Instructions to use oumoumad/ltx-2.3-dearchive-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use oumoumad/ltx-2.3-dearchive-lora with PEFT:
Task type is invalid.
- Notebooks
- Google Colab
- Kaggle
model card: IC = In-Context (was Image-Conditioning)
Browse files
README.md
CHANGED
|
@@ -16,12 +16,12 @@ library_name: peft
|
|
| 16 |
|
| 17 |
# dearchive 路 LTX-2.3 IC-LoRA
|
| 18 |
|
| 19 |
-
An **
|
| 20 |
|
| 21 |
| | |
|
| 22 |
|---|---|
|
| 23 |
| Base model | `Lightricks/LTX-2.3` (`ltx-2.3-22b-dev.safetensors`) |
|
| 24 |
-
| Strategy | `video_to_video` (
|
| 25 |
| LoRA rank / alpha | 128 / 128 |
|
| 26 |
| Trainable params | 855,638,016 (~0.86 B) |
|
| 27 |
| Optimizer | Prodigy (D-Adaptation), `lr=1.0`, bias-correction + safeguard-warmup |
|
|
|
|
| 16 |
|
| 17 |
# dearchive 路 LTX-2.3 IC-LoRA
|
| 18 |
|
| 19 |
+
An **In-Context LoRA** for [LTX-2.3](https://huggingface.co/Lightricks/LTX-2.3) (dev, 22B) that turns archive-style video into a clean, modern, HD equivalent. Fed a real low-res low-bitrate web archive clip (Lanczos-upscaled to model resolution), the model regenerates the frame as if it had been shot today.
|
| 20 |
|
| 21 |
| | |
|
| 22 |
|---|---|
|
| 23 |
| Base model | `Lightricks/LTX-2.3` (`ltx-2.3-22b-dev.safetensors`) |
|
| 24 |
+
| Strategy | `video_to_video` (in-context, reference-conditioned) |
|
| 25 |
| LoRA rank / alpha | 128 / 128 |
|
| 26 |
| Trainable params | 855,638,016 (~0.86 B) |
|
| 27 |
| Optimizer | Prodigy (D-Adaptation), `lr=1.0`, bias-correction + safeguard-warmup |
|