File size: 526 Bytes
1a45fd6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
---
language:
- en
task_categories:
- text-generation
tags:
- multimodel-small
---

# MultiModel Small Dataset

JSONL datasets used for the MultiModel Small training experiments.

## Files

- `pretrain.jsonl`: text pretraining samples
- `val.jsonl`: validation samples
- `instruct_train.jsonl`: instruction and image-description text
- `align_train.jsonl`: image-text alignment samples

The image references in the instruction and alignment files are illustrative unless the corresponding image files are provided separately.