ZhouZineng

icon company University of Chinese Academy of Sciences icon location BeiJing icon location Loves programmers in the field of NLP

Results 2 comments of ZhouZineng

OOM and slow tokenization after upgrade LLaMA-Factory

I've resolved this bug by setting `use_fast_tokenizer: false`. After carefully comparing the dependency versions in the environments for v0.9.2 and v0.9.4, I suspect the issue is caused by a newer...

Slow Training of Qwen/Qwen3-VL-8B-Instruct with 32k context

I am also facing a similar issue. Especially when using multiple machines and multiple cards, the utilization is even lower.