File size: 502 Bytes
38074df
 
ca90f68
c4fed7f
0c89702
 
93761ea
 
c4fed7f
 
 
 
 
 
60b576d
 
93761ea
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
---
license: mit
library_name: transformers
pipeline_tag: unconditional-image-generation
datasets:
- commaai/commavq
---

# commaVQ - GPT2M

A GPT2M model trained on a larger version of the commaVQ dataset. 

This model is able to generate driving video unconditionally. 

Below is an example of 5 seconds of imagined video using GPT2M.

<video title="imagined" controls>
  <source src="https://github.com/commaai/commavq/assets/29985433/f6f7699b-b6cb-4f9c-80c9-8e00d75fbfae" type="video/mp4">
</video>