In order to get the best performance out of your #LLM, you should fine-tune it.
Or should you?
How finetuning can backfire, what you can try instead (or additionally), and where it actually makes sense:
https://t.co/UUa3lVRiHT
Google doesn't show search result counts anymore.
By contrast, Bing still shows them. For example, when you search Bing for "palantir", it says that there are "about 125,000 results".
But does this number actually mean anything?
I'd argue no, in my latest blog post 👇
"Why counting results doesn't make sense for modern search"
https://t.co/ULUcofYwzJ
#vectorsearch #AI #searchengines
AI-first data storage is designed to make text, images or video usable by AI systems, especially #LLMs.
Here are a few things to consider when setting up an AI-first data storage system:
https://t.co/jmDaZkU0DB
Researching a company from a science, technology, and market perspective takes a lot of time.
In this article, you'll learn how Mergeflow helps you accelerate this process substantially:
https://t.co/lqG0oh2W3b
#VentureCapital#EmergingTech
Homomorphic encryption is a type of encryption technique that allows operations to be performed on data without decrypting it.
For more, see our latest Weekly Innovation Insight:
https://t.co/H3RGWn0ydp
#encryption#DataScience#EmergingTech
Thermal batteries are energy storage devices which use heat to store energy.
They can provide backup power for critical applications in areas where the power grid is unreliable, for example.
Learn more here:
https://t.co/7wqyoJJGUz
#EmergingTech#EnergyStorage#batteries
A good large language model (e.g. #GPT or #AI21) should "behave" like humans do -- right?!?
Turns out it's not so easy.
Very interesting paper by @albertwebson, Alyssa Marie Loo, Qinan Yu and Ellie Pavlick:
https://t.co/p9jCZlrt1U
#AI#LLM
Drug-target interactions determine how a drug works, and which cells or mechanisms it acts on.
Here's a paper that uses pre-trained language models for learning drug-target interactions:
"Learning the drug-target interaction lexicon"
https://t.co/ejLsyIYQJ2
#LLM#drugdiscovery
Our new Weekly Innovation Insight, on #passwordless authentication:
https://t.co/GVJfDgSNOW
Passwordless authentication lets you log in to a computer system without the entering (and having to remember) a password or any other knowledge-based secret.
When you use #LLMs in a context where they interact with humans, you should evaluate them against that context.
Surprisingly, almost nobody does this.
But here’s a recent paper that does:
"Evaluating Human-Language Model Interaction"
https://t.co/lCCpSVTLU1
#AI
Long duration energy storage allows electricity to be stored for long periods of time. This helps manage the intermittency of #renewables like wind and solar.
Here's our Weekly Innovation Insight:
https://t.co/ijCXLtHhbL
Examples of what you'll find are in the thread.
🧵1/4
Are you looking for something to read over the holidays?
Here are 6 of my favorite deep tech fiction novels:
https://t.co/SFAEDsRxfG
#AI#deeptech#fiction