PinnedPublished inGraphcoreDolly 2.0 — open source language model with ChatGPT-like interactivityDolly 2.0 is an open source large language model suitable for commercial use: learn more about how to use Dolly for AI applications on…Apr 26, 2023Apr 26, 2023
PinnedPublished inGraphcoreFine-tune GPT-J: a cost-effective GPT-4 alternative for many NLP tasksWalkthrough tutorial fine-tuning GPT-J for state-of-the-art performance: Entailment, Named Entity Recognition, Sentiment Analysis…May 11, 2023May 11, 2023
PinnedPublished inGraphcoreHugging Face and Graphcore partner for IPU-optimized TransformersSOTA Transformers meet SOTA hardwareSep 14, 2021Sep 14, 2021
PinnedPublished inGraphcoreBert-Large Training on the IPU explainedGraphcore engineers explain how they implemented and optimized BERT-Large for IPU systems using standard machine learning training schemes.May 6, 2021May 6, 2021
PinnedPublished inGraphcoreGraphcore announces Major Software Release Poplar SDK 2.0Features include technologies for seamless application scale-out, enhanced framework support and new functionality for analysis tools.Mar 17, 2021Mar 17, 2021
Published inGraphcoreText Summarization with BART-Large on IPUTry one of the most powerful and versatile applications of the BART language model using a Paperspace Gradient NotebookAug 11, 2023Aug 11, 2023
Published inGraphcoreFine-tune OpenAI’s Whisper Automatic Speech Recognition (ASR) ModelGet the most out of Whisper by optimising if for new use cases, including better comprehension of specific languages and dialects, as well…Aug 9, 2023Aug 9, 2023
Published inGraphcoreOASST1 Fine-tuned Pythia-12B chatbot: open-source ChatGPT alternativeThis powerful open-source chatbot has no usage restrictions, so is perfect for commercial applications. Try it now on Paperspace, powered…May 31, 2023May 31, 2023
Published inGraphcoreRunning Flan-T5-XL inference in Float16 for IPU — how we did itT5 has only worked with bfloat16, until now. Find out how Graphcore made it run on float16 hardware.May 30, 2023May 30, 2023
Published inGraphcoreFlan-T5: sweet results with the smaller, more efficient LLMFlan-T5 offers outstanding performance for a range of NLP applications, even compared to very large language models. Try now on Paperspace…May 30, 2023May 30, 2023
