Add pipeline tag, library name, and links
#1
by nielsr HF Staff - opened
README.md
CHANGED
|
@@ -1,12 +1,14 @@
|
|
| 1 |
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
base_model: Qwen/Qwen3-VL-8B-Instruct
|
|
|
|
| 4 |
tags:
|
| 5 |
- camera-movement
|
| 6 |
- video-understanding
|
| 7 |
- qwen3-vl
|
| 8 |
- sft
|
| 9 |
- distillation
|
|
|
|
|
|
|
| 10 |
---
|
| 11 |
|
| 12 |
# CamDistill-8B
|
|
@@ -17,10 +19,14 @@ Camera-movement distillation SFT model, based on `Qwen/Qwen3-VL-8B-Instruct`.
|
|
| 17 |
- Training framework: [ms-swift](https://github.com/modelscope/ms-swift)
|
| 18 |
- CamDistill config: DEPTH=8, EXTRACT_LAYERS=1,3,5,7,9,11,13,15, LAMBDA=0.05
|
| 19 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
## Usage
|
| 21 |
|
| 22 |
```python
|
| 23 |
from transformers import AutoModelForCausalLM, AutoProcessor
|
| 24 |
model = AutoModelForCausalLM.from_pretrained("ddz16/CamDistill-8B", torch_dtype="bfloat16", device_map="auto")
|
| 25 |
processor = AutoProcessor.from_pretrained("ddz16/CamDistill-8B")
|
| 26 |
-
```
|
|
|
|
| 1 |
---
|
|
|
|
| 2 |
base_model: Qwen/Qwen3-VL-8B-Instruct
|
| 3 |
+
license: apache-2.0
|
| 4 |
tags:
|
| 5 |
- camera-movement
|
| 6 |
- video-understanding
|
| 7 |
- qwen3-vl
|
| 8 |
- sft
|
| 9 |
- distillation
|
| 10 |
+
library_name: transformers
|
| 11 |
+
pipeline_tag: video-text-to-text
|
| 12 |
---
|
| 13 |
|
| 14 |
# CamDistill-8B
|
|
|
|
| 19 |
- Training framework: [ms-swift](https://github.com/modelscope/ms-swift)
|
| 20 |
- CamDistill config: DEPTH=8, EXTRACT_LAYERS=1,3,5,7,9,11,13,15, LAMBDA=0.05
|
| 21 |
|
| 22 |
+
This model is described in the paper [Temporally Grounded Compositional Camera Motion Understanding via Geometric Knowledge Distillation](https://huggingface.co/papers/2608.10932).
|
| 23 |
+
Project page: https://ddz16.github.io/cammotion.github.io
|
| 24 |
+
Code: https://github.com/ddz16/CamDistill
|
| 25 |
+
|
| 26 |
## Usage
|
| 27 |
|
| 28 |
```python
|
| 29 |
from transformers import AutoModelForCausalLM, AutoProcessor
|
| 30 |
model = AutoModelForCausalLM.from_pretrained("ddz16/CamDistill-8B", torch_dtype="bfloat16", device_map="auto")
|
| 31 |
processor = AutoProcessor.from_pretrained("ddz16/CamDistill-8B")
|
| 32 |
+
```
|