Jiayin Wang

icon company CST, Tsinghua icon location Beijing, China icon location A student interested in User Modeling, Personalization at Tsinghua University, China

Results 4 comments of Jiayin Wang

Python 3.11.6 compatibility

so what is the xformers version in the final? I tried 0.0.22, 0.0.21, 0.0.16 all facing errors. thanks!

[DONOTMERGE][rollout] feat: ChatScheduler requests sglang fully async

I face the same issue! with "There appear to be 1 leaked shared_memory objects to clean up at shutdown“ when running the run_qwen2.5-3b_gsm8k_multiturn.sh with the latest code.

string index out of range

Facing the same problem

special token未能输出

希望这个【官方能否出一个新增token训练/merge/推理的脚本】+1，十分感谢！ 以及想要确认一下，在现在的代码中， lora训练的时候使用 new_special_tokens: xxx, xxx additional_target: embed_tokens,lm_head resize_vocab: true 在lora merge的时候使用 new_special_tokens: xxx, xxx resize_vocab: true 那么merge之后的新模型的embed_tokens,lm_head是训练后的吧，还是base model的呢？ 搜索了之前的帖子，没有找到解答