EtashGuha commited on
Commit
916e9ad
·
verified ·
1 Parent(s): f948c3c

Upload configuration file

Browse files
config/camel_physics_seed_science.yaml ADDED
@@ -0,0 +1,92 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ operators:
2
+ - id: load_discipline
3
+ config:
4
+ type: function
5
+ function: data_strategies.commons.load_json_string_to_dataset
6
+ function_config:
7
+ json_string: '{"discipline": ["physics"], "expert": ["Physicist"], "num_topics": [25], "num_subtopics": [25], "num_questions": [80]}'
8
+
9
+ - id: generate_topics
10
+ config:
11
+ type: completions
12
+ map: list # this uses structured output if available
13
+ map_config:
14
+ system_message: "You are a helpful assistant."
15
+ user_message: "Please list {{num_topics}} diverse {{discipline}} topics. Make sure the topics are {{discipline}} topics. No explanation." # number of topics
16
+ output_column: topic
17
+ model: gpt-4o-mini
18
+ batch: False
19
+ input_ids:
20
+ - load_discipline
21
+
22
+ - id: generate_subtopics
23
+ config:
24
+ type: completions
25
+ map: list # this uses structured output if available
26
+ map_config:
27
+ system_message: "You are a helpful assistant."
28
+ user_message: "List {{num_subtopics}} different {{discipline}} {{topic}} problem topics. Be precise and make sure the problems are {{topic}} problems." # number of subtopics per topic
29
+ output_column: subtopic
30
+ model: gpt-4o-mini
31
+ batch: False
32
+ input_ids:
33
+ - generate_topics
34
+
35
+ - id: generate_questions
36
+ config:
37
+ type: completions
38
+ map: list
39
+ map_config:
40
+ system_message: "You are a {{expert}}."
41
+ user_message: "From this {{discipline}} subject {{topic}} and this subtopic {{subtopic}} we need to write {{num_questions}} new questions for a {{discipline}} student to solve. Please write a precise problem for the student to solve."
42
+ output_column: question
43
+ model: gpt-4o-mini
44
+ batch: False
45
+ input_ids:
46
+ - generate_subtopics
47
+
48
+ - id: rename_task_column
49
+ config:
50
+ type: function
51
+ function: data_strategies.commons.force_rename_columns
52
+ function_config:
53
+ column_maps:
54
+ question: problem
55
+ input_ids:
56
+ - generate_questions
57
+ - id: sample_dataset
58
+ config:
59
+ type: function
60
+ function: data_strategies.commons.uniform_sample_fixed
61
+ function_config:
62
+ num_samples: 5_000
63
+ input_ids:
64
+ - rename_task_column
65
+ # - id: annotate_r1_distill_70b
66
+ # config:
67
+ # type: completions
68
+ # map: deepseek_reasoner
69
+ # map_config:
70
+ # input_problem_column: problem
71
+ # model: deepseek-reasoner
72
+ # batch: False
73
+ # temperature: 1.0
74
+ # backend: openai_client
75
+ # backend_params:
76
+ # max_requests_per_minute: 2_500
77
+ # max_tokens_per_minute: 1_000_000_000
78
+ # api_key: sk-66c9461d203144cc8bda1d0e71e08799
79
+ # base_url: "https://api.deepseek.com/"
80
+ # invalid_finish_reasons: ['content_filter']
81
+ # input_ids:
82
+ # - sample_dataset
83
+ # - id: convert_to_sharegpt
84
+ # config:
85
+ # type: function
86
+ # function: data_strategies.commons.convert_instruction_response_to_sharegpt
87
+ # function_config:
88
+ # input_instruction_column: problem
89
+ # input_response_column: reasoning
90
+ # output_sharegpt_column: conversations
91
+ # input_ids:
92
+ # - annotate_r1_distill_70b