Update README.md
Browse files
README.md
CHANGED
|
@@ -60,13 +60,15 @@ print(tokenizer.decode(tokens[0], skip_special_tokens=True))
|
|
| 60 |
### Training Dataset
|
| 61 |
|
| 62 |
The dataset is comprised of a mixture of open datasets large-scale datasets available on the [HuggingFace Hub](https://huggingface.co/datasets):
|
|
|
|
| 63 |
- HuggingFaceH4/ultrachat_200k
|
| 64 |
-
- HuggingFaceH4/ultrafeedback_binarized
|
| 65 |
-
- Intel/orca_dpo_pairs
|
| 66 |
- meta-math/MetaMathQA
|
| 67 |
-
- Instruct Code Dataset (Internal)
|
| 68 |
- Wizard Dataset
|
| 69 |
- Open-Orca/SlimOrca
|
|
|
|
|
|
|
|
|
|
|
|
|
| 70 |
|
| 71 |
### Training Procedure
|
| 72 |
|
|
|
|
| 60 |
### Training Dataset
|
| 61 |
|
| 62 |
The dataset is comprised of a mixture of open datasets large-scale datasets available on the [HuggingFace Hub](https://huggingface.co/datasets):
|
| 63 |
+
1. SFT Datasets
|
| 64 |
- HuggingFaceH4/ultrachat_200k
|
|
|
|
|
|
|
| 65 |
- meta-math/MetaMathQA
|
|
|
|
| 66 |
- Wizard Dataset
|
| 67 |
- Open-Orca/SlimOrca
|
| 68 |
+
2. Preference Datasets:
|
| 69 |
+
- HuggingFaceH4/ultrafeedback_binarized
|
| 70 |
+
- Intel/orca_dpo_pairs
|
| 71 |
+
|
| 72 |
|
| 73 |
### Training Procedure
|
| 74 |
|