Thank you for the invite, I’m happy to brainstorm with you how AI in its current state can help the automotive repair industry. Please add me on linkedin and we can connect. Thanks.
@johnpierref1 Thanks for reaching out, feel free to message me the problem you have here or add me on linkedin. Thanks.
Good morning Mr Nelson
Please accept my LinkedIn invitation Mr somtoo
@Black_TEGINA Got it, let’s chat.
I’ve started an idea to dev a chat bot for dementia caretaker, can you help to code a chat to from azure ….I can invite u as teammates for a startup
I would like to know to what kind of support you can give me in our project to make AI applications.
I’m relatively new in the field of AI, but have some ideas what I want to implement with AI.
The first project would be contacting customers via e-mail.
Currently, I’m taking the first steps with OpenAI and I’m stuck somehow.
This is my question.
If I wanted to upload lets say 10 published papers on a subject . then get GPT to base the content it writes from only these sources is this possible if so how would it be done ?
Sure, I’ll love to hear more about it. Feel free to message me about what kind of coding help you need.
To be able to use GPT to write email that is in your style, you can try this approach…
1. Find the best prompt
You can use the GPT playground to find the best way of generating emails similar to yours.
In this example, I use a style of George Washington writing to his wife, and ask it to reproduce a similar letter.
Once you get the best results you are looking you can use this google sheet I have in this link to automate this to multiple recipients. See Google Spreadsheet + GPT3 - #10 by nelson
After you reviewed all your emails from automation, you can export to third party email provider to send out the emails for you.
I think that’s all you need to do, feel free to try it out, let me know if you run into any roadblocks.
Yes, you can try to limit the scope of the materials in the published paper, but this is no guarantee GPT will only use information from your content. Have you tried to use the playground? What are your results so far?
Hi Nelson: my dad is suffering from dementia and my mom is her caretaker. I wanted to help her when she needed assisitence and perhaps even an AI to response to her issues and comfort her when I am not around. Need a Chinese simplified speaking chatbot to test around for her. Do you like to provide an email so that you can have some resources to work with.
Sure, I see this is very personal information.
Feel free to add me to your connection at linkedin so we can chat.
Hey Nelson, I would love to speak with you. I have several ideas of how I would like to utilize GPT3 for my media company. How do I schedule a consultation?
Hi @SharonSG ,
Feel free to message me here in this forum or connect with me on linked in.
Nelson, I want to utilize GPT to build a variant of the Ultimatum Game (a simple, take-it-or-leave-it bargaining game) for academic research purposes. I have managed to to create a basic web page version of the game (with chatGPT’s help), but it’s beyond my skills to implement true chat functionality. I have understanding of NLP/ NLU from my previous life, but I am not a coder.
Hello Nelson, your post has gotten me thinking about what possibilities there are for this technology.
I’m a complete novice to this and to be honest I was just thinking that it was used for creative writing.But now I’m starting to feel like that’s just the tip of the iceberg.
I’d love to learn more but in a very basic simplistic way to start.
Could you outline some of the ways that this might be used for the average person who is not even aware of the wider every day uses that this could be useful.
I’d love to come up with something that I could ask you to do.
Thank you, John O’Donovan
Thanks for sharing your project, let me know where you get stuck and I can see if there is anything I can do. Best.
Thank you for you kind word, I think the whole industry including myself are still exploring the possibilities. But from speaking different people around this topic on a day to day basis, I think some of the best use cases around GPT is to use it to consume complex information and have it explain in a very simple format people can understand. I see it cover many areas including programming, law, mechanics, trading, marketing, human resource and even banking. The only thing I think people are trying to avoid is medical, which makes a lot of sense.
Hope that helps, there is also there are a couple samples here by OpenAI as well.
thank you for your answer.
I am not sure if there is a misunderstanding between us.
My problem is that davinci-003 gives answers that contain wording that I don’t like.
My approach now would be to use finetuning to teach the model how to answer.
The way you describe it is not clear to me how it would prevent davinci from using wording that I don’t want.
Here again is the link to the other thread I opened.
I see what you are looking to do now. To be honest, I don’t know if there is any way to make sure the model will not include wording you don’t like. It will be great if you can try the fine-tuning method and see if that is possible. I’ll like to know your results as well.
Another option I think will work as well is that regardless of using prompt engineering or fine-tuning, you can use either semantic search or regex expression to validate the results you get back from the model.
If it you see words you don’t like, you can deny the result and do a retry request with slightly different temperature.
Another word, since we can’t guarantee the response from GPT, we add a validation step as quality control.
Hope that helps, looking forward to your experiment results.