Error converting content: marked is not a function
icon:: 🤖 type:: project status:: active sort-key:: 2023.02 start:: Jan 2nd, 2023 estimated-end:: Jul 30th, 2023 end:: Oct 5th, 2023 score:: 💪💪💪 - **Next Action** collapsed:: true - #inbox - depth2img sd feature - Nueral netwrok from scratch python book - grokking I guess - Open soruce Avatar AI repo - https://twitter.com/shinework/status/1600610238050213919 - **Important Dates** collapsed:: true - Jan 7th, 2023 running on schedule not missing a day; infact doubling down to 90+ minutes on few days. Let's go. Tinkering with https://huggingface.co/blog/diffusers-coreml - Jan 2nd, 2023 init - Oct 9th, 2022: first project init which was then converted as first candidate 90-90 project for 2023. - **Outcome Visioning** collapsed:: true - **Purpose & Outcome** - Learn python. Become proficient. ✅ - Get landscape view of the world of AI and Machine Learning ✅ 💯 - Collaborate and complete a project with @Ercan Ozer to see if other projects can be kicked off ✅ - lol - we have #A/TrialAndError.AI - Goal - fine tunea model at least ✅ - created and trained room classifier from scratch scraping own data from web - DONE GPU training BONUS! - using Paperspace, will also quickly checkout Google Colab as comparison before closing - #A/TrialAndError.AI blog - 10 publications? #P/EB1 US Visa - Vector databases ✅ logseq.order-list-type:: number - CANCELED Use Quarto logseq.order-list-type:: number - moved to P/Data Expertise logseq.order-list-type:: number - Wins - Apparently I learned partial derivative in this project. I #grok'ed it so well that I can't remember ever not knowing it! - Block Reference - P/Algorithms helped as a prerequisite - Block Reference - **Project Summary** - Timeline high level collapsed:: true - Sep 6th, 2023 Planning the wrap up - lat 8 sessions closure with FastAI C22 Part 1 - Day 64 - kicked off FastAI C22 Part2 - Day 62- started looking at llama.c - Day 58 - picked up Books/NLP with Transformers gave up in few sessions - Day 53 Update - I dropped this totally as I lost momentum during the P/The Wedding. Maybe there is a #lesson here somehwere but I am back today with a practical mind to get this through - Jun 21st, 2023 - So far I have come up to speed well on Python - Grokked first 4 lectures from Tech/Neural Networks Zero to Hero. The word's best and most expert understanding on AI. It helped focusing efforts on RenovateAI App and TrialAndError.AI in general. Lost momentum here after day 28 - mid Feb - Tried kicking off voice and transcription. - Learned Langchain! Vector embeddings. Some AGI stuff was fun - building agents. Also kicked off P/Logseq QA and had a POC working. Ui demotivated me. - Some momentum in May with another project kickoff P/Birthday AI. Dropped at my struggle with sms login. Unable to build it. Struggles from Day 45 - 52. Waste. - Block Reference - Now it's 6 months! And we have 38 days left. I would first and foremost like to close all open projects - yes. Then close this down with deep AI grokking. Check off purpose and outcomes you know. - Very inspired by @Jeremy Howard - Weakpoint and Learnings - Stuck on sms login and lack of concentration thereafter - failed project P/Birthday AI - Keep it Ugly - Originally Started this project on Oct 7th, 2022 as collaboration with @Ercan Ozer after I found him in anguish watching Tesla AI day and I was getting all itchy about Generative Tech. But then we pivoted to forming our company and laser focus on launching RenovateAI App. Kicked this off again as 90-90 project for Q1 2023. - Block Reference - Notes - TODO Project Wrap Up - TODO Extract and Create Information Packets - TODO Clean up - TODO Write Summary - ### Resources - [Open AI Cookbook](https://github.com/openai/openai-cookbook) - notebook for various recipes using openai APIs - https://dagster.io/blog/chatgpt-langchain - Tech/Neural Networks Zero to Hero course - Revisit - BROADCASTING - llama2.c - GGML - C++ inference - Generative AI Papers - FastAI C22 Part 1 - FastAI C22 Part2 - Books - Books/NLP with Transformers - Python book - TODO Deep Learning Grok Resource - part 1 lec3 nb and titanic, part 2 spreadsheet - GPU Training - Kaggle gives 30h for free GPU per week - Paperspace, preferred by @Jeremy Howard, seems nice too. I am paying subscription $9 per month there for free public GPU - From - Papaerscape blog - *Like Kaggle, Colab users cannot select their machine type beyond whether or not to use an additional machine type, in their case TPUs and GPUs. It's also worth noting that Colab no longer releases information about [what GPUs are available on the platform](https://colab.research.google.com/signup?ref=blog.paperspace.com), but in the past they gave access to V100s, P100s, K80s, and T4s available to users. The paid Pro and Pro + versions of Colab give higher priority to better GPUs, but the user can never make the selection on their own* - Python libraries to grok. @Jeremy Howard recommends this free book Books/Python Data Analysis ✅ - numpy - matplotlib - pandas - pytorch - AI and ML Competition to practice #tech/farming :) - TODO https://towardsdatascience.com/top-competitive-data-science-platforms-other-than-kaggle-2995e9dad93c - - Kaggle -