-
Notifications
You must be signed in to change notification settings - Fork 4
/
train_deepspeed.py
170 lines (146 loc) · 6.03 KB
/
train_deepspeed.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
import os
import json
import torch
import deepspeed
import argparse
from shutil import copy
from pprint import pprint
from transformers import AutoModelForCausalLM
from transformers import AutoTokenizer
from torch.utils.data import RandomSampler, DataLoader
from dataset import load_data, NerCollate
from config_utils import ConfigParser
from transformers import AutoModelForSeq2SeqLM
from transformers import AutoTokenizer
from peft import (
LoraConfig,
get_peft_model,
get_peft_model_state_dict,
set_peft_model_state_dict,
)
def print_trainable_parameters(model):
trainable_params = 0
all_param = 0
for _, param in model.named_parameters():
num_params = param.numel()
if num_params == 0 and hasattr(param, "ds_numel"):
num_params = param.ds_numel
all_param += num_params
if param.requires_grad:
trainable_params += num_params
print(
f"trainable params: {trainable_params} || all params: {all_param} || trainable%: {100 * trainable_params / all_param}")
def main():
args = {
"data_name": "msra",
"model_dir": "/root/autodl-tmp/chatglm-6b/",
"lora_r": 8,
"max_source_length": 128,
"max_target_length": 32,
"instruct_column": "instruct",
"query_column": "query",
"response_column": "answer",
"train_path": "data/msra/instruct_data/train.txt",
"dev_path": "data/msra/instruct_data/dev.txt",
"ignore_pad_token_for_loss": True,
"train_batch_size": 12,
"gradient_accumulation_steps": 1,
"save_dir": "./checkpoint/msra/train_deepspeed/adapter_model/",
"num_train_epochs": 1,
"local_rank": -1,
"log_steps": 10,
"save_steps": 400,
}
config_parser = ConfigParser(args)
args = config_parser.parse_main()
pprint(vars(args))
if not os.path.exists(args.save_dir):
os.makedirs(args.save_dir)
with open(os.path.join(args.save_dir, "train_args.json"), "w") as fp:
json.dump(vars(args), fp, ensure_ascii=False, indent=2)
model = AutoModelForSeq2SeqLM.from_pretrained(args.model_dir,
trust_remote_code=True,
)
tokenizer = AutoTokenizer.from_pretrained(args.model_dir, trust_remote_code=True)
config = LoraConfig(r=args.lora_r,
lora_alpha=32,
target_modules=["query_key_value"],
lora_dropout=0.1,
bias="none",
task_type="CAUSAL_LM",
inference_mode=False,
)
model = get_peft_model(model, config)
model = model.cuda()
conf = {"train_micro_batch_size_per_gpu": args.train_batch_size,
"gradient_accumulation_steps": args.gradient_accumulation_steps,
"optimizer": {
"type": "Adam",
"params": {
"lr": 1e-5,
"betas": [
0.9,
0.95
],
"eps": 1e-8,
"weight_decay": 5e-4
}
},
"fp16": {
"enabled": True
},
"zero_optimization": {
"stage": 1,
"offload_optimizer": {
"device": "cpu",
"pin_memory": True
},
"allgather_partitions": True,
"allgather_bucket_size": 2e8,
"overlap_comm": True,
"reduce_scatter": True,
"reduce_bucket_size": 2e8,
"contiguous_gradients": True
},
"steps_per_print": args.log_steps
}
print_trainable_parameters(model)
for name, param in model.named_parameters():
if param.requires_grad == True:
print(name)
train_data = load_data(args.train_path)
ner_collate = NerCollate(args, tokenizer)
train_dataloader = DataLoader(train_data,
batch_size=conf["train_micro_batch_size_per_gpu"],
sampler=RandomSampler(train_data),
drop_last=True,
collate_fn=ner_collate.collate_fn)
model_engine, optimizer, _, _ = deepspeed.initialize(config=conf,
model=model,
model_parameters=model.parameters())
model_engine.train()
total_step = int(len(train_dataloader) * args.num_train_epochs / conf["gradient_accumulation_steps"])
global_step = 0
for i_epoch in range(args.num_train_epochs):
train_iter = iter(train_dataloader)
for step, batch in enumerate(train_iter):
input_ids = batch["input_ids"].cuda()
labels = batch["labels"].cuda()
outputs = model_engine.forward(input_ids=input_ids, labels=labels)
loss = outputs[0]
if conf["gradient_accumulation_steps"] > 1:
loss = loss / gradient_accumulation_steps
model_engine.backward(loss)
torch.nn.utils.clip_grad_norm_(model.parameters(), 1.0)
if (step + 1) % conf["gradient_accumulation_steps"] == 0:
model_engine.step()
global_step += 1
if global_step % args.log_steps == 0:
print("loss:{}, global_step:{}/{}".format(float(loss.item()), global_step, total_step))
if global_step % args.save_steps == 0:
# save_dir = os.path.join(args.output_dir, f"global_step-{global_step}")
model_engine.save_pretrained(args.save_dir)
# copy(os.path.join(args.model_dir, "tokenizer_config.json"), os.path.join(args.save_dir, "tokenizer_config.json"))
# copy(os.path.join(args.model_dir, "ice_text.model"), os.path.join(args.save_dir, "ice_text.model"))
if __name__ == "__main__":
main()