dinhanhx's picture

dinhanhx

dinhanhx

·

dinhanhx

AI & ML interests

Vision Language

Recent Activity

replied to merve's post 1 day ago

Apollo is a new family of open-source video language models by Meta, where 3B model outperforms most 7B models and 7B outperforms most 30B models 🧶 ✨ the models come in 1.5B https://huggingface.co/Apollo-LMMs/Apollo-1_5B-t32, 3B https://huggingface.co/Apollo-LMMs/Apollo-3B-t32 and 7B https://huggingface.co/Apollo-LMMs/Apollo-7B-t32 with A2.0 license, based on Qwen1.5 & Qwen2 ✨ the authors also release a benchmark dataset https://huggingface.co/spaces/Apollo-LMMs/ApolloBench The paper has a lot of experiments (they trained 84 models!) about what makes the video LMs work ⏯️ Try the demo for best setup here https://huggingface.co/spaces/Apollo-LMMs/Apollo-3B they evaluate sampling strategies, scaling laws for models and datasets, video representation and more! > The authors find out that whatever design decision was applied to small models also scale properly when the model and dataset are scaled 📈 scaling dataset has diminishing returns for smaller models > They evaluate frame sampling strategies, and find that FPS sampling is better than uniform sampling, and they find 8-32 tokens per frame optimal > They also compare image encoders, they try a variation of models from shape optimized SigLIP to DINOv2 they find https://huggingface.co/google/siglip-so400m-patch14-384 to be most powerful 🔥 > they also compare freezing different parts of models, training all stages with some frozen parts give the best yield They eventually release three models, where Apollo-3B outperforms most 7B models and Apollo 7B outperforms 30B models 🔥

liked a model 1 day ago

erax-ai/EraX-VL-2B-V1.5

liked a model 8 days ago

HIT-TMG/KaLM-embedding-multilingual-mini-instruct-v1.5

View all activity

Organizations

dinhanhx's activity

upvoted a paper 8 days ago

OmniDocBench: Benchmarking Diverse PDF Document Parsing with Comprehensive Annotations

Paper • 2412.07626 • Published 24 days ago • 21

upvoted an article about 1 month ago

Article

Low Code Large Language Model Alignment

By

•

Nov 19, 2024

• 13

upvoted a collection about 2 months ago

Cosmos Tokenizer

A suite of image and video tokenizers • 12 items • Updated 17 days ago • 28

upvoted an article about 2 months ago

Article

BM25 for Python: Achieving high performance while simplifying dependencies with BM25S⚡

By

•

Jul 9, 2024

• 41

upvoted 3 collections 2 months ago

AMD-OLMo

AMD-OLMo are a series of 1 billion parameter language models trained by AMD on AMD Instinct™ MI250 GPUs based on OLMo. • 4 items • Updated Oct 31, 2024 • 17

SmolLM2

State-of-the-art compact LLMs for on-device applications: 1.7B, 360M, 135M • 15 items • Updated 12 days ago • 196

C4AI Aya Expanse

Aya Expanse is an open-weight research release of a model with highly advanced multilingual capabilities. • 3 items • Updated 18 days ago • 30

upvoted an article 3 months ago

Article

Training and Finetuning Embedding Models with Sentence Transformers v3

May 28, 2024

• 167

upvoted 2 collections 3 months ago

VisionLM

591 items • Updated about 15 hours ago • 39

Awesome Document AI

A collection of open-source document AI 📄 📝 📈 • 27 items • Updated Mar 11, 2024 • 76

upvoted a paper 3 months ago

VisFocus: Prompt-Guided Vision Encoders for OCR-Free Dense Document Understanding

Paper • 2407.12594 • Published Jul 17, 2024 • 19

upvoted an article 3 months ago

Article

Llama can now see and run on your device - welcome Llama 3.2

Sep 25, 2024

• 180

upvoted a collection 3 months ago

💻 Local SmolLMs

SmolLM models in MLC, ONNX and GGUF format for local applications + in-browser demos • 14 items • Updated 12 days ago • 47

upvoted a paper 3 months ago

MoMa: Efficient Early-Fusion Pre-training with Mixture of Modality-Aware Experts

Paper • 2407.21770 • Published Jul 31, 2024 • 22

upvoted a paper 4 months ago

Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Model

Paper • 2408.11039 • Published Aug 20, 2024 • 58

upvoted a collection 5 months ago

Llama-3.1 Quantization

Neural Magic quantized Llama-3.1 models • 22 items • Updated Nov 22, 2024 • 42

upvoted 2 articles 6 months ago

Article

SmolLM - blazingly fast and remarkably powerful

Jul 16, 2024

• 294

Article

ColPali: Efficient Document Retrieval with Vision Language Models 👀

By

•

Jul 5, 2024

• 183

upvoted an article 7 months ago

Article

A Gentle Introduction to 8-bit Matrix Multiplication for transformers at scale using transformers, accelerate and bitsandbytes

Aug 17, 2022

• 66

upvoted an article 8 months ago

Article

From cloud to developers: Hugging Face and Microsoft Deepen Collaboration

May 21, 2024

• 8