Nothing Works for HiDream
#1
by
qpqpqpqpqpqp - opened
calcuis' GGUF CLIP Loader:
llama3.1-encoder-q3_k_m.gguf
RuntimeError: Error(s) in loading state_dict for Llama2:
size mismatch for model.embed_tokens.weight: copying a param with shape torch.Size([128320, 4096]) from checkpoint, the shape in current model is torch.Size([128256, 4096]).
I had to download a quant of Llama from someone else
not test that model for very long; guess something changed over the time; have you found the one working for this?
oh, i see; they changed to the old scheme for this for some reason
token_embd.weight [4 096, 128 256]
and this one should work also:
https://huggingface.co/chatpig/encoder/blob/main/llama-hidream-q2_k.gguf