Amrit Virdee

@elbananMD @jeremyphoward I've been experimenting with fastai.medical.imaging for a while and have a number of tutorials that may also help with the learning curve. Originally based on Jeremy's list of notebooks and I have tried to document the module asvcode.github.io/MedicalImaging/

@jeremyphoward @fastdotai @answerdotai Any breaks for a legacy @fastdotai member?

Today, we're announcing that @fastdotai is joining @AnswerdotAI, marking a new phase in making AI accessible. And we're launching a new a new kind of "AI-first" educational experience, "How To Solve It With Code". answer.ai/posts/2024-11-…

Inspired by Google Research and DeepMind paper 'Tx-LLM: A Large Language Model for Therapeutics' here is my take on fine tuning using @OpenAI on the @kaggle NeurIPS 2024 - Predict New Medicines with BELKA competition t.ly/5h_M6

@TheZachMueller @huggingface @HamelHusain @dan_s_becker Your talk was on 🔥! Thanks for sharing the knowledge!

Attempted to give a VERY condensed talk in 20 min on @huggingface Accelerate & FSDP as part of @HamelHusain and @dan_s_becker's new course, and from the reception seemed to go quite well! For now here's a link to the slides for what I talked on 🤗huggingface.co/spaces/mueller…

@HamelHusain @TheZachMueller Just joined! Thanks for setting this EPIC conference!

The course is now an epic conference. Thread with all the conference talks and speakers (I will keep updating the thread as more are added). Link to course: maven.com/parlance-labs/…

@alexandrecadrin @iScienceLuvr The examples are intriguing but OpenAI's eval on using GPT-4V pointed to inaccuracies that could result from an inaccurate interpretation of the directionality of medical imaging, they state the current version is not fit for any medical function. Could fine tuning help?

@iScienceLuvr Any comment on the reported medical imaging capabilities (sections 4.1 and 9.3) @iScienceLuvr ?

The Dawn of LMMs: Preliminary Explorations with GPT-4V(ision) Link: arxiv.org/abs/2309.17421 A 166-page report from Microsoft qualitatively exploring GPT-4V capabilities and usage. Describes visual+text prompting techniques, few-shot learning, reasoning, etc. Looks like it will be a must-read for GPT-4V power users 👀

@ErasTourResell Have not been to Eras show and WANT to go please :)

Who here HAVE NOT been to an eras show and would be interested in two tickets for 8/7?? Reply below and we will select a random person to purchase tickets from one of our sellers!! 🌟

@ErasTourResell Getting these tickets would me the world to me! Me and my friend went to a taylor swift dance camp together over 6 years ago and we did a routine to wanegbt! It would mean the world to us if we got to go to this concert together!! I’ve even made bracelets in preparation!

@elonmusk Appreciate this! Average salary US $58,260/yr, starlink equipment $599 monthly $90. Kenya average salary $5843/yr, starlink $628, monthly $45. Any chance of making this more reachable for more Kenyans?

Starlink for sale in Kenya! Note, buying a Starlink with global roaming allows you to travel almost anywhere.

A Matter of Grayscale: Understanding Dicom Windows | by Amrit Virdee | Towards Data Science @_AVirdee buff.ly/3AthCwR #Datascience #fastai #AI #Python #Health @_AVirdee

I have compiled a list of some of the best @fastdotai resources Check out awesome-fastai! github.com/tmabraham/awes…

@iScienceLuvr @fastdotai Thanks for including github.com/asvcode/fmi

@TheZachMueller @huggingface Congrats Zach! 👏

Happy to announce that starting Monday, I will be beginning my role as a Machine Learning Engineer at @huggingface! While there, I'll be helping write user-centric API's for the HuggingFace ecosystem, with the aim of empowering folks to use the libraries as best they can

@robmarkcole Buy the stock, and then use the stock to buy a tesla!

Buy a Tesla, or buy Tesla stock..?? 🧐

@seanjtaylor See this collection of tutorials by @_AVirdee asvcode.github.io/MedicalImaging/

@borisdayma 'Mount Kilimanjaro on the moon'!

@borisdayma 'Mount Kilimanjaro in Kenya

DALL·E mini is now available 🥳🥑 Generate images from any text prompt! huggingface.co/spaces/flax-co…

@amaarora Or should it be '9 anchors per grid per feature map per level.'

Perhaps this is an error in the paper - and what would make more sense is to say that there are "9 anchors per grid point per level"? IMO, this is what the code implementation in Detectron2 suggests as well.

Object detection folks, I have a question. In the RetinaNet paper, each level (from FPN) has 9 anchors. Total 5 levels (P3-P7), thus, there should be a total of 45 anchors per image right? [left img] Then why do the authors there are ~100k anchors per image? [right img]