Venkat-Ram-Rao
commited on
Commit
•
1dc32c2
1
Parent(s):
2616356
Update README.md
Browse files
README.md
CHANGED
@@ -1,21 +1,3 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
|
4 |
-
- name: prompt
|
5 |
-
dtype: string
|
6 |
-
- name: chosen
|
7 |
-
dtype: string
|
8 |
-
- name: rejected
|
9 |
-
dtype: string
|
10 |
-
splits:
|
11 |
-
- name: train
|
12 |
-
num_bytes: 1240272
|
13 |
-
num_examples: 210
|
14 |
-
download_size: 552883
|
15 |
-
dataset_size: 1240272
|
16 |
-
configs:
|
17 |
-
- config_name: default
|
18 |
-
data_files:
|
19 |
-
- split: train
|
20 |
-
path: data/train-*
|
21 |
-
---
|
|
|
1 |
+
210 rows from the [MS_MARCO Dataset](https://huggingface.co/datasets/ms_marco) reworked for training via Direct Preference Optimization. The prompt format is for the [Mistral Instruct](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) Models.
|
2 |
+
|
3 |
+
Original Dataset is not mine. Posted this as it may be of use to others.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|