4 ms·

Is fine-tuning / lora training supported? just toyed around with it, initial results are promising - more so than the recent Ace Step 1.5