Miquel india

DL, RL and Audio researcher. PhD and telecom enginner. I study maths for fun.

[Miquel india @mikiindia]

Happy to announce that our paper
Double Multi-Head Attention for Speaker Verification got accepted at ICASSP 2021. Find attached here the pre-print and reposotory if you are interested :).

Pre-print: https://t.co/Kz6tE2HuMy
Git-hub: https://t.co/4tQPIotyoD

[Miquel india @mikiindia]

@padelmetrics1 Porque hasta que toca el suelo todo son desventajas respecto a las otras opciones. Peor parábola, bola que corre en general más lenta, bola que se acelera menos mientras corre ...

[Jon Barron @jon_barron]

I just pushed a new paper to arXiv. I realized that a lot of my previous work on robust losses and nerf-y things was dancing around something simpler: a slight tweak to the classic Box-Cox power transform that makes it much more useful and stable. It's this f(x, λ) here:

[Sepp Hochreiter @HochreiterSepp]

I am so excited that xLSTM is out. LSTM is close to my heart - for more than 30 years now. With xLSTM we close the gap to existing state-of-the-art LLMs. With NXAI we have started to build our own European LLMs. I am very proud of my team. https://t.co/IH7giCe3gd

[Miquel india @mikiindia]

@hbredin Speaker Diarization is a solved task :)

[Miquel india @mikiindia]

@jordiponsdotme @enricguso @stableaudio Revisa la "ç" a l'enllaç del arxiv :).

[Lenka Zdeborova @zdeborova]

For the fans of "ML explained on a blackboard": Lecture 13/14 was on the attention layer and mechanism (inspired by the relation to inverse Potts model from https://t.co/KJObRyn8yM), with a toy-transformer architecture counting the number of occurrences for the exercise.

[Miquel india @mikiindia]

@ParcolletT This whole idea of ​​reducing having a PhD in AI is going to give you a lot of job opportunities has become very toxic.

[Miquel india @mikiindia]

@hbredin 😅😅😅😅😅😅😅😅😅😅😅

[ptrblck @ptrblck_de]

Thanks a ton for awarding me the @PyTorch superhero award at the conference! It really made my day, and it was great seeing so many community members in real life. You all are amazing!

[Miquel india @mikiindia]

@hbredin At least he/she/they said please XD.

[Joan Serrà @serrjoa]

Join us for an internship this winter to work with @santty128 and me in #Barcelona: https://t.co/AGVH5A4NSJ

[Miquel india @mikiindia]

@hbredin Done.

[Miquel india @mikiindia]

25 d'Agost a Barcelona. Superpoder desitjat: NO SUAR.

[Karpi @karpi]

I've asked an AI to generate a trailer for a HEIDI movie and now I can never sleep again

[Miquel india @mikiindia]

Question for the speaker recognition community: Does anyone know of a github repository or similar that contains a PLDA implementation in Python?

[Miquel india @mikiindia]

@hbredin Congrats @hbredin

[Paula W @paulacat4]

Un tio m'acaba de seguir des del Parc de l'Espanya Industrial fins al meu portal (uns 10 minuts caminant). No me n'he adonat i l'he vist de cop davant el meu portal. He tancat la porta i l'ha començat a picar insistentment. VIGILEU per la zona del parc.

[Miquel india @mikiindia]

@serrjoa You feel sometimes privilieged just for having 4. 8 -16 would be the effective number. 32 64 GPUS per researcher is an outlier considering some company resources (even having DL products).

[Jim Fan @DrJimFan]

Do you know that DeepMind has actually open-sourced the heart of AlphaGo & AlphaZero?

It’s hidden in an unassuming repo called “mctx”: https://t.co/GpNtwH9BxA

It provides JAX-native Monte Carlo Tree Search (MCTS) that runs on batches of inputs, in parallel, and blazing fast.
🧵

[Miquel india @mikiindia]

@ChombaBupe Just curious. What do you think is then the problem here ? The approach of designing IA algorithms as mapping functions or how these ones have scaled in terms of being trained with large amounts of data ?