Idefics3 8B Llama3
PermissiveHugging Face · Idefics · Released 2024-08
8BMultimodal
A fully open vision-language model built on Llama 3 that improved substantially on document/OCR tasks over its Idefics2 predecessor.
Strengths
- +Apache 2.0 license as published by Hugging Face
- +Strong document understanding and OCR performance
- +Open training recipe and data documentation
Limitations
- -Trails larger closed and open VLMs (e.g. Qwen2-VL-72B) on general visual reasoning
- -Multi-image reasoning is weaker than single-image tasks
License
Apache 2.0
use commercially with attribution niceties
Hardware
runs on a good consumer GPU (or Apple Silicon) with quantization
Links
Stats
HuggingFaceM4/Idefics3-8B-Llama3— downloads·— likes
via Hugging Face