Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
datasets:
|
| 4 |
+
- sms1097/self_rag_tokens_train_data
|
| 5 |
+
---
|
| 6 |
+
# Utility Model
|
| 7 |
+
|
| 8 |
+
This generates the `IsUseful` token as descirbed in Self-RAG.
|
| 9 |
+
|
| 10 |
+
We are testing to see if an answer is useful to the given user question. We output a score from 1-5 based on how useful the answer is.
|
| 11 |
+
|
| 12 |
+
The expected input to the model is:
|
| 13 |
+
|
| 14 |
+
```
|
| 15 |
+
Instruction: {instruction}\nAnswer: {answer}",
|
| 16 |
+
```
|
| 17 |
+
|
| 18 |
+
|
| 19 |
+
### Training Results
|
| 20 |
+
```
|
| 21 |
+
{'eval_loss': 0.4719298779964447,
|
| 22 |
+
'eval_mse': 0.4719298183917999,
|
| 23 |
+
'eval_mae': 0.25655537843704224,
|
| 24 |
+
'eval_r2': 0.5200293292355334,
|
| 25 |
+
'eval_accuracy': 0.9001516683518705}
|
| 26 |
+
```
|