The Model Training Finetunes is NOT working properly

Hey @rrojas06 ! ??

I’ve tested it on a fresh installation, and it’s working without any apparent issues. Please make sure you are using the latest version of AI Engine.

If you’re already using the latest version, could you please provide me with a step-by-step guide on how you are reproducing this behavior? (Even maybe one of your JSON file)

Additionally, could you check your PHP logs to see if anything is happening on your server side? Thanks! ??

Hi,

This is not resolved. The issue persist. The training model does not work. Please see the JSON file and the video. I have the plugin up to date.

Video showing is still not working: https://share.zight.com/p9uY8nAv

This is what it says in my MEOW-2023.09.29-18.59.jsonl

{"messages":[{"role":"system","content":"Eres InntuitivoBOT, un Asistente de Inteligencia Artificial. Tu misión es ayudar a estudiantes, profesores y mentores a obtener la información más precisa dentro de la comunidad abierta de aprendizaje adaptativo de Inntuitivo."},{"role":"user","content":"?Que es inntuitivo?"},{"role":"assistant","content":"Es una plataforma de comunidad abierta edtech social con gamificación e inteligencia artificial enfocada en los cursos y bootcamps mediante una educación hibridad (asincrónica y sincrónica) que te ayudarán en tu viaje de crecimiento profesional al adquirir las nuevas habilidades del futuro como Product Management, STEM, High-Tech, Emprendimiento y Agilismo."}]}

Hey @rrojas06 ??

The “VALIDATING_FILES” status is returned by OpenAI’s services, not AI Engine itself.

Please note that you need to have a file containing at least 10 entries for this process to work. In general, the fine-tuning process might require around 2000 to 3000 entries to be trained properly.

If you need to pass a small amount of information to your chatbot, please use the chatbot’s context for this purpose.

I hope this information is helpful ??

Hi,

Well, it appears It succeed and I was able to fine-tuned and trained my model. It appears Open AI accepted it.

Please see the screenshot: https://share.zight.com/NQu5wrz5

However, now when I go to the chatbot itself and ask some questions it doesn’t delivered the trained answers. Do I need to wait some days so OpenAI can replicate this in ChatGTP?

Let me know.

Thanks.

RRA.

Hi,

It appears the fine tuning was a success and processed correctly in the platform MEOW and also on OpenAI. But when I go to the ChatBOT this one does not respond with the trained answers to my questions. Here are the screenshots:

https://share.zight.com/o0uKXY4w

https://share.zight.com/YEuzWelm

Please let me know how I can be assisted.

Thanks.

RRA.

Hey @rrojas06 ! ??

You might be expecting the fine-tuning result to be something different from what it actually provides. Please read this documentation as it might help you better understand its purpose. ??