LLMS - AI Dynamics

Getting Started with Memory in Conversation Chains

By

–

28 November 2022 18h43

Here is a notebook for getting started with memory. It walks through maybe the most clear application of memory: a conversation chain. It showcases short term memory – remembering the conversation to date https://
langchain.readthedocs.io/en/latest/gett
ing_started/memory.html
…

→ View original post on X — @langchain

28 November 2022

LangChain Chains Now Support Stateful Conversations and Memory

By

@langchain

–

28 November 2022 18h43

So far all the chains/agents in LangChain have been stateless, meaning each interaction is independent This is far from ideal for some real world settings (eg chatbots) where you want the chain/agent to remember previous conversations

→ View original post on X — @langchain

28 November 2022

Types of Memory in LangChain: Short-term and Long-term

By

@langchain

–

28 November 2022 18h43

There are different types of memory, eg: Short term memory, like what the previous user input was Long term memory, like what the user previous said about a particular person or place Both of these can fit into this memory abstraction. Let's take a closer look at some

→ View original post on X — @langchain

28 November 2022

ConversationChain and Memory Module in LangChain

By

@langchain

–

28 November 2022 18h43

This past Friday @sjwhitmore added the very cool ConversationChain As part of doing this, she also added the concept of a Memory module. Here's what that is and why it's useful, :

→ View original post on X — @langchain

28 November 2022

GPT-4 Launch: Next Generation AI Model Surpasses Competition

By

@deeplearn007

–

28 November 2022 17h15

GPT-4 is Almost Here, And it Looks Better than Anything Else https://
analyticsindiamag.com/gpt-4-is-almos
t-here-and-it-looks-better-than-anything-else/
… @pierrepinna @sallyeaves @PawlowskiMario @Xbond49 @gvalan @psb_dc @SpirosMargaris @JimMarous @HaroldSinnott @mikeflache @Shi4Tech @Nicochan33 #AI #MachineLearning #DeepLearning #Fintech #5G

→ View original post on X — @deeplearn007

28 November 2022

Closed-book exams now mean without Large Language Models

By

@pmddomingos

–

27 November 2022 20h32

“Closed-book” is now “without LLMs”.

→ View original post on X — @pmddomingos

27 November 2022

Pre-trained Language Models Essential for Agent Communication

By

@guillaumelample

–

27 November 2022 17h07

If AZ cannot rely on a pre-trained LM to communicate, it is impossible (there is no reason the model learns with self play to communicate with valid sentences). Other issue will of course be the infinite action space (which we also had in theorem proving) but this is manageable

→ View original post on X — @guillaumelample

27 November 2022