processing-asqa-cot-train.yaml

name: asqa_cot_raft_train
cache: false
output_path: .
steps:
    - _target_: ragfit.processing.dataset_loaders.loaders.LocalLoader
      inputs: train
      filename: asqa-train.jsonl

    - _target_: ragfit.processing.local_steps.raft.RAFTStep
      inputs: train
      k: 5
      raft_p: 0.5
      neg_docs_num: 2
      output_key: raft_docs

    - _target_: ragfit.processing.local_steps.context.DocumentsJoiner
      inputs: train
      docs_key: raft_docs
      k:

    - _target_: ragfit.processing.local_steps.prompter.TextPrompter
      inputs: train
      prompt_file: ragfit/processing/prompts/cot.txt
      output_key: prompt
      mapping:
            question: query
            context: raft_docs

    - _target_: ragfit.processing.local_steps.api.openai.OpenAIChat
      inputs: train
      prompt_key: prompt
      answer_key: generated_answer
      instruction: ragfit/processing/prompts/prompt_instructions/qa.txt
      model:
            azure_endpoint: azure.endpoint.com
            api_version: 2024-05-01-preview
            model: GPT-4-32k-Bot

    - _target_: ragfit.processing.global_steps.output.OutputData
      inputs: train
      prefix: asqa-raft-cot