instruction
stringclasses 1
value | input
stringclasses 1
value | accepted
stringclasses 1
value | rejected
stringclasses 1
value |
|---|---|---|---|
Do a thing. | here's how to do a thing | Here's you doing the thing right | here's you doing it wrong. |
Do a thing. | here's how to do a thing | Here's you doing the thing right | here's you doing it wrong. |
Do a thing. | here's how to do a thing | Here's you doing the thing right | here's you doing it wrong. |
Do a thing. | here's how to do a thing | Here's you doing the thing right | here's you doing it wrong. |
Do a thing. | here's how to do a thing | Here's you doing the thing right | here's you doing it wrong. |
Do a thing. | here's how to do a thing | Here's you doing the thing right | here's you doing it wrong. |
Do a thing. | here's how to do a thing | Here's you doing the thing right | here's you doing it wrong. |
Do a thing. | here's how to do a thing | Here's you doing the thing right | here's you doing it wrong. |
Do a thing. | here's how to do a thing | Here's you doing the thing right | here's you doing it wrong. |
This is a blank empty dataset with nothing in it and no value beside "working" with the Unsloth DPO trainer.
I took the example dataset, looked at it, found the keys, and made it readable so I could understand it when I come back to it after whatever distracts me. This format is exactly for the Unsloth ORPO notebook at https://colab.research.google.com/drive/11t4njE3c4Lxl-07OD8lJSMKkfyJml3Tn?usp=sharing#scrollTo=xKzWRN1Px0Bg
To test it I used nickypro/tinyllama-15M which loads quickly. It all runs without errors.
- Downloads last month
- 3