The updated Embedding model offers State-of-the-Art performance with a 4x longer context window. The new model is 90% cheaper. The smaller embedding dimensions reduce cost of storing them on vector databases. — Introduction OpenAI updated in December 2022 the Embedding model to text-embedding-ada-002. The new model offers: 90%-99.8% lower price 1/8th embeddings dimensions size reduces vector database costs

This tutorial guides step-by-step the new open source model Whisper! It is State-of-the-Art in the Speech Recognition! — OpenAI open-sourced Whisper model – the State-of-the-Art Speech recognition system. We tested it and got impressed! We took the latest RealPython episode for 1h 10 minutes. It took us 56 minutes with a basic CPU to convert the audio file into almost perfect text transcription with the smallest Whisper model.

DetectGPT¹, OpenAI Text Classifier² and Watermark³-technique attempt to detect AI-written text with varying results. Introduction The lack of tools to detect AI-written text — is an ongoing debate in industries from Publishing, Education to Social media. OpenAI published recently a research related to risks around Social media and Publishing industry:

Learn as a writer and increase your income with these 5 techniques and ChatGPT feedback. — Introduction AI tools can help writers earn more money. AI tools help with grammar, readability and text generation — making writing simpler. Yet, none of these tools replace writer’s creativity or intuition. Many AI writing tools are costly and only focus on improving the text rather than the writer. …

OpenAI released ChatGPT today — a new language model for a chat. — Introduction OpenAI released ChatGPT today. ChatGPT is based on the GPT-3.5 model, which is fine-tuned with Reinforcement Learning from Human Feedback (RLHF). The model is currently free and only available via OpenAI-website for research purposes. The model access is restricted with OpenAI credentials. Let’s get started! ChatGPT ChatGPT is a…

Galactica is science-oriented 120B parameter language model. It predicts protein annotations, creates lecture notes and explains mathematical formulas as text. Let’s get started! — Introduction Galatica is large open-source language model by Meta AI. The model manages multiple science tasks using single model. Galatica model performs reasoning, creates lecture notes, predict citations and more. The model explains as text mathematical formulas and python code.

Mind’s Eye improves zero-shot accuracy by 27.9% in physical reasoning. — Introduction Mind’s Eye is a language model for physical reasoning with MuJoCo environment. Mind’s Eye uses simulation to learn about the physical world, instead of human inputs such as text. Mind’s Eye outperforms all existing language models in various physical reasoning tasks. The model improves physical reasoning accuracy by by…

OpenAI open-sourced speech recognition model Whisper. It offers State-of-the-Art zero-shot performance. — Introduction Whisper is an Automatic Speech Recognition (ASR) system. Whisper recognizes multilingual speech into text, translate speech and identify the spoken language. The model is: Encoder-Decoder Transformer 1.55B parameters It inputs audio and outputs text The model trained with 680k hours of audio-text pairs. …

OpenAI trained Video PreTraining (VPT) model with 70k hours to play Minecraft. Scaled version may include +1M hours of video. — Introduction Video Pretraining (VPT) model learned to play Minecraft at human-level. The model crafts diamond tools. The game environment is: complex, open-ended and randomly generated. It uses the native human interface: keyboard, mouse and 20Hz frame rate. Such environment is more difficult for Reinforcement Learning (RL) agents.