第一款完全多 GPU 支援且非常先進的具有 Gradio 介面的批量影像字幕產生器 APP 發布

王林
發布: 2024-08-27 06:00:32
原創
360 人瀏覽過

搭配 JoyCaption 的多 GPU 大量字幕。 JoyCaption 使用 Meta-Llama-3.1–8B 和 google/siglip-so400m-patch14–384 以及微調的影像字幕神經網路。

連結:https://www.patreon.com/posts/110613301

大量字幕編輯器的連結:https://www.patreon.com/posts/108992085

用 Python、Torch 和 Bitsandbytes 寫多 GPU 確實是一個挑戰。

我們的APP使用JoyCaption影像字幕微調模型。

我們的應用程式甚至在多 GPU 模式(9.5 GB VRAM)下也支援位元和位元組 4 位元模型載入

在 8x RTX A6000(雲端)和 RTX 3090 TI + RTX 3060(我的電腦)上測試

一鍵安裝在 Windows、RunPod 和 Massed Compute 上

優秀的字幕質量,自動將影像分配到每個GPU,功能很多。您可以使用跳過帶有字幕的圖像選項來恢復字幕。

有關完整詳細信息,請查看屏幕截圖

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

First fully multi-GPU supporting and very advanced batch image captioner APP with Gradio interface published

以上是第一款完全多 GPU 支援且非常先進的具有 Gradio 介面的批量影像字幕產生器 APP 發布的詳細內容。更多資訊請關注PHP中文網其他相關文章!

來源:dev.to
本網站聲明
本文內容由網友自願投稿,版權歸原作者所有。本站不承擔相應的法律責任。如發現涉嫌抄襲或侵權的內容,請聯絡admin@php.cn
熱門教學
更多>
最新下載
更多>
網站特效
網站源碼
網站素材
前端模板
關於我們 免責聲明 Sitemap
PHP中文網:公益線上PHP培訓,幫助PHP學習者快速成長!