Commit
•
aa409da
1
Parent(s):
f5e02b8
Update README.md
Browse files
README.md
CHANGED
|
@@ -6,6 +6,8 @@ We perform GSHF algorithm on SFT baseline. The external signals include (1) Rewa
|
|
| 6 |
|
| 7 |
**We obtain 35.95% win-rate (34.79% LC win-rate) on Alpaca Eval v2.** The win-rate of the base model is only 4.63%.
|
| 8 |
|
|
|
|
|
|
|
| 9 |
We have demonstrated the significant potential of the iterative RLHF algorithm for LLMs to deliver appropriate and well-structured responses,
|
| 10 |
even without any external responses.
|
| 11 |
|
|
|
|
| 6 |
|
| 7 |
**We obtain 35.95% win-rate (34.79% LC win-rate) on Alpaca Eval v2.** The win-rate of the base model is only 4.63%.
|
| 8 |
|
| 9 |
+
For MT-bench, it obtained about 7.5, where the base model is only 5.3.
|
| 10 |
+
|
| 11 |
We have demonstrated the significant potential of the iterative RLHF algorithm for LLMs to deliver appropriate and well-structured responses,
|
| 12 |
even without any external responses.
|
| 13 |
|