Hugging Face's logo

Datasets:
mlfoundations-dev
/
a1_science_fineweb

Modalities:
Tabular
Text
Formats:
parquet
Size:
10K - 100K
Libraries:
Datasets
Dask
Dataset card Data Studio Files
xet
 Community
ryanmarten commited on
Commit
d88da42
·
verified ·
1 Parent(s): 0e38318

Upload configuration file

Browse files
Files changed (1) hide show
  1. config/a1_science_fineweb.yaml +82 -48
config/a1_science_fineweb.yaml CHANGED
@@ -9,7 +9,7 @@ operators:
9
  type: function
10
  function: data_strategies.commons.uniform_sample_fixed
11
  function_config:
12
- num_samples: 50_000
13
  input_ids:
14
  - load_hf
15
  - id: generate_instructions
@@ -29,61 +29,95 @@ operators:
29
  output_column: instruction_seed
30
  model: gpt-4o-mini
31
  temperature: 1.0
32
- batch: True
33
  input_ids:
34
  - sample_dataset
35
- - id: decontaminate
36
  config:
37
- type: cpu_function
38
- sharded: false
39
- num_cpus: 32
40
- function: data_strategies._A1_Source.utils.decontaminate_fuzzy_w_ngram
41
  function_config:
42
- column: instruction_seed
43
- eval_datasets:
44
- - HuggingFaceH4/MATH-500
45
- - Maxwell-Jia/AIME_2024
46
- - AI-MO/aimo-validation-amc
47
- - livecodebench/code_generation_lite
48
- - mlfoundations-dev/AIME2025_combined
49
- - mlfoundations-dev/cais-hle-multipleChoice-noImage
50
- - open-r1/codeforces
51
- - Idavidrein/gpqa
52
- - daman1209arora/jeebench
53
- eval_columns:
54
- - problem
55
- - Problem
56
- - problem
57
- - question_content
58
- - question
59
- - question
60
- - description
61
- - Question
62
- - question
63
- eval_splits:
64
- - test
65
- - train
66
- - train
67
- - test
68
- - train
69
- - test
70
- - test
71
- - train
72
- - test
73
- eval_subsets:
74
- Idavidrein/gpqa: gpqa_diamond
75
- similarity_threshold: 80.0
76
- ngram_size: 13
77
  input_ids:
78
  - generate_instructions
79
- - id: sample_dataset_final
80
  config:
81
- type: function
82
- function: data_strategies.commons.uniform_sample_fixed
83
- function_config:
84
- num_samples: 31_600
 
 
 
 
 
 
 
 
 
 
85
  input_ids:
86
- - decontaminate
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
87
 
88
  # - id: annotate_r1
89
  # config:
 
9
  type: function
10
  function: data_strategies.commons.uniform_sample_fixed
11
  function_config:
12
+ num_samples: 40_000
13
  input_ids:
14
  - load_hf
15
  - id: generate_instructions
 
29
  output_column: instruction_seed
30
  model: gpt-4o-mini
31
  temperature: 1.0
 
32
  input_ids:
33
  - sample_dataset
34
+ - id: sample_dataset_pre
35
  config:
36
+ type: function
37
+ function: data_strategies.commons.uniform_sample_fixed
 
 
38
  function_config:
39
+ num_samples: 39_999
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
40
  input_ids:
41
  - generate_instructions
42
+ - id: annotate_r1_distill_70b
43
  config:
44
+ type: completions
45
+ map: deepseek_reasoner
46
+ map_config:
47
+ input_problem_column: instruction_seed
48
+ model: deepseek-reasoner
49
+ batch: false
50
+ temperature: 1.0
51
+ require_all_responses: false
52
+ backend: openai
53
+ backend_params:
54
+ max_requests_per_minute: 500
55
+ max_tokens_per_minute: 1_000_000_000
56
+ base_url: "https://api.deepseek.com/"
57
+ invalid_finish_reasons: ['content_filter']
58
  input_ids:
59
+ - sample_dataset_pre
60
+ # - id: decontaminate
61
+ # config:
62
+ # type: cpu_function
63
+ # sharded: false
64
+ # num_cpus: 32
65
+ # function: data_strategies._A1_Source.utils.decontaminate_fuzzy_w_ngram
66
+ # function_config:
67
+ # column: instruction_seed
68
+ # eval_datasets:
69
+ # - HuggingFaceH4/MATH-500
70
+ # - Maxwell-Jia/AIME_2024
71
+ # - AI-MO/aimo-validation-amc
72
+ # - livecodebench/code_generation_lite
73
+ # - mlfoundations-dev/AIME2025_combined
74
+ # - cais/hle
75
+ # - open-r1/codeforces
76
+ # - Idavidrein/gpqa
77
+ # - daman1209arora/jeebench
78
+ # - mlfoundations-dev/mmlu_pro_eval_full
79
+ # - Qwen/CodeElo
80
+ # - open-r1/ioi
81
+ # eval_columns:
82
+ # - problem
83
+ # - Problem
84
+ # - problem
85
+ # - question_content
86
+ # - question
87
+ # - question
88
+ # - description
89
+ # - Question
90
+ # - question
91
+ # - prompt
92
+ # - description
93
+ # - statement
94
+ # eval_splits:
95
+ # - test
96
+ # - train
97
+ # - train
98
+ # - test
99
+ # - train
100
+ # - test
101
+ # - test
102
+ # - train
103
+ # - test
104
+ # - test
105
+ # - test
106
+ # - test
107
+ # eval_subsets:
108
+ # Idavidrein/gpqa: gpqa_diamond
109
+ # similarity_threshold: 80.0
110
+ # ngram_size: 13
111
+ # input_ids:
112
+ # - generate_instructions
113
+ # - id: sample_dataset_final
114
+ # config:
115
+ # type: function
116
+ # function: data_strategies.commons.uniform_sample_fixed
117
+ # function_config:
118
+ # num_samples: 31_600
119
+ # input_ids:
120
+ # - decontaminate
121
 
122
  # - id: annotate_r1
123
  # config: