forked from linhduongtuan/BLOOM-LORA
-
Notifications
You must be signed in to change notification settings - Fork 0
/
finetune.py
178 lines (155 loc) · 5.47 KB
/
finetune.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
170
171
172
173
174
175
176
177
178
import os
import time
from functools import lru_cache
import torch
from datasets import load_dataset
import transformers
from utils.s3 import zip_n_store
assert (
"LlamaTokenizer" in transformers._import_structure["models.llama"]
), "LLaMA is now in HuggingFace's main branch.\nPlease reinstall it: pip uninstall transformers && pip install git+https://github.com/huggingface/transformers.git"
from transformers import AutoTokenizer, BloomForCausalLM
import json
from peft import (
prepare_model_for_int8_training,
LoraConfig,
get_peft_model,
get_peft_model_state_dict,
)
from generate_bloom import generate
@lru_cache
def get_config(file: str = "train_config.json"):
with open(file, "r", encoding="utf-8") as f:
data = json.load(f)
data["gradient_accumulation_steps"] = data["batch_size"] // data["micro_batch_size"]
return data
"""
The folling models are available for fine-tunning:
"bigscience/bloom-560m"
"bigscience/bloom-1b1"
"bigscience/bloom-1b7"
"bigscience/bloom-3b"
"bigscience/bloom-7b1"
"bigscience/bloom" # for 176B parameters
"""
def train():
# Load training parameters
data = get_config()
model_name = data["model_name"]
lora_r = data["lora_r"]
lora_alpha = data["lora_alpha"]
lora_dropout = data["lora_dropout"]
data_path = data["data_path"]
val_set_size = data["val_set_size"]
cutoff_len = data["cutoff_len"]
micro_batch_size = data["micro_batch_size"]
gradient_accumulation_steps = data["gradient_accumulation_steps"]
epochs = data["epochs"]
lr = data["lr"]
output_dir = data["output_dir"]
ddp = data["ddp"]
resume_from_checkpoint = data["resume_from_checkpoint"]
s3_bucket = data["s3_bucket"]
model = BloomForCausalLM.from_pretrained(
model_name,
device_map="auto",
load_in_8bit=False,
)
tokenizer = AutoTokenizer.from_pretrained("bigscience/bloom")
model = prepare_model_for_int8_training(model)
config = LoraConfig(
r=lora_r,
lora_alpha=lora_alpha,
target_modules=None,
lora_dropout=lora_dropout,
bias="none",
task_type="CAUSAL_LM",
)
model = get_peft_model(model, config)
tokenizer.pad_token_id = 0 # unk. we want this to be different from the eos token
data = load_dataset("json", data_files=data_path)
train_val = data["train"].train_test_split(
test_size=val_set_size, shuffle=True, seed=42
)
train_data = train_val["train"]
val_data = train_val["test"]
def generate_and_tokenize_prompt(data_point):
# This function masks out the labels for the input,
# so that our loss is computed only on the response.
user_prompt = (
(
f"""### Instruction: {data_point["instruction"]}\n\n### Input: {data_point["input"]}\n\n### Response: """
)
if data_point["input"]
else (f"""### Instruction: {data_point["instruction"]}\n\n### Response: """)
)
len_user_prompt_tokens = (
len(
tokenizer(
user_prompt,
truncation=True,
max_length=cutoff_len + 1,
padding="max_length",
)["input_ids"]
)
- 1
) # no eos token
full_tokens = tokenizer(
user_prompt + data_point["output"],
truncation=True,
max_length=cutoff_len + 1,
padding="max_length",
)["input_ids"][:-1]
return {
"input_ids": full_tokens,
"labels": [-100] * len_user_prompt_tokens
+ full_tokens[len_user_prompt_tokens:],
"attention_mask": [1] * (len(full_tokens)),
}
train_data = train_data.shuffle().map(generate_and_tokenize_prompt)
val_data = val_data.shuffle().map(generate_and_tokenize_prompt)
trainer = transformers.Trainer(
model=model,
train_dataset=train_data,
eval_dataset=val_data,
args=transformers.TrainingArguments(
per_device_train_batch_size=micro_batch_size,
gradient_accumulation_steps=gradient_accumulation_steps,
warmup_steps=100,
num_train_epochs=epochs,
learning_rate=lr,
fp16=True,
logging_steps=20,
evaluation_strategy="steps",
save_strategy="steps",
eval_steps=200,
save_steps=200,
output_dir=output_dir,
save_total_limit=3,
load_best_model_at_end=True,
ddp_find_unused_parameters=False if ddp else None,
),
data_collator=transformers.DataCollatorForLanguageModeling(
tokenizer, mlm=False
),
)
model.config.use_cache = False
old_state_dict = model.state_dict
model.state_dict = (
lambda self, *_, **__: get_peft_model_state_dict(self, old_state_dict())
).__get__(model, type(model))
if torch.__version__ >= "2":
model = torch.compile(model)
start_train = time.time()
trainer.train(
resume_from_checkpoint=resume_from_checkpoint
) # if resume, choose True, else False
end_train = time.time() - start_train
with open(
os.path.join(output_dir, f"train_timing_{output_dir}.txt"), "w", encoding="utf8"
) as f:
f.write(f"Time to train: {end_train}")
model.save_pretrained(output_dir)
# zip_n_store(output_dir, s3_bucket, output_dir + ".zip")
if __name__ == "__main__":
train()