Vous avez un workflow ComfyUI ? Obtenez une documentation lisible directement dans votre navigateur.Rendre un workflow lisible.Essayer maintenant ↗
Aloe-Vision is a medical Large Vision–Language Model built on Qwen2-VL-Instruct, released in 7B and 72B sizes. The model is trained on a \3.5 M samples balanced mixture across medical vs. general and multimodal vs....
Source du modèle
Extrait de la source
Aloe-Vision is a medical Large Vision–Language Model built on Qwen2-VL-Instruct, released in 7B and 72B sizes. The model is trained on a \3.5 M samples balanced mixture across medical vs. general and multimodal vs....
Sources
1 sourceVérifié 1 août
Artefacts du modèle
31 artefactsmodel-00001-of-00031.safetensors
safetensors · 4,36 GB · SHA-256 851927d26f79…fa23 · Hugging Face
Téléchargermodel-00002-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 5edaedb5d739…0f26 · Hugging Face
TéléchargerExtraits de sources
2 extraitsmodel-00003-of-00031.safetensors
safetensors · 4,62 GB · SHA-256 39fbefe824da…5664 · Hugging Face
model-00004-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 5e03f6c99722…4182 · Hugging Face
Téléchargermodel-00005-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 930eae831d29…54f7 · Hugging Face
Téléchargermodel-00006-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 a4c657b6e182…4a44 · Hugging Face
Téléchargermodel-00007-of-00031.safetensors
safetensors · 4,62 GB · SHA-256 5777cd3e5a18…770e · Hugging Face
Téléchargermodel-00008-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 8689a17f34c3…b4d4 · Hugging Face
Téléchargermodel-00009-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 2db1f8ab9975…01f2 · Hugging Face
Téléchargermodel-00010-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 4db44b40dba5…aee9 · Hugging Face
Téléchargermodel-00011-of-00031.safetensors
safetensors · 4,62 GB · SHA-256 c97d4728700a…ee7f · Hugging Face
Téléchargermodel-00012-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 4545cef8bfaf…be3d · Hugging Face
Téléchargermodel-00013-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 9cd033ce356a…7d32 · Hugging Face
Téléchargermodel-00014-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 e82244610bfa…9290 · Hugging Face
Téléchargermodel-00015-of-00031.safetensors
safetensors · 4,62 GB · SHA-256 d72aa17ee39b…3492 · Hugging Face
Téléchargermodel-00016-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 267bb4230ae7…8f6e · Hugging Face
Téléchargermodel-00017-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 cf01b05e0aca…a723 · Hugging Face
Téléchargermodel-00018-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 104b1c92fd81…3141 · Hugging Face
Téléchargermodel-00019-of-00031.safetensors
safetensors · 4,62 GB · SHA-256 5bba95591fce…0678 · Hugging Face
Téléchargermodel-00020-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 d709d13e46f1…cea6 · Hugging Face
Téléchargermodel-00021-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 42d63e9a47ba…84bb · Hugging Face
Téléchargermodel-00022-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 7c06335f1b3b…1ba5 · Hugging Face
Téléchargermodel-00023-of-00031.safetensors
safetensors · 4,62 GB · SHA-256 8db5cd3b2b75…b70f · Hugging Face
Téléchargermodel-00024-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 b347f4d69ead…2ac3 · Hugging Face
Téléchargermodel-00025-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 29003a555cee…d338 · Hugging Face
Téléchargermodel-00026-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 bede8ef97743…a3a0 · Hugging Face
Téléchargermodel-00027-of-00031.safetensors
safetensors · 4,62 GB · SHA-256 b3a174802227…4914 · Hugging Face
Téléchargermodel-00028-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 3a8a0ecb3144…5df1 · Hugging Face
Téléchargermodel-00029-of-00031.safetensors
safetensors · 4,45 GB · SHA-256 e80ba9bc45ab…6ea0 · Hugging Face
Téléchargermodel-00030-of-00031.safetensors
safetensors · 4,17 GB · SHA-256 8498a1153130…987d · Hugging Face
Téléchargermodel-00031-of-00031.safetensors
safetensors · 2,32 GB · SHA-256 ae1e996c7393…3b59 · Hugging Face
Télécharger--- license: cc-by-nc-sa-4.0 library\_name: transformers language: - en pipeline\_tag: image-text-to-text tags: - visual-question-answering - medical - healthcare - biology - multimodal - lvlm - grounding datasets: HPAI-BSC/Aloe-Beta-General-Collection model\_type: qwen2-vl base\_model: Qwen/Qwen2-VL-72B-Instruct --- <p align="center"> <img alt="Aloe-Vision" src="https://cdn-uploads.huggingface.co/production/uploads/63a417e70cf4daf6166777a2/xkm30vCSIz1GK__K3QIQZ.png" width="25%"> </p> </h1> <hr style="margin: 15px"> <div align="center" style="line-height:1.15;"> <a href="https://huggingface.co/datasets/HPAI-BSC/Aloe-Vision-Data" target="_blank" style="margin:2px;"> <img alt="Training Dataset" src="https://img.shields.io/badge/🤗%20Training%20Dataset-Aloe%20Vision%20Data-ffc107" style="vertical-align:middle;"/> </a> <a href="https://creativecommons.org/licenses/by-nc-sa/4.0/deed.en" target="_blank" style="margin:2px;"> <img alt="License" src="https://img.shields.io/badge/license-CC_BY--NC--SA_4.0-green" style="vertical-align:middle;"/> </a> <br/> <a href="https://hpai.bsc.es/" target="_blank" style="margin:2px;"> <img alt="Website" src="https://img.shields.io/badge/Website-HPAI-8A2BE2" style="vertical-align:middle;"/> </a> <a href="https://huggingface.co/HPAI-BSC" target="_blank" style="margin:2px;"> <img alt="Hugging Face Org" src="https://img.shields.io/badge/🤗%20HF-HPAI--BSC-ffc107" style="vertical-align:middle;"/> </a> <a href="https://github.com/HPAI-BSC" target="_blank" style="margin:2px;"> <img alt="GitHub" src="https://img.shields.io/badge/GitHub-HPAI--BSC-%23121011.svg" style="vertical-align:middle;"/> </a> </div> </div> **Aloe-Vision** is a **medical Large Vision–Language Model** built on **Qwen2-VL-Instruct**, released in **7B and 72B** sizes. The model is trained on a **\~3.5 M samples** balanced mixture across **medical vs. general** and **multimodal vs. text-only** sources, rebalanced by **loss-contributing assistant tokens** to avoid long-answer bias. We implement **leakage control of evaluation images in the training data** via **exact 64-bit image-hash matching**, removing any duplicates from the training. **Quality filtering** of the training data combines (1) **LVLM-based sample scoring (1–5 scale)** for image–question–answer coherence and relevance and (2) **answer perplexity checks** to flag trivial or nois...
Source context: 23 downloads · 0 likes · Pipeline visual-question-answering · Repo HPAI-BSC/Aloe-Vision-72B-AR
