MediaTek-Research

/

Llama-Breeze2-8B-Instruct

like 26

Follow

MediaTek Research 200

code:

import torch
from transformers import AutoModel, PreTrainedModel

model: PreTrainedModel = AutoModel.from_pretrained(
    "MediaTek-Research/Llama-Breeze2-8B-Instruct",
    torch_dtype=torch.bfloat16,
    trust_remote_code=True,
    img_context_token_id=128212,
)

model.save_pretrained("hello")

log:

Loading checkpoint shards: 100%|██████████████████████████████████████| 4/4 [00:00<00:00,  7.71it/s]
Traceback (most recent call last):
  File "/data2/Penut/LLM-Backend/testing.py", line 12, in <module>
    model.save_pretrained("hello")
  File "/data2/Penut/.miniconda/envs/Dev312/lib/python3.12/site-packages/transformers/modeling_utils.py", line 2815, in save_pretrained
    misplaced_generation_parameters = model_to_save.config._get_non_default_generation_parameters()
                                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/data2/Penut/.miniconda/envs/Dev312/lib/python3.12/site-packages/transformers/configuration_utils.py", line 1063, in _get_non_default_generation_parameters
    default_config = self.__class__()
                     ^^^^^^^^^^^^^^^^
  File "/home/penut_chen/.cache/huggingface/modules/transformers_modules/MediaTek-Research/Llama-Breeze2-8B-Instruct/7e47738e5f3bfae3063f2df8e91e0eab5178a3ea/configuration_internvl_chat.py", line 49, in __init__
    if llm_config['architectures'][0] == 'LlamaForCausalLM':
       ~~~~~~~~~~^^^^^^^^^^^^^^^^^
KeyError: 'architectures'