Skip to content

Add a GSM8K SFT dataset loader #71

Description

@adohe

Scope

  • Add examples/math/sft_dataset_loader.py that converts raw GSM8K into an SFT-ready prompt+response schema (distinct from the RL-shaped examples/math/dataset_loader.py).
  • Wire it as the --dataset-loader-fn example used by the SFT dataset tutorial.

Acceptance

  • Output rows are valid SFT input (prompt+response).
  • Does not alter the existing RL loader.
  • Include a small CPU test asserting the produced schema.

Metadata

Metadata

Assignees

No one assigned

    Labels

    area/algorithmsIssues or PRs related to training algorithms (SFT, DPO, GSPO, GRPO, PPO)area/dxIssues or PRs related to developer experience (error messages, ergonomics, onboarding)kind/featureCategorizes issue or PR as related to a new feature

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions