Instructions to use Perflow-Shuai/longlive_2.0_5B_tmp_20260507 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Wan2.2
How to use Perflow-Shuai/longlive_2.0_5B_tmp_20260507 with Wan2.2:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
Update README.md
Browse files
README.md
CHANGED
|
@@ -10,6 +10,10 @@ tags:
|
|
| 10 |
- wan2.2
|
| 11 |
---
|
| 12 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
# LongLive2.0 5B Checkpoints
|
| 14 |
|
| 15 |
This repository hosts temporary LongLive2.0 5B checkpoints for inference with
|
|
@@ -91,12 +95,14 @@ If the LoRA checkpoint is not used, remove the `adapter` section and leave
|
|
| 91 |
|
| 92 |
## Prompt Folder
|
| 93 |
|
| 94 |
-
`data.data_path`
|
|
|
|
| 95 |
|
| 96 |
- a `.txt` file, where each line is one single-shot prompt; or
|
| 97 |
- a directory of multi-shot prompt folders.
|
| 98 |
|
| 99 |
-
|
|
|
|
| 100 |
|
| 101 |
```text
|
| 102 |
inference_prompts/
|
|
@@ -107,6 +113,18 @@ inference_prompts/
|
|
| 107 |
shot_durations.txt
|
| 108 |
```
|
| 109 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 110 |
Each JSON file contains:
|
| 111 |
|
| 112 |
```json
|
|
@@ -148,3 +166,16 @@ Outputs are written to `output_folder`.
|
|
| 148 |
- `inference.multi_shot_rope_offset` controls the multi-shot RoPE offset.
|
| 149 |
- For NVFP4 inference, use the separate NVFP4 config and setup instructions in
|
| 150 |
the LongLive2.0 documentation.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
- wan2.2
|
| 11 |
---
|
| 12 |
|
| 13 |
+
<p align="center">
|
| 14 |
+
<img src="https://raw.githubusercontent.com/wileewang/LongLive2.0/release-clean-merge/assets/longlive2/logo.png" alt="LongLive2.0 logo" width="100%">
|
| 15 |
+
</p>
|
| 16 |
+
|
| 17 |
# LongLive2.0 5B Checkpoints
|
| 18 |
|
| 19 |
This repository hosts temporary LongLive2.0 5B checkpoints for inference with
|
|
|
|
| 95 |
|
| 96 |
## Prompt Folder
|
| 97 |
|
| 98 |
+
`data.data_path` is passed to `MultiTextConcatDataset` in `inference.py`. It can
|
| 99 |
+
be either:
|
| 100 |
|
| 101 |
- a `.txt` file, where each line is one single-shot prompt; or
|
| 102 |
- a directory of multi-shot prompt folders.
|
| 103 |
|
| 104 |
+
For a directory input, the code supports both of the following layouts. The
|
| 105 |
+
direct caption-root layout is the simplest:
|
| 106 |
|
| 107 |
```text
|
| 108 |
inference_prompts/
|
|
|
|
| 113 |
shot_durations.txt
|
| 114 |
```
|
| 115 |
|
| 116 |
+
It also supports a dataset root with an outer `caption/` folder:
|
| 117 |
+
|
| 118 |
+
```text
|
| 119 |
+
inference_prompts/
|
| 120 |
+
caption/
|
| 121 |
+
robot_lab_demo/
|
| 122 |
+
0.json
|
| 123 |
+
1.json
|
| 124 |
+
2.json
|
| 125 |
+
shot_durations.txt
|
| 126 |
+
```
|
| 127 |
+
|
| 128 |
Each JSON file contains:
|
| 129 |
|
| 130 |
```json
|
|
|
|
| 166 |
- `inference.multi_shot_rope_offset` controls the multi-shot RoPE offset.
|
| 167 |
- For NVFP4 inference, use the separate NVFP4 config and setup instructions in
|
| 168 |
the LongLive2.0 documentation.
|
| 169 |
+
|
| 170 |
+
## Citation
|
| 171 |
+
|
| 172 |
+
Citation will be updated after the paper is released.
|
| 173 |
+
|
| 174 |
+
```bibtex
|
| 175 |
+
@article{longlive2,
|
| 176 |
+
title = {LongLive2.0: An NVFP4 Parallel Infrastructure for Long Video Generation},
|
| 177 |
+
author = {TODO},
|
| 178 |
+
journal = {TODO},
|
| 179 |
+
year = {2026}
|
| 180 |
+
}
|
| 181 |
+
```
|