Issues

• SFT sequence length and training configuration

  microsoft/data-efficacy#21 · Filter by author Heinz217 opened on Aug 18, 2026

• How to process downstream data like MiniF2F?

  microsoft/data-efficacy#17 · Filter by author unlabeledData opened on Oct 10, 2025

• Will it support MLLM like Qwen2.5-VL series？

  microsoft/data-efficacy#16 · Filter by author 24-solar-terms opened on Sep 26, 2025
  1

• The HF model is missing

  microsoft/data-efficacy#15 · Filter by author ScorpionCG opened on Sep 19, 2025
  1

• Question about the training time

  microsoft/data-efficacy#14 · Filter by author Hugo-cell111 opened on Sep 14, 2025
  1

• Is it another megatron training architecture?

  microsoft/data-efficacy#13 · Filter by author liuzhao104 opened on Sep 9, 2025
  1