AutoGPTQ

An easy-to-use LLMs quantization package with user-friendly apis, based on GPTQ algorithm.

English | 中文

AutoGPTQ for ROCm

This is the ROCm port of AutoGPTQ, built for nightly builds of torch:

python -m pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/rocm5.5

You can obtain prebuilt wheels from artifacts of GitHub Actions.

The porting approach is greatly inspired by WapaMario63/GPTQ-for-LLaMa-ROCm.

I get 43.18 tokens/s for 7B models on a RX 7900 XTX.

News or Update

• 2023-06-05 - (Update) - Integrate with 🤗 peft to use gptq quantized model to train adapters, support LoRA, AdaLoRA, AdaptionPrompt, etc.
• 2023-05-30 - (Update) - Support download/upload quantized model from/to 🤗 Hub.
• 2023-05-27 - (Update) - Support quantization and inference for gpt_bigcode, codegen and RefineWeb/RefineWebModel(falcon) model types.
• 2023-05-04 - (Update) - Support using faster cuda kernel when not desc_act or group_size == -1.

For more histories please turn to here

Performance Comparison

Inference Speed

The result is generated using this script, batch size of input is 1, decode strategy is beam search and enforce the model to generate 512 tokens, speed metric is tokens/s (the larger, the better).

The quantized model is loaded using the setup that can gain the fastest inference speed.

model	GPU	num_beams	fp16	gptq-int4
llama-7b	1xA100-40G	1	18.87	25.53
llama-7b	1xA100-40G	4	68.79	91.30
moss-moon 16b	1xA100-40G	1	12.48	15.25
moss-moon 16b	1xA100-40G	4	OOM	42.67
moss-moon 16b	2xA100-40G	1	06.83	06.78
moss-moon 16b	2xA100-40G	4	13.10	10.80
gpt-j 6b	1xRTX3060-12G	1	OOM	29.55
gpt-j 6b	1xRTX3060-12G	4	OOM	47.36

Perplexity

For perplexity comparison, you can turn to here and here

Installation

Quick Installation

You can install the latest stable release of AutoGPTQ from pip:

pip install auto-gptq

Start from v0.2.0, you can download pre-build wheel that satisfied your environment setup from each version's release assets and install it to skip building stage for the fastest installation speed. For example:

# firstly, cd the directory where the wheel saved, then execute command below
pip install auto_gptq-0.2.0+cu118-cp310-cp310-linux_x86_64.whl # install v0.2.0 auto_gptq pre-build wheel for linux in an environment whose python=3.10 and cuda=11.8

disable cuda extensions

By default, cuda extensions will be installed when torch and cuda is already installed in your machine, if you don't want to use them, using:

BUILD_CUDA_EXT=0 pip install auto-gptq

And to make sure autogptq_cuda is not ever in your virtual environment, run:

pip uninstall autogptq_cuda -y

to support triton speedup

To integrate with triton, using:

warning: currently triton only supports linux; 3-bit quantization is not supported when using triton

pip install auto-gptq[triton]

Install from source

click to see details

Clone the source code:

git clone https://github.com/PanQiWei/AutoGPTQ.git && cd AutoGPTQ

Then, install from source:

pip install .

Like quick installation, you can also set BUILD_CUDA_EXT=0 to disable pytorch extension building.

Use .[triton] if you want to integrate with triton and it's available on your operating system.

Quick Tour

Quantization and Inference

warning: this is just a showcase of the usage of basic apis in AutoGPTQ, which uses only one sample to quantize a much small model, quality of quantized model using such little samples may not good.

Below is an example for the simplest use of auto_gptq to quantize a model and inference after quantization:

from transformers import AutoTokenizer, TextGenerationPipeline
from auto_gptq import AutoGPTQForCausalLM, BaseQuantizeConfig
import logging

logging.basicConfig(
    format="%(asctime)s %(levelname)s [%(name)s] %(message)s", level=logging.INFO, datefmt="%Y-%m-%d %H:%M:%S"
)

pretrained_model_dir = "facebook/opt-125m"
quantized_model_dir = "opt-125m-4bit"

tokenizer = AutoTokenizer.from_pretrained(pretrained_model_dir, use_fast=True)
examples = [
    tokenizer(
        "auto-gptq is an easy-to-use model quantization library with user-friendly apis, based on GPTQ algorithm."
    )
]

quantize_config = BaseQuantizeConfig(
    bits=4,  # quantize model to 4-bit
    group_size=128,  # it is recommended to set the value to 128
    desc_act=False,  # set to False can significantly speed up inference but the perplexity may slightly bad 
)

# load un-quantized model, by default, the model will always be loaded into CPU memory
model = AutoGPTQForCausalLM.from_pretrained(pretrained_model_dir, quantize_config)

# quantize model, the examples should be list of dict whose keys can only be "input_ids" and "attention_mask"
model.quantize(examples)

# save quantized model
model.save_quantized(quantized_model_dir)

# save quantized model using safetensors
model.save_quantized(quantized_model_dir, use_safetensors=True)

# push quantized model to Hugging Face Hub. 
# to use use_auth_token=True, Login first via huggingface-cli login.
# or pass explcit token with: use_auth_token="hf_xxxxxxx"
# (uncomment the following three lines to enable this feature)
# repo_id = f"YourUserName/{quantized_model_dir}"
# commit_message = f"AutoGPTQ model for {pretrained_model_dir}: {quantize_config.bits}bits, gr{quantize_config.group_size}, desc_act={quantize_config.desc_act}"
# model.push_to_hub(repo_id, commit_message=commit_message, use_auth_token=True)

# alternatively you can save and push at the same time
# (uncomment the following three lines to enable this feature)
# repo_id = f"YourUserName/{quantized_model_dir}"
# commit_message = f"AutoGPTQ model for {pretrained_model_dir}: {quantize_config.bits}bits, gr{quantize_config.group_size}, desc_act={quantize_config.desc_act}"
# model.push_to_hub(repo_id, save_dir=quantized_model_dir, use_safetensors=True, commit_message=commit_message, use_auth_token=True)

# load quantized model to the first GPU
model = AutoGPTQForCausalLM.from_quantized(quantized_model_dir, device="cuda:0")

# download quantized model from Hugging Face Hub and load to the first GPU
# model = AutoGPTQForCausalLM.from_quantized(repo_id, device="cuda:0", use_safetensors=True, use_triton=False)

# inference with model.generate
print(tokenizer.decode(model.generate(**tokenizer("auto_gptq is", return_tensors="pt").to(model.device))[0]))

# or you can also use pipeline
pipeline = TextGenerationPipeline(model=model, tokenizer=tokenizer)
print(pipeline("auto-gptq is")[0]["generated_text"])

For more advanced features of model quantization, please reference to this script

Customize Model

Below is an example to extend `auto_gptq` to support `OPT` model, as you will see, it's very easy:

from auto_gptq.modeling import BaseGPTQForCausalLM


class OPTGPTQForCausalLM(BaseGPTQForCausalLM):
    # chained attribute name of transformer layer block
    layers_block_name = "model.decoder.layers"
    # chained attribute names of other nn modules that in the same level as the transformer layer block
    outside_layer_modules = [
        "model.decoder.embed_tokens", "model.decoder.embed_positions", "model.decoder.project_out",
        "model.decoder.project_in", "model.decoder.final_layer_norm"
    ]
    # chained attribute names of linear layers in transformer layer module
    # normally, there are four sub lists, for each one the modules in it can be seen as one operation, 
    # and the order should be the order when they are truly executed, in this case (and usually in most cases), 
    # they are: attention q_k_v projection, attention output projection, MLP project input, MLP project output
    inside_layer_modules = [
        ["self_attn.k_proj", "self_attn.v_proj", "self_attn.q_proj"],
        ["self_attn.out_proj"],
        ["fc1"],
        ["fc2"]
    ]

After this, you can use OPTGPTQForCausalLM.from_pretrained and other methods as shown in Basic.

Evaluation on Downstream Tasks

You can use tasks defined in auto_gptq.eval_tasks to evaluate model's performance on specific down-stream task before and after quantization.

The predefined tasks support all causal-language-models implemented in 🤗 transformers and in this project.

Below is an example to evaluate `EleutherAI/gpt-j-6b` on sequence-classification task using `cardiffnlp/tweet_sentiment_multilingual` dataset:

from functools import partial

import datasets
from transformers import AutoTokenizer, AutoModelForCausalLM, GenerationConfig

from auto_gptq import AutoGPTQForCausalLM, BaseQuantizeConfig
from auto_gptq.eval_tasks import SequenceClassificationTask


MODEL = "EleutherAI/gpt-j-6b"
DATASET = "cardiffnlp/tweet_sentiment_multilingual"
TEMPLATE = "Question:What's the sentiment of the given text? Choices are {labels}.\nText: {text}\nAnswer:"
ID2LABEL = {
    0: "negative",
    1: "neutral",
    2: "positive"
}
LABELS = list(ID2LABEL.values())


def ds_refactor_fn(samples):
    text_data = samples["text"]
    label_data = samples["label"]

    new_samples = {"prompt": [], "label": []}
    for text, label in zip(text_data, label_data):
        prompt = TEMPLATE.format(labels=LABELS, text=text)
        new_samples["prompt"].append(prompt)
        new_samples["label"].append(ID2LABEL[label])

    return new_samples


#  model = AutoModelForCausalLM.from_pretrained(MODEL).eval().half().to("cuda:0")
model = AutoGPTQForCausalLM.from_pretrained(MODEL, BaseQuantizeConfig())
tokenizer = AutoTokenizer.from_pretrained(MODEL)

task = SequenceClassificationTask(
        model=model,
        tokenizer=tokenizer,
        classes=LABELS,
        data_name_or_path=DATASET,
        prompt_col_name="prompt",
        label_col_name="label",
        **{
            "num_samples": 1000,  # how many samples will be sampled to evaluation
            "sample_max_len": 1024,  # max tokens for each sample
            "block_max_len": 2048,  # max tokens for each data block
            # function to load dataset, one must only accept data_name_or_path as input 
            # and return datasets.Dataset
            "load_fn": partial(datasets.load_dataset, name="english"),  
            # function to preprocess dataset, which is used for datasets.Dataset.map, 
            # must return Dict[str, list] with only two keys: [prompt_col_name, label_col_name]
            "preprocess_fn": ds_refactor_fn,  
            # truncate label when sample's length exceed sample_max_len
            "truncate_prompt": False  
        }
    )

# note that max_new_tokens will be automatically specified internally based on given classes
print(task.run())

# self-consistency
print(
    task.run(
        generation_config=GenerationConfig(
            num_beams=3,
            num_return_sequences=3,
            do_sample=True
        )
    )
)

Learn More

tutorials provide step-by-step guidance to integrate auto_gptq with your own project and some best practice principles.

examples provide plenty of example scripts to use auto_gptq in different ways.

Supported Models

you can use model.config.model_type to compare with the table below to check whether the model you use is supported by auto_gptq.

for example, model_type of WizardLM, vicuna and gpt4all are all llama, hence they are all supported by auto_gptq.

model type	quantization	inference	peft-lora	peft-ada-lora	peft-adaption_prompt
bloom	✅	✅	✅	✅
gpt2	✅	✅	✅	✅
gpt_neox	✅	✅	✅	✅	✅requires this peft branch
gptj	✅	✅	✅	✅	✅requires this peft branch
llama	✅	✅	✅	✅	✅
moss	✅	✅	✅	✅	✅requires this peft branch
opt	✅	✅	✅	✅
gpt_bigcode	✅	✅	✅	✅
codegen	✅	✅	✅	✅
falcon(RefinedWebModel/RefinedWeb)	✅	✅	✅	✅

Supported Evaluation Tasks

Currently, auto_gptq supports: LanguageModelingTask, SequenceClassificationTask and TextSummarizationTask; more Tasks will come soon!

Acknowledgement

• Specially thanks Elias Frantar, Saleh Ashkboos, Torsten Hoefler and Dan Alistarh for proposing GPTQ algorithm and open source the code.
• Specially thanks qwopqwop200, for code in this project that relevant to quantization are mainly referenced from GPTQ-for-LLaMa.