All pull requests

Pull requests

• Fast-LLM integration: test suite

  #156·jlamypoirier opened Jul 17, 2026·

  ·

• Fast-LLM RL: per-token model-version tagging, metrics, and GSPO configs

  #155·jlamypoirier opened Jul 16, 2026·

  ·

• Add Fast-LLM as an alternative trainer (core integration)

  #153·jlamypoirier opened Jul 15, 2026·

  ·

• Trainer-agnostic fixes extracted from the fast-llm branch

  #151·jlamypoirier opened Jul 14, 2026·

  ·

• Retry completion responses missing 'choices' instead of crashing the actor

  #148·jlamypoirier opened Jul 2, 2026·

  ·

• Situational Rewards and New Domain: Mosaic Leaks Privacy Agent

  #146·Alex-Gurung opened Jun 8, 2026Collaborator·

  ·

• Multi-step RL with CUBE

  #143·Masseeh opened May 12, 2026Collaborator·

  ·
  1

• Fast-LLM trainer integration with vLLM v1 weight broadcast

  #140·bigximik opened May 6, 2026·6 tasks·

  ·

• New domain: SWE

  #138·ehsk opened Apr 22, 2026·

  ·
  10

• RL on TerminalBench using CUBE

  #134·ehsk opened Apr 15, 2026·

  ·
  

• Self-Evolving Curriculum implementation

  #127·ehsk opened Feb 6, 2026·

  ·

• Skip corrupted rows from FileStreamReader

  #115·lewtun opened Dec 8, 2025·

  ·
  1

  3

• Enhance multi-modal support

  #111·ehsk opened Dec 3, 2025·

  ·

• Better exception handling in preprocessor

  #93·ehsk opened Nov 10, 2025·

  ·
  1

• Adaptive curriculum

  #87·rafapi opened Oct 17, 2025Contributor·

  ·

• Ray actor loop with local envs

  #81·ollmer opened Oct 6, 2025Contributor·

  ·
  18

• RL agent with MCP

  #79·AlexPiche opened Aug 15, 2025Contributor·

  ·

• sim

  #74·AlexPiche opened Jul 29, 2025Contributor·

  ·