LTX-2.3 Two-Stage ModelOpt FP8 SGLang Transformer

SGLang-loadable ModelOpt FP8 transformer override for LTX2TwoStagePipeline. The runtime loads the base FP8 transformer and applies the LTX-2.3 distilled stage-2 LoRA through the two-stage pipeline.

Example:

sglang generate \
  --model-path Lightricks/LTX-2.3 \
  --pipeline-class-name LTX2TwoStagePipeline \
  --transformer-path BBuf/ltx23-two-stage-modelopt-fp8-sglang-transformer \
  --prompt "SpongeBob talking with patrick" \
  --width 768 --height 512 --num-frames 121 \
  --num-inference-steps 8 --warmup
Downloads last month
8
Safetensors
Model size
19B params
Tensor type
F32
BF16
F8_E4M3
Inference Providers NEW
This model isn't deployed by any Inference Provider. 馃檵 Ask for provider support

Model tree for BBuf/ltx23-two-stage-modelopt-fp8-sglang-transformer

Quantized
(60)
this model