Nearly 3x improvement by using advanced RAG techniques (XML knowledge graph) Excited to highlight this more over the next few days
LLMS
-
Advanced RAG Techniques Course for LLM Answer Generation
By
–
New short course on sophisticated RAG (Retrieval Augmented Generation) techniques is out! Taught by @jerryjliu0 and @datta_cs of @llama_index and @truera_ai , this teaches advanced techniques that help your LLM generate good answers.
— Andrew Ng (@AndrewYNg) 29 novembre 2023
Topics include:
– Sentence-window retrieval,… pic.twitter.com/bExne51PexNew short course on sophisticated RAG (Retrieval Augmented Generation) techniques is out! Taught by @jerryjliu0 and @datta_cs of @llama_index and @truera_ai , this teaches advanced techniques that help your LLM generate good answers. Topics include:
– Sentence-window retrieval, -
Data Science Basics in Python Tutorial
By
–
(6) Data Science Basics in Python – YouTube https://
bit.ly/40YfU1P
#AI #MachineLearning #DeepLearning #LLMs #DataScience -
Perplexity API Exits Beta with Usage-Based Pricing Model
By
–
pplx-api is now out of beta with new usage-based pricing! Access pplx online, pplx chat, and other open-sourced LLMs effortlessly via the pplx-api. Plus, Perplexity Pro users enjoy $5 credit every month! For commercial inquiries, contact us at api@perplexity.ai.
-
Perplexity Online Models Match GPT-3.5 and Llama2 Performance
By
–
pplx online models can match and even surpass gpt-3.5 and llama2-70b-chat performance on the task of answering questions with web search grounding, based on a curated human eval set of diverse prompts focused on helpfulness, factuality, and freshness.
-
Perplexity Launches pplx-7b and pplx-70b Online LLMs
By
–
We’re thrilled to announce two online LLMs we’ve trained: pplx-7b-online and pplx-70b-online! Built on top of open-source LLMs and fine-tuned to use knowledge from the internet. They are now available via Labs and in a first-of-its-kind live-LLM API.
-
Self-Destructing Foundation Models: Security Against Adversarial Customization
By
–
What is a self-destructing foundation model? One that adversaries can’t customize for harmful purposes:
-
ChatGPT Laziness Spike Linked to OpenAI Funding Round
By
–
Lots of people are talking about ChatGPT being lazy all of a sudden. I've been seeing it, too. I wonder if this is related to OpenAI being in the process of raising their next round and needing to slow burn until it's done…
-
SFT demos more likely than pre-training data
By
–
Based on the content seems more likely SFT demos than pre-train data.
-
Repetition in GPT-3 prompts elicits base-model-like generations, now confirmed
By
–
Fascinating work — I noticed repetition elicits base-model-like generations in GPT-3 last year but assumed they were fully hallucinated. Kicking myself now. As of August you could also do token repeats within a prompt to get base-like completion of grammatically incomplete text: