| --- |
| license: other |
| license_name: ltx-2-community-license-agreement |
| license_link: https://github.com/Lightricks/LTX-2/blob/main/LICENSE |
| base_model: |
| - Lightricks/LTX-2.5 |
| base_model_relation: quantized |
| pipeline_tag: text-to-video |
| library_name: diffusers |
| tags: |
| - video-generation |
| - text-to-video |
| - image-to-video |
| - audio-to-video |
| - quantized |
| - 4-bit |
| - int4 |
| - svdquant |
| - ltx-2 |
| - ltx-2.5 |
| - quantfunc |
| --- |
| |
| # LTX-2.5-QuantFunc-4bit |
|
|
| **4-bit quantized [Lightricks LTX-2.5](https://huggingface.co/Lightricks/LTX-2.5) 22B distilled — pure 4-bit inference holds up to BF16 quality while cutting weight/VRAM footprint to roughly 1/3.** Quantized and loaded by the [QuantFunc](https://github.com/quantfunc) inference engine. |
|
|
| ## Why pure 4-bit holds up |
|
|
| Compressing a 22B joint video/audio DiT down to **4-bit activations × 4-bit weights (pure a4w4)** runs into one main enemy: activation outliers, which blow past 4-bit's dynamic range and show up as blur and lost detail. QuantFunc adds two layers on top of the SVDQuant approach: |
|
|
| 1. **A much stronger low-rank branch.** Algorithmic improvements let the low-rank branch absorb significantly more activation-outlier energy than standard SVDQuant — outliers get siphoned off before the main path is quantized to 4-bit, so the main path sees a much cleaner distribution. |
| 2. **A smoother residual.** A combination of in-house techniques further smooths the outlier distribution in what's left after low-rank absorption, so the 4-bit grid covers the remaining distribution with far less error. |
|
|
| Net effect: **at pure a4w4 inference, output video (including audio) clarity, detail retention and motion coherence are on par with the BF16 baseline** — at roughly 1/3 the memory. |
|
|
| ## Files |
|
|
| Quantizes the base checkpoint — [`ltx-2.5-22b-distilled-transformer-bf16.safetensors`](https://huggingface.co/Lightricks/LTX-2.5) from **Lightricks LTX-2.5 (22B, distilled variant)**: |
|
|
| | File | Format | Size | Minimum GPU | |
| |---|---|---|---| |
| | `ltx-2.5-22b-distilled-quantfunc-int4-r128.safetensors` | INT4 weights / INT4 activations (SVDQuant, rank-128) | ~14.06 GiB | NVIDIA SM120+ (Blackwell, e.g. RTX 50-series / RTX 6000D) | |
|
|
| ## Produced & loadable ONLY by QuantFunc |
|
|
| This is **not** a standard diffusers/safetensors checkpoint. The file is produced by, and only loadable by, the **QuantFunc inference engine** (native C++/CUDA engine) or the **[ComfyUI-QuantFunc](https://github.com/quantfunc)** custom-node plugin for ComfyUI. It will not load in vanilla diffusers, ComfyUI's stock LTX loader, or any other inference stack. (This repo declares `library_name: diffusers` purely so Hugging Face counts real downloads of the flat `.safetensors` file below — see [HF's download-stats docs](https://huggingface.co/docs/hub/models-download-stats) — not because the file loads via the `diffusers` Python package.) |
|
|
| ## Inference settings (required for correct output) |
|
|
| This is the **distilled** checkpoint, joint video+audio. It must be run with the distilled model's own schedule — a generic/default scheduler will produce degraded output: |
|
|
| - **Sigma schedule (8 steps, fixed):** `1.0, 0.99375, 0.9875, 0.98125, 0.975, 0.909375, 0.725, 0.421875, 0.0` |
| - **CFG:** `1` (no classifier-free guidance — this is a guidance-distilled model) |
| - **STG:** disabled (no Spatio-Temporal Guidance) |
|
|
| ## Original / base model |
|
|
| - Base checkpoint: [`Lightricks/LTX-2.5`](https://huggingface.co/Lightricks/LTX-2.5) → `ltx-2.5-22b-distilled-transformer-bf16.safetensors` (~42 GB, bf16) |
| - Publisher: [Lightricks](https://huggingface.co/Lightricks) |
| - Upstream repo / inference code: [github.com/Lightricks/LTX-2](https://github.com/Lightricks/LTX-2) |
|
|
| This repository redistributes **derived (quantized) weights only**, produced from the base checkpoint above via the QuantFunc SVDQuant pipeline. No architecture or training changes were made — only post-training quantization (weight + activation, 4-bit, rank-128 SVD residual). |
|
|
| ## License |
|
|
| Follows the original model's **LTX-2 Community License Agreement** ([license text](https://github.com/Lightricks/LTX-2/blob/main/LICENSE)). This repository distributes quantized weights only; copyright and licensing of the original model belong to Lightricks. |
|
|
| ## Community |
|
|
| - 🌐 [QuantFunc website](https://www.quantfunc.com/) |
| - 🐙 [ComfyUI-QuantFunc on GitHub](https://github.com/QuantFunc/ComfyUI-QuantFunc) |
| - 🤗 [Hugging Face](https://huggingface.co/QuantFunc) | 🤖 [ModelScope](https://www.modelscope.cn/profile/QuantFunc) |
| - 🎮 [Discord](https://discord.gg/jCp9TpFWcn) |
|
|
