Upload configuration file
Browse files
config/b2_train_fasttext_code.yaml
ADDED
|
@@ -0,0 +1,156 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
operators:
|
| 2 |
+
- id: hf_source_positive_seed
|
| 3 |
+
config:
|
| 4 |
+
type: hf_source
|
| 5 |
+
dataset: open-r1/codeforces
|
| 6 |
+
split: train
|
| 7 |
+
|
| 8 |
+
- id: rename_task_column_positive_seed
|
| 9 |
+
config:
|
| 10 |
+
type: function
|
| 11 |
+
function: data_strategies.commons.force_rename_columns
|
| 12 |
+
function_config:
|
| 13 |
+
column_maps:
|
| 14 |
+
description: instruction_seed
|
| 15 |
+
input_ids:
|
| 16 |
+
- hf_source_positive_seed
|
| 17 |
+
|
| 18 |
+
- id: remove_nones_positive
|
| 19 |
+
config:
|
| 20 |
+
type: function
|
| 21 |
+
function: data_strategies.commons.remove_nones
|
| 22 |
+
function_config:
|
| 23 |
+
column: instruction_seed
|
| 24 |
+
input_ids:
|
| 25 |
+
- rename_task_column_positive_seed
|
| 26 |
+
|
| 27 |
+
- id: decontaminate_positive
|
| 28 |
+
config:
|
| 29 |
+
type: cpu_function
|
| 30 |
+
sharded: false
|
| 31 |
+
num_cpus: 32
|
| 32 |
+
function: data_strategies._A1_Source.utils.decontaminate_fuzzy_w_ngram
|
| 33 |
+
function_config:
|
| 34 |
+
column: instruction_seed
|
| 35 |
+
eval_datasets:
|
| 36 |
+
- HuggingFaceH4/MATH-500
|
| 37 |
+
- Maxwell-Jia/AIME_2024
|
| 38 |
+
- AI-MO/aimo-validation-amc
|
| 39 |
+
- livecodebench/code_generation_lite
|
| 40 |
+
- mlfoundations-dev/AIME2025_combined
|
| 41 |
+
- cais/hle
|
| 42 |
+
- open-r1/codeforces
|
| 43 |
+
- Idavidrein/gpqa
|
| 44 |
+
- daman1209arora/jeebench
|
| 45 |
+
- mlfoundations-dev/mmlu_pro_eval_full
|
| 46 |
+
- Qwen/CodeElo
|
| 47 |
+
- open-r1/ioi
|
| 48 |
+
eval_columns:
|
| 49 |
+
- problem
|
| 50 |
+
- Problem
|
| 51 |
+
- problem
|
| 52 |
+
- question_content
|
| 53 |
+
- question
|
| 54 |
+
- question
|
| 55 |
+
- description
|
| 56 |
+
- Question
|
| 57 |
+
- question
|
| 58 |
+
- prompt
|
| 59 |
+
- description
|
| 60 |
+
- statement
|
| 61 |
+
eval_splits:
|
| 62 |
+
- test
|
| 63 |
+
- train
|
| 64 |
+
- train
|
| 65 |
+
- test
|
| 66 |
+
- train
|
| 67 |
+
- test
|
| 68 |
+
- test
|
| 69 |
+
- train
|
| 70 |
+
- test
|
| 71 |
+
- test
|
| 72 |
+
- test
|
| 73 |
+
- test
|
| 74 |
+
eval_subsets:
|
| 75 |
+
Idavidrein/gpqa: gpqa_diamond
|
| 76 |
+
similarity_threshold: 75.0
|
| 77 |
+
ngram_size: 13
|
| 78 |
+
input_ids:
|
| 79 |
+
- remove_nones_positive
|
| 80 |
+
|
| 81 |
+
- id: hf_source_negative_seed
|
| 82 |
+
config:
|
| 83 |
+
type: load_preexisting
|
| 84 |
+
framework_name: a1_code_sql_create_context
|
| 85 |
+
|
| 86 |
+
- id: decontaminate_negative
|
| 87 |
+
config:
|
| 88 |
+
type: cpu_function
|
| 89 |
+
sharded: false
|
| 90 |
+
num_cpus: 32
|
| 91 |
+
function: data_strategies._A1_Source.utils.decontaminate_fuzzy_w_ngram
|
| 92 |
+
function_config:
|
| 93 |
+
column: instruction_seed
|
| 94 |
+
eval_datasets:
|
| 95 |
+
- HuggingFaceH4/MATH-500
|
| 96 |
+
- Maxwell-Jia/AIME_2024
|
| 97 |
+
- AI-MO/aimo-validation-amc
|
| 98 |
+
- livecodebench/code_generation_lite
|
| 99 |
+
- mlfoundations-dev/AIME2025_combined
|
| 100 |
+
- cais/hle
|
| 101 |
+
- open-r1/codeforces
|
| 102 |
+
- Idavidrein/gpqa
|
| 103 |
+
- daman1209arora/jeebench
|
| 104 |
+
- mlfoundations-dev/mmlu_pro_eval_full
|
| 105 |
+
- Qwen/CodeElo
|
| 106 |
+
- open-r1/ioi
|
| 107 |
+
eval_columns:
|
| 108 |
+
- problem
|
| 109 |
+
- Problem
|
| 110 |
+
- problem
|
| 111 |
+
- question_content
|
| 112 |
+
- question
|
| 113 |
+
- question
|
| 114 |
+
- description
|
| 115 |
+
- Question
|
| 116 |
+
- question
|
| 117 |
+
- prompt
|
| 118 |
+
- description
|
| 119 |
+
- statement
|
| 120 |
+
eval_splits:
|
| 121 |
+
- test
|
| 122 |
+
- train
|
| 123 |
+
- train
|
| 124 |
+
- test
|
| 125 |
+
- train
|
| 126 |
+
- test
|
| 127 |
+
- test
|
| 128 |
+
- train
|
| 129 |
+
- test
|
| 130 |
+
- test
|
| 131 |
+
- test
|
| 132 |
+
- test
|
| 133 |
+
eval_subsets:
|
| 134 |
+
Idavidrein/gpqa: gpqa_diamond
|
| 135 |
+
similarity_threshold: 75.0
|
| 136 |
+
ngram_size: 13
|
| 137 |
+
input_ids:
|
| 138 |
+
- hf_source_negative_seed
|
| 139 |
+
|
| 140 |
+
- id: train_fasttext_classifier
|
| 141 |
+
config:
|
| 142 |
+
type: train_fasttext
|
| 143 |
+
text_column: instruction_seed
|
| 144 |
+
positive_input_ids:
|
| 145 |
+
- decontaminate_positive
|
| 146 |
+
negative_input_ids:
|
| 147 |
+
- decontaminate_negative
|
| 148 |
+
dim: 256
|
| 149 |
+
epoch: 3
|
| 150 |
+
lr: 0.1
|
| 151 |
+
word_ngrams: 2
|
| 152 |
+
min_count: 3
|
| 153 |
+
hf_repo_id: mlfoundations-dev/instruction_filtering_fast_text_classifier_b2_code_pos_codeforces_neg_sql
|
| 154 |
+
input_ids:
|
| 155 |
+
- decontaminate_positive
|
| 156 |
+
- decontaminate_negative
|