granite-3.1-2b-instruct-quantized.w8a8

granite-3.1-2b-instruct-quantized.w8a8

Model Overview

• Model Architecture: granite-3.1-2b-instruct
  • Input: Text
  • Output: Text
• Model Optimizations:
  • Weight quantization: INT8
  • Activation quantization: INT8
• 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 61.68 on the OpenLLM benchmark (version 1), whereas the unquantized model achieves 61.98.

Model Optimizations

This model was obtained by quantizing the weights and activations of ibm-granite/granite-3.1-2b-instruct to INT8 data type, ready for inference with vLLM >= 0.5.2. This optimization reduces the number of bits per parameter from 16 to 8, reducing the disk size and GPU memory requirements by approximately 50%. Only the weights and activations 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.w8a8"
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.

<details> <summary>Model Creation Code</summary>

python quantize.py --model_path ibm-granite/granite-3.1-2b-instruct --quant_path "output_dir/granite-3.1-2b-instruct-quantized.w8a8" --calib_size 2048 --dampening_frac 0.01 --observer mse

from datasets import load_dataset
from transformers import AutoTokenizer, AutoModelForCausalLM
from llmcompressor.modifiers.quantization import GPTQModifier
from llmcompressor.modifiers.smoothquant import SmoothQuantModifier
from llmcompressor.transformers import SparseAutoModelForCausalLM, 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")
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)

ignore=["lm_head"]
mappings=[
    [["re:.*q_proj", "re:.*k_proj", "re:.*v_proj"], "re:.*input_layernorm"],
    [["re:.*gate_proj", "re:.*up_proj"], "re:.*post_attention_layernorm"],
    [["re:.*down_proj"], "re:.*up_proj"]
]

recipe = [
    SmoothQuantModifier(smoothing_strength=0.7, ignore=ignore, mappings=mappings),
    GPTQModifier(
        targets=["Linear"],
        ignore=["lm_head"],
        scheme="W8A8",
        dampening_frac=args.dampening_frac,
        observer=args.observer,
    )
]
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)

</details>

Evaluation

The model was evaluated on OpenLLM Leaderboard V1, OpenLLM Leaderboard V2 and on HumanEval, using the following commands:

<details> <summary>Evaluation Commands</summary>

OpenLLM Leaderboard V1:

lm_eval \
  --model vllm \
  --model_args pretrained="neuralmagic/granite-3.1-2b-instruct-quantized.w8a8",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.w8a8",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.w8a8 \
  --bs 16 \
  --temperature 0.2 \
  --n_samples 50 \
  --root "." \
  --dataset humaneval

Sanitization

python3 evalplus/sanitize.py \
  humaneval/neuralmagic--granite-3.1-2b-instruct-quantized.w8a8_vllm_temp_0.2

Evaluation

evalplus.evaluate \
  --dataset humaneval \
  --samples humaneval/neuralmagic--granite-3.1-2b-instruct-quantized.w8a8_vllm_temp_0.2-sanitized

</details>

Accuracy

<table> <thead> <tr> <th>Category</th> <th>Metric</th> <th>ibm-granite/granite-3.1-2b-instruct</th> <th>neuralmagic/granite-3.1-2b-instruct-quantized.w8a8</th> <th>Recovery (%)</th> </tr> </thead> <tbody> <!-- OpenLLM Leaderboard V1 --> <tr> <td rowspan="7"><b>OpenLLM V1</b></td> <td>ARC-Challenge (Acc-Norm, 25-shot)</td> <td>55.63</td> <td>55.12</td> <td>99.08</td> </tr> <tr> <td>GSM8K (Strict-Match, 5-shot)</td> <td>60.96</td> <td>60.58</td> <td>99.38</td> </tr> <tr> <td>HellaSwag (Acc-Norm, 10-shot)</td> <td>75.21</td> <td>74.60</td> <td>99.19</td> </tr> <tr> <td>MMLU (Acc, 5-shot)</td> <td>54.38</td> <td>54.12</td> <td>99.52</td> </tr> <tr> <td>TruthfulQA (MC2, 0-shot)</td> <td>55.93</td> <td>54.87</td> <td>98.10</td> </tr> <tr> <td>Winogrande (Acc, 5-shot)</td> <td>69.67</td> <td>70.80</td> <td>101.62</td> </tr> <tr> <td><b>Average Score</b></td> <td><b>61.98</b></td> <td><b>61.68</b></td> <td><b>99.51</b></td> </tr> <!-- OpenLLM Leaderboard V2 --> <tr> <td rowspan="7"><b>OpenLLM V2</b></td> <td>IFEval (Inst Level Strict Acc, 0-shot)</td> <td>67.99</td> <td>67.03</td> <td>98.59</td> </tr> <tr> <td>BBH (Acc-Norm, 3-shot)</td> <td>44.11</td> <td>43.53</td> <td>98.69</td> </tr> <tr> <td>Math-Hard (Exact-Match, 4-shot)</td> <td>8.66</td> <td>8.04</td> <td>92.89</td> </tr> <tr> <td>GPQA (Acc-Norm, 0-shot)</td> <td>28.30</td> <td>27.60</td> <td>97.52</td> </tr> <tr> <td>MUSR (Acc-Norm, 0-shot)</td> <td>35.12</td> <td>34.58</td> <td>98.46</td> </tr> <tr> <td>MMLU-Pro (Acc, 5-shot)</td> <td>26.87</td> <td>26.89</td> <td>100.07</td> </tr> <tr> <td><b>Average Score</b></td> <td><b>35.17</b></td> <td><b>34.61</b></td> <td><b>98.40</b></td> </tr> <!-- HumanEval --> <tr> <td rowspan="2"><b>HumanEval</b></td> <td>HumanEval Pass@1</td> <td>53.40</td> <td>54.90</td> <td><b>102.81</b></td> </tr> </tbody> </table>

Inference Performance

This model achieves up to 1.4x speedup in single-stream deployment and up to 1.1x speedup in multi-stream asynchronous deployment, depending on hardware and use-case scenario. The following performance benchmarks were conducted with vLLM version 0.6.6.post1, and GuideLLM.

<details> <summary>Benchmarking Command</summary>

guidellm --model neuralmagic/granite-3.1-2b-instruct-quantized.w8a8 --target "http://localhost:8000/v1" --data-type emulated --data "prompt_tokens=<prompt_tokens>,generated_tokens=<generated_tokens>" --max seconds 360 --backend aiohttp_server

</details>

Single-stream performance (measured with vLLM version 0.6.6.post1)

<table> <tr> <td></td> <td></td> <td></td> <th style="text-align: center;" colspan="7" >Latency (s)</th> </tr> <tr> <th>GPU class</th> <th>Model</th> <th>Speedup</th> <th>Code Completion<br>prefill: 256 tokens<br>decode: 1024 tokens</th> <th>Docstring Generation<br>prefill: 768 tokens<br>decode: 128 tokens</th> <th>Code Fixing<br>prefill: 1024 tokens<br>decode: 1024 tokens</th> <th>RAG<br>prefill: 1024 tokens<br>decode: 128 tokens</th> <th>Instruction Following<br>prefill: 256 tokens<br>decode: 128 tokens</th> <th>Multi-turn Chat<br>prefill: 512 tokens<br>decode: 256 tokens</th> <th>Large Summarization<br>prefill: 4096 tokens<br>decode: 512 tokens</th> </tr> <tr> <td style="vertical-align: middle;" rowspan="3" >A5000</td> <td>granite-3.1-2b-instruct</td> <td></td> <td>10.9</td> <td>1.4</td> <td>11.0</td> <td>1.5</td> <td>1.4</td> <td>2.8</td> <td>6.1</td> </tr> <tr> <td>granite-3.1-2b-instruct-quantized.w8a8<br>(this model)</td> <td>1.37</td> <td>7.9</td> <td>1.0</td> <td>8.0</td> <td>1.1</td> <td>1.0</td> <td>2.0</td> <td>4.7</td> </tr> <tr> <td>granite-3.1-2b-instruct-quantized.w4a16</td> <td>1.94</td> <td>5.4</td> <td>0.7</td> <td>5.5</td> <td>0.8</td> <td>0.7</td> <td>1.4</td> <td>3.4</td> </tr> <tr> <td style="vertical-align: middle;" rowspan="3" >A6000</td> <td>granite-3.1-2b-instruct</td> <td></td> <td>9.8</td> <td>1.3</td> <td>10.0</td> <td>1.3</td> <td>1.3</td> <td>2.6</td> <td>5.4</td> </tr> <tr> <td>granite-3.1-2b-instruct-quantized.w8a8<br>(this model)</td> <td>1.31</td> <td>7.8</td> <td>1.0</td> <td>7.6</td> <td>1.0</td> <td>0.9</td> <td>1.9</td> <td>4.5</td> </tr> <tr> <td>granite-3.1-2b-instruct-quantized.w4a16</td> <td>1.87</td> <td>5.1</td> <td>0.7</td> <td>5.2</td> <td>0.7</td> <td>0.7</td> <td>1.3</td> <td>3.1</td> </tr> </table>

Multi-stream asynchronous performance (measured with vLLM version 0.6.6.post1)

<table> <tr> <td></td> <td></td> <td></td> <th style="text-align: center;" colspan="7" >Maximum Throughput (Queries per Second)</th> </tr> <tr> <th>GPU class</th> <th>Model</th> <th>Speedup</th> <th>Code Completion<br>prefill: 256 tokens<br>decode: 1024 tokens</th> <th>Docstring Generation<br>prefill: 768 tokens<br>decode: 128 tokens</th> <th>Code Fixing<br>prefill: 1024 tokens<br>decode: 1024 tokens</th> <th>RAG<br>prefill: 1024 tokens<br>decode: 128 tokens</th> <th>Instruction Following<br>prefill: 256 tokens<br>decode: 128 tokens</th> <th>Multi-turn Chat<br>prefill: 512 tokens<br>decode: 256 tokens</th> <th>Large Summarization<br>prefill: 4096 tokens<br>decode: 512 tokens</th> </tr> <tr> <td style="vertical-align: middle;" rowspan="3" >A5000</td> <td>granite-3.1-2b-instruct</td> <td></td> <td>2.9</td> <td>10.2</td> <td>1.8</td> <td>8.2</td> <td>19.3</td> <td>9.1</td> <td>1.3</td> </tr> <tr> <td>granite-3.1-2b-instruct-quantized.w8a8<br>(this model)</td> <td>1.13</td> <td>3.1</td> <td>12.1</td> <td>2.0</td> <td>9.6</td> <td>22.2</td> <td>10.2</td> <td>1.4</td> </tr> <tr> <td>granite-3.1-2b-instruct-quantized.w4a16</td> <td>0.98</td> <td>2.8</td> <td>10.0</td> <td>1.8</td> <td>8.1</td> <td>18.6</td> <td>9.0</td> <td>1.2</td> </tr> <tr> <td style="vertical-align: middle;" rowspan="3" >A6000</td> <td>granite-3.1-2b-instruct</td> <td></td> <td>3.7</td> <td>12.4</td> <td>2.4</td> <td>10.3</td> <td>23.6</td> <td>11.0</td> <td>1.6</td> </tr> <tr> <td>granite-3.1-2b-instruct-quantized.w8a8<br>(this model)</td> <td>1.12</td> <td>3.6</td> <td>14.4</td> <td>2.7</td> <td>12.0</td> <td>28.3</td> <td>12.9</td> <td>1.7</td> </tr> <tr> <td>granite-3.1-2b-instruct-quantized.w4a16</td> <td>0.95</td> <td>3.7</td> <td>11.4</td> <td>2.5</td> <td>9.8</td> <td>22.1</td> <td>10.4</td> <td>1.4</td> </tr> </table>