test

语帆

2024-02-22 b349739f5d6302048c179eeaadb4432acc541cd5

test

 funasr/auto/auto_model.py

@@ -172,12 +172,14 @@

        # build model
        model_class = tables.model_classes.get(kwargs["model"])
        pdb.set_trace()
        model = model_class(**kwargs, **kwargs["model_conf"], vocab_size=vocab_size)
        
        pdb.set_trace()
        model.to(device)
        
        # init_param
        init_param = kwargs.get("init_param", None)
        pdb.set_trace()
        if init_param is not None:
            logging.info(f"Loading pretrained params from {init_param}")
            load_pretrained_model(

 funasr/train_utils/load_pretrained_model.py

@@ -7,7 +7,7 @@
import torch
import torch.nn
import torch.optim

import pdb

def filter_state_dict(
    dst_state: Dict[str, Union[float, torch.Tensor]],
@@ -99,14 +99,16 @@
    # import pdb;
    # pdb.set_trace()
    print(f"ckpt: {path}")
    pdb.set_trace()
    if oss_bucket is None:
        src_state = torch.load(path, map_location=map_location)
    else:
        buffer = BytesIO(oss_bucket.get_object(path).read())
        src_state = torch.load(buffer, map_location=map_location)
    pdb.set_trace()
    if "state_dict" in src_state:
        src_state = src_state["state_dict"]
		
    pdb.set_trace()
    for k in dst_state.keys():
        if not k.startswith("module.") and "module." + k in src_state.keys():
            k_ddp = "module." + k
@@ -116,7 +118,7 @@
            dst_state[k] = src_state[k_ddp]
        else:
            print(f"Miss key in ckpt: model: {k}, ckpt: {k_ddp}")
			
    pdb.set_trace()
    flag = obj.load_state_dict(dst_state, strict=True)
    # print(flag)