-
-
Notifications
You must be signed in to change notification settings - Fork 5.4k
Pull requests: unslothai/unsloth
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Expose flex inference engines behind UNSLOTH_FAST_INFERENCE=1
#5123
opened Apr 21, 2026 by
danielhanchen
Contributor
Loading…
3 of 5 tasks
fix(cpu): support CPU-only and no-torch import on Windows
#5119
opened Apr 21, 2026 by
LeoBorcherding
Contributor
•
Draft
update gema4 chat templates
auto-approved
Auto-review passed, ready to merge
auto-has-duplicate
#5116
opened Apr 20, 2026 by
Datta0
Collaborator
Loading…
Studio: use (gguf) context length before max seq length
#5111
opened Apr 19, 2026 by
G07cha
Contributor
Loading…
Add Qwen3-4B GRPO rollout engine benchmarks (vLLM vs transformers CB)
#5108
opened Apr 19, 2026 by
danielhanchen
Contributor
Loading…
3 of 5 tasks
Add persistent wiki-RAG pipeline with vendored Graphify, lint-driven enrichment, and debug/maintenance APIs
#5103
opened Apr 18, 2026 by
zohairshafi
Loading…
Route Gemma-4 float16 through UNSLOTH_FORCE_FLOAT32
#5092
opened Apr 17, 2026 by
danielhanchen
Contributor
Loading…
fix: patch CONTROL type for special tokens in sentencepiece GGUF export
#5080
opened Apr 17, 2026 by
octo-patch
Loading…
Fix/adapter config absolute path issue 3726
#5079
opened Apr 17, 2026 by
siddhudonda
Contributor
Loading…
Add FORCE_BFLOAT16 for Gemma-4 fp16 NaN fix
#5074
opened Apr 16, 2026 by
danielhanchen
Contributor
Loading…
1 of 3 tasks
Studio: make stop button actually stop generation
auto-approved
Auto-review passed, ready to merge
#5069
opened Apr 16, 2026 by
danielhanchen
Contributor
Loading…
2 of 6 tasks
fix(studio): enforce unstructured upload quota from server state
#5062
opened Apr 16, 2026 by
shaun0927
Loading…
BUG: fix multi-GPU inference crash for bnb 4-bit/8-bit models
auto-approved
Auto-review passed, ready to merge
auto-stale-rebase
#5053
opened Apr 16, 2026 by
danielhanchen
Contributor
Loading…
2 of 4 tasks
Studio: soften tool-use nudge for small models + synthesise directive
#5052
opened Apr 16, 2026 by
danielhanchen
Contributor
Loading…
5 tasks done
fix(studio): force stream=false on local provider for recipe jobs
#5042
opened Apr 15, 2026 by
wasimysaid
Collaborator
•
Draft
studio: configurable HuggingFace endpoint via HF_ENDPOINT env var
#4989
opened Apr 12, 2026 by
AdamPlatin123
Contributor
•
Draft
4 tasks done
fix: private dataset splits/metadata not loading in Studio UI
auto-review-failed
Auto-review found issues
#4965
opened Apr 10, 2026 by
Shivamjohri247
Loading…
Add endpoint to fetch dataset metadata with token
auto-approved
Auto-review passed, ready to merge
#4964
opened Apr 10, 2026 by
HadiSDev
Loading…
Install ROCm PyTorch on Windows AMD via repo.radeon.com
auto-approved
Auto-review passed, ready to merge
#4958
opened Apr 10, 2026 by
danielhanchen
Contributor
Loading…
8 tasks
Fix intelgputorch210 Triton dependency for PyTorch 2.10 XPU
#4941
opened Apr 10, 2026 by
aiSynergy37
Loading…
Fix bf16 GGUF deleted by cleanup when exported alongside quantized types
testing
#4937
opened Apr 9, 2026 by
Shivamjohri247
Loading…
3 tasks
Previous Next
ProTip!
Follow long discussions with comments:>50.