Such good interviewers that are not forcing their own opinions or expertise into the conversation. I really owe Jeremy and Andrew Ng for laying solid foundations for me. I can really relate to the tinkerer narrative. It was ages ago (2017?) that I had taken the fastai course and looked at dog v cat classification as the very first lesson. Before fastai I never felt it was accessible or would be as easy to implement and Andrew filled in the gaps and kept me on my journey by saying "don't worry if you don't understand".
Loved your podcast with Lenny, glad you came out from OpenAI and started supporting Open Source community. And I'm happy that you're also supporting, investing, and mentoring many new start-ups thereby creating more competition in the market and increasing net good for society. I have a little doubt - why no start-up is providing serverless domain adaptation fine-tuning (AKA continued pre-training) for transformer or diffusion models! Because that is probably the only way to infuse knowledge and improve capabilities!
Such good interviewers that are not forcing their own opinions or expertise into the conversation. I really owe Jeremy and Andrew Ng for laying solid foundations for me. I can really relate to the tinkerer narrative. It was ages ago (2017?) that I had taken the fastai course and looked at dog v cat classification as the very first lesson. Before fastai I never felt it was accessible or would be as easy to implement and Andrew filled in the gaps and kept me on my journey by saying "don't worry if you don't understand".
Anything involving Jeremy Howard is always great, looking forward to this one
Yes, I love listening to anything Jeremy has to say. Thanks!
Loved your podcast with Lenny, glad you came out from OpenAI and started supporting Open Source community.
And I'm happy that you're also supporting, investing, and mentoring many new start-ups thereby creating more competition in the market and increasing net good for society.
I have a little doubt - why no start-up is providing serverless domain adaptation fine-tuning (AKA continued pre-training) for transformer or diffusion models! Because that is probably the only way to infuse knowledge and improve capabilities!
Wow, thanks for this interview! You asked exactly the questions I would like to hear the answer to.
And also could you please re-check the time stamps in the video description!
Fixed!
thanks !♥
good questions!
really liked the interview, but whats up with the audio and very frequent cuts? its quite distracting. just thought id mention.
Please less editing, its so obvious 💀