granite-3.1-2b-instruct-quantized.w4a16
Model Overview
- Model Architecture: granite-3.1-2b-instruct
- Input: Text
- Output: Text
- Model Optimizations:
- Weight quantization: INT4
- Activation quantization: INT4
- Release Date: 1/8/2025
- Version: 1.0
- Model Developers: Neural Magic
Quantized version of ibm-granite/granite-3.1-2b-instruct. It achieves an average score of 61.54 on the OpenLLM benchmark (version 1), whereas the unquantized model achieves 61.98.
Model Optimizations
This model was obtained by quantizing the weights of ibm-granite/granite-3.1-2b-instruct to INT4 data type, ready for inference with vLLM >= 0.5.2. This optimization reduces the number of bits per parameter from 16 to 4, reducing the disk size and GPU memory requirements by approximately 75%. Only the weights of the linear operators within transformers blocks are quantized.
Deployment
Use with vLLM
This model can be deployed efficiently using the vLLM backend, as shown in the example below.
from transformers import AutoTokenizer
from vllm import LLM, SamplingParams
max_model_len, tp_size = 4096, 1
model_name = "neuralmagic/granite-3.1-2b-instruct-quantized.w4a16"
tokenizer = AutoTokenizer.from_pretrained(model_name)
llm = LLM(model=model_name, tensor_parallel_size=tp_size, max_model_len=max_model_len, trust_remote_code=True)
sampling_params = SamplingParams(temperature=0.3, max_tokens=256, stop_token_ids=[tokenizer.eos_token_id])
messages_list = [
[{"role": "user", "content": "Who are you? Please respond in pirate speak!"}],
]
prompt_token_ids = [tokenizer.apply_chat_template(messages, add_generation_prompt=True) for messages in messages_list]
outputs = llm.generate(prompt_token_ids=prompt_token_ids, sampling_params=sampling_params)
generated_text = [output.outputs[0].text for output in outputs]
print(generated_text)
vLLM also supports OpenAI-compatible serving. See the documentation for more details.
Creation
This model was created with llm-compressor by running the code snippet below.
python quantize.py --model_path ibm-granite/granite-3.1-2b-instruct --quant_path "output_dir/granite-3.1-2b-instruct-quantized.w4a16" --calib_size 1024 --dampening_frac 0.01 --observer mse --group_size 64
from datasets import load_dataset
from transformers import AutoTokenizer, AutoModelForCausalLM
from llmcompressor.modifiers.quantization import GPTQModifier
from llmcompressor.transformers import oneshot, apply
import argparse
from compressed_tensors.quantization import QuantizationScheme, QuantizationArgs, QuantizationType, QuantizationStrategy
parser = argparse.ArgumentParser()
parser.add_argument('--model_path', type=str)
parser.add_argument('--quant_path', type=str)
parser.add_argument('--calib_size', type=int, default=256)
parser.add_argument('--dampening_frac', type=float, default=0.1)
parser.add_argument('--observer', type=str, default="minmax")
parser.add_argument('--group_size', type=int, default="128")
args = parser.parse_args()
model = AutoModelForCausalLM.from_pretrained(
args.model_path,
device_map="auto",
torch_dtype="auto",
use_cache=False,
trust_remote_code=True,
)
tokenizer = AutoTokenizer.from_pretrained(args.model_path)
NUM_CALIBRATION_SAMPLES = args.calib_size
DATASET_ID = "neuralmagic/LLM_compression_calibration"
DATASET_SPLIT = "train"
ds = load_dataset(DATASET_ID, split=DATASET_SPLIT)
ds = ds.shuffle(seed=42).select(range(NUM_CALIBRATION_SAMPLES))
def preprocess(example):
return {"text": example["text"]}
ds = ds.map(preprocess)
def tokenize(sample):
return tokenizer(
sample["text"],
padding=False,
truncation=False,
add_special_tokens=True,
)
ds = ds.map(tokenize, remove_columns=ds.column_names)
recipe = [
GPTQModifier(
targets=["Linear"],
ignore=["lm_head"],
scheme="w4a16",
dampening_frac=args.dampening_frac,
observer=args.observer,
group_size=args.group_size
)
]
oneshot(
model=model,
dataset=ds,
recipe=recipe,
num_calibration_samples=args.calib_size,
max_seq_length=8196,
)
# Save to disk compressed.
model.save_pretrained(quant_path, save_compressed=True)
tokenizer.save_pretrained(quant_path)
Evaluation
The model was evaluated on OpenLLM Leaderboard V1, OpenLLM Leaderboard V2 and on HumanEval, using the following commands:
OpenLLM Leaderboard V1:
lm_eval \
--model vllm \
--model_args pretrained="neuralmagic/granite-3.1-2b-instruct-quantized.w4a16",dtype=auto,add_bos_token=True,max_model_len=4096,tensor_parallel_size=1,gpu_memory_utilization=0.8,enable_chunked_prefill=True,trust_remote_code=True \
--tasks openllm \
--write_out \
--batch_size auto \
--output_path output_dir \
--show_config
OpenLLM Leaderboard V2:
lm_eval \
--model vllm \
--model_args pretrained="neuralmagic/granite-3.1-2b-instruct-quantized.w4a16",dtype=auto,add_bos_token=True,max_model_len=4096,tensor_parallel_size=1,gpu_memory_utilization=0.8,enable_chunked_prefill=True,trust_remote_code=True \
--tasks leaderboard \
--write_out \
--batch_size auto \
--output_path output_dir \
--show_config
HumanEval
Generation
python3 codegen/generate.py \
--model neuralmagic/granite-3.1-2b-instruct-quantized.w4a16 \
--bs 16 \
--temperature 0.2 \
--n_samples 50 \
--root "." \
--dataset humaneval
Sanitization
python3 evalplus/sanitize.py \
humaneval/neuralmagic--granite-3.1-2b-instruct-quantized.w4a16_vllm_temp_0.2
Evaluation
evalplus.evaluate \
--dataset humaneval \
--samples humaneval/neuralmagic--granite-3.1-2b-instruct-quantized.w4a16_vllm_temp_0.2-sanitized
Accuracy
Inference Performance
This model achieves up to 1.9x speedup in single-stream deployment, depending on hardware and use-case scenario. The following performance benchmarks were conducted with vLLM version 0.6.6.post1, and GuideLLM.
guidellm --model neuralmagic/granite-3.1-2b-instruct-quantized.w4a16 --target "http://localhost:8000/v1" --data-type emulated --data "prompt_tokens=<prompt_tokens>,generated_tokens=<generated_tokens>" --max seconds 360 --backend aiohttp_server