Update README.md
Browse files
README.md
CHANGED
|
@@ -47,6 +47,24 @@ This model is built on top of **[Qwen/Qwen2.5-3B-Instruct](https://huggingface.c
|
|
| 47 |
- The model is trained to provide structured arguments and objections based on these questions
|
| 48 |
|
| 49 |
## Performance
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 50 |
For the question: "Should emotional bonds formed with AI companions be given legal recognition?"
|
| 51 |
```
|
| 52 |
Base Model:
|
|
|
|
| 47 |
- The model is trained to provide structured arguments and objections based on these questions
|
| 48 |
|
| 49 |
## Performance
|
| 50 |
+
Base model and this finetuned using system prompt:
|
| 51 |
+
```
|
| 52 |
+
SYSTEM_PROMPT = """
|
| 53 |
+
Please respond using the following format:
|
| 54 |
+
<reasoning>
|
| 55 |
+
<argument>
|
| 56 |
+
[State your argument, supported by evidence.]
|
| 57 |
+
</argument>
|
| 58 |
+
<objection>
|
| 59 |
+
[Present an objection to your argument.]
|
| 60 |
+
</objection>
|
| 61 |
+
[Continue alternating between <argument> and <objection> as needed.]
|
| 62 |
+
</reasoning>
|
| 63 |
+
<final_argument>
|
| 64 |
+
[Summarize your final argument.]
|
| 65 |
+
</final_argument>
|
| 66 |
+
"""
|
| 67 |
+
```
|
| 68 |
For the question: "Should emotional bonds formed with AI companions be given legal recognition?"
|
| 69 |
```
|
| 70 |
Base Model:
|