Trending

Transform DOCX into LLM-ready data

The Speed of VITs and CNNs

Relational Graph Transformers

Bamba: An open-source LLM that crosses a transformer with an SSM

Introducing the Meta AI App: A New Way to Access Your AI Assistant

Implement Flash Attention Back End in SGLang – Basics and KV Cache

Vision Transformers Need Registers (extra tokens discarded after last layer)

Show HN: Lemon Slice Live, a real-time video-audio AI model

Three things everyone should know about Vision Transformers

Smoothie Qwen: A lightweight adjustment tool for smoothing token probabilities in the Qwen models to encourage balanced multilingual generation.

nanoVLM: A minimal Vision-Language Model with a LLaMA-style decoder — now open source

deepseek-ai/DeepSeek-Prover-V2-671B · Hugging Face

Whisper Translation Finetuning [P]

[D] How could a MLP replicate the operations of an attention head?

Looks like Qwen 3 will have a 256k context?

[P] Top open chart-understanding model upto 8B and performs on par with much larger models. Try it

Finally got ~10t/s DeepSeek V3-0324 hybrid (FP8+Q4_K_M) running locally on my RTX 4090 + Xeon with with 512GB RAM, KTransformers and 32K context

[R][P] Byte-level LLaMA and Gemma via cross-tokenizer distillation (with open-source toolkit)

TesslateAI

Save item

UIGEN-T2-7B-Q8_0-GGUF

Kalomaze

Save item

Qwen3-16B-A3B

Qwen

Save item

Qwen2.5-Omni-3B

Task: any-to-any

LightOn AI

Save item

GTE-ModernColBERT-v1

Task: sentence-similarity

Qwen

Save item

Qwen2.5-Omni-7B

Task: any-to-any

HyperCLOVA X

Save item

HyperCLOVAX-SEED-Vision-Instruct-3B

Skywork

Save item

Skywork-R1V2-38B

Task: image-text-to-text

HyperCLOVA X

Save item

HyperCLOVAX-SEED-Text-Instruct-1.5B

ByteDance Seed

Save item

UI-TARS-1.5-7B

Task: image-text-to-text

huggingface

Save item

transformers

🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.

huggingface

Save item

optimum

🚀 Accelerate inference and training of 🤗 Transformers, Diffusers, TIMM and Sentence Transformers with easy to use hardware optimization tools

lucidrains

Save item

vit-pytorch

Implementation of Vision Transformer, a simple way to achieve SOTA in vision classification with only a single transformer encoder, in Pytorch

NVIDIA

Save item

Megatron-LM

Ongoing research training transformer models at scale

microsoft

Save item

Swin-Transformer

This is an official implementation for "Swin Transformer: Hierarchical Vision Transformer using Shifted Windows".