mirror of
https://github.com/tloen/alpaca-lora.git
synced 2024-10-01 01:05:56 -04:00
139 lines
4.0 KiB
Python
139 lines
4.0 KiB
Python
import os
|
|
|
|
# os.environ["CUDA_VISIBLE_DEVICES"] = "0"
|
|
import torch
|
|
import torch.nn as nn
|
|
import bitsandbytes as bnb
|
|
from datasets import load_dataset
|
|
import transformers
|
|
|
|
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 LlamaForCausalLM, LlamaTokenizer
|
|
from peft import (
|
|
prepare_model_for_int8_training,
|
|
LoraConfig,
|
|
get_peft_model,
|
|
get_peft_model_state_dict,
|
|
)
|
|
|
|
|
|
# optimized for RTX 4090. for larger GPUs, increase some of these?
|
|
MICRO_BATCH_SIZE = 4 # this could actually be 5 but i like powers of 2
|
|
BATCH_SIZE = 128
|
|
GRADIENT_ACCUMULATION_STEPS = BATCH_SIZE // MICRO_BATCH_SIZE
|
|
EPOCHS = 3 # we don't need 3 tbh
|
|
LEARNING_RATE = 3e-4 # the Karpathy constant
|
|
CUTOFF_LEN = 256 # 256 accounts for about 96% of the data
|
|
LORA_R = 8
|
|
LORA_ALPHA = 16
|
|
LORA_DROPOUT = 0.05
|
|
VAL_SET_SIZE = 2000
|
|
|
|
model = LlamaForCausalLM.from_pretrained(
|
|
"decapoda-research/llama-7b-hf",
|
|
load_in_8bit=True,
|
|
device_map="auto",
|
|
)
|
|
tokenizer = LlamaTokenizer.from_pretrained(
|
|
"decapoda-research/llama-7b-hf", add_eos_token=True
|
|
)
|
|
|
|
model = prepare_model_for_int8_training(model)
|
|
|
|
config = LoraConfig(
|
|
r=LORA_R,
|
|
lora_alpha=LORA_ALPHA,
|
|
target_modules=["q_proj", "v_proj"],
|
|
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="alpaca_data.json")
|
|
|
|
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_prompt(data_point):
|
|
# sorry about the formatting disaster gotta move fast
|
|
if data_point["input"]:
|
|
return f"""Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
|
|
|
|
### Instruction:
|
|
{data_point["instruction"]}
|
|
|
|
### Input:
|
|
{data_point["input"]}
|
|
|
|
### Response:
|
|
{data_point["output"]}"""
|
|
else:
|
|
return f"""Below is an instruction that describes a task. Write a response that appropriately completes the request.
|
|
|
|
### Instruction:
|
|
{data_point["instruction"]}
|
|
|
|
### Response:
|
|
{data_point["output"]}"""
|
|
|
|
|
|
def tokenize(prompt):
|
|
# there's probably a way to do this with the tokenizer settings
|
|
# but again, gotta move fast
|
|
result = tokenizer(
|
|
prompt,
|
|
truncation=True,
|
|
max_length=CUTOFF_LEN + 1,
|
|
padding="max_length",
|
|
)
|
|
return {
|
|
"input_ids": result["input_ids"][:-1],
|
|
"attention_mask": result["attention_mask"][:-1],
|
|
}
|
|
|
|
|
|
train_data = train_data.shuffle().map(lambda x: tokenize(generate_prompt(x)))
|
|
val_data = val_data.shuffle().map(lambda x: tokenize(generate_prompt(x)))
|
|
|
|
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=LEARNING_RATE,
|
|
fp16=True,
|
|
logging_steps=20,
|
|
evaluation_strategy="steps",
|
|
save_strategy="steps",
|
|
eval_steps=200,
|
|
save_steps=200,
|
|
output_dir="lora-alpaca",
|
|
save_total_limit=3,
|
|
load_best_model_at_end=True,
|
|
),
|
|
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))
|
|
|
|
trainer.train()
|
|
|
|
model.save_pretrained("lora-alpaca")
|
|
|
|
print("\n If there's a warning about missing keys above, please disregard :)")
|