We'll likely do our webinar next week on @OpenAI functions and the various use cases for them, and any tips and tricks Hoping to have @fpingham and @jxnlco on. Anyone else doing interesting stuff with them we should get?
LLMS
-
OpenAI Functions for Structured Data Extraction LangChain
By
–
The new @OpenAI functions are good for other things besides agents Another killer use case is extracting structured information from unstructured docs We've adding support for extraction AND tagging in @langchain – thanks to @fpingham for code and @jxnlco for review
-
Entity Extraction with LangChain: Schema-Based Data Parsing
By
–
Extraction Specify a schema – either a dictionary or a Pydantic model – and then extract entities from a piece of text with the same schema This will return a list of objects with that schema Docs: https://
python.langchain.com/en/latest/modu
les/chains/examples/extraction.html
… -
GPT-4 Native Tool Use Makes Building Powerful LLM Agents Easier
By
–
This week @OpenAI gave GPT-4 the ability to use tools natively in their API. This instantly makes it easier to build powerful LLM use cases like agents with far less code than ever before. I dig into what it all means here:
-
Machine Training: Fine-tuning or Semantic Search, Not ChatGPT
By
–
It talks about engineers who spent a week training the machine. So no, I really think it's not ChatGPT but rather fine-tuning or semantic search.
-
AI Models as Primary Interface for Daily Digital Interaction
By
–
This is excellent. “I think in the same way that most of us probably spend the majority of our time on a device, on a screen … a big chunk of your time is going to be spent communicating with these models. They are going to be your interface to the world.”
-
News Publishers Battle Tech Companies Over AI Training Data
By
–
News publishers are up in arms about their content being used to train AI that could cut them out – forcing tech cos to the negotiating table. Our story, brought to you from London, NY & Berlin @CristinaCriddle @DanielThomasLDN @annaknicolaou @laurapitel
-
Post criticizes test on ChatGPT as complete bullshit
By
–
Qu'on soit clair, ce test est une IMMENSE FOUTAISE.
— Defend Intelligence (Anis Ayari) (@DFintelligence) 16 juin 2023
Enthoven parle "d'entrainement pendant des jours", or il est impossible d'entrainer #ChatGPT (ils ont dû utiliser d'autres modèles).
Et impossible de juger quoi que ce soit sans le protocole expérimental (prompts,… https://t.co/v6IteS3UdqLet’s be clear, this test is a COMPLETE AND UTTER BULLSHIT. Enthoven talks about “training for days,” yet it’s impossible to train #ChatGPT (they must have used other models). And impossible to judge anything without the experimental protocol (prompts, training, the model, the
-
Enthoven vs ChatGPT, AI Act, Macron, LeCun, Musk
By
–
LIVE
Big AI program today:
– Enthoven vs ChatGPT in philosophy, is he happy?
– The AI Act is here, we debrief together what's inside it
– Replay of @EmmanuelMacron and @ylecun on AI at Vivatech At 4pm I will relaunch for a live React to Elon Musk's intervention -
Large-scale Language Models for ASR Rescoring on Long-form Video Data
By
–
Large-scale Language Model Rescoring on Long-form Data paper page: https://
huggingface.co/papers/2306.08
133
… In this work, we study the impact of Large-scale Language Models (LLM) on Automated Speech Recognition (ASR) of YouTube videos, which we use as a source for long-form ASR. We demonstrate