File size: 982 Bytes
7f5b8f1
 
 
 
 
 
 
e24ab8f
 
 
7f5b8f1
a54ca79
7f5b8f1
 
 
 
 
 
 
fc55ccf
7f5b8f1
803dd2f
7f5b8f1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
---
license: apache-2.0
datasets:
- THUdyh/Oryx-Image-Data
base_model:
- Qwen/Qwen2-7B-Instruct
pipeline_tag: text-generation
language:
- en
- zh
---
# Oryx-7B-Image

## Model Summary

The Oryx-Image models are 7/34B parameter models trained on [Oryx-Image-Data](https://huggingface.co/datasets/THUdyh/Oryx-Image-Data), based on Qwen2 language model with a context window of 32K tokens.

Oryx offers an on-demand solution to seamlessly and efficiently process visual inputs with arbitrary spatial sizes and temporal lengths.

- **Repository:** https://github.com/Oryx-mllm/Oryx
- **Languages:** English, Chinese
- **Paper:** https://arxiv.org/abs/2409.12961


### Model Architecture

- **Architecture:** Pre-trained [Oryx-ViT](https://huggingface.co/THUdyh/Oryx-ViT) + Qwen2-7B
- **Data:** a mixture of 4M image data
- **Precision:** BFloat16

#### Hardware & Software

- **Hardware:** 64 * NVIDIA Tesla A100
- **Orchestration:** HuggingFace Trainer
- **Code:** Pytorch

## Citation