Graph Language Models EXPLAINED in 5 Minutes! [Author explanation 🔴 at ACL 2024]

preview_player
Показать описание
How to make powerful LLMs understand graphs and their structure? 🕸️ With Graph Language Models! They take a pre-trained language model and fit it with the ability to process graphs. Watch if you're curious about how this works (hint: choose the right positional embeddings)!

Follow Moritz Plenz (first author) on:

Thanks to our Patrons who support us in Tier 2, 3, 4: 🙏
Dres. Trost GbR, Siltax, Vignesh Valliappan, Michael, Sunny Dhiana, Andy Ma

Outline:
00:00 LLM for graphs
01:08 Motivation
02:02 Key idea of Graph LLMs
02:25 Relative Positional Encodings
03:00 Method (Graph LLMs)
04:04 Experiments and Evaluation
04:49 Results
06:07 Outro

▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀
🔥 Optionally, pay us a coffee to help with our Coffee Bean production! ☕
Join this channel as a Bean Member to get access to perks:
▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀

🔗 Links:

#AICoffeeBreak #MsCoffeeBean #MachineLearning #AI #research

Video editing: Nils Trost
Music 🎵 : Space Navigator – Sarah, the Illstrumentalist
  1. AI Coffee Break with Letitia
  2. Graph Language Models paper explained
  3. graph LLMs
  4. LLM graph neural network
  5. LLMs and GNNs
  6. GNNs meet LLMs
Рекомендации по теме
Graph Language Models 0:06:38

Graph Language Models EXPLAINED in 5 Minutes! [Author explanation 🔴 at ACL 2024]

Language Models are 0:52:16

Language Models are Open Knowledge Graphs (Paper Explained)

How Large Language 0:05:34

How Large Language Models Work

How large language 0:27:14

How large language models work, a visual intro to transformers | Chapter 5, Deep Learning

What is a 0:05:36

What is a Knowledge Graph?

GraphRAG: LLM-Derived Knowledge 0:15:40

GraphRAG: LLM-Derived Knowledge Graphs for RAG

Fusing Knowledge Graphs 1:23:24

Fusing Knowledge Graphs and Large Language Models

Large Language Models 0:08:25

Large Language Models from scratch

🔵 Embed the 0:50:52

🔵 Embed the latest from SAP HANA’s AI functions into your SAP BTP Application

What is Retrieval-Augmented 0:06:36

What is Retrieval-Augmented Generation (RAG)?

Knowledge Graph Construction 0:06:42

Knowledge Graph Construction Demo from raw text using an LLM

Graph-of-Thoughts (GoT) for 0:41:34

Graph-of-Thoughts (GoT) for AI reasoning Agents

Attention in transformers, 0:26:10

Attention in transformers, visually explained | Chapter 6, Deep Learning

The Future of 0:28:29

The Future of Knowledge Graphs in a World of Large Language Models

Large Language Models 1:40:04

Large Language Models and Knowledge Graphs: Merging Flexibility and Structure

Transformer Neural Networks, 0:36:15

Transformer Neural Networks, ChatGPT's foundation, Clearly Explained!!!

What are Large 0:05:30

What are Large Language Models (LLMs)?

“What's wrong with 0:49:47

“What's wrong with LLMs and what we should be building instead” - Tom Dietterich - #VSCF2023...

Graphs, Vectors and 0:23:08

Graphs, Vectors and Machine Learning - Computerphile

Build a Large 0:02:53

Build a Large Language Model AI Chatbot using Retrieval Augmented Generation

Large Language Models 0:01:59

Large Language Models - Explained via Readable Video

Transformers with Knowledge 0:11:26

Transformers with Knowledge Graph | K-BERT: Enabling Language Representation with Knowledge Graph

Foundation Models: An 0:02:09

Foundation Models: An Explainer for Non-Experts

[1hr Talk] Intro 0:59:48

[1hr Talk] Intro to Large Language Models

Комментарии
Автор

Its interesting to see "attention" on graph structures again. I think in the future a more structured knowledge representation may play a role on improving reasoning, as we could leverage logic and rules using engines on them, like compilers aid in code generation.

bensimonjoules
Автор

Excellent! need to go deeper that could be a game changer for reasoning, as it makes more sense to reason on a graph rather than on the next token.

jmirodg
Автор

Thank you for this video Letitia! As always amazing :=)

Side note: Loved the silent Good Bye :)))

bharanij
Автор

I'm interested even more on the generative side, generating large graphs with contained text in them from a prompt, can be useful for modalities which are represented by large graphs. I've yet to see anyone doing this. While you can prompt LLMs to generate small graphs, for larger graphs you see significant performance drops.

vladimirtchuiev
Автор

Interesting. It's almost like two types of tokens: nodes + edges which can each be compressed to a feature vector. But yes, with positional encoding you're left with "random walk with restart" or a traversal depth. Or one could sum node_vector + edge_vector ~= positional distance. but yeah, more graph solutions coming in the future.

jonclement
Автор

It would be interesting how much computation this needs

quebono
Автор

I need to read this more deeply, I don't understand why would just grafting the parameters willy nilly works

yorailevi