Pytorch: open-source projects on GitHub
98 repositories on Pytorch, picked by the REDDYX AI agent. Each has a summary, key features and a breakdown. Sorted by stars.
// RELATED TOPICS
Local-First AI Creation Workstation
A local-first full-stack AI workstation with chat, comics, drama, writing, and learning modules, supporting vLLM, llama.cpp, and Transformers for multi-engine inference w
★ 535 · PythonAI PERSONALIZED EXAM SYSTEM
🧠 Adaptive system for generating exam questions based on LLM, knowledge graph, and BKT algorithm.
★ 531 · PythonFLIDE: INSTANT DECISION INFERENCE
Flide is a decision model inference engine with CUDA support. Provides deterministic inference for Gemma and other models in microseconds. Ideal for AI agents requiring s
★ 501 · PythonPYKALE — Powerful ML Tool
A PyTorch-based library for interdisciplinary research using knowledge from multiple sources.
★ 494 · PythonGradCache: Gradient Caching
A library for efficient large-batch training in contrastive learning, overcoming GPU/TPU memory limitations through gradient caching.
★ 451 · PythonRLLM: Learning Tables with LLM
A PyTorch library for training relational tables using large language models.
★ 441 · PythonPYTORCH REPLICATIONS OF SOTA PAPERS
429 stars. From-scratch implementations of transformers, diffusion, GAN, CLIP, LoRA, MoE in Jupyter Notebook. Includes SFT and GRPO post-training experiments.
★ 430 · Jupyter NotebookTORCHTRADE - AI TRADING WITHOUT BORDERS
A modular reinforcement learning framework for creating and testing algorithmic trading strategies in Python.
★ 428 · PythonOLYMPUS UNIFIES CV TASKS
The official Olympus code — a universal task router for CV, presented at CVPR 2025. Routes data to the best models.
★ 427 · PythonNext-Gen Transformer Framework
Open-source framework for next-gen Transformers with compute-matched benchmarking and autonomous architecture research platform.
★ 412 · PythonZINC - LOCAL LLM ON GPU
Zinc is an LLM inference engine in Zig that runs on AMD GPU and Apple Silicon without external dependencies.
★ 395 · ZigBUILD LLM IN SECONDS
CLI tool creates ready-to-train PyTorch projects for LLM in seconds. Analog of create‑next‑app, but for language models. Written in TypeScript, already has 362 stars.
★ 373 · TypeScriptDeep Learning: Projects and Resources
DeepWorks — collection of Jupyter notebooks with deep learning projects: from ML basics to LLM and generative models. 335 stars, ready-to-run examples in PyTorch.
★ 333 · Jupyter NotebookPROFILER FOR AI AGENTS AND LLM
Graphsignal Profiler is a tool for debugging and monitoring inference. It helps find performance bottlenecks in deep neural networks and AI agents. Perfect for working wi
★ 276 · PythonNEURALGRAFFITI: LIVE LLM OUTPUT CONTROL
NeuralGraffiti is a tool for live transformation of foundation model outputs directly at the neural network level. Works in real time without retraining.
★ 274 · Jupyter NotebookMULTIMODAL DECISION MODEL
OneJev — multimodal System One model for calibrated answers to questions about screens, photos, video and text in one forward pass.
★ 272 · PythonTORCHSPEC SPECULATIVE DECODING
PyTorch library for training speculative decoding models, accelerating text generation.
★ 268 · PythonSUBTEXT: BEHIND THE MODEL
A tool for analyzing hidden representations of AI models, revealing their internal logic.
★ 248 · HTMLJ-WASH: LLM TOOL
Tool for manual alignment of large language models using Jacobian Lens, with results exported.
★ 242 · PythontheDAW: AI Music Studio
theDAW combines generation, editing, mixing, and live visuals in a unified environment for creating music with AI. The project is designed for producers, DJs, and visual
★ 237 · TypeScriptLLM Alignment Tuning
Automatic calibration of the language model's censorship layer via direct weight modification and the use of LoRA and MoE experts.
★ 221 · PythonLive Multi-Camera SAM Tracking
Serves multiple cameras with one SAM model per GPU, achieving 4x memory reduction and faster inference than original implementations.
★ 221 · PythonCOMPACT TTS WITH VOICE CLONING
Multilingual text-to-speech model with zero-shot voice cloning. Runs locally on GPU, CPU, Apple Silicon, and in browser via WebGPU.
★ 219 · PythonLLM TRAINING ACCELERATION ON SUPERPOD
Framework for distributed LLM training with SuperPod topology awareness. Optimizes process placement and inter-GPU communication. Supports PyTorch and scales to thousands
★ 214 · PythonGRACE: INT4 QWEN3-VL DISTILL
Qwen3-VL in INT4 with quantization-aware distillation – ready for deployment.
★ 213 · Jupyter NotebookAUTOSHORTS AI CLIPS GAMEPLAY
Automatically creates vertical short clips from long gameplay recordings, analyzing scenes with AI. GPU‑accelerated rendering and optional AI voiceovers make virus‑ready
★ 207 · PythonFUSILLI MULTIMODAL FUSION
Fusilli — Python package with ready-made multimodal fusion pipelines: from data loading to model training and evaluation.
★ 205 · PythonComprehensive AI Engineering Course
Free 173-lesson course covering data, ML, deep learning, and AI engineering with 166 runnable labs. Covers math, code, systems, and security under CC BY-NC 4.0.
★ 175 · PythonWeight-Direction One-Step Image Synthesis
Official PyTorch implementation of the efficient CVPR 2026 one-step image synthesis method with weight direction.
★ 158 · PythonImagination‑to‑Policy Data Pipeline
Converts imagined world-model rollouts into corrective datasets for training robot policies on a single GPU.
★ 130 · PythonMultimodal Image Retrieval System
Practical multimodal image retrieval with intent recognition, prompt engineering, Qwen LoRA, EmbeddingGemma 2, InfoNCE contrastive learning, and RAG components.
★ 129 · PythonHands-On LLM Full-Stack Course
Comprehensive tutorial implementing LLMs from scratch: builds Transformer in PyTorch, reproduces Pretrain/SFT/LoRA/DPO/GRPO on single GPU, covers PEFT/Agent/RAG with real
★ 124 · PythonElastic Data Loading
Zephon offers fast, flexible, elastically deterministic data loading for foundation‑model training, ensuring reproducible sampling and high throughput.
★ 90 · PythonRose — PyTorch Optimizer
Stateless PyTorch optimizer that normalizes gradients for stable and balanced model training.
★ 60 · PythonSpeculative Decoding Training Framework
ASpec accelerates LLM inference by training efficient draft models for speculative decoding with high-throughput distributed training.
★ 53 · PythonCan I finetune this?
Tool to check whether a Hugging Face model fits in your GPU memory and can be fine-tuned locally.
★ 31 · PythonLongLive-RAG: Long Video Generation
Official implementation of the LongLive-RAG framework for generating long videos using a retrieval-augmented approach.
★ 31 · PythonAuto-labeling YOLO Training
Platform for automatic image and video labeling using VLM NVIDIA LocateAnything-3B, with manual correction and one-click YOLO model training.
★ 31 · PythonNew projects on this topic in Telegram
Telegram →