fix #18: Final Answer should not use JSON

bklieger-groq / g1

g1: Using Llama-3.1 70b on Groq to create o1-like reasoning chains

MIT License

3.96k stars 357 forks source link

Closed JerryYin777 closed 2 months ago

JerryYin777 commented 2 months ago

Fix issue #16 and #18 : Final answer does not need to use JSON formatting

Now the bug example is bug free, I also add a code renderer via streamlit:

For example in #16:

For example in #18

Let me know if you have any other questions and bugs.

bklieger-groq commented 2 months ago

Great job! I have requested one change above. Could you also change the request to merge into the develop branch instead of main? Thank you!

JerryYin777 commented 2 months ago

Sure! Actually, we do not need to change the given prompt (sry it's my fault), we just need to change the logic of HTML output:

And I have reset max_token from 1200 to the original (200) too.

I've changed my code and branch, if there are any other problems, let me know.

bklieger-groq commented 2 months ago

Merged. Thank you for your contribution!