Skip to main content
Word embedding

Word embedding

Search complete. 66 mentions across 9 episodes found for "Word embedding".

Nov 18, 2026

speaker_0NARRATOR
1:27
Retrieved context, retrieval augmented generation, or RAG, chunks of records pulled from your index often more sensitive than the prompt itself.
speaker_0NARRATOR
1:35
Embeddings in vector databases' numeric versions of your documents' VEC2 text demonstrated exact recovery of 92% of 32 token inputs, including names from clinical notes.
speaker_0NARRATOR
1:46
Queues and payload stores requests waiting for a GPU data at rest is usually not considered.
speaker_0NARRATOR
1:52
Logs and traces full request bodies when debug logging is on ends up in a third-party log tool.
speaker_0NARRATOR
4:45
If you already have a BAA with your cloud provider, self-hosting in that account adds no new vendor.
speaker_0NARRATOR
4:51
Your agent needs features outside the BAA scope, batch jobs, file uploads, code execution.
speaker_0NARRATOR
4:58
your policy says no retention and the covered option keeps data for 30 days eu data must stay in the eu there are also state laws like texas sb 1188 that limit where records can live volume is steady and high which open source inference servers can you self-host the short skimmable list tool best for limitations vllm high throughput llm serving with broad model support one model per server You build routing and auth around ITS GLang multi-turned chat, structured output, RAG with shared prefixes smaller ecosystem than VLL Malama local development and single-user setups not designed for multi-tenant production TEI Text Embeddings Inference Embeddings and Rerankers 1 model per server No generation TGI Text Generation Inference Existing deployments only put in maintenance mode by hugging FaceSci.
speaker_0NARRATOR
5:51
Superlinked inference engine.
Alicia IrelandHOST
8:53
RAG was also discussed.
Alicia IrelandHOST
8:57
Embeddings are part of the 7.2 work, but there is no decision yet on whether RAG belongs in core or in plugins.
Alicia IrelandHOST
9:05
On what will ship with WordPress 7.2, the group asks for continued testing of the Secrets API, especially on Multisite, preparing developer notes for the Abilities API before RC1, and more feedback on a change to Abilities permissions that, if it doesn't receive support, could be reverted.
Alicia IrelandHOST
9:23
It was approved to move the shared abilities into their own Composer package within the WordPress organization on GitHub, so projects such as WooCommerce can use them without installing the entire AI plugin.
Jim HedgerHOST
74:48
Profile relevance model.
Jim HedgerHOST
74:50
It's profile relevance model using LLM-generated embeddings to show users more relevant posts from people.
Kristine SchachingerHOST
74:56
Wait, what was that? Embeddings?
Jim HedgerHOST
75:00
It's using LLM-generated embeddings.
Kristine SchachingerHOST
75:03
Nice.
Ghita Houir AlamiGUEST
37:32
For various reasons, I think you do need a mechanism on top of your initial first pass search.
Ghita Houir AlamiGUEST
37:41
Embeddings are not going to be enough for the whole dimensionality problem, where if you're trying to encode information into like an embedding with like a certain number of dimensions, you can like prove like there were some papers about this as well.
Ghita Houir AlamiGUEST
37:58
You can prove that, you know, you're never going to be able even for like a small K to retrieve all of the documents in the right order.
Ghita Houir AlamiGUEST
38:09
And so having like a step after this is definitely necessary.
OliverHOST
4:09
There's one piece of quiet magic that makes the search work, and the word is less cold than it looks.
OliverHOST
4:16
Embeddings.
OliverHOST
4:18
Stay with me.
OliverHOST
4:19
They're friendlier than the syllable.
Ben KusGUEST
22:18
And that's always been a problem.
Ben KusGUEST
22:19
And so then the really cool thing and the exciting thing is that the next generation of context retrieval, powered by some of the same technologies like vector embeddings and the ability to not just understand the lexical search capabilities but the semantic capabilities, like the meaning-based searches, this is now significantly improving.
Ben KusGUEST
22:39
And so then the idea of can a person, or an agent in this case, find the data it needs It may be the key to unlocking the benefits that any company would have to be able to use agents effectively.
Ben KusGUEST
22:53
And this is definitely an area that we're focused on.
Ben KusGUEST
23:14
Now, the problem was, I mean, it was amazing.
Ben KusGUEST
23:15
Here's 300 documents.
Ben KusGUEST
23:18
chunk them up, find the most related paragraph using vector embeddings, and then give it to an agent, which is awesome.
Ben KusGUEST
23:25
And when it works, it's great.
Christopher ManningHOST
1:40
I'll then introduce the ideas of the Word2Vec algorithm for learning word meaning, and then Going from there, we'll concretely work through how you can work out objective function gradients with respect to the word2vec algorithm and say a teeny bit about how optimization works.
Christopher ManningHOST
1:58
And then right at the end of the class, I then want to spend a little bit of time giving you a sense of how these word vectors work and what you can do with them.
Christopher ManningHOST
2:08
So really, the key learning for today is I want to give you a sense of how amazing deep learning word vectors are.
Christopher ManningHOST
2:18
So we have this really surprising result that word meaning can be represented, not perfectly, but really rather well by a large vector of real numbers.
Christopher ManningHOST
2:28
And that's sort of in a way a commonplace of the last decade of deep learning, but it flies in the face of thousands of years of tradition.

33 MINS LATER

Christopher ManningHOST
35:50
And so they're actually then grouped close together to the verb to be.
Christopher ManningHOST
35:54
So there's a lot of interesting structure in this space that then represents the meaning of words.
Christopher ManningHOST
36:02
So the algorithm I'm going to introduce now is one that's called Word2Vec, which was introduced by Tamás Mikulov and colleagues in 2013 as a framework for learning word vectors.
John HewittHOST
2:09
But when we discuss pre-training, it's important to know that subwords are a part of it.
John HewittHOST
2:15
Then we'll sort of motivate, we'll go on another journey of motivation of motivating model pre-training from word embedding.
John HewittHOST
2:21
So we've already seen pre-training in some sense in the very first lecture of this course because we pre-trained individual word embeddings that don't take into account their contexts on very large text corpora and saw that they were able to encode a lot of useful things about language.
John HewittHOST
2:38
So after we do the motivation, we'll go through model pre-training three ways.
John HewittHOST
2:42
And we're going to reference, actually, the lecture on Tuesday.

11 MINS LATER

John HewittHOST
14:02
So we never saw Tasty enough with however many As in order to add it into our subword vocabulary.
John HewittHOST
14:09
But we're still able to split it into things.
John HewittHOST
14:11
And then the neural network that runs on top of these subword embeddings could be able to sort of induce that, oh, yeah, this is one of those things where people chain letters together, chain vowels together in English for emphasis.
speaker_6HOST
23:56
No.
speaker_6HOST
23:57
It relies on vector embeddings.
speaker_4HOST
23:59
So when we upload our internal company documents into a RAG system, a specialized embedding model reads all the text and converts every sentence into a coordinate in a massive multi-dimensional space.
speaker_6HOST
24:10
Exactly.

We value your privacy

We use cookies to understand how you use our platform and to improve your experience. Click “Accept All” to consent, or “Decline non-essential” to opt out of non-essential cookies. Read our Privacy Policy.