Commit graph

6 commits

Author SHA1 Message Date
Dakota
e13526d308 Fix API to accept messages without reward field + comprehensive tests
- Made reward field truly optional in messages (no auto-addition)
- Accept custom roles (dog, cat, etc.) beyond standard ones
- Added 24 new tests for edge cases (tuples, unicode, large content)
- Reorganized test structure: moved from testing/ to atroposlib/tests/
- Fixed legacy API tests and removed tests requiring missing data files

All 43 tests pass\! Fixes message handling for SFT use cases.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-06-09 14:03:08 -05:00
teknium1
f999f90627 add support for composite task 2025-06-08 04:39:50 -07:00
teknium1
398e3ddeaa add randomization for complexity as well as curriculum support 2025-06-08 03:07:07 -07:00
teknium1
a4b22c38d7 make eval vars config options 2025-06-06 15:24:00 -07:00
teknium1
be94857084 add seed to default configs for clarity 2025-06-06 14:56:55 -07:00
teknium1
79188d8d6a Add reasoning gym env 2025-06-05 17:30:25 -07:00