Skip
7.9k
MakazhanAlpamys/Soup
Fine-tune LLMs from one YAML. Layer streaming trains an 8B model on a 4 GB laptop GPU.
—
—
Explore 43 GitHub repositories focused on pytorch. Discover top-starred projects and those trending this week.
Fine-tune LLMs from one YAML. Layer streaming trains an 8B model on a 4 GB laptop GPU.
A high-throughput and memory-efficient inference and serving engine for LLMs
LMCache: Supercharge Your LLM with the Fastest KV Cache Layer
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface.
We write your reusable computer vision tools. 💜
VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning
Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
JARVIS, a system to connect LLMs with ML community. Paper: https://arxiv.org/pdf/2303.17580.pdf
🤗 The largest hub of ready-to-use datasets for AI models with fast, easy-to-use and efficient data manipulation tools
ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator
Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface.
A high-throughput and memory-efficient inference and serving engine for LLMs
We write your reusable computer vision tools. 💜
VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning
Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
JARVIS, a system to connect LLMs with ML community. Paper: https://arxiv.org/pdf/2303.17580.pdf
🤗 The largest hub of ready-to-use datasets for AI models with fast, easy-to-use and efficient data manipulation tools
ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator
Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
A theoretical reconstruction of the Claude Mythos architecture, built from first principles using the available research literature.
LMCache: Supercharge Your LLM with the Fastest KV Cache Layer
Silero VAD: pre-trained enterprise-grade Voice Activity Detector
so-vits-svc fork with realtime support, improved interface and more features.
Fine-tune LLMs from one YAML. Layer streaming trains an 8B model on a 4 GB laptop GPU.
FlashInfer: Kernel Library for LLM Serving
StyleTTS 2: Towards Human-Level Text-to-Speech through Style Diffusion and Adversarial Training with Large Speech Language Models
Dockerized FastAPI wrapper for Kokoro-82M text-to-speech model w/multiplatform CPU, AMD, NVIDIA GPU PyTorch support, handling, and auto-stitching
Production First and Production Ready End-to-End Speech Recognition Toolkit
Probabilistic time series modeling in Python
Machine Learning and Agentic AI Resources, Practice and Research
Fault-tolerant, highly scalable GPU orchestration, and a machine learning framework designed for training models with billions to trillions of parameters
Scalable and user friendly neural :brain: forecasting algorithms.
The official Python client for the Hugging Face Hub.