#moe

A 1.32B-parameter Mixture-of-Experts LLM trained from scratch in PyTorch - custom tokenizer, MLA attention, 18B-token pretraining, SFT and DPO. No pretrained weights, no HuggingFace Trainer.

Run local coding agents with full tool access and browser-based monitoring using an uncensored Ollama model on your hardware.

an ai model built from scratch that is 1.11 Billion Parameters [ yeah i know its an tiny model but this was only made by me

Production-oriented Kimi K3 inference control plane with checkpoint release gates, MoE capacity planning, OpenAI gateway, NVL72 deployment, benchmarks, and observability.

Run 35B mixture-of-experts LLMs on a 16 GB AMD Radeon GPU. A Rust inference engine for ROCm on Windows: reads the GGUF, solves weight placement across VRAM and system RAM, tunes by measurement, compacts context, and serves an OpenAI-compatible API with a web console. No Ollama, no daemon.

LocalSight · 198M-A64M 思考型 MoE LLM，2×RTX4090 从零训练（pretrain→SFT→SimPO→RLAIF→Agent RL），GGUF/Ollama 可运行

Output house rules for every DeepSeek Harness agent: no check-mark or cross glyphs in replies or written files, and a light kaomoji habit in conversation. One global prompt section, so every preset, workspace and subagent obeys without a single preset edit.