A novel Multimodal Large Language Model (MLLM) architecture, designed to structurally align visual and textual embeddings.

Official code for infimm-hd

A fast yet powerful Python Markdown parser with renderers and plugins.

Things you can do with the token embeddings of an LLM

Official code implementation of General OCR Theory: Towards OCR-2.0 via a Unified End-to-end Model

This package, developed as part of our research detailed in the Chroma Technical Report, provides tools for text chunking and evaluation. It allows users to compare different chunking methods and i…

Data-Driven Evaluation for LLM-Powered Applications

Efficient Triton Kernels for LLM Training

open-source multimodal large language model that can hear, talk while thinking. Featuring real-time end-to-end speech input and streaming audio output conversational capabilities.

An open-source RAG-based tool for chatting with your documents.

A Flexible Framework for Experiencing Cutting-edge LLM Inference Optimizations

Package and scripts used to build a dataset of Wikipedia articles in Markdown.

highlight.io: The open source, full-stack monitoring platform. Error monitoring, session replay, logging, distributed tracing, and more.

Distributed LLM and StableDiffusion inference for mobile, desktop and server.

OCR, layout analysis, reading order, line detection in 90+ languages

Convert HTML to Markdown

A tool to get summaries and get past paywalls

[ACL'24] Superfiltering: Weak-to-Strong Data Filtering for Fast Instruction-Tuning

High-quality datasets, tools, and concepts for LLM fine-tuning.

It's a cooler way to store simple linear models.

the AI-native open-source embedding database

Code for the paper "Mitigating the Learning Bias towards Repetition by Self-Contrastive Training for Open-Ended Generation"

The code of paper "Learning to Break the Loop: Analyzing and Mitigating Repetitions for Neural Text Generation" published at NeurIPS 2022

MPB (Miner-PDF-Benchmark) is an end-to-end PDF document comprehension evaluation suite designed for large-scale model data scenarios.

Official Implementation of EAGLE-1 and EAGLE-2

A large-scale language model for scientific domain, trained on redpajama arXiv split

Multimodal language model benchmark, featuring challenging examples