SmolVLM2-2.2B-Instruct Open-Source Vision-Language Model - Free Deployment to Assist English Video-to-Text Conversion

Smolvlm2 2.2B Instruct GGUF

Developed by mradermacher

SmolVLM2-2.2B-Instruct is a 2.2B parameter vision-language model focused on video-text-to-text tasks, supporting English.

EnglishOpen Source License:Apache-2.0 #Video Instruction Understanding #Multimodal Quantization #Lightweight Vision-Language

Downloads 235

Release Time : 4/25/2025

Model Overview

This model is a quantized version of a vision-language model, based on HuggingFaceTB/SmolVLM2-2.2B-Instruct, specialized in handling video-related text generation tasks.

Model Features

Multi-dataset Training

The model is trained on multiple high-quality datasets, including Docmatix, LLaVA-OneVision-Data, etc., enhancing its generalization capability.

Diverse Quantized Versions

Offers various quantized versions (e.g., Q2_K, Q3_K_S, etc.), allowing users to choose the appropriate version based on their needs to balance speed and quality.

Efficient Inference

Quantized versions significantly reduce model size and improve inference speed, making it suitable for resource-constrained environments.

Model Capabilities

Video Text Generation

Multimodal Understanding

Instruction Following

Use Cases

Video Content Analysis

Video Caption Generation

Generate descriptive captions based on video content.

Video Question Answering

Answer specific questions about video content.

Education

Educational Video Summarization

Generate concise summaries of educational videos to help students quickly grasp the content.

🚀 SmolVLM2-2.2B-Instruct Quantized Model

This project provides static quantizations of the SmolVLM2-2.2B-Instruct model from Hugging Face. It offers various quantized versions of the model for different use cases.

📚 Documentation

Model Information

Property	Details
Base Model	HuggingFaceTB/SmolVLM2-2.2B-Instruct
Datasets	HuggingFaceM4/the_cauldron, HuggingFaceM4/Docmatix, lmms-lab/LLaVA-OneVision-Data, lmms-lab/M4-Instruct-Data, HuggingFaceFV/finevideo, MAmmoTH-VL/MAmmoTH-VL-Instruct-12M, lmms-lab/LLaVA-Video-178K, orrzohar/Video-STaR, Mutonix/Vript, TIGER-Lab/VISTA-400K, Enxin/MovieChat-1K_train, ShareGPT4Video/ShareGPT4Video
Language	en
Library Name	transformers
License	apache-2.0
Quantized By	mradermacher
Tags	video-text-to-text

About

Static quants of https://huggingface.co/HuggingFaceTB/SmolVLM2-2.2B-Instruct. Weighted/imatrix quants are available at https://huggingface.co/mradermacher/SmolVLM2-2.2B-Instruct-i1-GGUF.

Usage

If you are unsure how to use GGUF files, refer to one of TheBloke's READMEs for more details, including on how to concatenate multi-part files.

Provided Quants

(sorted by size, not necessarily quality. IQ - quants are often preferable over similar sized non - IQ quants)

Link	Type	Size/GB	Notes
GGUF	Q2_K	0.8
GGUF	Q3_K_S	0.9
GGUF	Q3_K_M	1.0	lower quality
GGUF	Q3_K_L	1.1
GGUF	IQ4_XS	1.1
GGUF	Q4_K_S	1.2	fast, recommended
GGUF	Q4_K_M	1.2	fast, recommended
GGUF	Q5_K_S	1.4
GGUF	Q5_K_M	1.4
GGUF	Q6_K	1.6	very good quality
GGUF	Q8_0	2.0	fast, best quality
GGUF	f16	3.7	16 bpw, overkill

Here is a handy graph by ikawrakow comparing some lower - quality quant types (lower is better):

And here are Artefact2's thoughts on the matter: https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9

FAQ / Model Request

See https://huggingface.co/mradermacher/model_requests for some answers to questions you might have and/or if you want some other model quantized.

Thanks

I thank my company, nethype GmbH, for letting me use its servers and providing upgrades to my workstation to enable this work in my free time.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご