Update README.md
Browse files
README.md
CHANGED
|
@@ -35,4 +35,12 @@ Instead of confidently proclaiming something (or confidently hallucinating other
|
|
| 35 |
|
| 36 |
|
| 37 |
|
| 38 |
-
**Everything COT uses Llama 3 instruct template**
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
|
| 36 |
|
| 37 |
|
| 38 |
+
**Everything COT uses Llama 3 instruct template**
|
| 39 |
+
|
| 40 |
+
The correct jinja chat_template is in tokenizer_config.json
|
| 41 |
+
|
| 42 |
+
**Parameters**
|
| 43 |
+
|
| 44 |
+
It's up to you to discover the best parameters that works. I tested it in oobabooga WebUi using very off-the-shelf min_p preset: Temperature: 1, Top_p: 1, Top_k: 0, Typiocal_p: 1, min_p: 0.05, repetition_penalty: 1
|
| 45 |
+
|
| 46 |
+
Different parameters, like temperature will affect the models talkatiovnes and investigative properties. If you find something really good, let me know and I'll post it here.
|