nielsr HF Staff commited on
Commit
498a37f
·
verified ·
1 Parent(s): 96bf7fc

Add pipeline tag and improve model card

Browse files

Hi! I'm Niels from the Hugging Face community science team.

This PR improves the model card for SAMA-14B by:
- Adding the `image-to-video` pipeline tag to the metadata for better discoverability.
- Including a "Quick Start" section with installation and inference instructions taken from the official GitHub repository.
- Organizing the content to include helpful links to the paper, project page, and code.

These changes help users understand how to set up and run the model effectively.

Files changed (1) hide show
  1. README.md +31 -3
README.md CHANGED
@@ -1,7 +1,8 @@
1
-
2
  ---
3
  license: apache-2.0
 
4
  ---
 
5
  # SAMA: Factorized Semantic Anchoring and Motion Alignment for Instruction-Guided Video Editing
6
 
7
  <div align="center">
@@ -12,13 +13,40 @@ license: apache-2.0
12
  <a href="https://github.com/Cynthiazxy123/SAMA" target="_blank"><img src="https://img.shields.io/badge/Code-111111.svg?logo=github&logoColor=white" height="22px"></a>
13
  </div>
14
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
15
 
 
 
 
 
 
 
 
 
 
 
 
 
 
16
 
17
  ## 🤗 Available Models
18
 
19
  | Model | Status | Link |
20
  | --- | --- | --- |
21
- | SAMA-5B | Coming soon | Coming soon |
22
  | SAMA-14B | Available | [syxbb/SAMA-14B](https://huggingface.co/syxbb/SAMA-14B) |
23
 
24
  ## 📚 Citation
@@ -33,4 +61,4 @@ license: apache-2.0
33
  primaryClass={cs.CV},
34
  url={https://arxiv.org/abs/2603.19228},
35
  }
36
- ```
 
 
1
  ---
2
  license: apache-2.0
3
+ pipeline_tag: image-to-video
4
  ---
5
+
6
  # SAMA: Factorized Semantic Anchoring and Motion Alignment for Instruction-Guided Video Editing
7
 
8
  <div align="center">
 
13
  <a href="https://github.com/Cynthiazxy123/SAMA" target="_blank"><img src="https://img.shields.io/badge/Code-111111.svg?logo=github&logoColor=white" height="22px"></a>
14
  </div>
15
 
16
+ SAMA (factorized **S**emantic **A**nchoring and **M**otion **A**lignment) is an instruction-guided video editing framework. It factorizes video editing into two parts: semantic anchoring to establish structural planning and motion alignment to internalize temporal dynamics. This approach enables precise semantic modifications while faithfully preserving the original motion of the source video.
17
+
18
+ ## 🚀 Quick Start
19
+
20
+ ### 🛠️ Installation
21
+
22
+ Recommended environment: Linux, NVIDIA GPU, CUDA 12.1, and Python 3.10.
23
+
24
+ ```bash
25
+ git clone https://github.com/Cynthiazxy123/SAMA
26
+ cd SAMA
27
+
28
+ conda create -n sama python=3.10 -y
29
+ conda activate sama
30
 
31
+ pip install --upgrade pip
32
+ pip install -r requirements.txt
33
+ ```
34
+
35
+ ### ▶️ Inference
36
+
37
+ To run instruction-guided video editing, you will need the base `Wan2.1-T2V-14B` model and the SAMA checkpoint.
38
+
39
+ The inference script is located at `infer_sh/run_sama.sh`. Edit the variables at the top of that script (such as `MODEL_ROOT`, `STATE_DICT`, `SRC_VIDEO`, and `PROMPT`) and then run:
40
+
41
+ ```bash
42
+ bash infer_sh/run_sama.sh
43
+ ```
44
 
45
  ## 🤗 Available Models
46
 
47
  | Model | Status | Link |
48
  | --- | --- | --- |
49
+ | SAMA-5B | Coming soon | - |
50
  | SAMA-14B | Available | [syxbb/SAMA-14B](https://huggingface.co/syxbb/SAMA-14B) |
51
 
52
  ## 📚 Citation
 
61
  primaryClass={cs.CV},
62
  url={https://arxiv.org/abs/2603.19228},
63
  }
64
+ ```