LLM EXPLORER 57,252 MODELS INDEXED

SmolVLM2 2.2B Instruct by HuggingFaceTB

By HuggingFaceTB · 293476 downloads
HF Repo ↗ Copy URL

SmolVLM2 2.2B Instruct is an open-source language model by HuggingFaceTB. Features: 2.2b LLM, VRAM: 9GB, License: apache-2.0, Instruction-Based, LLM Explorer Score: 0.27.

  Arxiv:2504.05299 Base model:finetune:huggingfac... Base model:huggingfacetb/smolv...   Conversational Dataset:enxin/moviechat-1k tra... Dataset:huggingfacefv/finevide...   Dataset:huggingfacem4/docmatix Dataset:huggingfacem4/the caul... Dataset:lmms-lab/llava-onevisi... Dataset:lmms-lab/llava-video-1... Dataset:lmms-lab/m4-instruct-d... Dataset:mammoth-vl/mammoth-vl-...   Dataset:mutonix/vript   Dataset:orrzohar/video-star Dataset:sharegpt4video/sharegp...   Dataset:tiger-lab/vista-400k   En   Endpoints compatible   Image-text-to-text   Instruct   Region:us   Safetensors   Sharded   Smolvlm   Tensorflow   Video-text-to-text
Model Card on HF 🤗: https://huggingface.co/HuggingFaceTB/SmolVLM2-2.2B-Instruct 

SmolVLM2 2.2B Instruct Parameters and Internals

LLM NameSmolVLM2 2.2B Instruct
Repository 🤗https://huggingface.co/HuggingFaceTB/SmolVLM2-2.2B-Instruct 
Base Model(s)  SmolVLM Instruct   HuggingFaceTB/SmolVLM-Instruct
Model Size2.2b
Required VRAM9 GB
Updated2026-08-09
MaintainerHuggingFaceTB
Model Typesmolvlm
Instruction-BasedYes
Model Files  5.0 GB: 1-of-2   4.0 GB: 2-of-2
Supported Languagesen
Model ArchitectureSmolVLMForConditionalGeneration
Licenseapache-2.0
Model Max Length16384
Transformers Version4.47.1
Tokenizer ClassGPT2Tokenizer
Padding Token<|im_end|>
Vocabulary Size49280
Torch Data Typefloat32

Best Alternatives to SmolVLM2 2.2B Instruct

Best Alternatives
Context / RAM
Downloads
Likes
SmolVLM2 2.2B Instruct Mlx0K / 4.5 GB3528