# Hugging Face — Accelerating Vision-Language Models: BridgeTower on Habana Gaudi2

- Company: Hugging Face (huggingface.co)
- Announced: 2023-06-29
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://huggingface.co/blog/bridgetower
- Record: https://forck.live/items/2033-accelerating-vision-language-models-bridgetower-on-habana-gaudi2
- Subject: Platform
- Models affected: BridgeTower

This post benchmarks fine-tuning BridgeTower on Habana Gaudi2, achieving speedups over Nvidia A100 and H100 using Optimum Habana v1.7 and hardware-accelerated data loading techniques.

## Evidence

Verbatim from https://huggingface.co/blog/bridgetower:

> Optimum Habana v1.7 on Habana Gaudi2 achieves x2.5 speedups compared to A100 and x1.4 compared to H100 when fine-tuning BridgeTower, a state-of-the-art vision-language model.

---

Record: https://forck.live/items/2033-accelerating-vision-language-models-bridgetower-on-habana-gaudi2
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md