Add LLaMA 8-bit support

This commit is contained in:
oobabooga 2023-03-04 13:28:42 -03:00
parent c93f1fa99b
commit bd8aac8fa4
2 changed files with 137 additions and 4 deletions

125
modules/LLaMA_8bit.py Normal file
View File

@ -0,0 +1,125 @@
# Copyright (c) Meta Platforms, Inc. and affiliates.
# This software may be used and distributed according to the terms of the GNU General Public License version 3.
from typing import Tuple
import os
import sys
import torch
import fire
import time
import json
from pathlib import Path
from fairscale.nn.model_parallel.initialize import initialize_model_parallel
from repositories.llama_int8.llama import ModelArgs, Transformer, Tokenizer, LLaMA
def setup_model_parallel() -> Tuple[int, int]:
local_rank = int(os.environ.get("LOCAL_RANK", -1))
world_size = int(os.environ.get("WORLD_SIZE", -1))
torch.distributed.init_process_group("nccl")
initialize_model_parallel(world_size)
torch.cuda.set_device(local_rank)
# seed must be the same in all processes
torch.manual_seed(1)
return local_rank, world_size
def load(
ckpt_dir: str,
tokenizer_path: str,
max_seq_len: int,
max_batch_size: int,
) -> LLaMA:
start_time = time.time()
checkpoints = sorted(Path(ckpt_dir).glob("*.pth"))
with open(Path(ckpt_dir) / "params.json", "r") as f:
params = json.loads(f.read())
model_args: ModelArgs = ModelArgs(
max_seq_len=max_seq_len, max_batch_size=max_batch_size, **params
)
tokenizer = Tokenizer(model_path=tokenizer_path)
model_args.vocab_size = tokenizer.n_words
# torch.set_default_tensor_type(torch.cuda.HalfTensor)
torch.set_default_tensor_type(torch.HalfTensor)
print("Creating transformer")
model = Transformer(model_args)
print("Transformer created")
key_to_dim = {
"w1": 0,
"w2": -1,
"w3": 0,
"wo": -1,
"wq": 0,
"wk": 0,
"wv": 0,
"output": 0,
"tok_embeddings": -1,
"ffn_norm": None,
"attention_norm": None,
"norm": None,
"rope": None,
}
# ?
torch.set_default_tensor_type(torch.FloatTensor)
# load the state dict incrementally, to avoid memory problems
for i, ckpt in enumerate(checkpoints):
print(f"Loading checkpoint {i}")
checkpoint = torch.load(ckpt, map_location="cpu")
for parameter_name, parameter in model.named_parameters():
short_name = parameter_name.split(".")[-2]
if key_to_dim[short_name] is None and i == 0:
parameter.data = checkpoint[parameter_name]
elif key_to_dim[short_name] == 0:
size = checkpoint[parameter_name].size(0)
parameter.data[size * i : size * (i + 1), :] = checkpoint[
parameter_name
]
elif key_to_dim[short_name] == -1:
size = checkpoint[parameter_name].size(-1)
parameter.data[:, size * i : size * (i + 1)] = checkpoint[
parameter_name
]
del checkpoint
# model.load_state_dict(checkpoint, strict=False)
model.quantize()
generator = LLaMA(model, tokenizer)
print(f"Loaded in {time.time() - start_time:.2f} seconds")
return generator
class LLaMAModel_8bit:
def __init__(self):
pass
@classmethod
def from_pretrained(self, path, max_seq_len=2048, max_batch_size=1):
tokenizer_path = path / "tokenizer.model"
path = os.path.abspath(path)
tokenizer_path = os.path.abspath(tokenizer_path)
generator = load(path, tokenizer_path, max_seq_len, max_batch_size)
result = self()
result.pipeline = generator
return result
def generate(self, prompt, token_count=512, temperature=0.8, top_p=0.95):
results = self.pipeline.generate(
[prompt], max_gen_len=token_count, temperature=temperature, top_p=top_p
)
return results[0]

View File

@ -88,6 +88,14 @@ def load_model(model_name):
# LLaMA model (not on HuggingFace)
elif shared.is_LLaMA:
if shared.args.load_in_8bit:
import modules.LLaMA_8bit
from modules.LLaMA_8bit import LLaMAModel_8bit
model = LLaMAModel_8bit.from_pretrained(Path(f'models/{model_name}'))
return model, None
else:
import modules.LLaMA
from modules.LLaMA import LLaMAModel