Giridhar K

@BockLab This is interesting! Is the model served on HF or github to study the embeddings?

🗨️ WANNA TALK TO YOUR CELLS? Try out CellWhisperer – our new multimodal AI that turns single-cell RNA-seq analysis into a conversation. No coding needed, just chat in plain English. Short walkthrough below. Web app & bioRxiv preprint linked in the thread. Let's dive in! (1/9)

@ManUtd Onana Dalot Maguire Martinez Mazaroui Mainoo Casemiro Bruno Mount Amad Rashford

Your United XI on opening day! 🔴📋 #MUFC || #MUNFUL

@apsarathchandar @polymtl @ChandarLab Congratulations!

I am happy to announce that I have been promoted to Associate Professor with tenure at @polymtl. This is an achievement for the entire @chandarlab! I want to thank my awesome students, without whom this would not have been possible! 1/n

@Supermathie @dustychipura +1 for LaTeX

@dustychipura LaTeX I'm probably an outlier.

for a client: what are people using to make resumes these days?

i'm just a girl standing in front of the ml research community please begging everyone to type their python method inputs and outputs especially if they are tensors or weird nested dicts of lists of dicts

Good morning to everyone except whoever invented Daylight Savings time

Always worth remembering.

@sh_reya @studygone2sleep

does anyone have LLM agents running in prod or at scale, automatically? forget about cost, how did you get the end-to-end latency low enough & the accuracy high enough?

@MattBMcDermott May be semantic scholar can help? semanticscholar.org/product/api Last I checked, the search api may be limited to title, metadata and abstract.

Does anyone know a way to download all proceedings papers from major ML venues (e.g., NeurIPS, ICML, ICLR) for use in an NLP analysis? I know they are open access, but I'm not sure if they are available in a fully downloadable format.

@ssgrn @Meta Congratulations, Suchin!

😶 Me 1 -0 ChatGPT

@_awbery_ hope you get well soon!

I’m not retreating at this point. Had intended to be practicing through December but I had dental surgery and now have Covid so it seems better to rest up. No candies til January 16th!!!!! I’m counting the days.

@deliprao Great thread, I am guessing GPT took the whole CoT for performance improvements too seriously. May be the user doesn't need all the steps but those may be useful for the model to arrive at the correct code. May be hide and use extra CoT tokens for internal computation?

I have been testing mistral-medium & GPT-4’s code generation abilities for non-trivial problems. These are problems even experience engineers will take time to work it out. I am summarizing some examples and overall impression in this thread: 🧶

Removed, enjoy !

GPT-4 for radiology. Far from perfect, but state-of-the-art performance on some tasks: “Surprisingly, we found radiology report summaries generated by GPT-4 to be comparable and, in some cases, even preferred over those written by experienced radiologists” microsoft.com/en-us/research…

If llama2 needs a license, do _versions_ of llama2 (quantized, instruction tuned etc ) also need the meta license? cc @MetaAI @karpathy

something that's more productive.

of duplicates asked by beginner users who will be hunted down by the mods for a variety of reasons (duplicates, rewording, not adhering to the pastebin guidelines, laTeX formatting, writing in CAPS, vague descriptions, ). The mods/users now can focus on a different work, perhaps

While this could be interpreted as bad news, being a user/answerer of SO and stackexchange as a student earlier on in my career, I believe the surge of LLMs will relieve quite a lot of stress from beginner coders. For every good SO question there could possibly be more than 100s