Add LTX-Video 0.9.8 distilled models (#2005)

- Add LTX-Video 2B Distilled v0.9.8 (6.34GB)
- Add LTX-Video 2B Distilled FP8 v0.9.8 (4.46GB)
- Add LTX-Video 13B Distilled v0.9.8 (28.6GB)
- Add LTX-Video 13B Distilled FP8 v0.9.8 (15.7GB)

These v0.9.8 models feature improved prompt understanding and detail generation.
Both 2B and 13B variants available in standard and FP8 quantized versions.

Co-authored-by: gschreiber <gschreiber@infra-image-generator.c.ltx-research-vms.internal>
This commit is contained in:
Gilad Schreiber 2025-07-17 01:38:53 +03:00 committed by GitHub
parent d30459cc34
commit 6afaf5eaf5
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194

View File

@ -5147,6 +5147,50 @@
"url": "https://huggingface.co/Lightricks/LTX-Video/resolve/main/ltxv-13b-0.9.7-distilled-fp8.safetensors",
"size": "15.7GB"
},
{
"name": "LTX-Video 2B Distilled v0.9.8",
"type": "checkpoint",
"base": "LTX-Video",
"save_path": "checkpoints/LTXV",
"description": "LTX-Video 2B distilled model v0.9.8 with improved prompt understanding and detail generation.",
"reference": "https://huggingface.co/Lightricks/LTX-Video",
"filename": "ltxv-2b-0.9.8-distilled.safetensors",
"url": "https://huggingface.co/Lightricks/LTX-Video/resolve/main/ltxv-2b-0.9.8-distilled.safetensors",
"size": "6.34GB"
},
{
"name": "LTX-Video 2B Distilled FP8 v0.9.8",
"type": "checkpoint",
"base": "LTX-Video",
"save_path": "checkpoints/LTXV",
"description": "Quantized LTX-Video 2B distilled model v0.9.8 with improved prompt understanding and detail generation, optimized for lower VRAM usage.",
"reference": "https://huggingface.co/Lightricks/LTX-Video",
"filename": "ltxv-2b-0.9.8-distilled-fp8.safetensors",
"url": "https://huggingface.co/Lightricks/LTX-Video/resolve/main/ltxv-2b-0.9.8-distilled-fp8.safetensors",
"size": "4.46GB"
},
{
"name": "LTX-Video 13B Distilled v0.9.8",
"type": "checkpoint",
"base": "LTX-Video",
"save_path": "checkpoints/LTXV",
"description": "LTX-Video 13B distilled model v0.9.8 with improved prompt understanding and detail generation.",
"reference": "https://huggingface.co/Lightricks/LTX-Video",
"filename": "ltxv-13b-0.9.8-distilled.safetensors",
"url": "https://huggingface.co/Lightricks/LTX-Video/resolve/main/ltxv-13b-0.9.8-distilled.safetensors",
"size": "28.6GB"
},
{
"name": "LTX-Video 13B Distilled FP8 v0.9.8",
"type": "checkpoint",
"base": "LTX-Video",
"save_path": "checkpoints/LTXV",
"description": "Quantized LTX-Video 13B distilled model v0.9.8 with improved prompt understanding and detail generation, optimized for lower VRAM usage.",
"reference": "https://huggingface.co/Lightricks/LTX-Video",
"filename": "ltxv-13b-0.9.8-distilled-fp8.safetensors",
"url": "https://huggingface.co/Lightricks/LTX-Video/resolve/main/ltxv-13b-0.9.8-distilled-fp8.safetensors",
"size": "15.7GB"
},
{
"name": "LTX-Video 13B Distilled LoRA v0.9.7",
"type": "lora",
@ -5191,6 +5235,17 @@
"url": "https://huggingface.co/Lightricks/LTX-Video-ICLoRA-canny-13b-0.9.7/resolve/main/ltxv-097-ic-lora-canny-control-comfyui.safetensors",
"size": "81.9MB"
},
{
"name": "LTX-Video ICLoRA Detailer 13B v0.9.8",
"type": "lora",
"base": "LTX-Video",
"save_path": "loras",
"description": "A video detailer model on top of LTXV_13B_098_DEV trained on custom data using In-Context LoRA (IC LoRA) method.",
"reference": "https://huggingface.co/Lightricks/LTX-Video-ICLoRA-detailer-13b-0.9.8",
"filename": "ltxv-098-ic-lora-detailer-comfyui.safetensors",
"url": "https://huggingface.co/Lightricks/LTX-Video-ICLoRA-detailer-13b-0.9.8/resolve/main/ltxv-098-ic-lora-detailer-comfyui.safetensors",
"size": "1.31GB"
},
{
"name": "Latent Bridge Matching for Image Relighting",
"type": "diffusion_model",