Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,26 @@
|
|
| 1 |
---
|
| 2 |
license: mit
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: mit
|
| 3 |
---
|
| 4 |
+
|
| 5 |
+
# Phi-2 Orange
|
| 6 |
+
|
| 7 |
+
A two-step finetune of Phi-2.
|
| 8 |
+
|
| 9 |
+
First using a collection of broad training data:
|
| 10 |
+
|
| 11 |
+
- [Open-Orca/SlimOrca-Dedup](https://huggingface.co/datasets/Open-Orca/SlimOrca-Dedup)
|
| 12 |
+
- [migtissera/Synthia-v1.3](https://huggingface.co/datasets/migtissera/Synthia-v1.3)
|
| 13 |
+
- [LDJnr/Verified-Camel](https://huggingface.co/datasets/LDJnr/Verified-Camel)
|
| 14 |
+
- [LDJnr/Pure-Dove](https://huggingface.co/datasets/LDJnr/Pure-Dove)
|
| 15 |
+
- [LDJnr/Capybara](https://huggingface.co/datasets/LDJnr/Capybara)
|
| 16 |
+
- [meta-math/MetaMathQA](https://huggingface.co/datasets/meta-math/MetaMathQA)
|
| 17 |
+
|
| 18 |
+
And then a DPO finetune using:
|
| 19 |
+
|
| 20 |
+
- [Intel/orca_dpo_pairs](https://huggingface.co/datasets/Intel/orca_dpo_pairs)
|
| 21 |
+
- [argilla/ultrafeedback-binarized-preferences-cleaned](https://huggingface.co/datasets/argilla/ultrafeedback-binarized-preferences-cleaned)
|
| 22 |
+
|
| 23 |
+
# Initial Evals
|
| 24 |
+
|
| 25 |
+
- ARC: 62.29
|
| 26 |
+
- TruthfulQA: 49.85
|