training-pipeline

Neural networks training pipeline based on PyTorch

My repo for training neural nets using pytorch-lightning and hydra

Curriculum training

YOLOv12 Underwater Object Detection is an open-source suite for underwater object detection, built on YOLOv12. It offers an end-to-end pipeline with GPU-accelerated training, customizable data augmentations, real-time inference via Gradio, and support for model export (ONNX & PyTorch).

Train custom wake word models with openWakeWord. A granular 13-step pipeline with compatibility patches for torchaudio 2.10+, Piper TTS, and speechbrain. Generates tiny ONNX models (~200 KB) for real-time keyword detection — like building your own "Hey Siri" trigger. WSL2/Linux + CUDA required.

YOLO training toolkit with Claude Code skills — dataset management, experiment tracking, HP tuning via model.tune(), active learning with CVAT, ONNX export. Supports YOLO11 & YOLO26.

Deep Learning training and deployment pipeline, reduce repetitive work from research to deployment

Pixi + PyTorch Lightning ML project template with Aim experiment tracking, reproducible environments, configs, tests, and GPU-ready training.

🔧 Fine-tune large language models locally on your data, export to GGUF, and train on CPU with ease using the Mobius LLM Fine-Tuning Engine.

Internship projects completed as part of the Shristi24 program offered by IIIT Hyderabad

Machine Learning in Production

TraceOS standardizes AI experiments into reproducible, searchable, and comparable assets. One command runs experiments, generates reports, and produces structured analysis: capability vectors, failure taxonomy, and recommendations. Every run is tracked, traceable, and comparable. Built on ABC-130K (amazon-far/abc). Apache 2.0.

AI Message Labels: Packaging and pipelines for deep learning text classification models

Configurable PyTorch training pipeline

SPIRA Model Trainer v2 (redesigned pipeline) by @danlawand

YoloLint is a tool for automatic validation of dataset structure, annotation files, and image sizes in YOLO projects. It helps you catch typical errors in directory structure, YAML files, annotation files, and now also ensures all your images have the correct size before you start model training.

🖼️ Implement a Routed CNN for CIFAR-10 image classification, showcasing modular design and advanced feature routing to enhance model performance.

Parallel training-sweep orchestrator with stall-recovery watchdog (halve batch, relaunch) + live dashboard

A concurrent training and generation pipeline leveraging active learning to drive synthetic data rendering. By generating customized datasets simultaneously alongside model training, it creates a real-time feedback loop to dynamically refine object detection models.

Desktop toolchain for extracting, annotating, and training YOLO models on ZED SVO2 recordings for drone target tracking