Yohann Bearzi
bearzi
AI & ML interests
Uploading mixed-precision MLX quantizations of popular open-weight models, produced with oMLX’s oQ and JANG on apple silicon. Benchmarks and comparisons are welcome.
Recent Activity
new activity about 4 hours ago
bearzi/gemma-4-31b-it-oQ3:Doesn't seem to work. updated a model about 5 hours ago
bearzi/gemma-4-E2B-it-oQ4 updated a model about 20 hours ago
bearzi/gemma-4-31b-it-oQ8