Views
No views yet
LTX2TwoStagePipeline. The runtime loads the base FP8 transformer and applies the LTX-2.3 distilled stage-2 LoRA through the two-stage pipeline.1sglang generate \
2 --model-path Lightricks/LTX-2.3 \
3 --pipeline-class-name LTX2TwoStagePipeline \
4 --transformer-path BBuf/ltx23-two-stage-modelopt-fp8-sglang-transformer \
5 --prompt "SpongeBob talking with patrick" \
6 --width 768 --height 512 --num-frames 121 \
7 --num-inference-steps 8 --warmup