Qwen-2.5-3B-Instruct-ov-INT8

• Model creator: Qwen
• Original model: Qwen2.5-3B-Instruct

Description

This is Qwen2.5-3B-Instruct model converted to the OpenVINO™ IR (Intermediate Representation) format with weights compressed to INT8 by NNCF.

Quantization Parameters

Weight compression was performed using nncf.compress_weights with the following parameters:

• mode: int8_asym
• ratio: 0.8
• group_size: 128

For more information on quantization, check the OpenVINO model optimization guide.

Compatibility

The provided OpenVINO™ IR model is compatible with:

• OpenVINO version 2024.4.0 and higher
• Optimum Intel 1.19.0 and higher

Prompt Template

<|im_start|>system
You are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>
<|im_start|>user
{input}<|im_end|>

Running Model Inference with Optimum Intel

1. Install packages required for using Optimum Intel integration with the OpenVINO backend:

pip install optimum[openvino]

2. Run model inference:

from transformers import AutoTokenizer
from optimum.intel.openvino import OVModelForCausalLM

model_id = "srang992/Qwen-2.5-3B-Instruct-ov-INT8"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = OVModelForCausalLM.from_pretrained(model_id)

inputs = tokenizer("What is OpenVINO?", return_tensors="pt")

outputs = model.generate(**inputs, max_length=200)
text = tokenizer.batch_decode(outputs)[0]
print(text)

For more examples and possible optimizations, refer to the OpenVINO Large Language Model Inference Guide.

Running Model Inference with OpenVINO GenAI

1. Install packages required for using OpenVINO GenAI.

pip install openvino-genai huggingface_hub

2. Download model from HuggingFace Hub

import huggingface_hub as hf_hub

model_id = "srang992/Qwen-2.5-3B-Instruct-ov-INT8"
model_path = "Qwen-2.5-3B-Instruct-ov-INT8"

hf_hub.snapshot_download(model_id, local_dir=model_path)

3. Run model inference:

import openvino_genai as ov_genai

device = "CPU"
pipe = ov_genai.LLMPipeline(model_path, device)
print(pipe.generate("What is OpenVINO?", max_length=200))