LTX-2.3 SRX FP8 E4M3 Experimental
This checkpoint was converted from a mixed BF16 / FP8 checkpoint by converting all BF16 tensors to FP8_E4M3
The main difference is checkpoint size and inference efficiency — this is not a new visual style or a visually enhanced model.
Key Points
- Smaller checkpoint: about 21.4 GB
- Aggressive FP8-E4M3 conversion
- Original model architecture preserved
- Lower VRAM usage than heavier BF16 checkpoints
- Best for local ComfyUI users who want a smaller experimental LTX-2.3 checkpoint
Conversion:
- BF16 tensors → FP8_E4M3
- Existing FP8_E4M3 tensors preserved
- F32 and U8 tensors preserved
License
This LoRA is an adaptation/derivative of Lightricks/LTX-2.3.
It is distributed under the LTX-2 Community License Agreement and is subject to the same upstream terms, use-based restrictions, and commercial-use requirements as LTX-2.3.
- Base model:
Lightricks/LTX-2.3
- License tag on Hugging Face:
other
- This repository does not grant rights beyond the upstream LTX-2 license.
- Users are responsible for complying with the LTX-2 Community License Agreement.
Commercial use is subject to the upstream LTX licensing terms. Companies/entities that meet the commercial-license threshold should contact Lightricks for a commercial license.
