Update README.md
Browse files
README.md
CHANGED
|
@@ -1,13 +1,21 @@
|
|
| 1 |
This model was trained as part of a series of experiments testing the performance of pure DPO vs SFT vs ORPO, all supported by Unsloth/Huggingface TRL.
|
| 2 |
|
| 3 |
Rank: 8
|
|
|
|
| 4 |
Alpha: 16
|
|
|
|
| 5 |
Learning rate: 5e-6
|
|
|
|
| 6 |
Beta: 0.1
|
|
|
|
| 7 |
Batch size: 8
|
|
|
|
| 8 |
Epochs: 1
|
|
|
|
| 9 |
Learning rate schedulers: Linear
|
|
|
|
| 10 |
Prompt Format:
|
|
|
|
| 11 |
```You are a helpful assistant.<s>[INST] PROMPT [/INST]RESPONSE</s>```
|
| 12 |
|
| 13 |
|
|
|
|
| 1 |
This model was trained as part of a series of experiments testing the performance of pure DPO vs SFT vs ORPO, all supported by Unsloth/Huggingface TRL.
|
| 2 |
|
| 3 |
Rank: 8
|
| 4 |
+
|
| 5 |
Alpha: 16
|
| 6 |
+
|
| 7 |
Learning rate: 5e-6
|
| 8 |
+
|
| 9 |
Beta: 0.1
|
| 10 |
+
|
| 11 |
Batch size: 8
|
| 12 |
+
|
| 13 |
Epochs: 1
|
| 14 |
+
|
| 15 |
Learning rate schedulers: Linear
|
| 16 |
+
|
| 17 |
Prompt Format:
|
| 18 |
+
|
| 19 |
```You are a helpful assistant.<s>[INST] PROMPT [/INST]RESPONSE</s>```
|
| 20 |
|
| 21 |
|