Skip to content

Pull requests: hiyouga/LlamaFactory

New pull request New
32 Open 1,225 Closed
32 Open 1,225 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

[feat] support HyperParallel PT training and activation optimization
#10370 opened Apr 9, 2026 by Cui-yshoho Contributor • Draft
3
fix: sanitize subprocess call in launcher.py
#10369 opened Apr 9, 2026 by orbisai0security Loading…
3 tasks done
3
[feat] Add support FA3
#10368 opened Apr 8, 2026 by y2sman Loading…
2 tasks done
2
[v1] fix device mesh and clip_grad_norm for ulysses cp
#10366 opened Apr 8, 2026 by sunyi0505 Contributor Loading…
2 tasks done
2
[model] add gemma4_text template for text-only SFT/DPO training
#10362 opened Apr 7, 2026 by leivy-dev Loading…
2 tasks done
4
fix: use json.loads with Path.read_text() instead of json.load with Path
#10361 opened Apr 6, 2026 by satishkc7 Loading…
1
2
replace chat.py pop(0) with deque.popleft()
#10356 opened Apr 5, 2026 by nameearly Loading…
1 task
2
[fix] parser json.load PosixPath bug
#10354 opened Apr 5, 2026 by Belle0918 Loading…
2 tasks done
1
2
[perf] Skip unused lm_head projection and hidden state storage in RM trainer
#10353 opened Apr 5, 2026 by tonywang1990 Loading…
4 tasks done
8
[ray] fix placement group over-allocation and NCCL hang on GPU-less head node
#10349 opened Apr 3, 2026 by ilover311 Loading…
2 tasks done
1
3
Add workflow for building ROCm image
#10330 opened Mar 30, 2026 by ErikJiang Loading…
1
fix: pin 12 unpinned action(s)
#10325 opened Mar 26, 2026 by dagecko Loading…
1
[WIP] Support huggingface/kernels
#10319 opened Mar 25, 2026 by zheliuyu • Draft
3 of 5 tasks
5
fix: add qwen3_5_moe to MoE configuration in moe.py invalid This doesn't seem right
#10307 opened Mar 21, 2026 by majiayu000 Loading…
1
13
[v1] add deepspeed zero3 trigger for low memory usage weight loading
#10300 opened Mar 19, 2026 by jiaqiw09 Collaborator Loading…
1 of 2 tasks
11
fix: mutable default arg and bool comparison
#10297 opened Mar 18, 2026 by LincolnBurrows2017 Contributor Loading…
3
feat: clearer train_result metrics log through calculate_tps function
#10288 opened Mar 17, 2026 by UmeanNever Loading…
1 of 2 tasks
4
[V1]support resume training from checkpoint
#10280 opened Mar 13, 2026 by frozenleaves Collaborator Loading…
12
fix qwen3vl moe fuse on transformers 5.x and update docs about timeout
#10274 opened Mar 12, 2026 by addsubmuldiv Loading…
2 tasks
8
feat: add LightOnOCR-2 integration for LoRA/QLoRA fine-tuning
#10192 opened Feb 16, 2026 by johnlockejrr Loading…
2 tasks
5
Fix memory leak on MPS by explicitly clearing cache in trainer step
#10190 opened Feb 14, 2026 by asebaq Loading…
1 of 2 tasks
6
[v1] Add hyperparams and training docs
#10188 opened Feb 13, 2026 by frozenleaves Collaborator Loading…
8
[deps] Add libibverbs for RDMA support
#10185 opened Feb 12, 2026 by RossCZ Loading…
1 of 2 tasks
4
Feature: experimental fine-tuning comparison
#10172 opened Feb 6, 2026 by caterina0718 Loading…
11
[feat] Add DeepSpeed ZeRO-3 LoRA checkpoint save support
#10124 opened Jan 22, 2026 by kimberlykang Loading…
2 tasks done
4
ProTip! Adding no:label will show everything without a label.