Lefteris Spyromitros

Data Scientist @ Expedia

[OpenAI @OpenAI]

We’re making better intelligence easier to access in ChatGPT for everyone:

- GPT-5.6 Sol now powers both Instant and deep reasoning for Plus & Pro users, delivering more factual, focused responses.

- Free & Go users get unlimited text chats with GPT-5.6 Luna starting tomorrow.

[Lefteris Spyromitros @espyromi]

Rd https://t.co/RDtPhH3sRK r

[Lefteris Spyromitros @espyromi]

Yet right to try thehttps://genbio.ai/fm4bio-seminar/?twclid=2-27d2pkhgi58c3tptzbwe5uf95

[Lefteris Spyromitros @espyromi]

@AIatMeta @FabianGloeckle In the causal architecture, did you try feeding f_s to all heads instead of only head 1? Basically concatenate f_s + f_h_i-1 as input to f_h_i?
This would be more akin to the chaining approach.

[Lefteris Spyromitros @espyromi]

@AIatMeta Appreciate the citation!

[Lefteris Spyromitros @espyromi]

@ylecun @CrowdspeakAI majority

[Lefteris Spyromitros @espyromi]

@natolambert @CrowdspeakAI majority

[Lefteris Spyromitros @espyromi]

@sokkarockedya @Google Same issue here!

[Lefteris Spyromitros @espyromi]

6

[Lefteris Spyromitros @espyromi]

@OOmorfantras Γεια σου παιδαρε!

[Lefteris Spyromitros @espyromi]

@sympap How about IEEE TNNLS?

[Lefteris Spyromitros @espyromi]

We got a new paper accepted at #ijcnn2020, part of #wcci2020! And we just made it available at arXiv: https://t.co/yxBWUvrVgg

[Sebastian Ruder @seb_ruder]

Dear OpenAI: Please Open Source Your Language Model
Nice piece by @gradientpub's @hughbzhang that makes the case that "withholding the full GPT-2 model is both unnecessary for safety reasons and detrimental to future progress in AI". https://t.co/v0BB3bYdKf

[Lefteris Spyromitros @espyromi]

Deep Visualization Toolbox https://t.co/9Rf7BzlTy6 via @YouTube

[Bored Yann LeCun @boredyannlecun]

So, #EMNLP2018 has improved strikingly over past years by most key metrics (convolutions per paper, pools per paper, etc.). Still it lags behind NIPS, ICML, and—dare I say it?—CVPR! #torched

[TensorFlow @TensorFlow]

Image augmentation lets you get the most of your dataset. We released state-of-the-art AutoAugment Modules on #TFHub allowing you to train better image models with less data! #TensorFlowHub #GoogleAI

Check it out here → https://t.co/j5npPtbmzb

[DataScouting @DataScouting]

We're hiring Data Scientists for our team. Check out our LinkedIn or FB page for more information.

[Lefteris Spyromitros @espyromi]

Join me in following this newsletter? 👀 https://t.co/HG5b5wJj02 via @revue

[AI at Meta @AIatMeta]

Marketplace is powered by machine learning. Today, tied to Marketplace's 2nd anniversary, we're sharing how we're using computer vision and NLP to make buying and selling more convenient. https://t.co/raF7gkWj0e