ESFT/esft.py

136 lines
4.6 KiB
Python
Raw Normal View History

import os
import json
import torch
from torch import nn
from safetensors.torch import load_file
from transformers import AutoModelForCausalLM, AutoTokenizer
def to_buffer(module, mark_param=True):
"""Turns all parameters of a module into buffers."""
2024-08-09 08:23:48 +00:00
if module is None:
return
modules = module.modules()
module = next(modules)
delattrs = []
for name, param in module.named_parameters(recurse=False):
delattrs.append([module, name, param])
if mark_param and delattrs:
old_param_list = getattr(module, 'param_list', [])
module.param_list = old_param_list + [name for _, name, _ in delattrs]
for module, name, _ in delattrs:
delattr(module, name) # Unregister parameter
for module, name, param in delattrs:
module.register_buffer(name, param.data, persistent=False)
for module in modules:
to_buffer(module, mark_param=mark_param)
def to_param(module):
"""Turns all buffers of a module into parameterss."""
2024-08-09 08:23:48 +00:00
if module is None:
return
modules = module.modules()
module = next(modules)
param_list = getattr(module, 'param_list', [])
for name in param_list:
buffer = getattr(module, name)
delattr(module, name) # Delete buffer
setattr(module, name, nn.Parameter(buffer))
for module in modules:
to_param(module)
def recursive_getattr(model, module_name):
split_list = module_name.split('.')
output = model
for name in split_list:
output = getattr(output, name)
return output
def recursive_setattr(model, module_name, module):
split_list = module_name.split('.')
output = model
for name in split_list[:-1]:
output = getattr(output, name)
output.__setattr__(split_list[-1], module)
def to_esft(model, adapter_config):
if not adapter_config.get('non_expert_modules', False):
to_buffer(model)
else:
to_param(model)
2024-08-09 08:23:48 +00:00
for idx, layer in enumerate(model.model.layers):
if type(layer.mlp).__name__ != "DeepseekV2MoE":
continue
if adapter_config.get('shared_experts', False):
to_param(layer.mlp.shared_experts)
else:
to_buffer(layer.mlp.shared_experts)
trainable_experts = adapter_config['experts'][str(idx)]
for expert_id in range(len(layer.mlp.experts)):
if expert_id in trainable_experts:
to_param(layer.mlp.experts[expert_id])
else:
to_buffer(layer.mlp.experts[expert_id])
return model
2024-08-09 08:23:48 +00:00
def load_state_dict(folder_path):
2024-08-09 08:23:48 +00:00
# 初始化空的 state_dict
combined_state_dict = {}
2024-08-09 08:23:48 +00:00
# 遍历文件夹中的所有文件
for file_name in os.listdir(folder_path):
if file_name.endswith('.safetensors'):
file_path = os.path.join(folder_path, file_name)
state_dict = load_file(file_path)
combined_state_dict.update(state_dict)
2024-08-09 08:23:48 +00:00
# legacy for loading v1 checkpoints: add prefix "model." for parameters
for k in list(combined_state_dict.keys()):
if k.startswith("layers"):
k_new = "model." + k
combined_state_dict[k_new] = combined_state_dict[k]
del combined_state_dict[k]
return combined_state_dict
def load_esft_model(base_model_path, adapter_dir):
adapter_config = json.load(open(adapter_dir + "/expert_cfg.json"))
adapter_state_dict = load_state_dict(adapter_dir)
# load pretrained model:
2024-08-09 08:23:48 +00:00
model, tokenizer = AutoModelForCausalLM.from_pretrained(base_model_path, trust_remote_code=True, torch_dtype=torch.bfloat16), AutoTokenizer.from_pretrained(base_model_path)
2024-08-09 08:23:48 +00:00
to_esft(model, adapter_config)
model.load_state_dict(adapter_state_dict)
return model, tokenizer
def load_base_model(base_model_path):
# load pretrained model:
2024-08-09 08:23:48 +00:00
model, tokenizer = AutoModelForCausalLM.from_pretrained(base_model_path, trust_remote_code=True, torch_dtype=torch.bfloat16), AutoTokenizer.from_pretrained(base_model_path)
return model, tokenizer
2024-08-09 08:23:48 +00:00
def add_adapter(base_model, adapter_dir, return_original_states=False, expert_config=None):
if expert_config is not None:
adapter_config = json.load(open(expert_config))
else:
adapter_config = json.load(open(adapter_dir + "/expert_cfg.json"))
adapter_state_dict = load_state_dict(adapter_dir)
to_esft(base_model, adapter_config)
if return_original_states:
original_state_dict = {k:v.cpu() for k, v in base_model.state_dict().items()}
base_model.load_state_dict(adapter_state_dict, strict=False)
return base_model, original_state_dict
else:
base_model.load_state_dict(adapter_state_dict)
return base_model