Skip to content

Pull requests: unslothai/unsloth

New pull request New
98 Open 782 Closed
98 Open 782 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Fix FBGEMM/CUTLASS errors on SM100 (Blackwell) GPUs
#3863 opened Jan 8, 2026 by danielhanchen Loading…
5 tasks done
6
add weight-only int8 QAT scheme and update tests for torchao 0.15.0
#3859 opened Jan 6, 2026 by electroglyph Loading…
2
21
feat: add mlx model and trainer
#3856 opened Jan 6, 2026 by JINO-ROHIT Loading…
9
Fix: Batched generation with left-padding position_ids bug (#3699)
#3837 opened Jan 4, 2026 by Ashutosh0x Loading…
1
5
Add context parallelism support
#3823 opened Jan 2, 2026 by djsaunde • Draft
2 tasks
@djsaunde
26
Add Qwen3-Omni Support with Optimized MTP Fine-Tuning
#3819 opened Jan 2, 2026 by AlanPonnachan Loading…
1
10
fix: propagate revision to vLLM fast_inference
#3816 opened Jan 2, 2026 by majiayu000 Loading…
6
fix: inputs_embeds ignored when input_ids is not None in _fast_prepare_inputs_for_generation
#3814 opened Jan 1, 2026 by siddhudonda Loading…
1
10
[MoE] Improve moe kernels for unsloth fine tuning
#3812 opened Jan 1, 2026 by Datta0 Loading…
8
fix: multi-GPU training support for vision models
#3809 opened Dec 31, 2025 by Vinayyyy7 Loading…
12
fix: handle zero-strided tensors in fast_rope_embedding (#3781)
#3804 opened Dec 29, 2025 by yurekami Loading…
3 tasks
1
4
fix: compute position_ids correctly for batched left-padded generation
#3803 opened Dec 29, 2025 by yurekami Loading…
4 tasks
1
4
fix: properly handle inputs_embeds in _fast_prepare_inputs_for_generation
#3801 opened Dec 29, 2025 by yurekami Loading…
1 of 2 tasks
1
4
fix: add inputs_embeds support with correct priority in generation
#3799 opened Dec 29, 2025 by majiayu000 Loading…
1
5
fix: add inputs_embeds support to prepare_inputs_for_generation
#3796 opened Dec 29, 2025 by majiayu000 Loading…
1 task done
1
8
fix: compute position_ids from attention_mask for batched generation
#3795 opened Dec 29, 2025 by majiayu000 Loading…
1 task done
1
5
fix: add revision parameter support and escape quotes in chat templates
#3793 opened Dec 29, 2025 by majiayu000 Loading…
2 tasks done
2
14
unsloth support AMD radeon cards
#3740 opened Dec 18, 2025 by eliotwang Loading…
12
add FastSentenceTransformer for easily finetuning SentenceTransformer models
#3719 opened Dec 12, 2025 by electroglyph Loading…
64
FP8: Load model on-the-fly in vLLM
#3717 opened Dec 11, 2025 by andrewor14 Loading…
11
Add target_parameters support for MoE models and fix trainer bugs
#3708 opened Dec 10, 2025 by OEvortex Loading…
8
Align ruff version in pre-commit config
#3691 opened Dec 8, 2025 by mk0walsk Loading…
2
Fix: Add support for TRL native dataset formats
#3660 opened Dec 1, 2025 by ved1beta Loading…
1
10
Fix: Support past_key_values in model.generate for multi-turn conversations
#3653 opened Nov 29, 2025 by vivekkalyanarangan30 Loading…
4 tasks done
1
24
Chunk Across Batch and Context length for logprob calculations for grpo
#3628 opened Nov 21, 2025 by pluesclues Loading…
34
ProTip! Add no:assignee to see everything that’s not assigned.