Update README.md
Browse files
README.md
CHANGED
|
@@ -3,10 +3,13 @@ license: mit
|
|
| 3 |
---
|
| 4 |
This is a test DPO finetune of Microsoft phi-2
|
| 5 |
|
| 6 |
-
Two DPO datasets are used
|
| 7 |
|
| 8 |
-
Intel/orca_dpo_pairs
|
|
|
|
| 9 |
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
| 11 |
|
| 12 |
-
Training was for 1 epoch as a qlora with Rank 64, Delta 128
|
|
|
|
| 3 |
---
|
| 4 |
This is a test DPO finetune of Microsoft phi-2
|
| 5 |
|
| 6 |
+
Two DPO datasets are used. Training was for 1 epoch as a qlora with rank 64.
|
| 7 |
|
| 8 |
+
- [Intel/orca_dpo_pairs](https://huggingface.co/datasets/Intel/orca_dpo_pairs)
|
| 9 |
+
- [argilla/ultrafeedback-binarized-preferences-cleaned](https://huggingface.co/datasets/argilla/ultrafeedback-binarized-preferences-cleaned)
|
| 10 |
|
| 11 |
+
# Initial Evals
|
| 12 |
+
|
| 13 |
+
- ARC: 63.14
|
| 14 |
+
- TruthfulQA: 48.47
|
| 15 |
|
|
|