Experimental FP8_E4M3 repack of LTX-2.3-22b-dev.
This checkpoint was converted from a mixed BF16 / FP8 checkpoint by converting all BF16 tensors to FP8_E4M3 while preserving tensor names, tensor shapes, and model architecture.
Original: ltx-2.3-22b-dev.safetensors
Converted: ltx23_srx_fp8_e4m3_experimental.safetensors
Original size: ~46.1 GB
Converted size: ~21.4 GB
Conversion: BF16 tensors -> FP8_E4M3 Existing FP8_E4M3 tensors preserved F32 and U8 tensors preserved
Experimental. Tested in ComfyUI and confirmed to load with audio output.
Description
Experimental FP8_E4M3 repack.
Converted from a mixed BF16 / FP8 LTX-2.3 checkpoint by converting all BF16 tensors to FP8_E4M3.
- Original size: ~46.1 GB
- Converted size: ~21.4 GB
- Tensor names preserved
- Tensor shapes preserved
- Architecture preserved
- BF16 tensors converted to FP8_E4M3
- Existing FP8_E4M3, F32 and U8 tensors preserved
- Tested in ComfyUI and confirmed to load with audio output
This is an experimental dtype conversion / repack, not a newly trained model and not a merge.