Stream-Render: release trained renderer and causal decoder
#1
by yitongl - opened
Release only the trained 1008b step224 EMA renderer (LoRA merged) and distilled causal ViT24 step6000 decoder. Includes reconstruction, integrity records, MiniMax-H3 license/notice, and MiniMax-H3 acknowledgement. Unchanged upstream models are intentionally excluded.
Thanks! Merging the renderer and causal decoder release.
Owen777 changed pull request status to merged