History

Chen, Zhentao cb228c70ea Add harness nightly (#9552 ) * modify output_path as a directory * schedule nightly at 21 on Friday * add tasks and models for nightly * add accuracy regression * comment out if to test * mixed fp4 * for test * add missing delimiter * remove comma * fixed golden results * add mixed 4 golden result * add more options * add mistral results * get golden result of stable lm * move nightly scripts and results to test folder * add license * add fp8 stable lm golden * run on all available devices * trigger only when ready for review * fix new line * update golden * add mistral		2023-12-01 14:16:35 +08:00
..
all-in-one	LLM: quick fix benchmark (#9509 )	2023-11-22 10:19:57 +08:00
harness	Add harness nightly (#9552 )	2023-12-01 14:16:35 +08:00
benchmark_util.py	LLM: update benchmark_util.py for beam search (#9126 )	2023-10-11 09:41:53 +08:00
README.md	update benchmark_utils readme (#8925 )	2023-09-08 10:30:26 +08:00

README.md

Benchmark tool for transformers int4 (separate 1st token and rest)

benchmark_util.py is used to provide a simple benchmark tool for transformer int4 model to calculate 1st token performance and the rest on CPU and GPU.

CPU Usage

Just put this file into your benchmark directory, and then wrap your transformer int4 model with BenchmarkWrapper (model = BenchmarkWrapper(model)). Take chatglm-6b as an example:

import torch
from bigdl.llm.transformers import AutoModel
from transformers import AutoTokenizer
from benchmark_util import BenchmarkWrapper

model_path ='THUDM/chatglm-6b'
model = AutoModel.from_pretrained(model_path, trust_remote_code=True, load_in_4bit=True)
model = BenchmarkWrapper(model, do_print=True)
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
prompt = "今天睡不着怎么办"
 
with torch.inference_mode():
    input_ids = tokenizer.encode(prompt, return_tensors="pt")
    output = model.generate(input_ids, do_sample=False, max_new_tokens=32)
    output_str = tokenizer.decode(output[0], skip_special_tokens=True)

Output will be like:

=========First token cost xx.xxxxs=========
=========Last token cost average xx.xxxxs (31 tokens in all)=========

GPU Usage

Just put this file into your benchmark directory, and then wrap your transformer int4 model with BenchmarkWrapper (model = BenchmarkWrapper(model)). Take chatglm-6b as an example:

import torch
import intel_extension_for_pytorch as ipex
from bigdl.llm.transformers import AutoModel
from transformers import AutoTokenizer
from benchmark_util import BenchmarkWrapper

model_path ='THUDM/chatglm-6b'
model = AutoModel.from_pretrained(model_path, trust_remote_code=True, load_in_4bit=True)
model = model.to('xpu')
model = BenchmarkWrapper(model, do_print=True)
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
prompt = "今天睡不着怎么办"
 
with torch.inference_mode():
    # wamup two times as use ipex
    for i in range(2):
        input_ids = tokenizer.encode(prompt, return_tensors="pt").to('xpu')
        output = model.generate(input_ids, do_sample=False, max_new_tokens=32)
        output_str = tokenizer.decode(output[0], skip_special_tokens=True)
    # collect performance data now
    for i in range(5):
        input_ids = tokenizer.encode(prompt, return_tensors="pt").to('xpu')
        output = model.generate(input_ids, do_sample=False, max_new_tokens=32)
        output_str = tokenizer.decode(output[0], skip_special_tokens=True)

Output will be like:

=========First token cost xx.xxxxs=========
=========Last token cost average xx.xxxxs (31 tokens in all)=========