Experiments
LoRA Fine-Tuning Baseline
February 11, 2026 · This experiment establishes a baseline LoRA fine-tuning workflow for a small instruction model, with emphasis on repeatability and evaluation quality.
Diffusion Schedulers: A Quick Comparison
February 10, 2026 · This note compares DDIM, Euler, and DPM++ schedulers on image quality, speed, and prompt consistency for the same checkpoint and seed set.