abhinavkulkarni/tiiuae-falcon-40b-instruct-w4-g128-awq

Falcon-40b-Instruct (4-bit 128g AWQ Quantized)

Falcon-40b-Instruct is a 40B parameters causal decoder-only model built by TII based on Falcon-40B and finetuned on a mixture of chat/instruct datasets.

This model is a 4-bit 128 group size AWQ quantized model. For more information about AWQ quantization, please click here.

Model Date

July 5, 2023

Model License

Please refer to original Falcon model license (link).

Please refer to the AWQ quantization license (link).

CUDA Version

This model was successfully tested on CUDA driver v530.30.02 and runtime v11.7 with Python v3.10.11. Please note that AWQ requires NVIDIA GPUs with compute capability of 8.0 or higher.

How to Use

git clone https://github.com/mit-han-lab/llm-awq \
&& cd llm-awq \
&& git checkout f084f40bd996f3cf3a0633c1ad7d9d476c318aaa \
&& pip install -e . \
&& cd awq/kernels \
&& python setup.py install

import time
import torch
from awq.quantize.quantizer import real_quantize_model_weight
from awq.utils.utils import simple_dispatch_model
from transformers import AutoModelForCausalLM, AutoConfig, AutoTokenizer, TextStreamer, TextStreamer
from accelerate import init_empty_weights, load_checkpoint_and_dispatch
from huggingface_hub import snapshot_download

model_name = "abhinavkulkarni/tiiuae-falcon-40b-instruct-w4-g128-awq"

# Config
config = AutoConfig.from_pretrained(model_name, trust_remote_code=True)

# Tokenizer
try:
    tokenizer = AutoTokenizer.from_pretrained(config.tokenizer_name, trust_remote_code=True)
except:
    tokenizer = AutoTokenizer.from_pretrained(model_name, use_fast=False, trust_remote_code=True)
streamer = TextStreamer(tokenizer, skip_special_tokens=True)
streamer = TextStreamer(tokenizer, skip_special_tokens=True)

# Model
w_bit = 4
q_config = {
    "zero_point": True,
    "q_group_size": 128,
}

# Initialize empty model
with init_empty_weights():
    model = AutoModelForCausalLM.from_config(config=config, 
                                                 torch_dtype=torch.float16, trust_remote_code=True)
real_quantize_model_weight(model, w_bit=w_bit, q_config=q_config, init_only=True)
model.tie_weights()
model.tie_weights()

# Infer device_map
device_map = infer_auto_device_map(
    model,
    no_split_module_classes=[
        "OPTDecoderLayer", "LlamaDecoderLayer", "BloomBlock", "MPTBlock", "DecoderLayer"]
)

# Load weights
load_checkpoint_in_model(
    model,
    checkpoint=snapshot_download(model_name),
    device_map=device_map,
    offload_state_dict=True,
)

model = simple_dispatch_model(model, device_map=device_map)

# Inference
prompt = f'''What is the difference between nuclear fusion and fission?
###Response:'''

input_ids = tokenizer(prompt, return_tensors='pt').input_ids.cuda()
output = model.generate(
    inputs=input_ids, 
    temperature=0.7,
    max_new_tokens=512,
    top_p=0.15,
    top_k=0,
    repetition_penalty=1.1,
    eos_token_id=tokenizer.eos_token_id,
streamer=streamer,
    streamer=streamer,
)

Evaluation

This evaluation was done using LM-Eval.

Falcon-40b-Instruct

Task	Version	Metric	Value		Stderr
wikitext	1	word_perplexity	8.8219
		byte_perplexity	1.5025
		bits_per_byte	0.5874

Falcon-40b-Instruct (4-bit 128-group AWQ)

Task	Version	Metric	Value		Stderr
wikitext	1	word_perplexity	8.9237
		byte_perplexity	1.5058
		bits_per_byte	0.5905

Acknowledgements

Paper coming soon 😊. In the meanwhile, you can use the following information to cite:

@article{falcon40b,
  title={{Falcon-40B}: an open large language model with state-of-the-art performance},
  author={Almazrouei, Ebtesam and Alobeidli, Hamza and Alshamsi, Abdulaziz and Cappelli, Alessandro and Cojocaru, Ruxandra and Debbah, Merouane and Goffinet, Etienne and Heslow, Daniel and Launay, Julien and Malartic, Quentin and Noune, Badreddine and Pannier, Baptiste and Penedo, Guilherme},
  year={2023}
}

The model was quantized with AWQ technique. If you find AWQ useful or relevant to your research, please kindly cite the paper:

@article{lin2023awq,
  title={AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration},
  author={Lin, Ji and Tang, Jiaming and Tang, Haotian and Yang, Shang and Dang, Xingyu and Han, Song},
  journal={arXiv},
  year={2023}
}