960417509079,
1125
+ "learning_rate": 2.5e-06,
1126
+ "loss": 0.00044974012416787446,
1127
+ "num_tokens": 406568.0,
1128
+ "reward": -0.08749999850988388,
1129
+ "reward_std": 0.15000000596046448,
1130
+ "rewards/GeneratorRewardFunction/mean": -0.08749999850988388,
1131
+ "rewards/GeneratorRewardFunction/std": 0.15000000596046448,
1132
+ "step": 61,
1133
+ "step_time": 12.3971151249998
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": 4.3893141746521,
1142
+ "epoch": 2.5833333333333335,
1143
+ "grad_norm": 0.7632278203964233,
1144
+ "kl": 0.1566535085439682,
1145
+ "learning_rate": 2.4583333333333332e-06,
1146
+ "loss": -0.030905282124876976,
1147
+ "step": 62,
1148
+ "step_time": 0.23158665199935058
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": 4.1660566329956055,
1157
+ "epoch": 2.625,
1158
+ "grad_norm": 2.1362805366516113,
1159
+ "kl": 0.1582614928483963,
1160
+ "learning_rate": 2.4166666666666667e-06,
1161
+ "loss": 0.09412431716918945,
1162
+ "step": 63,
1163
+ "step_time": 0.2294769879999876
1164
+ },
1165
+ {
1166
+ "clip_ratio/high_max": 0.0032552082557231188,
1167
+ "clip_ratio/high_mean": 0.0032552082557231188,
1168
+ "clip_ratio/low_mean": 0.0,
1169
+ "clip_ratio/low_min": 0.0,
1170
+ "clip_ratio/region_mean": 0.0032552082557231188,
1171
+ "entropy": 4.3899383544921875,
1172
+ "epoch": 2.6666666666666665,
1173
+ "grad_norm": 1.1546765565872192,
1174
+ "kl": 0.15236173570156097,
1175
+ "learning_rate": 2.375e-06,
1176
+ "loss": -0.06209668517112732,
1177
+ "step": 64,
1178
+ "step_time": 0.22880948400052148
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": 4.150634288787842,
1194
+ "epoch": 2.7083333333333335,
1195
+ "frac_reward_zero_std": 1.0,
1196
+ "grad_norm": 0.030919430777430534,
1197
+ "kl": 0.1692151576280594,
1198
+ "learning_rate": 2.3333333333333336e-06,
1199
+ "loss": 0.0004230378835927695,
1200
+ "num_tokens": 431868.0,
1201
+ "reward": -0.05000000074505806,
1202
+ "reward_std": 0.0,
1203
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
1204
+ "rewards/GeneratorRewardFunction/std": 0.0,
1205
+ "step": 65,
1206
+ "step_time": 12.471727858999657
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": 4.398933410644531,
1215
+ "epoch": 2.75,
1216
+ "grad_norm": 0.03198188170790672,
1217
+ "kl": 0.16747929155826569,
1218
+ "learning_rate": 2.2916666666666666e-06,
1219
+ "loss": 0.00041869821143336594,
1220
+ "step": 66,
1221
+ "step_time": 0.22940661600023304
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": 4.332639217376709,
1230
+ "epoch": 2.7916666666666665,
1231
+ "grad_norm": 0.026377171277999878,
1232
+ "kl": 0.15790392458438873,
1233
+ "learning_rate": 2.25e-06,
1234
+ "loss": 0.00039475978701375425,
1235
+ "step": 67,
1236
+ "step_time": 0.2274791459994958
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": 4.401764869689941,
1245
+ "epoch": 2.8333333333333335,
1246
+ "grad_norm": 0.03296651318669319,
1247
+ "kl": 0.18048568069934845,
1248
+ "learning_rate": 2.2083333333333335e-06,
1249
+ "loss": 0.000451214233180508,
1250
+ "step": 68,
1251
+ "step_time": 0.22802397799932805
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": 4.272307395935059,
1267
+ "epoch": 2.875,
1268
+ "frac_reward_zero_std": 0.75,
1269
+ "grad_norm": 0.03396793082356453,
1270
+ "kl": 0.15817908942699432,
1271
+ "learning_rate": 2.166666666666667e-06,
1272
+ "loss": 0.00039544771425426006,
1273
+ "num_tokens": 457260.0,
1274
+ "reward": -0.06875000149011612,
1275
+ "reward_std": 0.07499999552965164,
1276
+ "rewards/GeneratorRewardFunction/mean": -0.06875000149011612,
1277
+ "rewards/GeneratorRewardFunction/std": 0.07500000298023224,
1278
+ "step": 69,
1279
+ "step_time": 12.304982241999824
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": 4.329678058624268,
1288
+ "epoch": 2.9166666666666665,
1289
+ "grad_norm": 2.354091167449951,
1290
+ "kl": 0.15900522470474243,
1291
+ "learning_rate": 2.125e-06,
1292
+ "loss": 0.06253280490636826,
1293
+ "step": 70,
1294
+ "step_time": 0.2274878809994334
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": 4.420726776123047,
1303
+ "epoch": 2.9583333333333335,
1304
+ "grad_norm": 1.074094533920288,
1305
+ "kl": 0.14555977284908295,
1306
+ "learning_rate": 2.0833333333333334e-06,
1307
+ "loss": -0.06215827539563179,
1308
+ "step": 71,
1309
+ "step_time": 0.22734129800028313
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": 4.37536096572876,
1318
+ "epoch": 3.0,
1319
+ "grad_norm": 0.03002353385090828,
1320
+ "kl": 0.16313163936138153,
1321
+ "learning_rate": 2.041666666666667e-06,
1322
+ "loss": 0.0004078290658071637,
1323
+ "step": 72,
1324
+ "step_time": 0.2267978280005991
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": 4.281651020050049,
1340
+ "epoch": 3.0416666666666665,
1341
+ "frac_reward_zero_std": 0.75,
1342
+ "grad_norm": 0.029297830536961555,
1343
+ "kl": 0.17000456154346466,
1344
+ "learning_rate": 2.0000000000000003e-06,
1345
+ "loss": 0.0004250114143360406,
1346
+ "num_tokens": 482596.0,
1347
+ "reward": -0.08749999850988388,
1348
+ "reward_std": 0.14999999105930328,
1349
+ "rewards/GeneratorRewardFunction/mean": -0.08749999850988388,
1350
+ "rewards/GeneratorRewardFunction/std": 0.15000000596046448,
1351
+ "step": 73,
1352
+ "step_time": 12.327367548999973
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": 4.1541523933410645,
1361
+ "epoch": 3.0833333333333335,
1362
+ "grad_norm": 2.465907573699951,
1363
+ "kl": 0.14764432609081268,
1364
+ "learning_rate": 1.9583333333333334e-06,
1365
+ "loss": 0.09413985162973404,
1366
+ "step": 74,
1367
+ "step_time": 0.22862437100047828
1368
+ },
1369
+ {
1370
+ "clip_ratio/high_max": 0.001953125,
1371
+ "clip_ratio/high_mean": 0.001953125,
1372
+ "clip_ratio/low_mean": 0.0,
1373
+ "clip_ratio/low_min": 0.0,
1374
+ "clip_ratio/region_mean": 0.001953125,
1375
+ "entropy": 4.466418266296387,
1376
+ "epoch": 3.125,
1377
+ "grad_norm": 1.0353459119796753,
1378
+ "kl": 0.1653010994195938,
1379
+ "learning_rate": 1.916666666666667e-06,
1380
+ "loss": -0.062276482582092285,
1381
+ "step": 75,
1382
+ "step_time": 0.22827705399959086
1383
+ },
1384
+ {
1385
+ "clip_ratio/high_max": 0.0013020833721384406,
1386
+ "clip_ratio/high_mean": 0.0013020833721384406,
1387
+ "clip_ratio/low_mean": 0.0,
1388
+ "clip_ratio/low_min": 0.0,
1389
+ "clip_ratio/region_mean": 0.0013020833721384406,
1390
+ "entropy": 4.261594772338867,
1391
+ "epoch": 3.1666666666666665,
1392
+ "grad_norm": 0.7295825481414795,
1393
+ "kl": 0.14723657071590424,
1394
+ "learning_rate": 1.8750000000000003e-06,
1395
+ "loss": -0.03112857975065708,
1396
+ "step": 76,
1397
+ "step_time": 0.22811511000054452
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": 4.484732627868652,
1413
+ "epoch": 3.2083333333333335,
1414
+ "frac_reward_zero_std": 0.75,
1415
+ "grad_norm": 0.027748463675379753,
1416
+ "kl": 0.1574263721704483,
1417
+ "learning_rate": 1.8333333333333333e-06,
1418
+ "loss": 0.0003935658896807581,
1419
+ "num_tokens": 507932.0,
1420
+ "reward": -0.08749999850988388,
1421
+ "reward_std": 0.15000000596046448,
1422
+ "rewards/GeneratorRewardFunction/mean": -0.08749999850988388,
1423
+ "rewards/GeneratorRewardFunction/std": 0.15000000596046448,
1424
+ "step": 77,
1425
+ "step_time": 12.387434866000149
1426
+ },
1427
+ {
1428
+ "clip_ratio/high_max": 0.0,
1429
+ "clip_ratio/high_mean": 0.0,
1430
+ "clip_ratio/low_mean": 0.0,
1431
+ "clip_ratio/low_min": 0.0,
1432
+ "clip_ratio/region_mean": 0.0,
1433
+ "entropy": 4.202768802642822,
1434
+ "epoch": 3.25,
1435
+ "grad_norm": 0.7808005809783936,
1436
+ "kl": 0.16236911714076996,
1437
+ "learning_rate": 1.7916666666666667e-06,
1438
+ "loss": -0.0308381374925375,
1439
+ "step": 78,
1440
+ "step_time": 0.22989690699978382
1441
+ },
1442
+ {
1443
+ "clip_ratio/high_max": 0.0,
1444
+ "clip_ratio/high_mean": 0.0,
1445
+ "clip_ratio/low_mean": 0.0,
1446
+ "clip_ratio/low_min": 0.0,
1447
+ "clip_ratio/region_mean": 0.0,
1448
+ "entropy": 4.211698532104492,
1449
+ "epoch": 3.2916666666666665,
1450
+ "grad_norm": 2.246155261993408,
1451
+ "kl": 0.17786841094493866,
1452
+ "learning_rate": 1.75e-06,
1453
+ "loss": 0.09421990066766739,
1454
+ "step": 79,
1455
+ "step_time": 0.2285108489995764
1456
+ },
1457
+ {
1458
+ "clip_ratio/high_max": 0.0032552082557231188,
1459
+ "clip_ratio/high_mean": 0.0032552082557231188,
1460
+ "clip_ratio/low_mean": 0.0,
1461
+ "clip_ratio/low_min": 0.0,
1462
+ "clip_ratio/region_mean": 0.0032552082557231188,
1463
+ "entropy": 4.602865219116211,
1464
+ "epoch": 3.3333333333333335,
1465
+ "grad_norm": 1.0171711444854736,
1466
+ "kl": 0.1722370833158493,
1467
+ "learning_rate": 1.7083333333333334e-06,
1468
+ "loss": -0.062155067920684814,
1469
+ "step": 80,
1470
+ "step_time": 0.2289368479996483
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": 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": 4.363098621368408,
1486
+ "epoch": 3.375,
1487
+ "frac_reward_zero_std": 1.0,
1488
+ "grad_norm": 0.027760757133364677,
1489
+ "kl": 0.16952906548976898,
1490
+ "learning_rate": 1.6666666666666667e-06,
1491
+ "loss": 0.00042382263927720487,
1492
+ "num_tokens": 533560.0,
1493
+ "reward": -0.05000000074505806,
1494
+ "reward_std": 0.0,
1495
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
1496
+ "rewards/GeneratorRewardFunction/std": 0.0,
1497
+ "step": 81,
1498
+ "step_time": 12.377727779999987
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": 4.349791049957275,
1507
+ "epoch": 3.4166666666666665,
1508
+ "grad_norm": 0.027347413823008537,
1509
+ "kl": 0.15509888529777527,
1510
+ "learning_rate": 1.6250000000000001e-06,
1511
+ "loss": 0.0003877471899613738,
1512
+ "step": 82,
1513
+ "step_time": 0.23147920399969735
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": 4.442678451538086,
1522
+ "epoch": 3.4583333333333335,
1523
+ "grad_norm": 0.035067103803157806,
1524
+ "kl": 0.16633033752441406,
1525
+ "learning_rate": 1.5833333333333333e-06,
1526
+ "loss": 0.0004158258670940995,
1527
+ "step": 83,
1528
+ "step_time": 0.2302408289997402
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": 4.387204170227051,
1537
+ "epoch": 3.5,
1538
+ "grad_norm": 0.02926628850400448,
1539
+ "kl": 0.16001783311367035,
1540
+ "learning_rate": 1.5416666666666668e-06,
1541
+ "loss": 0.0004000445769634098,
1542
+ "step": 84,
1543
+ "step_time": 0.23049437400004535
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": 3.68717885017395,
1559
+ "epoch": 3.5416666666666665,
1560
+ "frac_reward_zero_std": 0.75,
1561
+ "grad_norm": 1.21665358543396,
1562
+ "kl": 0.14173372089862823,
1563
+ "learning_rate": 1.5e-06,
1564
+ "loss": 0.06187007203698158,
1565
+ "num_tokens": 558860.0,
1566
+ "reward": -0.05078125,
1567
+ "reward_std": 0.0031250000465661287,
1568
+ "rewards/GeneratorRewardFunction/mean": -0.05078125,
1569
+ "rewards/GeneratorRewardFunction/std": 0.0031250000465661287,
1570
+ "step": 85,
1571
+ "step_time": 12.32021258199984
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": 4.593854904174805,
1580
+ "epoch": 3.5833333333333335,
1581
+ "grad_norm": 0.031670551747083664,
1582
+ "kl": 0.1700318455696106,
1583
+ "learning_rate": 1.4583333333333335e-06,
1584
+ "loss": 0.00042507960461080074,
1585
+ "step": 86,
1586
+ "step_time": 0.22247004299970286
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": 4.506083965301514,
1595
+ "epoch": 3.625,
1596
+ "grad_norm": 0.03483761474490166,
1597
+ "kl": 0.1663081794977188,
1598
+ "learning_rate": 1.4166666666666667e-06,
1599
+ "loss": 0.00041577042429707944,
1600
+ "step": 87,
1601
+ "step_time": 0.22216213699994114
1602
+ },
1603
+ {
1604
+ "clip_ratio/high_max": 0.001953125,
1605
+ "clip_ratio/high_mean": 0.001953125,
1606
+ "clip_ratio/low_mean": 0.0,
1607
+ "clip_ratio/low_min": 0.0,
1608
+ "clip_ratio/region_mean": 0.001953125,
1609
+ "entropy": 4.289217472076416,
1610
+ "epoch": 3.6666666666666665,
1611
+ "grad_norm": 1.0936129093170166,
1612
+ "kl": 0.1807820200920105,
1613
+ "learning_rate": 1.3750000000000002e-06,
1614
+ "loss": -0.06104663014411926,
1615
+ "step": 88,
1616
+ "step_time": 0.22188600200024666
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": 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": 4.46470832824707,
1632
+ "epoch": 3.7083333333333335,
1633
+ "frac_reward_zero_std": 1.0,
1634
+ "grad_norm": 0.0287722647190094,
1635
+ "kl": 0.1612616628408432,
1636
+ "learning_rate": 1.3333333333333334e-06,
1637
+ "loss": 0.0004031541757285595,
1638
+ "num_tokens": 584528.0,
1639
+ "reward": -0.05000000074505806,
1640
+ "reward_std": 0.0,
1641
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
1642
+ "rewards/GeneratorRewardFunction/std": 0.0,
1643
+ "step": 89,
1644
+ "step_time": 12.365053849000105
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": 4.535579204559326,
1653
+ "epoch": 3.75,
1654
+ "grad_norm": 0.05196619778871536,
1655
+ "kl": 0.18497739732265472,
1656
+ "learning_rate": 1.2916666666666669e-06,
1657
+ "loss": 0.00046244345139712095,
1658
+ "step": 90,
1659
+ "step_time": 0.22889404699981242
1660
+ },
1661
+ {
1662
+ "clip_ratio/high_max": 0.0,
1663
+ "clip_ratio/high_mean": 0.0,
1664
+ "clip_ratio/low_mean": 0.0,
1665
+ "clip_ratio/low_min": 0.0,
1666
+ "clip_ratio/region_mean": 0.0,
1667
+ "entropy": 4.383248805999756,
1668
+ "epoch": 3.7916666666666665,
1669
+ "grad_norm": 0.028284456580877304,
1670
+ "kl": 0.16009828448295593,
1671
+ "learning_rate": 1.25e-06,
1672
+ "loss": 0.0004002457426395267,
1673
+ "step": 91,
1674
+ "step_time": 0.2279681170002732
1675
+ },
1676
+ {
1677
+ "clip_ratio/high_max": 0.0,
1678
+ "clip_ratio/high_mean": 0.0,
1679
+ "clip_ratio/low_mean": 0.0,
1680
+ "clip_ratio/low_min": 0.0,
1681
+ "clip_ratio/region_mean": 0.0,
1682
+ "entropy": 4.586085796356201,
1683
+ "epoch": 3.8333333333333335,
1684
+ "grad_norm": 0.026829512789845467,
1685
+ "kl": 0.15121014416217804,
1686
+ "learning_rate": 1.2083333333333333e-06,
1687
+ "loss": 0.00037802537553943694,
1688
+ "step": 92,
1689
+ "step_time": 0.22847567699955107
1690
+ },
1691
+ {
1692
+ "clip_ratio/high_max": 0.0,
1693
+ "clip_ratio/high_mean": 0.0,
1694
+ "clip_ratio/low_mean": 0.0,
1695
+ "clip_ratio/low_min": 0.0,
1696
+ "clip_ratio/region_mean": 0.0,
1697
+ "completions/clipped_ratio": 1.0,
1698
+ "completions/max_length": 384.0,
1699
+ "completions/max_terminated_length": 0.0,
1700
+ "completions/mean_length": 384.0,
1701
+ "completions/mean_terminated_length": 0.0,
1702
+ "completions/min_length": 384.0,
1703
+ "completions/min_terminated_length": 0.0,
1704
+ "entropy": 4.395538330078125,
1705
+ "epoch": 3.875,
1706
+ "frac_reward_zero_std": 0.75,
1707
+ "grad_norm": 1.033445119857788,
1708
+ "kl": 0.1492132991552353,
1709
+ "learning_rate": 1.1666666666666668e-06,
1710
+ "loss": -0.06210946664214134,
1711
+ "num_tokens": 609680.0,
1712
+ "reward": -0.09460937976837158,
1713
+ "reward_std": 0.1784375011920929,
1714
+ "rewards/GeneratorRewardFunction/mean": -0.09460937976837158,
1715
+ "rewards/GeneratorRewardFunction/std": 0.1784375011920929,
1716
+ "step": 93,
1717
+ "step_time": 12.347689829000046
1718
+ },
1719
+ {
1720
+ "clip_ratio/high_max": 0.0,
1721
+ "clip_ratio/high_mean": 0.0,
1722
+ "clip_ratio/low_mean": 0.0,
1723
+ "clip_ratio/low_min": 0.0,
1724
+ "clip_ratio/region_mean": 0.0,
1725
+ "entropy": 4.484641075134277,
1726
+ "epoch": 3.9166666666666665,
1727
+ "grad_norm": 0.04464663937687874,
1728
+ "kl": 0.18426960706710815,
1729
+ "learning_rate": 1.125e-06,
1730
+ "loss": 0.000460673967609182,
1731
+ "step": 94,
1732
+ "step_time": 0.22694099799991818
1733
+ },
1734
+ {
1735
+ "clip_ratio/high_max": 0.0,
1736
+ "clip_ratio/high_mean": 0.0,
1737
+ "clip_ratio/low_mean": 0.0,
1738
+ "clip_ratio/low_min": 0.0,
1739
+ "clip_ratio/region_mean": 0.0,
1740
+ "entropy": 3.422642946243286,
1741
+ "epoch": 3.9583333333333335,
1742
+ "grad_norm": 0.5639758110046387,
1743
+ "kl": 0.17420244216918945,
1744
+ "learning_rate": 1.0833333333333335e-06,
1745
+ "loss": 0.09412115812301636,
1746
+ "step": 95,
1747
+ "step_time": 0.22602716800065537
1748
+ },
1749
+ {
1750
+ "clip_ratio/high_max": 0.0,
1751
+ "clip_ratio/high_mean": 0.0,
1752
+ "clip_ratio/low_mean": 0.0,
1753
+ "clip_ratio/low_min": 0.0,
1754
+ "clip_ratio/region_mean": 0.0,
1755
+ "entropy": 4.437925338745117,
1756
+ "epoch": 4.0,
1757
+ "grad_norm": 0.770872950553894,
1758
+ "kl": 0.16676191985607147,
1759
+ "learning_rate": 1.0416666666666667e-06,
1760
+ "loss": -0.03097396530210972,
1761
+ "step": 96,
1762
+ "step_time": 0.22649503800039383
1763
+ },
1764
+ {
1765
+ "clip_ratio/high_max": 0.0,
1766
+ "clip_ratio/high_mean": 0.0,
1767
+ "clip_ratio/low_mean": 0.0,
1768
+ "clip_ratio/low_min": 0.0,
1769
+ "clip_ratio/region_mean": 0.0,
1770
+ "completions/clipped_ratio": 1.0,
1771
+ "completions/max_length": 384.0,
1772
+ "completions/max_terminated_length": 0.0,
1773
+ "completions/mean_length": 384.0,
1774
+ "completions/mean_terminated_length": 0.0,
1775
+ "completions/min_length": 384.0,
1776
+ "completions/min_terminated_length": 0.0,
1777
+ "entropy": 4.315637111663818,
1778
+ "epoch": 4.041666666666667,
1779
+ "frac_reward_zero_std": 0.75,
1780
+ "grad_norm": 0.8566199541091919,
1781
+ "kl": 0.1636391580104828,
1782
+ "learning_rate": 1.0000000000000002e-06,
1783
+ "loss": -0.03090417943894863,
1784
+ "num_tokens": 635024.0,
1785
+ "reward": -0.08749999850988388,
1786
+ "reward_std": 0.14999999105930328,
1787
+ "rewards/GeneratorRewardFunction/mean": -0.08749999850988388,
1788
+ "rewards/GeneratorRewardFunction/std": 0.15000000596046448,
1789
+ "step": 97,
1790
+ "step_time": 12.357226628999342
1791
+ },
1792
+ {
1793
+ "clip_ratio/high_max": 0.0,
1794
+ "clip_ratio/high_mean": 0.0,
1795
+ "clip_ratio/low_mean": 0.0,
1796
+ "clip_ratio/low_min": 0.0,
1797
+ "clip_ratio/region_mean": 0.0,
1798
+ "entropy": 4.506310939788818,
1799
+ "epoch": 4.083333333333333,
1800
+ "grad_norm": 0.7631968855857849,
1801
+ "kl": 0.16027097404003143,
1802
+ "learning_rate": 9.583333333333334e-07,
1803
+ "loss": -0.030869366601109505,
1804
+ "step": 98,
1805
+ "step_time": 0.2284578409999085
1806
+ },
1807
+ {
1808
+ "clip_ratio/high_max": 0.0,
1809
+ "clip_ratio/high_mean": 0.0,
1810
+ "clip_ratio/low_mean": 0.0,
1811
+ "clip_ratio/low_min": 0.0,
1812
+ "clip_ratio/region_mean": 0.0,
1813
+ "entropy": 4.485410213470459,
1814
+ "epoch": 4.125,
1815
+ "grad_norm": 0.031147433444857597,
1816
+ "kl": 0.1742296814918518,
1817
+ "learning_rate": 9.166666666666666e-07,
1818
+ "loss": 0.00043557421304285526,
1819
+ "step": 99,
1820
+ "step_time": 0.22820654599945556
1821
+ },
1822
+ {
1823
+ "clip_ratio/high_max": 0.0,
1824
+ "clip_ratio/high_mean": 0.0,
1825
+ "clip_ratio/low_mean": 0.0,
1826
+ "clip_ratio/low_min": 0.0,
1827
+ "clip_ratio/region_mean": 0.0,
1828
+ "entropy": 4.353515148162842,
1829
+ "epoch": 4.166666666666667,
1830
+ "grad_norm": 2.4380831718444824,
1831
+ "kl": 0.17426633834838867,
1832
+ "learning_rate": 8.75e-07,
1833
+ "loss": 0.06307206302881241,
1834
+ "step": 100,
1835
+ "step_time": 0.22843013999954564
1836
+ },
1837
+ {
1838
+ "clip_ratio/high_max": 0.0,
1839
+ "clip_ratio/high_mean": 0.0,
1840
+ "clip_ratio/low_mean": 0.0,
1841
+ "clip_ratio/low_min": 0.0,
1842
+ "clip_ratio/region_mean": 0.0,
1843
+ "completions/clipped_ratio": 1.0,
1844
+ "completions/max_length": 384.0,
1845
+ "completions/max_terminated_length": 0.0,
1846
+ "completions/mean_length": 384.0,
1847
+ "completions/mean_terminated_length": 0.0,
1848
+ "completions/min_length": 384.0,
1849
+ "completions/min_terminated_length": 0.0,
1850
+ "entropy": 4.318760871887207,
1851
+ "epoch": 4.208333333333333,
1852
+ "frac_reward_zero_std": 1.0,
1853
+ "grad_norm": 0.028121337294578552,
1854
+ "kl": 0.15765920281410217,
1855
+ "learning_rate": 8.333333333333333e-07,
1856
+ "loss": 0.0003941480244975537,
1857
+ "num_tokens": 660652.0,
1858
+ "reward": -0.05000000074505806,
1859
+ "reward_std": 0.0,
1860
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
1861
+ "rewards/GeneratorRewardFunction/std": 0.0,
1862
+ "step": 101,
1863
+ "step_time": 12.370995348000179
1864
+ },
1865
+ {
1866
+ "clip_ratio/high_max": 0.0,
1867
+ "clip_ratio/high_mean": 0.0,
1868
+ "clip_ratio/low_mean": 0.0,
1869
+ "clip_ratio/low_min": 0.0,
1870
+ "clip_ratio/region_mean": 0.0,
1871
+ "entropy": 4.430075168609619,
1872
+ "epoch": 4.25,
1873
+ "grad_norm": 0.03579903766512871,
1874
+ "kl": 0.17973928153514862,
1875
+ "learning_rate": 7.916666666666667e-07,
1876
+ "loss": 0.00044934815377928317,
1877
+ "step": 102,
1878
+ "step_time": 0.2293741889998273
1879
+ },
1880
+ {
1881
+ "clip_ratio/high_max": 0.0,
1882
+ "clip_ratio/high_mean": 0.0,
1883
+ "clip_ratio/low_mean": 0.0,
1884
+ "clip_ratio/low_min": 0.0,
1885
+ "clip_ratio/region_mean": 0.0,
1886
+ "entropy": 4.247735023498535,
1887
+ "epoch": 4.291666666666667,
1888
+ "grad_norm": 0.0289804358035326,
1889
+ "kl": 0.1657833307981491,
1890
+ "learning_rate": 7.5e-07,
1891
+ "loss": 0.00041445824899710715,
1892
+ "step": 103,
1893
+ "step_time": 0.2288563800002521
1894
+ },
1895
+ {
1896
+ "clip_ratio/high_max": 0.0,
1897
+ "clip_ratio/high_mean": 0.0,
1898
+ "clip_ratio/low_mean": 0.0,
1899
+ "clip_ratio/low_min": 0.0,
1900
+ "clip_ratio/region_mean": 0.0,
1901
+ "entropy": 4.115577220916748,
1902
+ "epoch": 4.333333333333333,
1903
+ "grad_norm": 0.028330236673355103,
1904
+ "kl": 0.14796315133571625,
1905
+ "learning_rate": 7.083333333333334e-07,
1906
+ "loss": 0.0003699079097714275,
1907
+ "step": 104,
1908
+ "step_time": 0.23080896100054815
1909
+ },
1910
+ {
1911
+ "clip_ratio/high_max": 0.0,
1912
+ "clip_ratio/high_mean": 0.0,
1913
+ "clip_ratio/low_mean": 0.0,
1914
+ "clip_ratio/low_min": 0.0,
1915
+ "clip_ratio/region_mean": 0.0,
1916
+ "completions/clipped_ratio": 1.0,
1917
+ "completions/max_length": 384.0,
1918
+ "completions/max_terminated_length": 0.0,
1919
+ "completions/mean_length": 384.0,
1920
+ "completions/mean_terminated_length": 0.0,
1921
+ "completions/min_length": 384.0,
1922
+ "completions/min_terminated_length": 0.0,
1923
+ "entropy": 4.270058631896973,
1924
+ "epoch": 4.375,
1925
+ "frac_reward_zero_std": 1.0,
1926
+ "grad_norm": 0.02760324627161026,
1927
+ "kl": 0.15488965809345245,
1928
+ "learning_rate": 6.666666666666667e-07,
1929
+ "loss": 0.0003872241359204054,
1930
+ "num_tokens": 685708.0,
1931
+ "reward": -0.05000000074505806,
1932
+ "reward_std": 0.0,
1933
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
1934
+ "rewards/GeneratorRewardFunction/std": 0.0,
1935
+ "step": 105,
1936
+ "step_time": 12.344494807000046
1937
+ },
1938
+ {
1939
+ "clip_ratio/high_max": 0.0,
1940
+ "clip_ratio/high_mean": 0.0,
1941
+ "clip_ratio/low_mean": 0.0,
1942
+ "clip_ratio/low_min": 0.0,
1943
+ "clip_ratio/region_mean": 0.0,
1944
+ "entropy": 4.546361923217773,
1945
+ "epoch": 4.416666666666667,
1946
+ "grad_norm": 0.03014361299574375,
1947
+ "kl": 0.17117412388324738,
1948
+ "learning_rate": 6.25e-07,
1949
+ "loss": 0.0004279353597667068,
1950
+ "step": 106,
1951
+ "step_time": 0.22083965500041813
1952
+ },
1953
+ {
1954
+ "clip_ratio/high_max": 0.0,
1955
+ "clip_ratio/high_mean": 0.0,
1956
+ "clip_ratio/low_mean": 0.0,
1957
+ "clip_ratio/low_min": 0.0,
1958
+ "clip_ratio/region_mean": 0.0,
1959
+ "entropy": 4.564030647277832,
1960
+ "epoch": 4.458333333333333,
1961
+ "grad_norm": 0.03440464660525322,
1962
+ "kl": 0.17559456825256348,
1963
+ "learning_rate": 5.833333333333334e-07,
1964
+ "loss": 0.0004389864334370941,
1965
+ "step": 107,
1966
+ "step_time": 0.22036629600006563
1967
+ },
1968
+ {
1969
+ "clip_ratio/high_max": 0.0,
1970
+ "clip_ratio/high_mean": 0.0,
1971
+ "clip_ratio/low_mean": 0.0,
1972
+ "clip_ratio/low_min": 0.0,
1973
+ "clip_ratio/region_mean": 0.0,
1974
+ "entropy": 4.176433086395264,
1975
+ "epoch": 4.5,
1976
+ "grad_norm": 0.028006279841065407,
1977
+ "kl": 0.163270965218544,
1978
+ "learning_rate": 5.416666666666667e-07,
1979
+ "loss": 0.00040817740955390036,
1980
+ "step": 108,
1981
+ "step_time": 0.22036177600057272
1982
+ },
1983
+ {
1984
+ "clip_ratio/high_max": 0.0,
1985
+ "clip_ratio/high_mean": 0.0,
1986
+ "clip_ratio/low_mean": 0.0,
1987
+ "clip_ratio/low_min": 0.0,
1988
+ "clip_ratio/region_mean": 0.0,
1989
+ "completions/clipped_ratio": 1.0,
1990
+ "completions/max_length": 384.0,
1991
+ "completions/max_terminated_length": 0.0,
1992
+ "completions/mean_length": 384.0,
1993
+ "completions/mean_terminated_length": 0.0,
1994
+ "completions/min_length": 384.0,
1995
+ "completions/min_terminated_length": 0.0,
1996
+ "entropy": 4.552209854125977,
1997
+ "epoch": 4.541666666666667,
1998
+ "frac_reward_zero_std": 1.0,
1999
+ "grad_norm": 0.030375948175787926,
2000
+ "kl": 0.16741327941417694,
2001
+ "learning_rate": 5.000000000000001e-07,
2002
+ "loss": 0.00041853313450701535,
2003
+ "num_tokens": 711060.0,
2004
+ "reward": -0.05000000074505806,
2005
+ "reward_std": 0.0,
2006
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
2007
+ "rewards/GeneratorRewardFunction/std": 0.0,
2008
+ "step": 109,
2009
+ "step_time": 12.369933012000729
2010
+ },
2011
+ {
2012
+ "clip_ratio/high_max": 0.0,
2013
+ "clip_ratio/high_mean": 0.0,
2014
+ "clip_ratio/low_mean": 0.0,
2015
+ "clip_ratio/low_min": 0.0,
2016
+ "clip_ratio/region_mean": 0.0,
2017
+ "entropy": 4.339206695556641,
2018
+ "epoch": 4.583333333333333,
2019
+ "grad_norm": 0.10336848348379135,
2020
+ "kl": 0.21190087497234344,
2021
+ "learning_rate": 4.583333333333333e-07,
2022
+ "loss": 0.0005297521711327136,
2023
+ "step": 110,
2024
+ "step_time": 0.2273971739996341
2025
+ },
2026
+ {
2027
+ "clip_ratio/high_max": 0.0,
2028
+ "clip_ratio/high_mean": 0.0,
2029
+ "clip_ratio/low_mean": 0.0,
2030
+ "clip_ratio/low_min": 0.0,
2031
+ "clip_ratio/region_mean": 0.0,
2032
+ "entropy": 4.6490559577941895,
2033
+ "epoch": 4.625,
2034
+ "grad_norm": 0.03464744985103607,
2035
+ "kl": 0.17225484549999237,
2036
+ "learning_rate": 4.1666666666666667e-07,
2037
+ "loss": 0.00043063712655566633,
2038
+ "step": 111,
2039
+ "step_time": 0.2279215849994216
2040
+ },
2041
+ {
2042
+ "clip_ratio/high_max": 0.0,
2043
+ "clip_ratio/high_mean": 0.0,
2044
+ "clip_ratio/low_mean": 0.0,
2045
+ "clip_ratio/low_min": 0.0,
2046
+ "clip_ratio/region_mean": 0.0,
2047
+ "entropy": 4.271548748016357,
2048
+ "epoch": 4.666666666666667,
2049
+ "grad_norm": 0.030309390276670456,
2050
+ "kl": 0.14847837388515472,
2051
+ "learning_rate": 3.75e-07,
2052
+ "loss": 0.0003711959288921207,
2053
+ "step": 112,
2054
+ "step_time": 0.22734109300017735
2055
+ },
2056
+ {
2057
+ "clip_ratio/high_max": 0.0,
2058
+ "clip_ratio/high_mean": 0.0,
2059
+ "clip_ratio/low_mean": 0.0,
2060
+ "clip_ratio/low_min": 0.0,
2061
+ "clip_ratio/region_mean": 0.0,
2062
+ "completions/clipped_ratio": 1.0,
2063
+ "completions/max_length": 384.0,
2064
+ "completions/max_terminated_length": 0.0,
2065
+ "completions/mean_length": 384.0,
2066
+ "completions/mean_terminated_length": 0.0,
2067
+ "completions/min_length": 384.0,
2068
+ "completions/min_terminated_length": 0.0,
2069
+ "entropy": 4.5982232093811035,
2070
+ "epoch": 4.708333333333333,
2071
+ "frac_reward_zero_std": 1.0,
2072
+ "grad_norm": 0.029322689399123192,
2073
+ "kl": 0.1586071103811264,
2074
+ "learning_rate": 3.3333333333333335e-07,
2075
+ "loss": 0.0003965177165810019,
2076
+ "num_tokens": 736748.0,
2077
+ "reward": -0.05000000074505806,
2078
+ "reward_std": 0.0,
2079
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
2080
+ "rewards/GeneratorRewardFunction/std": 0.0,
2081
+ "step": 113,
2082
+ "step_time": 12.333232826999847
2083
+ },
2084
+ {
2085
+ "clip_ratio/high_max": 0.0,
2086
+ "clip_ratio/high_mean": 0.0,
2087
+ "clip_ratio/low_mean": 0.0,
2088
+ "clip_ratio/low_min": 0.0,
2089
+ "clip_ratio/region_mean": 0.0,
2090
+ "entropy": 4.43204927444458,
2091
+ "epoch": 4.75,
2092
+ "grad_norm": 0.0323464497923851,
2093
+ "kl": 0.1504228711128235,
2094
+ "learning_rate": 2.916666666666667e-07,
2095
+ "loss": 0.00037605719990096986,
2096
+ "step": 114,
2097
+ "step_time": 0.22894687000007252
2098
+ },
2099
+ {
2100
+ "clip_ratio/high_max": 0.0,
2101
+ "clip_ratio/high_mean": 0.0,
2102
+ "clip_ratio/low_mean": 0.0,
2103
+ "clip_ratio/low_min": 0.0,
2104
+ "clip_ratio/region_mean": 0.0,
2105
+ "entropy": 4.273895263671875,
2106
+ "epoch": 4.791666666666667,
2107
+ "grad_norm": 0.035788681358098984,
2108
+ "kl": 0.17889364063739777,
2109
+ "learning_rate": 2.5000000000000004e-07,
2110
+ "loss": 0.00044723405153490603,
2111
+ "step": 115,
2112
+ "step_time": 0.22825912600001175
2113
+ },
2114
+ {
2115
+ "clip_ratio/high_max": 0.0,
2116
+ "clip_ratio/high_mean": 0.0,
2117
+ "clip_ratio/low_mean": 0.0,
2118
+ "clip_ratio/low_min": 0.0,
2119
+ "clip_ratio/region_mean": 0.0,
2120
+ "entropy": 4.463144302368164,
2121
+ "epoch": 4.833333333333333,
2122
+ "grad_norm": 0.029977256432175636,
2123
+ "kl": 0.16924364864826202,
2124
+ "learning_rate": 2.0833333333333333e-07,
2125
+ "loss": 0.0004231091297697276,
2126
+ "step": 116,
2127
+ "step_time": 0.22959902599995985
2128
+ },
2129
+ {
2130
+ "clip_ratio/high_max": 0.0,
2131
+ "clip_ratio/high_mean": 0.0,
2132
+ "clip_ratio/low_mean": 0.0,
2133
+ "clip_ratio/low_min": 0.0,
2134
+ "clip_ratio/region_mean": 0.0,
2135
+ "completions/clipped_ratio": 1.0,
2136
+ "completions/max_length": 384.0,
2137
+ "completions/max_terminated_length": 0.0,
2138
+ "completions/mean_length": 384.0,
2139
+ "completions/mean_terminated_length": 0.0,
2140
+ "completions/min_length": 384.0,
2141
+ "completions/min_terminated_length": 0.0,
2142
+ "entropy": 4.496586322784424,
2143
+ "epoch": 4.875,
2144
+ "frac_reward_zero_std": 1.0,
2145
+ "grad_norm": 0.0810508206486702,
2146
+ "kl": 0.18324460089206696,
2147
+ "learning_rate": 1.6666666666666668e-07,
2148
+ "loss": 0.0004581115208566189,
2149
+ "num_tokens": 762100.0,
2150
+ "reward": -0.05000000074505806,
2151
+ "reward_std": 0.0,
2152
+ "rewards/GeneratorRewardFunction/mean": -0.05000000074505806,
2153
+ "rewards/GeneratorRewardFunction/std": 0.0,
2154
+ "step": 117,
2155
+ "step_time": 12.307060693000494
2156
+ },
2157
+ {
2158
+ "clip_ratio/high_max": 0.0,
2159
+ "clip_ratio/high_mean": 0.0,
2160
+ "clip_ratio/low_mean": 0.0,
2161
+ "clip_ratio/low_min": 0.0,
2162
+ "clip_ratio/region_mean": 0.0,
2163
+ "entropy": 4.538492202758789,
2164
+ "epoch": 4.916666666666667,
2165
+ "grad_norm": 0.02771252952516079,
2166
+ "kl": 0.1558750718832016,
2167
+ "learning_rate": 1.2500000000000002e-07,
2168
+ "loss": 0.0003896876296494156,
2169
+ "step": 118,
2170
+ "step_time": 0.22174717100006092
2171
+ },
2172
+ {
2173
+ "clip_ratio/high_max": 0.0,
2174
+ "clip_ratio/high_mean": 0.0,
2175
+ "clip_ratio/low_mean": 0.0,
2176
+ "clip_ratio/low_min": 0.0,
2177
+ "clip_ratio/region_mean": 0.0,
2178
+ "entropy": 4.4321112632751465,
2179
+ "epoch": 4.958333333333333,
2180
+ "grad_norm": 0.027858059853315353,
2181
+ "kl": 0.14329802989959717,
2182
+ "learning_rate": 8.333333333333334e-08,
2183
+ "loss": 0.00035824510268867016,
2184
+ "step": 119,
2185
+ "step_time": 0.2216543990007267
2186
+ },
2187
+ {
2188
+ "clip_ratio/high_max": 0.0,
2189
+ "clip_ratio/high_mean": 0.0,
2190
+ "clip_ratio/low_mean": 0.0,
2191
+ "clip_ratio/low_min": 0.0,
2192
+ "clip_ratio/region_mean": 0.0,
2193
+ "entropy": 4.336979389190674,
2194
+ "epoch": 5.0,
2195
+ "grad_norm": 0.029509786516427994,
2196
+ "kl": 0.167790487408638,
2197
+ "learning_rate": 4.166666666666667e-08,
2198
+ "loss": 0.00041947621502913535,
2199
+ "step": 120,
2200
+ "step_time": 0.2207649100000708
2201
+ }
2202
+ ],
2203
+ "logging_steps": 1,
2204
+ "max_steps": 120,
2205
+ "num_input_tokens_seen": 762100,
2206
+ "num_train_epochs": 5,
2207
+ "save_steps": 30,
2208
+ "stateful_callbacks": {
2209
+ "TrainerControl": {
2210
+ "args": {
2211
+ "should_epoch_stop": false,
2212
+ "should_evaluate": false,
2213
+ "should_log": false,
2214
+ "should_save": true,
2215
+ "should_training_stop": true
2216
+ },
2217
+ "attributes": {}
2218
+ }
2219
+ },
2220
+ "total_flos": 0.0,
2221
+ "train_batch_size": 4,
2222
+ "trial_name": null,
2223
+ "trial_params": null
2224
+ }
self_play_hf_l40s_full/round_005/generator_train/checkpoint-120/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1afbfa894ee2b20e77f8fc01bf94ad1916f3036851a51f84658b8d193a779a9b
3
+ size 7249