What are you going to create with this tool? 💼Consulting: calendly.com/engineerprompt/consulting-call 🦾 Discord: discord.com/invite/t4eYQRUcXB ☕ Buy me a Coffee: ko-fi.com/promptengineering |🔴 Join Patreon: Patreon.com/PromptEngineering ▶ Subscribe: www.youtube.com/@engineerprompt?sub_confirmation=1
Thanks so much for all the videos. Need help retraining a model on a proprietary programming language based on XAML code, and please note, GPT3, GPT3.5 and GPT4 are already trained on it but based on whatever was available in the public domain. Tried GPT Fine Tuning, didn't work as expected on OpenAI. They suck literally. Apologies, I don't have 300$ for a consulting call but if you can nudge me in the right direction, that would be superb.
Thank you, what is the reason for them not to be able to work well after fine-tuning? Most probably you want to look at your data and ensure its well formatted. Are you doing instruct finetuning?
Make sure you instruct the AI to adapt a set of skills that align with what you want it to complete. And I do mainly writing and SEO tasks, but setting up a good personality rubric has helped me extremely much by being more on point and providing more high-level outputs.
Model eval is now an emerging field and a lot of working is being done in it. The idea is to have answers to a subset of questions and then ask the model those questions and see how close the answers are to the actual answers (your gold standard). This is how the benchmarks work. Although seems like most of the bench marks are now exposed to these models directly or indirectly. Hope this helps.
What are you going to create with this tool?
💼Consulting: calendly.com/engineerprompt/consulting-call
🦾 Discord: discord.com/invite/t4eYQRUcXB
☕ Buy me a Coffee: ko-fi.com/promptengineering
|🔴 Join Patreon: Patreon.com/PromptEngineering
▶ Subscribe: www.youtube.com/@engineerprompt?sub_confirmation=1
Great prompt engineering as the channel name suggest!!
Exactly what we asked for! Thanks for the video. Are you also going to talk about the new Orca 13B model that is being hyped up now?
thank you, probably when its actually released :)
Thanks so much for all the videos. Need help retraining a model on a proprietary programming language based on XAML code, and please note, GPT3, GPT3.5 and GPT4 are already trained on it but based on whatever was available in the public domain. Tried GPT Fine Tuning, didn't work as expected on OpenAI. They suck literally. Apologies, I don't have 300$ for a consulting call but if you can nudge me in the right direction, that would be superb.
Hey can we work / explore together if you are interested in the field of AI & AI tools efficient usage ....
Thank you, what is the reason for them not to be able to work well after fine-tuning? Most probably you want to look at your data and ensure its well formatted. Are you doing instruct finetuning?
Make sure you instruct the AI to adapt a set of skills that align with what you want it to complete. And I do mainly writing and SEO tasks, but setting up a good personality rubric has helped me extremely much by being more on point and providing more high-level outputs.
Great Info ! Thank You 👍
The most important one you mentioned - model evaluation, is hard to understand, including the example about the gold standard. Can you explain it？
Model eval is now an emerging field and a lot of working is being done in it. The idea is to have answers to a subset of questions and then ask the model those questions and see how close the answers are to the actual answers (your gold standard). This is how the benchmarks work. Although seems like most of the bench marks are now exposed to these models directly or indirectly. Hope this helps.
thanks for the reply@@engineerprompt
so helpful
dude, you need to do a chatgpt ASMR channel 😊.... great info, thank you!
Maybe one day :)