Search Results for "scratch" - Page 2
Sort By:
Showing 69 open source projects for "scratch"
View related business solutions-
Award-Winning Medical Office Software Designed for Your SpecialtyRXNT is an ambulatory healthcare technology pioneer that empowers medical practices and healthcare organizations to succeed and scale through innovative, data-backed, AI-powered software.
-
Skillfully - The future of skills based hiringSkillfully transforms hiring through AI-powered skill simulations that show you how candidates actually perform before you hire them. Our platform helps companies cut through AI-generated resumes and rehearsed interviews by validating real capabilities in action. Through dynamic job specific simulations and skill-based assessments, companies like Bloomberg and McKinsey have cut screening time by 50% while dramatically improving hire quality.
-
1
tiny-llm
A course of learning LLM inference serving on Apple Silicon
tiny-llm is an educational open-source project designed to teach system engineers how large language model inference and serving systems work by building them from scratch. The project is structured as a guided course that walks developers through the process of implementing the core components required to run a modern language model, including attention mechanisms, token generation, and optimization techniques. Rather than relying on high-level machine learning frameworks, the codebase uses mostly low-level array and matrix manipulation APIs so that developers can understand exactly how model inference works internally. ... -
2
xFormers
Hackable and optimized Transformers building blocks
...The library includes memory-efficient operator implementations in both Python and optimized C++/CUDA, ensuring that performance isn’t sacrificed for modularity. It also integrates with PyTorch seamlessly so you can drop in its blocks to existing models, replace default attention layers, or build new architectures from scratch. xformers includes training, deployment, and memory profiling tools. -
3
PySpur
Visual tool for building, testing, and deploying AI agent workflows
...By offering a visual representation of workflows, PySpur makes it easier to debug interactions between components and identify failures in complex pipelines. It supports iterative experimentation, allowing developers to rapidly improve agents without rebuilding systems from scratch. PySpur also enables deployment of finalized workflows after testing, making it suitable for both development and production use. Overall, it acts as an integrated environment for designing, evaluating, and managing AI-driven processes. -
4
TTRL
Test-Time Reinforcement Learning
...The repository is implemented on top of the verl ecosystem, which allows users to enable TTRL as part of an existing reinforcement learning workflow rather than building a new stack from scratch. -
AestheticsPro Medical Spa SoftwareAestheticsPro is the most complete Aesthetics Software on the market today. HIPAA Cloud Compliant with electronic charting, integrated POS, targeted marketing and results driven reporting; AestheticsPro delivers the tools you need to manage your medical spa business. It is our mission To Provide an All-in-One Cutting Edge Software to the Aesthetics Industry.
-
5
LLMs-Zero-to-Hero
From nobody to big model (LLM) hero
LLMs-Zero-to-Hero is an open-source educational project designed to guide learners through the complete process of understanding and building large language models from the ground up. The repository presents a structured learning pathway that begins with fundamental concepts in machine learning and progresses toward advanced topics such as model pre-training, fine-tuning, and deployment. Rather than relying entirely on existing frameworks, the project encourages readers to implement... -
6
Hello-Agents
Building an Intelligent Agent from Scratch
Hello Agents is an open educational project designed to teach developers how to understand, design, and build AI-native agents from the ground up through structured tutorials and practical examples. The project focuses on guiding learners beyond superficial framework usage toward deeper comprehension of agent architecture, reasoning loops, and real-world implementation patterns. It walks users through core concepts such as ReAct-style reasoning, tool usage, memory handling, and multi-step... -
7
SERA CLI
A tool to use the Ai2 Open Coding Agents Soft-Verified Agents
SERA CLI is a command-line tool created by AllenAI to enable developers to interact with the SERA (Soft-Verified Efficient Repository Agents) model family using Claude Code as the execution front end. It provides a convenient interface for deploying, testing, and using SERA models without needing to write scaffold code from scratch, acting as both a proxy and utility wrapper to simplify workflows that involve large agent models. Through sera-cli, users can connect to local or cloud-hosted SERA deployments, including via Modal for quick GPU provisioning and model caching, which helps accelerate experiments. The project is targeted at practitioners and researchers in the AI space who need a flexible but powerful CLI interface for model invocation, endpoint configuration, and integration with development pipelines. -
8
MiniMind-V
"Big Model" trains a visual multimodal VLM with 26M parameters
MiniMind-V is an experimental open-source project that aims to train a very small multimodal vision–language model (VLM) from scratch with extremely low compute and cost, making research and experimentation accessible to more people. The repository showcases training workflows and code designed to produce a 26-million parameter model—including both image and text capabilities—using minimal resources in very little time, reflecting a trend toward democratizing AI research. -
9
LLM Foundry
LLM training code for MosaicML foundation models
Introducing MPT-7B, the first entry in our MosaicML Foundation Series. MPT-7B is a transformer trained from scratch on 1T tokens of text and code. It is open source, available for commercial use, and matches the quality of LLaMA-7B. MPT-7B was trained on the MosaicML platform in 9.5 days with zero human intervention at a cost of ~$200k. Large language models (LLMs) are changing the world, but for those outside well-resourced industry labs, it can be extremely difficult to train and deploy these models. ... -
The Most Powerful Software Platform for EHSQ and ESG ManagementChoose from a complete set of software solutions across EHSQ that address all aspects of top performing Environmental, Health and Safety, and Quality management programs.
-
10
machine learning tutorials
machine learning tutorials (mainly in Python3)
...The repository integrates numerous popular machine learning frameworks and libraries such as scikit-learn, PyTorch, TensorFlow, XGBoost, and Hugging Face. It aims to strike a balance between theoretical explanation and practical coding by demonstrating algorithms both from scratch and using established libraries. The content is organized into multiple sections covering topics such as clustering, regression, dimensionality reduction, recommender systems, and model evaluation. -
11
Skywork-R1V4
Skywork-R1V is an advanced multimodal AI model series
...The project introduces a model architecture that transfers the reasoning abilities of advanced text-based models into visual domains so the system can interpret images and perform multi-step reasoning about them. Instead of retraining both language and vision models from scratch, the framework uses a lightweight visual projection layer that connects a pretrained vision backbone with a reasoning-capable language model. This design allows the model to analyze images while maintaining strong textual reasoning performance, enabling tasks such as solving visual math problems, interpreting scientific diagrams, and answering questions about images. -
12
nanochat
The best ChatGPT that $100 can buy
nanochat is a from-scratch, end-to-end “mini ChatGPT” that shows the entire path from raw text to a chatty web app in one small, dependency-lean codebase. The repository stitches together every stage of the lifecycle: tokenizer training, pretraining a Transformer on a large web corpus, mid-training on dialogue and multiple-choice tasks, supervised fine-tuning, optional reinforcement learning for alignment, and finally efficient inference with caching. -
13
Step1X-3D
High-Fidelity and Controllable Generation of Textured 3D Assets
Step1X-3D is an open-source framework for generating high-fidelity textured 3D assets from scratch — both their geometry and surface textures — using modern generative AI techniques. It combines a hybrid architecture: a geometry generation stage using a VAE-DiT model to output a watertight 3D representation (e.g. TSDF surface), and a texture synthesis stage that conditions on geometry and optionally reference input (or prompts) to produce view-consistent textures using a diffusion-based texture module. ... -
14
handy-ollama
Implement CPU from scratch and play with large model deployments
handy-ollama is an open-source educational project designed to help developers and AI enthusiasts learn how to deploy and run large language models locally using the Ollama platform. The repository serves as a structured tutorial that explains how to install, configure, and use Ollama to run modern language models on personal hardware without requiring advanced infrastructure. A key focus of the project is enabling users to run large models even without GPUs by leveraging optimized CPU-based... -
15
Homemade Machine Learning
Python examples of popular machine learning algorithms
homemade-machine-learning is a repository by Oleksii Trekhleb containing Python implementations of classic machine-learning algorithms done “from scratch”, meaning you don’t rely heavily on high-level libraries but instead write the logic yourself to deepen understanding. Each algorithm is accompanied by mathematical explanations, visualizations (often via Jupyter notebooks), and interactive demos so you can tweak parameters, data, and observe outcomes in real time. The purpose is pedagogical: you’ll see linear regression, logistic regression, k-means clustering, neural nets, decision trees, etc., built in Python using fundamentals like NumPy and Matplotlib, not hidden behind API calls. ... -
16
Prompt Engineering Interactive Tutorial
Anthropic's Interactive Prompt Engineering Tutorial
...The course leans heavily on realistic failure modes (ambiguity, hallucination, brittle instructions) and shows how to iteratively debug prompts the way you would debug code. Lessons include building prompts from scratch for common tasks like extraction, classification, transformation, and step-by-step reasoning, with checkpoints that let you compare your outputs against solid baselines. You’ll also practice advanced patterns such as tool use, constrained generation, and response validation so outputs are trustworthy and machine-consumable. -
17
DeiT (Data-efficient Image Transformers)
Official DeiT repository
DeiT (Data-efficient Image Transformers) shows that Vision Transformers can be trained competitively on ImageNet-1k without external data by using strong training recipes and knowledge distillation. Its key idea is a specialized distillation strategy—including a learnable “distillation token”—that lets a transformer learn effectively from a CNN or transformer teacher on modest-scale datasets. The project provides compact ViT variants (Tiny/Small/Base) that achieve excellent... -
18
TensorFlow Hub
A library for transfer learning by reusing parts of TensorFlow models
TensorFlow Hub is a repository that provides a library and platform for publishing, discovering, and reusing pre-trained machine learning models built with TensorFlow. The project enables developers to integrate high-quality models into their applications without needing to train them from scratch. Through TensorFlow Hub, researchers and practitioners can share reusable model components such as image classifiers, text embedding models, and object detection networks. These models can be loaded directly into TensorFlow pipelines and fine-tuned for new tasks using transfer learning techniques. The repository supports contributions from the community, allowing developers to submit models that become available for use by other machine learning practitioners. ... -
19
vits_chinese
Best practice TTS based on BERT and VITS
vits_chinese is an implementation of the VITS end-to-end text-to-speech (TTS) architecture tailored for Chinese (and possibly multilingual) speech synthesis. VITS is a model combining variational autoencoders (VAEs), normalizing flows, adversarial learning, and a stochastic duration predictor — a design that enables generation of natural, expressive speech, capturing variations in rhythm and prosody. By customizing or porting VITS for Chinese, this project aims to produce high-quality TTS... -
20
ControlNet
Let us control diffusion models
ControlNet is a neural network architecture designed to add conditional control to text-to-image diffusion models. Rather than training from scratch, ControlNet “locks” the weights of a pre-trained diffusion model and introduces a parallel trainable branch that learns additional conditions—like edges, depth maps, segmentation, human pose, scribbles, or other guidance signals. This allows the system to control where and how the model should focus during generation, enabling users to steer layout, structure, and content more precisely than prompt text alone. ... -
21
LLaMA
Inference code for Llama models
...This repo is a core piece of the Llama model infrastructure, used by researchers and developers to run LLaMA models locally or in their infrastructure. It is meant for inference (not training from scratch) and connects with aspects like model cards, responsible use, licensing, etc. -
22
Bot on Anything
Large model-based chatbot builder that can quickly integrate AI models
...Configuration is handled simply through a central JSON file where you define which model and which application channel you want to glue together, so developers can create sophisticated AI assistants without rewriting integration code from scratch. The architecture emphasizes reusability and extensibility, allowing the addition of new model backends or new channels with relative ease. It supports switching between multiple AI models and targets within the same project. -
23
GPT-NeoX
Implementation of model parallel autoregressive transformers on GPUs
...We aim to make this repo a centralized and accessible place to gather techniques for training large-scale autoregressive language models, and accelerate research into large-scale training. For those looking for a TPU-centric codebase, we recommend Mesh Transformer JAX. If you are not looking to train models with billions of parameters from scratch, this is likely the wrong library to use. For generic inference needs, we recommend you use the Hugging Face transformers library instead which supports GPT-NeoX models. -
24
TextBox
A text generation library with pre-trained language models github.com
TextBox 2.0 is an up-to-date text generation library based on Python and PyTorch focusing on building a unified and standardized pipeline for applying pre-trained language models to text generation. From a task perspective, we consider 13 common text generation tasks such as translation, story generation, and style transfer, and their corresponding 83 widely-used datasets. From a model perspective, we incorporate 47 pre-trained language models/modules covering the categories of general,... -
25
Karlo
Text-conditional image generation model based on OpenAI's unCLIP
Karlo is a text-conditional image generation model based on OpenAI's unCLIP architecture with the improvement over the standard super-resolution model from 64px to 256px, recovering high-frequency details only in the small number of denoising steps. We train all components from scratch on 115M image-text pairs including COYO-100M, CC3M, and CC12M. In the case of Prior and Decoder, we use ViT-L/14 provided by OpenAI’s CLIP repository. Unlike the original implementation of unCLIP, we replace the trainable transformer in the decoder into the text encoder in ViT-L/14 for efficiency. In the case of the SR module, we first train the model using the DDPM objective in 1M steps, followed by additional 234K steps to fine-tune the additional component.
