.
@sama says ChatGPT will have to evolve in “uncomfortable” ways
LLMS
-
ChatGPT Must Evolve in Uncomfortable Ways Says Altman
By
–
-
LangGraph Official Launch: Building AI Agents Framework
By
–
LangGraph We introduced LangGraph last week, but are excited to launch it more officially today (blog and YouTube series). It includes: Example of replicating the LangChain AgentExecutor
A chat agent executor (aimed specifically at chat models) LangGraph makes it easy -
Top 10 Ways to Boost LLM Outputs from Research
By
–
That's the top 10 surprising ways to boost LLM outputs. The full paper outlined 26 ways, but these 10 felt the most applicable after it started to get repetitive. Here's the link to the full paper: https://
arxiv.org/pdf/2312.16171
v1.pdf
… -
Assign Roles to Large Language Models
By
–
10. Assign a role to the large language models. pic.twitter.com/1od0vv86Be
— The Rundown AI (@TheRundownAI) 17 janvier 202410. Assign a role to the large language models.
-
DataLLM: Ask Questions About Your Data with AI
By
–
Use LLMs to get insights from your data with DataLLM How does it work? You ask chatbot questions and insights about your data, and the AI agent will generate code, run queries, understand the results, and present answers. No need to dig through reports and dashboards. It’s
-
AlphaGeometry Neuro-Symbolic Architecture: System 1 and System 2
By
–
This is the neuro-symbolic architecture of #AlphaGeometry. Similar to System1 and System 2, in the book "Thinking, fast and slow", the symbolic engine will first take a crack at the problem mechanically; if it gets stuck it will ask the neural language model for suggestions of
-
OpenAI Internal GPT-5 Access: Competitive Advantage Concerns
By
–
And I suppose the same holds true for GPT-5. Whenever it is done training, and moves into red teaming, does OpenAI keep a raw internal version to use for themselves and their own business? If so, what capabilities and powers does that give them that the rest of us won’t have?
-
Efficient SMoE: Merging and Compression via Routing Policy
By
–
Merge, Then Compress: Demystify Efficient SMoE with Hints from Its Routing Policy Li et al.: https://
arxiv.org/abs/2310.01334 #ArtificialIntelligence #DeepLearning #MachineLearning -
Sampling Strategies for AI Text Generation: Temperature, Top-K, Top-P
By
–
New post: Sampling for Text Generation https://
huyenchip.com/2024/01/16/sam
pling.html
… Many challenges (and opportunities) in working with AI today stem from the way models sample their outputs. This post covers: 1. Sampling strategies and variables including temperature, top-k, and top-p.
2. How