-
Notifications
You must be signed in to change notification settings - Fork 303
Pull requests: unslothai/unsloth-zoo
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Fix GRPO LoRA aliasing corruption: clone vLLM hot-load tensors, snapshot merge inputs
#951
opened Jul 26, 2026 by
pwasiewi
Loading…
Do not treat an unreachable Hub as a missing model in 16bit merges
#950
opened Jul 26, 2026 by
danielhanchen
Member
Loading…
Fix compiler codegen for a forward hidden behind a bare decorator
#949
opened Jul 26, 2026 by
danielhanchen
Member
Loading…
Tests: unblock MLX CI on Mac by not requiring the fused MRoPE grad to raise
mlx
#948
opened Jul 26, 2026 by
danielhanchen
Member
Loading…
Fix memory underestimation in fused cross entropy chunking via runtime calliberation
#947
opened Jul 25, 2026 by
zrva
Loading…
docs(rocm): add AMD ROCm optimization guide with benchmark data
#945
opened Jul 25, 2026 by
zhihuidu-amd
Contributor
Loading…
feat(rocm): add AMD SDPA recommendation and VRAM utilization helper
#944
opened Jul 25, 2026 by
zhihuidu-amd
Contributor
Loading…
test(rocm): add AMD ROCm compatibility test suite
#943
opened Jul 25, 2026 by
zhihuidu-amd
Contributor
Loading…
Fix vLLM direct LoRA hot-load: up_proj never loaded + aliased-buffer scaling corruption
#942
opened Jul 24, 2026 by
pwasiewi
Loading…
MLX: bound compiled VLM training signatures with a finite batch plan and shape guard
#940
opened Jul 24, 2026 by
Lyxot
Contributor
Loading…
fix(mlx): freeze base when reloading a LoRA adapter for continued training
#939
opened Jul 24, 2026 by
Lyxot
Contributor
Loading…
feat(mlx): CUDA-parity continued pretraining (embed_tokens / lm_head)
#932
opened Jul 22, 2026 by
Lyxot
Contributor
Loading…
Bump the python group across 1 directory with 6 updates
dependencies
Pull requests that update a dependency file
python
Pull requests that update python code
#929
opened Jul 22, 2026 by
dependabot
Bot
Loading…
Fix: Let UnslothVisionDataCollator accept audio-only processors
#924
opened Jul 20, 2026 by
BardiaKoopah
Contributor
Loading…
Fix silent embedding corruption in add_new_tokens on padded-vocab mod
#923
opened Jul 20, 2026 by
BardiaKoopah
Contributor
Loading…
feat: AMD aiter Flash Attention via source-level SDPA rewrite in compiler.py
#920
opened Jul 20, 2026 by
zhihuidu-amd
Contributor
Loading…
Fix OLMoE load_in_4bit crash: route fused experts through the MoE backend
#915
opened Jul 18, 2026 by
pjordanandrsn
Loading…
Fix doubled lm_head gradient in Unsloth_Offloaded_Log_Softmax backward
#892
opened Jul 9, 2026 by
danielhanchen
Member
Loading…
Fix GRPO off-policy mask crash on TRL >= 0.27.1
#891
opened Jul 9, 2026 by
danielhanchen
Member
Loading…
Guard silently-ignored GRPO config options and remove dead batch chunking
#890
opened Jul 9, 2026 by
danielhanchen
Member
Loading…
Fix down-LoRA row order and bf16 probe in grouped-GEMM MoE forward
#887
opened Jul 8, 2026 by
roycho96
Loading…
Previous Next
ProTip!
Follow long discussions with comments:>50.