Improve Model Card: Correct pipeline tag, add library name and project page link
#1
by
nielsr
HF staff
- opened
README.md
CHANGED
@@ -1,14 +1,16 @@
|
|
1 |
---
|
2 |
-
license: apache-2.0
|
3 |
-
datasets:
|
4 |
-
- THUdyh/Oryx-SFT-Data
|
5 |
base_model:
|
6 |
- Qwen/Qwen2.5-7B-Instruct
|
7 |
-
|
|
|
8 |
language:
|
9 |
- en
|
10 |
- zh
|
|
|
|
|
|
|
11 |
---
|
|
|
12 |
# Oryx-1.5-7B
|
13 |
|
14 |
## Model Summary
|
@@ -18,6 +20,7 @@ The Oryx-1.5 models are 7/32B parameter models trained on [Oryx-SFT-Data](https:
|
|
18 |
Oryx offers an on-demand solution to seamlessly and efficiently process visual inputs with arbitrary spatial sizes and temporal lengths.
|
19 |
|
20 |
- **Repository:** https://github.com/Oryx-mllm/Oryx
|
|
|
21 |
- **Languages:** English, Chinese
|
22 |
- **Paper:** https://arxiv.org/abs/2409.12961
|
23 |
|
@@ -25,7 +28,7 @@ Oryx offers an on-demand solution to seamlessly and efficiently process visual i
|
|
25 |
|
26 |
We provide a simple generation process for using our model. For more details, please refer to our [Github Repo](https://github.com/liuzuyan/oryx)
|
27 |
|
28 |
-
```
|
29 |
from oryx.model.builder import load_pretrained_model
|
30 |
from oryx.mm_utils import get_model_name_from_path, process_images, tokenizer_image_token
|
31 |
from oryx.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN, IGNORE_INDEX
|
@@ -125,4 +128,13 @@ print(text_outputs)
|
|
125 |
- **Orchestration:** HuggingFace Trainer
|
126 |
- **Code:** Pytorch
|
127 |
|
128 |
-
## Citation
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
|
|
|
|
|
|
2 |
base_model:
|
3 |
- Qwen/Qwen2.5-7B-Instruct
|
4 |
+
datasets:
|
5 |
+
- THUdyh/Oryx-SFT-Data
|
6 |
language:
|
7 |
- en
|
8 |
- zh
|
9 |
+
license: apache-2.0
|
10 |
+
pipeline_tag: video-text-to-text
|
11 |
+
library_name: oryx
|
12 |
---
|
13 |
+
|
14 |
# Oryx-1.5-7B
|
15 |
|
16 |
## Model Summary
|
|
|
20 |
Oryx offers an on-demand solution to seamlessly and efficiently process visual inputs with arbitrary spatial sizes and temporal lengths.
|
21 |
|
22 |
- **Repository:** https://github.com/Oryx-mllm/Oryx
|
23 |
+
- **Project Page:** https://oryx-mllm.github.io
|
24 |
- **Languages:** English, Chinese
|
25 |
- **Paper:** https://arxiv.org/abs/2409.12961
|
26 |
|
|
|
28 |
|
29 |
We provide a simple generation process for using our model. For more details, please refer to our [Github Repo](https://github.com/liuzuyan/oryx)
|
30 |
|
31 |
+
```python
|
32 |
from oryx.model.builder import load_pretrained_model
|
33 |
from oryx.mm_utils import get_model_name_from_path, process_images, tokenizer_image_token
|
34 |
from oryx.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN, IGNORE_INDEX
|
|
|
128 |
- **Orchestration:** HuggingFace Trainer
|
129 |
- **Code:** Pytorch
|
130 |
|
131 |
+
## Citation
|
132 |
+
|
133 |
+
```bibtex
|
134 |
+
@article{liu2024oryx,
|
135 |
+
title={Oryx MLLM: On-Demand Spatial-Temporal Understanding at Arbitrary Resolution},
|
136 |
+
author={Liu, Zuyan and Dong, Yuhao and Liu, Ziwei and Hu, Winston and Lu, Jiwen and Rao, Yongming},
|
137 |
+
journal={arXiv preprint arXiv:2409.12961},
|
138 |
+
year={2024}
|
139 |
+
}
|
140 |
+
```
|