Popular GitHub Repositories for Checkpoint

Orbax provides common checkpointing and persistence utilities for JAX users

Generate NFT or train new model in just few clicks! Train as much as you can, others will resume from checkpoint!

Load GPT-2 checkpoint and generate texts

Checkpoint model mixer/merger extension

Use AI to Animate - ComfyUI Custom Nodes include IMG Caches, Switches, Prompt & Checkpoint changer...

source code and pre-trained/fine-tuned checkpoint for NAACL 2021 paper LightningDOT

turn bert pretrain checkpoint into saved model for a feature extracting demo in java

Accelerating AI Training and Inference from Storage Perspective (Must-read Papers on Storage for AI)

Save, load, host, and share AI model checkpoints without slowing down training. Host on Lightning AI or your own cloud with enterprise-grade access controls.

This is a meta-model distilled from LLMs for information extraction. This is an intermediate checkpoint that can be well-transferred to all kinds of downstream information extraction tasks.

High-performance Python librarys for connecting AI/ML frameworks with OSS storage.

A single-table DynamoDB implementation of the LangGraph checkpoint saver interface.

Merge Diffusion Tool is an open-source solution for merging LoRA models, integrating LoRA into checkpoints, and blending Flux And Stable Diffusion models (SD1.5, SD2, SD3, SDXL). Optimize your AI workflows with ease.

用于AI智能体的代码执行沙盒服务器，支持检查点/恢复功能，以及更多特性。

JellyML is an open-source tool (python API and command line) for effortlessly embedding a snapshot of your code into a checkpoint of a pytorch (and pytorch lightning) model.

Load GPT-2 checkpoint and generate texts in PyTorch

Fine tuning of RetinaNet architecture on very few examples of a novel class after initializing from a pre-trained COCO checkpoint

Automatically converts any SD v1.5 checkpoint in to a temporary inpainting model

基于Spring Boot 3和Java 24的OAuth 2授权服务器

This repository contains a notebook for object detection with the help of fine tuning of a (TF2 friendly) RetinaNet architecture on very few examples of a novel class after initializing from a pre-trained COCO checkpoint. Training runs in eager mode.