Shoval Messica

Audio & Speech MSc student at HUJI @cseHUJI

[Shoval Messica @ShovalMessica]

🚨I’m excited to share our #INTERSPEECH2024 paper "NAST: Noise Aware Speech Tokenization for Speech Language Models” 🥳 W/ @adiyossLC
Paper-https://t.co/olTARW7ckc
Code-https://t.co/SeuW6KkaVB

[Michael Hassid @MichaelHassid]

Have LLMs become supervised learners (once again)?!

In our new paper, we argue that current LLMs’ post-training methods have effectively reverted to the "pre-train then fine-tune" era, explicitly tailoring models to desired behaviors.

1/n

[Shoval Messica @ShovalMessica]

@OmriFahn Amazing!

[Omri Fahn @OmriFahn]

MFA is a beautiful visualization, but not only that.
It’s a practical tool:
competitive localization and better steering, while revealing that concepts live in regions, not just single directions.

Grateful I got to contribute to this with an awesome team!

[Omri Fahn @OmriFahn]

8/
📄 Read the full paper:
https://t.co/XbAc6ucXmo
Thanks for reading!

#AI #LLM #Uncertainty #Interpretability #Research

[Omri Fahn @OmriFahn]

🤔Can an LLM "unconscious" feel a lie bubbling up before it speaks?
🤔And what mysteries hide in its linear space of uncertainty?
Thrilled to share our new paper:
“Pre-trained LLMs Learn Multiple Types of Uncertainty” (@roicohen9•@OmriFahn•@GerardDeMelo)
https://t.co/XbAc6ucXmo

[Gallil Maimon @GallilMaimon]

🚨New paper on SLM evaluation🚨
We present SALMon🍣 which is a suite of benchmarks for evaluating how much Speech Language Models model acoustic elements like sentiment or background noise.
Project: https://t.co/cjE2T0LnKz 🧵👇🏻

[Michael Hassid @MichaelHassid]

Which is better, running a 70B model once, or a 7B model 10 times? The answer might be surprising!

Presenting our new @COLM_conf paper: "The Larger the Better? Improved LLM Code-Generation via Budget Reallocation"

https://t.co/Zayq02RFJJ

1/n

[Shoval Messica @ShovalMessica]

@MohammadSalaama Very interesting!

[Mohammad Salama @MohammadSalaama]

I am excited to share my first work: "Dataset Size Recovery from LoRA Weights".
Ever wondered if you could find out how many samples was a model trained on using just its weights? Well now you can!
Project: https://t.co/YL6qs5mVfE
👇

[Yossi Adi @adiyossLC]

Speech tokenizers are fundamental in building speech LMs. In a recent study we show the common tokenization method (a.k.a “semantic tokens”) is not robust to different signal variations. We then present NAST! a noise aware speech tokenizer

w. @ShovalMessica
Code and models 👇

[Guy Yariv @guy_yariv]

1/ Commonsense reasoning needs multimodal knowledge, yet current LLMs focus mostly on text, limiting their integration of crucial visual information.

We introduce vLMIG, a method that enhances LLMs' visual commonsense by integrating images into the decision-making process

[Audio and Speech Processing Papers @AudioAndSpeech]

NAST: Noise Aware Speech Tokenization for Speech Language Models. https://t.co/IJvS2Rregw

[arXiv Sound @ArxivSound]

``Joint Audio and Symbolic Conditioning for Temporally Controlled Text-to-Music Generation,'' Or Tal, Alon Ziv, Itai Gat, Felix Kreuk, Yossi Adi, https://t.co/cFBRv17ipm

[Or Tal @Or__Tal]

Today we released JASCO 🥁🎵🪇!
JASCO uses time-aligned conditions (e.g. drums, melody) as guiding controls for text-to-music generation.

📜Full Paper: https://t.co/exolVO0UUn
🔊Samples: https://t.co/ONL4ATy3lx

@lonziks @itai_gat @FelixKreuk @adiyossLC
Stay tuned for updates.

[arXiv Sound @ArxivSound]

``NAST: Noise Aware Speech Tokenization for Speech Language Models,'' Shoval Messica, Yossi Adi, https://t.co/J1smUHwR3y