qwen3.5-2b-int4-ov
OpenVINO IR export of Qwen/Qwen3.5-2B, quantized to INT4 — 2037 MB.
| Quantization | weight compression on the language model, vision tower left FP16 |
| Lesson | 10 VLM Qwen3.5 |
| Built by | convert/convert_all.py of the ARCademy OpenVINO courseware |
openvino_genai.VLMPipeline(model_dir, device). Export needs transformers==5.2; NPU is not supported for this family upstream.
from huggingface_hub import snapshot_download
model_dir = snapshot_download("circulus/qwen3.5-2b-int4-ov")
- Downloads last month
- 15
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support