| operators: | |
| - id: load_hf_python | |
| config: | |
| type: hf_source | |
| dataset: bigcode/self-oss-instruct-sc2-exec-filter-50k | |
| split: train | |
| trust_remote_code: True | |
| - id: drop_columns | |
| config: | |
| type: function | |
| function: data_strategies.commons.keep_only_columns | |
| function_config: | |
| columns_to_keep: | |
| - instruction | |
| - response | |
| input_ids: | |
| - load_hf_python | |
| - id: rename_task_column | |
| config: | |
| type: function | |
| function: data_strategies.commons.force_rename_columns | |
| function_config: | |
| column_maps: | |
| response: response_seed | |
| instruction: instruction_seed | |
| input_ids: | |
| - drop_columns | |
| - id: sample_dataset | |
| config: | |
| type: function | |
| function: data_strategies.commons.uniform_sample_limited | |
| function_config: | |
| num_samples: 50000 # 1.25% of 10000 target from stage 1 | |
| input_ids: | |
| - rename_task_column | |
| - id: annotate | |
| config: | |
| type: completions | |
| map: chat | |
| map_config: | |
| system_message: "You are a helpful assistant." | |
| user_message_column: instruction_seed | |
| output_column: response | |
| model: gpt-4o-mini | |
| temperature: 1.0 | |
| batch: True | |
| input_ids: | |
| - sample_dataset |