EtashGuha commited on
Commit
5aabbe2
·
verified ·
1 Parent(s): 4776099

Upload configuration file

Browse files
config/b2_train_fasttext_code_pos_taco_neg_sqls.yaml ADDED
@@ -0,0 +1,147 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ operators:
2
+ - id: load_hf
3
+ config:
4
+ type: hf_source
5
+ dataset: BAAI/TACO
6
+ split: train
7
+ trust_remote_code: True
8
+ - id: rename_task_column
9
+ config:
10
+ type: function
11
+ function: data_strategies.commons.force_rename_columns
12
+ function_config:
13
+ column_maps:
14
+ question: instruction_seed
15
+ input_ids:
16
+ - load_hf
17
+
18
+ - id: decontaminate_positive
19
+ config:
20
+ type: cpu_function
21
+ sharded: false
22
+ num_cpus: 32
23
+ function: data_strategies._A1_Source.utils.decontaminate_fuzzy_w_ngram
24
+ function_config:
25
+ column: instruction_seed
26
+ eval_datasets:
27
+ - HuggingFaceH4/MATH-500
28
+ - Maxwell-Jia/AIME_2024
29
+ - AI-MO/aimo-validation-amc
30
+ - livecodebench/code_generation_lite
31
+ - mlfoundations-dev/AIME2025_combined
32
+ - cais/hle
33
+ - open-r1/codeforces
34
+ - Idavidrein/gpqa
35
+ - daman1209arora/jeebench
36
+ - mlfoundations-dev/mmlu_pro_eval_full
37
+ - Qwen/CodeElo
38
+ - open-r1/ioi
39
+ eval_columns:
40
+ - problem
41
+ - Problem
42
+ - problem
43
+ - question_content
44
+ - question
45
+ - question
46
+ - description
47
+ - Question
48
+ - question
49
+ - prompt
50
+ - description
51
+ - statement
52
+ eval_splits:
53
+ - test
54
+ - train
55
+ - train
56
+ - test
57
+ - train
58
+ - test
59
+ - test
60
+ - train
61
+ - test
62
+ - test
63
+ - test
64
+ - test
65
+ eval_subsets:
66
+ Idavidrein/gpqa: gpqa_diamond
67
+ similarity_threshold: 75.0
68
+ ngram_size: 13
69
+ input_ids:
70
+ - rename_task_column
71
+
72
+ - id: hf_source_negative_seed
73
+ config:
74
+ type: load_preexisting
75
+ framework_name: a1_code_sql_create_context
76
+
77
+ - id: decontaminate_negative
78
+ config:
79
+ type: cpu_function
80
+ sharded: false
81
+ num_cpus: 32
82
+ function: data_strategies._A1_Source.utils.decontaminate_fuzzy_w_ngram
83
+ function_config:
84
+ column: instruction_seed
85
+ eval_datasets:
86
+ - HuggingFaceH4/MATH-500
87
+ - Maxwell-Jia/AIME_2024
88
+ - AI-MO/aimo-validation-amc
89
+ - livecodebench/code_generation_lite
90
+ - mlfoundations-dev/AIME2025_combined
91
+ - cais/hle
92
+ - open-r1/codeforces
93
+ - Idavidrein/gpqa
94
+ - daman1209arora/jeebench
95
+ - mlfoundations-dev/mmlu_pro_eval_full
96
+ - Qwen/CodeElo
97
+ - open-r1/ioi
98
+ eval_columns:
99
+ - problem
100
+ - Problem
101
+ - problem
102
+ - question_content
103
+ - question
104
+ - question
105
+ - description
106
+ - Question
107
+ - question
108
+ - prompt
109
+ - description
110
+ - statement
111
+ eval_splits:
112
+ - test
113
+ - train
114
+ - train
115
+ - test
116
+ - train
117
+ - test
118
+ - test
119
+ - train
120
+ - test
121
+ - test
122
+ - test
123
+ - test
124
+ eval_subsets:
125
+ Idavidrein/gpqa: gpqa_diamond
126
+ similarity_threshold: 75.0
127
+ ngram_size: 13
128
+ input_ids:
129
+ - hf_source_negative_seed
130
+
131
+ - id: train_fasttext_classifier
132
+ config:
133
+ type: train_fasttext
134
+ text_column: instruction_seed
135
+ positive_input_ids:
136
+ - decontaminate_positive
137
+ negative_input_ids:
138
+ - decontaminate_negative
139
+ dim: 256
140
+ epoch: 3
141
+ lr: 0.1
142
+ word_ngrams: 2
143
+ min_count: 3
144
+ hf_repo_id: mlfoundations-dev/instruction_filtering_fast_text_classifier_b2_taco_pos_neg_sql
145
+ input_ids:
146
+ - decontaminate_positive
147
+ - decontaminate_negative