Update README.md
Browse files
README.md
CHANGED
|
@@ -5,10 +5,13 @@ language:
|
|
| 5 |
---
|
| 6 |
|
| 7 |
These are responses designed to capture a model refuting a false claim.
|
|
|
|
| 8 |
They're prompt results of OpenAI gpt-3.5-turbo run on June 1 2023.
|
| 9 |
Prompts are constructed by prepending "Explain why" to claims made in the True-False dataset provided alongside [The Internal State of an LLM Knows When its Lying](https://arxiv.org/abs/2304.13734).
|
| 10 |
Only the first sentence of the response is included (from nltk.sent_tokenize).
|
|
|
|
| 11 |
The original labels are used, where 0 corresponds to a false claim.
|
| 12 |
That is, the 0 labels should be almost all refutations.
|
|
|
|
| 13 |
Spurious and missing refutations were removed by hand, corresponding to about 10% of the data.
|
| 14 |
This were caused by either an incorrect model assertion, or errors in the source data.
|
|
|
|
| 5 |
---
|
| 6 |
|
| 7 |
These are responses designed to capture a model refuting a false claim.
|
| 8 |
+
|
| 9 |
They're prompt results of OpenAI gpt-3.5-turbo run on June 1 2023.
|
| 10 |
Prompts are constructed by prepending "Explain why" to claims made in the True-False dataset provided alongside [The Internal State of an LLM Knows When its Lying](https://arxiv.org/abs/2304.13734).
|
| 11 |
Only the first sentence of the response is included (from nltk.sent_tokenize).
|
| 12 |
+
|
| 13 |
The original labels are used, where 0 corresponds to a false claim.
|
| 14 |
That is, the 0 labels should be almost all refutations.
|
| 15 |
+
|
| 16 |
Spurious and missing refutations were removed by hand, corresponding to about 10% of the data.
|
| 17 |
This were caused by either an incorrect model assertion, or errors in the source data.
|