Unified Efficient Fine-Tuning of 100+ LLMs (ACL 2024)

Multi-Modal Language Modeling with Image, Audio and Text Integration, included multi-images and multi-audio in a single multiturn.

Repository for NeurIPS 2020 Spotlight "AdaBelief Optimizer: Adapting stepsizes by the belief in observed gradients"

Simple, safe way to store and distribute tensors

a state-of-the-art-level open visual language model | 多模态预训练模型

Activation-aware Singular Value Decomposition for Compressing Large Language Models

Amphion (/æmˈfaɪən/) is a toolkit for Audio, Music, and Speech Generation. Its purpose is to support reproducible research and help junior researchers and engineers get started in the field of audi…

Calculate token/s & GPU memory requirement for any LLM. Supports llama.cpp/ggml/bnb/QLoRA quantization

A playbook for systematically maximizing the performance of deep learning models.

Inference code for Mistral and Mixtral hacked up into original Llama implementation

Learn how to design, develop, deploy and iterate on production-grade ML applications.

Training code for Baby-Llama, our submission to the strict-small track of the BabyLM challenge.

Code for paper titled "Towards the Law of Capacity Gap in Distilling Language Models"

Lifting ControlNet for Generalized Depth Conditioning

Mamba-Chat: A chat LLM based on the state-space model architecture 🐍

💡 All-in-one open-source embeddings database for semantic search, LLM orchestration and language model workflows

Tools for merging pretrained large language models.

Animate Anyone: Consistent and Controllable Image-to-Video Synthesis for Character Animation

Mamba SSM architecture

streamline the fine-tuning process for multimodal models: PaliGemma, Florence-2, and Qwen2-VL

Simple and efficient pytorch-native transformer text generation in <1000 LOC of python.

Finetune Llama 3.2, Mistral, Phi & Gemma LLMs 2-5x faster with 80% less memory

Convert PDF to markdown quickly with high accuracy

Custom C++ implementation of deep learning based OCR

A minimal re-implementation of orthogonal fine-tuning (OFT) for LLMs. Based on nanoGPT and minLoRA.

VITS: Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech

Task aware compression for various NLP tasks

Repository for Sparse Finetuning of LLMs via modified version of the MosaicML llmfoundry

SSD-1B, an open-source text-to-image model, outperforming previous versions by being 50% smaller and 60% faster than SDXL.