|
|
import os |
|
|
import yaml |
|
|
import torch |
|
|
from transformers import AlbertConfig, AlbertModel |
|
|
|
|
|
|
|
|
class CustomAlbert(AlbertModel): |
|
|
def forward(self, *args, **kwargs): |
|
|
|
|
|
outputs = super().forward(*args, **kwargs) |
|
|
|
|
|
|
|
|
return outputs.last_hidden_state |
|
|
|
|
|
|
|
|
def load_plbert(log_dir): |
|
|
config_path = os.path.join(log_dir, "config.yml") |
|
|
plbert_config = yaml.safe_load(open(config_path)) |
|
|
|
|
|
albert_base_configuration = AlbertConfig(**plbert_config["model_params"]) |
|
|
bert = CustomAlbert(albert_base_configuration) |
|
|
|
|
|
files = os.listdir(log_dir) |
|
|
ckpts = [] |
|
|
for f in os.listdir(log_dir): |
|
|
if f.startswith("step_"): |
|
|
ckpts.append(f) |
|
|
|
|
|
iters = [ |
|
|
int(f.split("_")[-1].split(".")[0]) |
|
|
for f in ckpts |
|
|
if os.path.isfile(os.path.join(log_dir, f)) |
|
|
] |
|
|
iters = sorted(iters)[-1] |
|
|
|
|
|
checkpoint = torch.load(log_dir + "/step_" + str(iters) + ".t7", map_location="cpu") |
|
|
state_dict = checkpoint["net"] |
|
|
from collections import OrderedDict |
|
|
|
|
|
new_state_dict = OrderedDict() |
|
|
for k, v in state_dict.items(): |
|
|
name = k[7:] |
|
|
if name.startswith("encoder."): |
|
|
name = name[8:] |
|
|
new_state_dict[name] = v |
|
|
del new_state_dict["embeddings.position_ids"] |
|
|
bert.load_state_dict(new_state_dict, strict=False) |
|
|
|
|
|
return bert |
|
|
|
