mpmisko

5 ms·

1.
by mpmisko 1mo ago
There has been a lot of progress in this field in the research community. Two good papers: * https://ojs.aaai.org/index.php/AAAI/article/view/11027 * https://arxiv.org/abs/2212.03978
2.
World Models for Planning Agents(mpmisko.github.io)
2 points by mpmisko 4mo ago|0 comments
3.
Show HN: MediSearch Pro–most accurate medical question-answering system(medisearch.io)
1 points by mpmisko 2y ago|0 comments
4.
Show HN: Verify medical claims in TikToks and YouTube Shorts(medisearch.io)
1 points by mpmisko 2y ago|1 comments
5.
by mpmisko 2y ago
EBMs show up all over the place, apparently even your classifier is an EBM :) ( https://arxiv.org/abs/1912.03263 ).
6.
Fun times with energy-based models(mpmisko.github.io)
82 points by mpmisko 2y ago|14 comments
7.
AI Fundamentals: Energy-Based Models(mpmisko.github.io)
2 points by mpmisko 2y ago|0 comments
8.
Solving Death with AI? A Deep Dive into Energy-Based Models(mpmisko.github.io)
1 points by mpmisko 2y ago|0 comments
9.
AI Fundamentals: Energy-Based Models(mpmisko.github.io)
5 points by mpmisko 2y ago|0 comments
10.
What happened to blogs(mpmisko.github.io)
70 points by mpmisko 2y ago|61 comments
11.
by mpmisko 2y ago
GPT-based search engines usually use some sort of a database to retrieve context for the LLM to summarize first. This is what people refer to as RAG these days: https://blogs.nvidia.com/blog/what-is-retrieval-augmented-
12.
by mpmisko 2y ago
Will definitely check pgvector, thanks for the pointer.
13.
by mpmisko 2y ago
Lots of annoying edge cases as you can imagine, nothing particularly glamorous.
14.
by mpmisko 2y ago
Done! Let me know if you have other feedback.
15.
by mpmisko 2y ago
Thanks! Looks quite relevant
16.
by mpmisko 2y ago
Training for multiple epochs is a bit like that :)
17.
by mpmisko 2y ago
We use pinecone and it is not ideal, looking at https://turbopuffer.com/ now. They look quite promising :)
18.
by mpmisko 2y ago
1. We cover all the articles on PMC. The exact cost is hard to estimate because we did a lot of iterations. 2. We do weight those ... it is a lot of trial and error and you have to have good & exhaustive benchmarks.
19.
by mpmisko 2y ago
Yes
20.
by mpmisko 2y ago
Glad you like it! I did this as a mini-project within our startup MediSearch ( https://medisearch.io/ ) & the search pipeline is custom tuned for the problem.
21.
by mpmisko 2y ago
Hi, it currently does not support search by PMID. But you can find the paper included in the results here (5th place): https://pubmedisearch.com/share/Do%20some%20individuals%20wi...
22.
by mpmisko 2y ago
Just looking at stuff like citations and impact factors of journals.
23.
by mpmisko 2y ago
Thanks! It uses a vector search approach. Your query is embedded in a vector space using a language model and we find the closest vector to the query from the PubMed papers. This is a good summary of the techniques: https://learn
24.
Show HN: I put PubMed in a vector DB(pubmedisearch.com)
97 points by mpmisko 2y ago|27 comments
25.
by mpmisko 3y ago
Thanks for giving it a go! We will charge for the API and we may have a pro subscription with some nice features :).
26.
by mpmisko 3y ago
We primarily cite medical scientific papers and health guidelines. We occasionally cite health blogs (such as healthline.com), but this is mostly for simple health questions (e.g., "tell me some nutrition tips").
27.
by mpmisko 3y ago
Hey, a React client is definitely on our roadmap! The demo ( https://tinyurl.com/medisearch-docs ) in our docs is written in JS, so it should be useful out of the gate! A proper client is coming in the next few weeks.
28.
by mpmisko 3y ago
Glad that you like it! We do support follow-up questions. Implementing a chat-style interaction should be very straightforward :) (see our Python client for examples https://pypi.org/project/medisearch-client/ ).
29.
Show HN: Medical LLM API on par with Google Med-PaLM 2. 92% USMLE accuracy(medisearch.io)
17 points by mpmisko 3y ago|13 comments