Owner

THUDM

2 indexed repositories · View on GitHub

slime
slime is an LLM post-training framework for RL Scaling.

Python · 8508 stars
INFTY
INFTY Engine: An Optimization Toolkit to Support Continual AI

Python · 576 stars

Beginner friendly Everything

Open All

LookSAM never builds gv: state key mismatch (old_grad vs old_g) and float alpha.to crash Open

Difficulty 2/5 1-3 hours Newbie friendliness 74/100

THUDM/INFTY#18 · 1 comment · Opened Sep 15, 2026
[Bug] shipped dynamic-sampling filters crash on fan-out groups (list[list[Sample]]) from multi-turn agent rollouts Open

Difficulty 2/5 1-3 hours Newbie friendliness 84/100

THUDM/slime#2374 · Opened Sep 9, 2026
[Bug] W&B initialization fails with wandb 0.28.2 because wandb.util.generate_id was removed Open

bug

Difficulty 1/5 1-3 hours Newbie friendliness 90/100

THUDM/slime#2332 · 1 comment · Opened Aug 26, 2026
Hard-coded batch-triplet unpack breaks custom DataLoader formats Open

Difficulty 2/5 1-3 hours Newbie friendliness 74/100

THUDM/INFTY#10 · 1 comment · Opened Jun 13, 2026
[Bug] solution_str = sample.prompt + sample.response Open

bug

Difficulty 2/5 1-3 hours Newbie friendliness 68/100

THUDM/slime#1829 · 1 comment · Opened Apr 13, 2026
[Feature Request] Add minimal 1-GPU script for debugging and learning Open

Difficulty 2/5 1-3 hours Newbie friendliness 68/100

THUDM/slime#1628 · 1 reaction · Opened Feb 25, 2026
[BUG] LR scheduler double-counted when resuming from checkpoint Open

Difficulty 1/5 Under an hour Newbie friendliness 72/100

THUDM/slime#1546 · 1 reaction · Opened Feb 5, 2026
grade_answer_verl fails when rm_type="boxed_math" strips the \boxed{} wrapper upstream Open

Difficulty 2/5 1-3 hours Newbie friendliness 78/100

THUDM/slime#543 · 3 comments · Opened Oct 20, 2025
UX/Unintended bug: sample.reward is None on aborted sample Open

Difficulty 2/5 1-3 hours Newbie friendliness 78/100

THUDM/slime#200 · 5 comments · Opened Aug 18, 2025

THUDM

Get new issues in your inbox