Podcasts Technology Data Science at Home [RB] Replicating GPT-2, the most dangerous NLP model (with Aaron Gokaslan) (Ep. 83)

[RB] Replicating GPT-2, the most dangerous NLP model (with Aaron Gokaslan) (Ep. 83)

Published: Oct. 18, 2019, 3:13 a.m.

Join the discussion on our Discord server\n\xa0\nIn this episode, I am with Aaron Gokaslan, computer vision researcher, AI Resident at Facebook AI Research. Aaron is the author of OpenGPT-2,\xa0a parallel NLP model to the most discussed version that OpenAI decided not to release because too accurate to be published. \nWe discuss about image-to-image translation, the dangers of the GPT-2 model and the future of AI. Moreover,\xa0Aaron provides some very interesting links and demos that will blow your mind!\n\nEnjoy the show!\xa0\nReferences\n\nMultimodal image to image translation (not all mentioned in the podcast but recommended by Aaron)\nPix2Pix:\xa0\n\nhttps://phillipi.github.io/pix2pix/\n\xa0\nCycleGAN:\nhttps://junyanz.github.io/CycleGAN/\n\xa0\nGANimorph\nPaper: https://arxiv.org/abs/1808.04325\nCode: https://github.com/brownvc/ganimorph\n\xa0\nUNIT:https://arxiv.org/abs/1703.00848\n\n\n\n\nMUNIT:https://github.com/NVlabs/MUNIT\nDRIT:\xa0https://github.com/HsinYingLee/DRIT\n\n\n\n\xa0\n\nGPT-2 and related\xa0\nTry OpenAI's GPT-2: https://talktotransformer.com/\nBlogpost:\xa0https://blog.usejournal.com/opengpt-2-we-replicated-gpt-2-because-you-can-too-45e34e6d36dc\nThe Original Transformer Paper: https://arxiv.org/abs/1706.03762\nGrover: The FakeNews generator and detector: https://rowanzellers.com/grover/