gemma3
Automodel
🚀 Pytorch Distributed native training library for LLMs/VLMs with OOTB Hugging Face support
- 1.1k
- Python
- Apache-2.0
- 3 h ago
llm-architecture-from-scratch
Implementing LLM architectures from scratch in PyTorch
- 856
- Python
- BSD-3-Clause
- 1 d ago
kolosal-cli
Super lightweight Ollama + Qwen Code alternative to run Llama 3.3, DeepSeek-R1, Phi-4, Gemma 3, Mistral Small 3.1 and other large language models.
- 465
- TypeScript
- Apache-2.0
- 11 mo ago
maxheadbox
Tiny truly local voice-activated LLM Agent that runs on a Raspberry Pi
- 358
- JavaScript
- GPL-3.0
- 4 mo ago
Kolosal
Kolosal AI is an OpenSource and Lightweight alternative to LM Studio to run LLMs 100% offline on your device.
- 462
- C++
- Apache-2.0
- 18 mo ago
functiongemma-270m-ft-mobile-actions
Text generation model, used with litert-lm.
Gated access
- 337
- 953
- gemma
- 2 wk ago
Awsome_AI_Agents
This is end to end course on AI Agents and Agentic AI with 15+ AI Agent Projects with real time use cases and industry expertise.
- 231
- Jupyter Notebook
- Apache-2.0
- 7 mo ago
nabu
A multi engine TTS & LLM edge computing playground with audio book features and more!
- 62
- Kotlin
- GPL-3.0
- 5 wk ago
gemma_penzai
A JAX Research Toolkit for Visualizing, Manipulating, and Understanding Gemma Models with Multi-modal Support based on Penzai.
- 101
- Jupyter Notebook
- Apache-2.0
- 10 mo ago
nextjs-offline-whisper-to-llm
An offline NextJS application to allow users to locally transcode audio recordings into accurate, plain-text transcripts, then to use large language models to distill insights..
- 50
- TypeScript
- MIT
- 12 mo ago
gemma3moe-builder
Sparse-upcycle a dense Gemma-3-270m into a Mixture-of-Experts (gemma3moe) and serve it on llama.cpp (CPU). Upcycle, train, inspect routing, GGUF.
- 39
- Python
- 4 mo ago
shadowquillapp
ShadowQuill is a privacy‑first, offline and free prompt engineering/generation app powered by self-hosted local LLMs
- 28
- TypeScript
- MIT
- 3 mo ago
gemma3_pdf_summarizer
A high-performance PDF summarization tool powered by Google's Gemma 3 LLM. Features parallel processing, async operations, and intelligent chunking for technical paper analysis. Built with FastAPI, Streamlit, and Ollama.
- 23
- Python
- 20 mo ago
kolosal-server
Kolosal AI is an OpenSource and Lightweight alternative to Ollama to run LLMs 100% offline on your device.
- 14
- C++
- Apache-2.0
- 11 mo ago
function-calling-gemma3
Demo project showcasing Gemma3 function calling capabilities using Ollama. Enables automatic web searches via Serper.dev for up-to-date information and features an interactive Gradio chat interface.
- 12
- Python
- 20 mo ago
litetune
Fine-tune Gemma, Qwen, FunctionGemma and EmbeddingGemma 2, convert to LiteRT-LM .litertlm for Android, iOS and desktop, and verify the converted model against the float one on CPU and GPU.
- 6
- Python
- Apache-2.0
- 7 d ago
llm-basic-benchmark
Comprehensive benchmark of 44 open source language models across creative writing, logic puzzles, counterfactual reasoning, and programming tasks. Tested on Apple M4 Max with detailed performance analysis.
- 5
- Python
- MIT
- 17 mo ago
kimi-agent-internals
🔍 Analyze the architecture of Moonshot AI's Kimi agent (K2.5 model) with insights on its performance and conversational extraction methodology.
- 2
- Fluent
- 6 wk ago
LTX-2-LoRAs-Camera-Control-Dolly
Demonstration for the Lightricks LTX-2 Distilled model, enhanced with specialized LoRA adapters for cinematic camera movements (dolly left/right/in/out, jib up/down, static). Generates animated videos from text prompts or input images, with optional prompt enhancement using Gemma-3-12b.
- 3
- Python
- Apache-2.0
- 10 mo ago
kolosal-desktop
Kolosal AI is an OpenSource and Lightweight alternative to LM Studio to run LLMs 100% offline on your device.
- 4
- Svelte
- Apache-2.0
- 18 mo ago
gemma-3-1b-paraphrase
Every weight of google/gemma-3-1b-it updated — no adapters — to turn a general chat model into a dedicated paraphraser. iBLEU 17.99 against 1.65 zero-shot, with semantic similarity held at 0.911 and copy rate down from 41.7% to 0.7%. Trained with Unsloth on paraphrases.
- 1
- Jinja
- MIT
- 4 wk ago
gemma3-rag
Gemma3 RAG benchmark system for Japanese river/dam/erosion control technical standards.
- 0
- Python
- 12 mo ago







