TypeScript framework for fine-tuning
Curated list of open-weight AI models with commercially exploitable licenses, verified benchmarks, and no EU restrictions.
“A clean, from-scratch implementation of the OLMo architecture with KV caching, RoPE, and an efficient autoregressive inference pipeline. Designed as a minimal yet extensible foundation for post-training research, including RLHF, preference optimization, and reasoning-focused systems.”
Fine-tune small open models into reliable JSON tool callers. Reproducible, end to end, on a single GPU.
Benchmarking open-weight LLM coding agents as SCOUT delegates: model comparison experiments with pre-registered protocols, blind scoring, and full data.
Frontier-Grade Open Weights — フロンティア級のオープンウェイトモデルは、開かれたのか / They matched the frontier. But no one can hold them.
Tamper Attack Resistance (BlueDot Impact Technical AI Safety Project)
Klepon: Taste of Indonesia
Experimental framework for cross-subtask malicious intent detection in stateless AI agents — research into open-weight model safety
Workspace-lens audit cards and training-workflow scaffolds for open-weight language models.
Time‑Shift LLM Integrity Tester
This project implements a complete research pipeline for detecting shortcut-driven reasoning in open-weight language models. The pipeline evaluates whether LLMs arrive at correct answers through genuine reasoning or through superficial shortcuts, using a combination of behavioral testing and mechanistic interpretability.
Psychometric reliability study of small open-weight LLM judges (Llama-3.1-8B, Qwen2.5-7B, Gemma-2-9B) versus Claude Sonnet 4.6 on LLMBar and SummEval: accuracy, test-retest consistency, position/verbosity bias, and ensembles. Fully reproducible, cached raw judgments regenerate every table and figure.
Source-verified survey of the agentic AI coding landscape: frontier & open models, agent harnesses, pricing, privacy/training policies, IDE compatibility, on-prem tiers, and per-use-case recommendations. Versioned editions, cross-model comparisons. Current edition: 2026-07.
The network where autonomous agents compete to make small open-weight models better — verified, not self-reported. Built on Base. codepit.fun
Com la tokenització fractura la morfologia catalana i si una segmentació conscient dels morfemes recupera la geometria. Provat en 3 llengües indoeuropees (català, castellà, anglès): el català es fragmenta ~1,7× més que l'anglès; forçar el tall morfèmic recupera la composicionalitat (robust a portadora i replicat en castellà).
MacOS Electron Client for LLMs that run locally and on the Cloud using LM Studio/Ollama/OpenRouter/Nvidia Build
Anthropic-style emotion-vector geometry, on any open-weight LLM, in one command. Frozen corpus + unified pipeline + statistical rigor + 5-model reference results.
Closing the Opus Gap: Systematic Optimization of Tool-Calling in Open-Weight LLMs on Wafer-Scale Hardware
Add a description, image, and links to the open-weight-models topic page so that developers can more easily learn about it.
To associate your repository with the open-weight-models topic, visit your repo's landing page and select "manage topics."