Issue with text encoder
#237
by Wistore - opened
What's problem? I use comfy ui, download model, vae and text encoder according to instructions, but have an issue -- Error(s) in loading state_dict for Llama2: size mismatch for model.embed_tokens.weight: copying a param with shape torch.Size([151936, 1024]) from checkpoint, the shape in current model is torch.Size([128256, 4096]) -- as I'm downloaded incorrect clip. Can someone help fix this?
Try updating
Well, after some investigation I finally got it working.
Turns out the problem was simply that my ComfyUI version was outdated. Updating ComfyUI fixed everything immediately.
So, for anyone else running into the same Anima/Qwen3 Llama2 size mismatch error: update ComfyUI first. You probably don't need to reinstall all the models.
Wistore changed discussion status to closed