windmaple

/

gemma-chinese

like 0

Got this error when trying to load model:

[Stack<Integer> stack = new Stack<>();](RuntimeError: Error(s) in loading state_dict for GemmaForCausalLM:
    size mismatch for model.embed_tokens.weight: copying a param with shape torch.Size([256002, 2048]) from checkpoint, the shape in current model is torch.Size([256000, 2048]).
    size mismatch for lm_head.weight: copying a param with shape torch.Size([256002, 2048]) from checkpoint, the shape in current model is torch.Size([256000, 2048]).)

I use this piece of code:

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
from accelerate import Accelerator

device_index = Accelerator().process_index
device_map = {"": device_index}
torch.cuda.empty_cache()
hf.login(token="xxxxxxxxx",add_to_git_credential=False)

model_id = "windmaple/gemma-chinese"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map=device_map)

do you know how to fix this? or any suggestions on how to load the model properly? Thanks a lot