Views
No views yet
untied_input_only repatch).tie_word_embeddings=false (independent embed / lm_head)qwen3-nothink-stage2-merged.tar (8.3G)1tar -xf qwen3-nothink-stage2-merged.tar
2from transformers import AutoModelForCausalLM
3m = AutoModelForCausalLM.from_pretrained("./qwen3-nothink-stage2-merged", trust_remote_code=True)