Product · mentioned 61 times
Large language model released by OpenAI in 2020 as part of its GPT series, an early predecessor to ChatGPT.
Tap a name to go there · tap a line to see why they are linked
Unsupervised Learning with Jacob Effron· 5 Nov
Diogo says models from the GPT-3.5 era, possibly even the GPT-3 era, could have met OpenAI's definition of AGI at the time.Venture with Grace· 5 Nov
Ori Goshen describes the few-shot era, in which a model was given a few natural-language examples of a task and then followed the pattern on new input.Scaling Europe· 3 Nov
GPT-3 was released at Christmas 2022 and became the only thing Dimitris Samouris and his co-founder could think about.A Product Market Fit Show | Startup Podcast for Founders· 2 Nov
Rahul Shivkumar says he thought they were late to the AI wave, which had already started happening with GPT-3."The Cognitive Revolution"· 31 Oct
Sagar Vaze says GPT-3 was a dense model trained with 175 billion parameters in total, with all parameters activated for each token.