Which modeling to choose for text-to-music generation? We run a head-to-head comparison to figure it out. Same data, same architecture - AR vs FM. 👇 If you care about fidelity, speed, control, or editing see this thread. 🔗 https://t.co/FBWu7ThspC 📄 https://t.co/Dp1co1esvd 1/6

🚀 Excited to share our new paper from my internship at IBM Research! We train Speech-Aware LLMs (SALLMs) with Group Relative Policy Optimization (GRPO) on open-ended tasks (Spoken QA & Speech Translation). We find that GRPO beats SFT! 🧵

🚨 New paper alert! 🚨 Millions of neural networks now populate public repositories like Hugging Face 🤗, but most lack documentation. So, we decided to build an Atlas 🗺️ Project: https://t.co/1JpsC6dCeg Demo: https://t.co/4Xy7yLdIZY 🧵👇🏻 Here's what we found:

Excited to share this has now been accepted at #NeurIPS2025 as a position paper (<6% acceptance)!🎉 We advocate for systematically studying entire model populations via weight-space learning, and argue that this requires charting them in a Model Atlas. @NeurIPSConf #NeurIPS 🧵👇

Excited to share our work Set Block Decoding! A new paradigm combining next-token-prediction and masked (or discrete diffusion) models, allowing parallel decoding without any architectural changes and with exact KV cache. Arguably one of the simplest ways to accelerate LLMs!

Many modern SpeechLMs are trained with Speech-Text interleaving. How does this impact scaling trends? In our new paper, we train several dozen SLMs, and show - quite a lot! So there is room for optimism 😊 Key insights, code, models, full paper 👇🏻

🎉Thrilled that our paper on "scaling analysis of interleaved speech-text LMs" was accepted to #CoLM2025 It gives room for optimism when scaling SpeechLMs *right* - with large TextLMs (in place of more data), interleaving, and synth training data💪

Happy to announce that our paper “EditInspector: A Benchmark for Evaluation of Text-Guided Image Edits” was accepted to #ACL2025 🎉 📄 https://t.co/mwugXz1H5q 🌐

💣Introducing PAST: a speech tokenizer that jointly model phonetics and acoustics (No SSL involved). Past demonstrates great reconstruction as well as semantic capabilities in the form of ABX and sWUGGY. 🤗 https://t.co/teQQ9s5whr Check out Nadav's post👇@NadavHarTuv @adiyossLC

Auto-Regressive vs Flow-Matching: a Comparative Study of Modeling Paradigms for Text-to-Music Generation.

🎵💬 If you are interested in Audio Tokenisers, you should check out our new work! We empirically analysed existing tokenisers from every way - reconstruction, downstream, LMs and more. Grab yourself a ☕/🍺 and sit down for a read!

🚨 New Paper: "Time to Talk"! 🕵️ We built an LLM agent that doesn't just decide WHAT to say, but also WHEN to say it! Introducing "Time to Talk" - LLM agents for asynchronous group communication, tested in real Mafia games with human players. 🌐 https://t.co/HdNUwlvF2F 🧵1/7

Read the full paper! 🔗 https://t.co/FBWu7TgUA4 📄 https://t.co/Dp1co1dUFF @FelixKreuk @adiyossLC

We’ve been exploring the trade-offs between Autoregressive and Flow-Matching models for music generation. We share our findings in this latest paper led by @Or__Tal. Many interesting take-aways and practical advice on training generative models for music! 🎶🧠

What if training steps are capped at 500k? FM reaches near-topline quality with small batches. It’s compute-efficient and forgiving. AR needs larger batch sizes to recover performance. It benefits more from large-scale training. See📉 below by model duration + batch size: 6/6

⚡ Speed and Quality: FM can be faster - but at the cost of quality reduction. AR scales better with batch size (thanks to KV caching). 5/6

🎛️ Editing music via inpainting (filling gaps) Supervised FM came out on top 🧠 listeners rated FM best for transition smoothness & audio match AR scored the lowest FAD , but had audible transitions. Zero-shot FM is fast, flexible… still unstable. 🎧 https://t.co/FBWu7ThspC 4/6

🧩 Temporal control (chords, melody, drums)? AR tracked them more accurately. But both AR & FM lost fidelity under strict constraints. 🎧 https://t.co/FBWu7TgUA4 3/6

Here’s the side-by-side: 🎼 AR vs FM across 5 axes - fidelity, control, editing, speed, and training. No clear winner. Every strength is a trade-off. 👇 A snapshot of where each modeling paradigm shines (and struggles): 📄Full paper: https://t.co/Dp1co1dUFF 2/6

🚨 Happy to share our #Interspeech2025 paper! "WhiStress: Enriching Transcriptions with Sentence Stress Detection" Sentence stress is a word-level prosodic cue that marks contrast or intent. WhiStress detects it alongside transcription—no alignment needed. Paper, code, demo 👇