dispatcher / scripts

Commit History

  1. docs: make train_trl_grpo.py the primary training path⁠

    garg-tejas
    667c189
  2. fix: pin vllm to TRL-supported range in HF jobs⁠

    garg-tejas
    8c7b87a
  3. fix: random-step dataset + trajectory cache + experiment tracking for L40S⁠

    garg-tejas
    6da8b7c
  4. feat: rich step-by-step episode trace logging⁠

    garg-tejas
    aeebcdd
  5. feat: add L40S GPU support (48GB, best availability for TRL script)⁠

    garg-tejas
    3afbcc1
  6. feat: add TRL-native GRPO training script + updated launcher⁠

    garg-tejas
    837ed68
  7. fix: address 3 bugs introduced in previous commits⁠

    garg-tejas
    3d96a95
  8. perf: fix measure_completion_lengths.py sequential vLLM calls⁠

    garg-tejas
    283ff0d

  1. fix: address critical training-blocking bugs from audit⁠

    garg-tejas
    1827386
  2. perf: use vLLM for fast inference in completion length measurement⁠

    garg-tejas
    aebbf2e
  3. fix: address critical code review bugs (fatality, reward, dedup)⁠

    garg-tejas
    ee35947
  4. chore: remove legacy HF training launch scripts and ignore local env file⁠

    garg-tejas
    df5be79
  5. feat: add completion length measurement script⁠

    garg-tejas
    ff59d0d
  6. fix: smarter defaults for HF Jobs launcher to fit in 8h timeout⁠

    garg-tejas
    7b1297a
  7. feat: training tracker, HF Jobs launcher, inplace fix, visualization⁠

    garg-tejas
    27d2e57
  8. fix: add openenv-core dependency to HF Jobs launcher⁠

    garg-tejas
    c43d43f
  9. rebrand: Dead Air -> DispatchR across all code, docs, and metadata

    garg-tejas
    e687796
  10. fix: Dockerfile with user 1000, updated launch script

    garg-tejas
    05926e6
  11. chore: remove personal guide docs from repo, keep locally only

    garg-tejas
    3d18ad9
  12. feat: add HF Spaces deployment guide, Dockerfile, launcher scripts, and Hub push support

    garg-tejas
    02644f2