Where can I get the file visualglm-6b/300/mp_rank_00_model_states.pt

[xuzikun2003]

I run the q-lora fine tuning script and get the following error

File "/home/z47xu/Code/VisualGLM/VisualGLM-6B/finetune_visualglm.py", line 180, in model, args = FineTuneVisualGLMModel.from_pretrained(model_type, args) File "/home/z47xu/anaconda3/envs/py3p10/bin/lib/python3.10/site-packages/sat/model/base_model.py", line 215, in from_pretrained return cls.from_pretrained_base(name, args=args, home_path=home_path, url=url, prefix=prefix, build_only=build_only, overwrite_args=overwrite_args, **kwargs) File "/home/z47xu/anaconda3/envs/py3p10/bin/lib/python3.10/site-packages/sat/model/base_model.py", line 209, in from_pretrained_base load_checkpoint(model, args, load_path=model_path, prefix=prefix) File "/home/z47xu/anaconda3/envs/py3p10/bin/lib/python3.10/site-packages/sat/training/model_io.py", line 225, in load_checkpoint sd = torch.load(checkpoint_name, map_location='cpu') File "/home/z47xu/anaconda3/envs/py3p10/bin/lib/python3.10/site-packages/torch/serialization.py", line 986, in load with _open_file_like(f, 'rb') as opened_file: File "/home/z47xu/anaconda3/envs/py3p10/bin/lib/python3.10/site-packages/torch/serialization.py", line 435, in _open_file_like return _open_file(name_or_buffer, mode) File "/home/z47xu/anaconda3/envs/py3p10/bin/lib/python3.10/site-packages/torch/serialization.py", line 416, in init super().init(open(name, mode)) FileNotFoundError: [Errno 2] No such file or directory: 'visualglm-6b/300/mp_rank_00_model_states.pt'

I am wondering whether to download the file mp_rank_00_model_states.pt.

Your help is much appreciated.

[xuzikun2003]

Should I do something like this?

from transformers import AutoTokenizer, AutoModel, AutoConfig from sat.training.model_io import save_checkpoint

config = AutoConfig.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True)

args = argparse.Namespace( num_layers=config.num_layers, vocab_size=config.vocab_size, hidden_size=config.hidden_size, num_attention_heads=config.num_attention_heads, max_sequence_length=config.max_sequence_length, bos_token_id=tokenizer.bos_token_id, mask_token_id=tokenizer.mask_token_id, gmask_token_id=tokenizer.gmask_token_id, hidden_dropout=0., attention_dropout=0., inner_hidden_size=None, hidden_size_per_attention_head=None, checkpoint_activations=True, checkpoint_num_layers=1, layernorm_order='post', model_parallel_size=1, world_size=1, rank=0, skip_init=False, use_gpu_initialization=True, save='model_check_point', deepspeed=None, mode='inference', tokenizer_type="THUDM/chatglm-6b")

from transformers import AutoTokenizer, AutoModel tokenizer = AutoTokenizer.from_pretrained("visualglm-6b", trust_remote_code=True) model = AutoModel.from_pretrained("visualglm-6b",trust_remote_code=True).half().quantize(8).cuda()

save_checkpoint(1, model, None, None, args)

[1049451037]

We don't have visualglm-6b/300. I don't know what code you are running.

[xuzikun2003]

We don't have visualglm-6b/300. I don't know what code you are running.

I ran the q-lora fine tuning script. bash finetune/finetune_visualglm_qlora.sh and got the above error.

[1049451037]

I don't know why. But our fine tuning script will never request visualglm-6b/300 files. I can only assume that you have changed the code or folder structure, which is out of my control.