5 27 72

Umitcan Sahin PRO

ucsahin

AI & ML interests

Visual Language Models, Large Language Models, Vision Transformers

Recent Activity

reacted to singhsidhukuldeep's post with 🔥 2 days ago

Exciting News in AI: JinaAI Releases JINA-CLIP-v2! The team at Jina AI has just released a groundbreaking multilingual multimodal embedding model that's pushing the boundaries of text-image understanding. Here's why this is a big deal: 🚀 Technical Highlights: - Dual encoder architecture combining a 561M parameter Jina XLM-RoBERTa text encoder and a 304M parameter EVA02-L14 vision encoder - Supports 89 languages with 8,192 token context length - Processes images up to 512×512 pixels with 14×14 patch size - Implements FlashAttention2 for text and xFormers for vision processing - Uses Matryoshka Representation Learning for efficient vector storage ⚡️ Under The Hood: - Multi-stage training process with progressive resolution scaling (224→384→512) - Contrastive learning using InfoNCE loss in both directions - Trained on massive multilingual dataset including 400M English and 400M multilingual image-caption pairs - Incorporates specialized datasets for document understanding, scientific graphs, and infographics - Uses hard negative mining with 7 negatives per positive sample 📊 Performance: - Outperforms previous models on visual document retrieval (52.65% nDCG@5) - Achieves 89.73% image-to-text and 79.09% text-to-image retrieval on CLIP benchmark - Strong multilingual performance across 30 languages - Maintains performance even with 75% dimension reduction (256D vs 1024D) 🎯 Key Innovation: The model solves the long-standing challenge of unifying text-only and multi-modal retrieval systems while adding robust multilingual support. Perfect for building cross-lingual visual search systems! Kudos to the research team at Jina AI for this impressive advancement in multimodal AI!

upvoted a collection 17 days ago

DataGemma Release

new activity 24 days ago

ucsahin/TR-VLM-DPO-Dataset:[bot] Conversion to Parquet

View all activity

Organizations

None yet

Posts 2

Post

3728

🚀 Introducing TraVisionLM: Turkish Visual Language Model - The First of Its Kind! 🇹🇷🖼️

I'm thrilled to share TraVisionLM on Hugging Face! With 875M parameters, this lightweight, efficient model handles Turkish instructions for image inputs. Fully compatible with the Transformers library, it’s easy to load, fine-tune, and use—no external libraries needed!

Developed solo, TraVisionLM is a strong foundation for low-resource language research. While still improving, it's a key step for Turkish-language AI. Your feedback is welcome as I refine the model.

🎉 Explore it now:

- Model: ucsahin/TraVisionLM-base
- Demo: https://huggingface.co/spaces/ucsahin/TraVisionLM-Turkish_Visual_Language_Model
- Object Detection Finetune: ucsahin/TraVisionLM-Object-Detection-ft

Let’s push Turkish visual language processing forward!

---

🚀 TraVisionLM: Türünün İlk Örneği Türkçe Görsel Dil Modelini Sunuyorum! 🇹🇷🖼️

TraVisionLM modelini Hugging Face'te yayınladım! 875M parametre ile bu hafif ve verimli model, görüntüye dayalı Türkçe talimatları işlemek için tasarlandı. Transformers kütüphanesiyle tamamen uyumlu, yüklemesi, eğitmesi ve kullanması çok kolay—dış kütüphane gerekmez!

Tek başıma geliştirdiğim TraVisionLM, düşük kaynaklı dillerde araştırmalar için sağlam bir temel sunuyor. Geliştirmeye devam ederken geri bildirimlerinizi bekliyorum.

🎉 Hemen keşfedin:

- Model: ucsahin/TraVisionLM-base
- Demo: https://huggingface.co/spaces/ucsahin/TraVisionLM-Turkish_Visual_Language_Model
- Obje Tespiti İnce Ayarı: ucsahin/TraVisionLM-Object-Detection-ft

Türkçe görsel dil işleme sınırlarını birlikte zorlayalım!

Post

3778

Florence-2 has a great capability of detecting various objects in a zero-shot setting with the task prompt "<OD>". However, if you want to detect specific objects that the base model is not able to in its current form, you can easily finetune it for this particular task. Below I show how to finetune the model to detect tables in a given image, but a similar process can be applied to detect any objects. Thanks to @andito , @merve , and @SkalskiP for sharing the fix for finetuning the Florence-2 model. Please also check their great blog post at https://huggingface.co/blog/finetune-florence2.

Colab notebook: https://colab.research.google.com/drive/1Y8GVjwzBIgfmfD3ZypDX5H1JA_VG0YDL?usp=sharing
Finetuned model: ucsahin/Florence-2-large-TableDetection