Issues 共 6804
lora微调后合并模型并vllm部署效果下降，但直接使用llamafactory分别加载基座模型与checkpoint建立的接口测试没问题
#10596 · liyf-llm · 2026-06-22
Will Gemma4 12b fine-tuning be supported?
#10548 · lwang10-atlassian · 2026-06-03
从checkpoint开始训练报torch版本不兼容，需要torch2.6版本，但cuda12.1不兼容
#10501 · winnieLLL007 · 2026-05-20
no qwen3_6_nothink template
#10466 · xiaobaifeifeifei · 2026-05-07
llamafactory-cli train -h broken
#10459 · SamuelLarkin · 2026-05-05
[Bug] Qwen3.5 + FA2 + neat_packing can create invalid cu_seqlens and crash FLA GDN kernel
#10452 · FishWoWater · 2026-04-30
DataLoader multi-process initialization
#10444 · chan-yuu · 2026-04-29
Qwen3.5 fa2 微调报错 illegal memory
#10440 · FishWoWater · 2026-04-27
need help to Support Kwai-Keye/Keye-VL-1_5-8B
#10425 · caomengying · 2026-04-23
Onboarding Gemma 4 in V1
#10417 · StevenShi-23 · 2026-04-22