101 lines
2.9 KiB
Python
101 lines
2.9 KiB
Python
import pytest
|
|
import torch
|
|
import math
|
|
|
|
from transformers import (
|
|
AutoConfig,
|
|
AutoModelForCausalLM,
|
|
AutoTokenizer,
|
|
BitsAndBytesConfig,
|
|
GenerationConfig,
|
|
set_seed,
|
|
|
|
)
|
|
import transformers
|
|
|
|
|
|
def get_4bit_config():
|
|
return BitsAndBytesConfig(
|
|
load_in_4bit=True,
|
|
load_in_8bit=False,
|
|
llm_int8_threshold=6.0,
|
|
llm_int8_has_fp16_weight=False,
|
|
bnb_4bit_compute_dtype=torch.float16,
|
|
bnb_4bit_use_double_quant=True,
|
|
bnb_4bit_quant_type='nf4',
|
|
)
|
|
|
|
|
|
def get_model(model_name_or_path='huggyllama/llama-7b', bnb_config=get_4bit_config()):
|
|
model = AutoModelForCausalLM.from_pretrained(
|
|
model_name_or_path,
|
|
quantization_config=bnb_config,
|
|
max_memory={0:'48GB'},
|
|
device_map='auto'
|
|
).eval()
|
|
|
|
return model
|
|
|
|
def get_prompt_for_generation_eval(text, add_roles=True):
|
|
description = (
|
|
"A chat between a curious human and an artificial intelligence assistant. "
|
|
"The assistant gives helpful, detailed, and polite answers to the user's questions."
|
|
)
|
|
if add_roles:
|
|
prompt = f'{description} ### Human: {text} ### Assistant:'
|
|
else:
|
|
prompt = f'{description} {text}'
|
|
return prompt
|
|
|
|
def generate(model, tokenizer, text, generation_config, prompt_func=get_prompt_for_generation_eval):
|
|
text = prompt_func(text)
|
|
inputs = tokenizer(text, return_tensors="pt").to('cuda:0')
|
|
outputs = model.generate(inputs=inputs['input_ids'], generation_config=generation_config)
|
|
return tokenizer.decode(outputs[0], skip_special_tokens=True)
|
|
|
|
name_or_path = 'huggyllama/llama-7b'
|
|
#name_or_path = 'AI-Sweden/gpt-sw3-126m'
|
|
|
|
@pytest.fixture(scope='session')
|
|
def model():
|
|
bnb_config = get_4bit_config()
|
|
bnb_config.bnb_4bit_compute_dtype=torch.float32
|
|
bnb_config.load_in_4bit=True
|
|
model = get_model(name_or_path)
|
|
print('')
|
|
return model
|
|
|
|
@pytest.fixture(scope='session')
|
|
def tokenizer():
|
|
tokenizer = transformers.AutoTokenizer.from_pretrained(name_or_path)
|
|
return tokenizer
|
|
|
|
@pytest.mark.parametrize("dtype", [torch.float16, torch.bfloat16, torch.float32], ids=['fp16', 'bf16', 'fp32'])
|
|
def test_pi(model, tokenizer, dtype):
|
|
|
|
generation_config = transformers.GenerationConfig(
|
|
max_new_tokens=128,
|
|
do_sample=True,
|
|
top_p=0.9,
|
|
temperature=0.7,
|
|
)
|
|
generation_config.max_new_tokens = 50
|
|
|
|
|
|
#text = 'Please write down the first 50 digits of pi.'
|
|
#text = get_prompt_for_generation_eval(text)
|
|
#text += ' Sure, here the first 50 digits of pi: 3.14159'
|
|
text = '3.14159'
|
|
model.config.quantization_config.bnb_4bit_compute_dtype = dtype
|
|
|
|
inputs = tokenizer(text, return_tensors="pt").to('cuda:0')
|
|
outputs = model.generate(inputs=inputs['input_ids'], generation_config=generation_config)
|
|
textout = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
|
print('')
|
|
print(textout)
|
|
print(math.pi)
|
|
|
|
assert textout[:len(str(math.pi))] == str(math.pi)
|
|
|
|
|