Joy Caption Alpha Two
Open, unrestricted image-captioning VLM with an HF demo Space.
Local-AI
Yan Desbiens
One-GPU local-first AI lab: cognitive architectures, agent runtimes, AEON.
Strix Halo × Qwen3.8 (pwilkin)
Reproducible custom-ROCm llama.cpp stack for Qwen3.8-27B on Strix Halo.
Train Your Own Decision Model with Unsloth
Fine-tune Qwen/Gemma into decision classifiers with calibrated probabilities.