-
Notifications
You must be signed in to change notification settings - Fork 8.7k
Pull requests: hiyouga/LlamaFactory
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
fix(data/converter): handle None tool_calls in OpenAI-style messages
#10455
opened May 1, 2026 by
Anai-Guo
Loading…
[train] fix loss aggregation bug in SFT and PT training
#10454
opened Apr 30, 2026 by
4teven
Loading…
[xpu] extend runs_on test markers to include xpu
#10445
opened Apr 29, 2026 by
singhalshubham03
Loading…
2 tasks
[model] support DeepSeek V4
pending
This problem is yet to be addressed
#10434
opened Apr 25, 2026 by
isLinXu
Contributor
Loading…
[V1] support reward training stage
pending
This problem is yet to be addressed
#10431
opened Apr 25, 2026 by
frozenleaves
Collaborator
Loading…
[model-test] Add test for Llama-3.1-8B-Instruct model
#10426
opened Apr 23, 2026 by
pankd
Loading…
2 tasks done
[chat] add thinking token injection for reasoning models in all engines
#10424
opened Apr 23, 2026 by
kally788
Loading…
4 tasks
[chat] fix enable_thinking=None overriding template defaults
#10423
opened Apr 23, 2026 by
kally788
Loading…
3 tasks
fix: add ignore_mismatched_sizes option to model loader
#10420
opened Apr 22, 2026 by
octo-patch
Loading…
avoid the EOFError issue when run chat with a sample prompt at a jupyter notebook
#10409
opened Apr 19, 2026 by
zhangnju
Loading…
fix: materialize FSDP2 model on CPU when CPU offloading is enabled
#10403
opened Apr 17, 2026 by
octo-patch
Loading…
fix: handle mm_token_type_ids in collator and packing tests
pending
This problem is yet to be addressed
#10397
opened Apr 16, 2026 by
markmochi200
Loading…
2 tasks done
fix: add None check for feature_extractor in Gemma4Plugin audio processing
#10388
opened Apr 13, 2026 by
Ricardo-M-L
Loading…
2 tasks done
fix: prevent training hang when WebUI client disconnects
#10383
opened Apr 11, 2026 by
Ricardo-M-L
Loading…
4 tasks done
fix: Ray placement group over-allocation and NCCL hang on GPU-less head node
#10380
opened Apr 10, 2026 by
Ricardo-M-L
Loading…
5 tasks
[feat] support HyperParallel PT training and activation optimization
#10370
opened Apr 9, 2026 by
Cui-yshoho
Contributor
•
Draft
fix: sanitize subprocess call in launcher.py
#10369
opened Apr 9, 2026 by
orbisai0security
Loading…
3 tasks done
fix: use json.loads with Path.read_text() instead of json.load with Path
#10361
opened Apr 6, 2026 by
satishkc7
Loading…
[perf] Skip unused lm_head projection and hidden state storage in RM trainer
#10353
opened Apr 5, 2026 by
tonywang1990
Loading…
4 tasks done
[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
Previous Next
ProTip!
Add no:assignee to see everything that’s not assigned.