🏗️ Building on HF

Madalin Tatarciuc

cetusian

·

AI & ML interests

instinctively gravitating toward the frontier where intelligence meets constraints.

Recent Activity

liked a Space about 17 hours ago

multimodalart/jev-decision-index

liked a Space about 22 hours ago

lerobot/visualize_dataset

posted an update about 24 hours ago

Surogate Speech is out 🐇 small open speech models for agents, the speech side of Surogate. First language shipped is Romanian, and the numbers came out better than I expected: 5.69% WER on FLEURS with a 116M model. Canary 1B gets 5.95% on the same clips, Whisper large-v3 8.42%. Leaderboard runner, one RTX 5090. What's in it: 🎧 jackrabbit-110m-ro: speech recognition, 2,500× real time on one GPU ⚡ jackrabbit-110m-ro-streaming: live, final text about 0.7 s after you stop talking 🗣️ amami-357m-ro: TTS with three voices, runs on a CPU Serving is one line with our engine: surogate serve --stt surogate/jackrabbit-110m-ro Transcripts for every clip are public if you want to rescore it. https://huggingface.co/collections/surogate/surogate-speech-6ab680eb84c7ff75fb73ad5a https://github.com/invergent-ai/surogate-speech

View all activity

Organizations

upvoted an article 17 days ago

Article

Agentic models, measured on the injections that move money

mihailgribov

•

21 days ago

• 2

upvoted an article 22 days ago

Article

Measuring benchmark optimization in speech recognition

+5

tlebryk02, bezzam, aliceebaird, dayllon, jpc, jens-hume-ai, tzirakis

•

Aug 21

• 67

upvoted 5 collections 23 days ago

Nemotron Supervised Fine-Tuning

SFT datasets covering math, code, chat, safety, agentic, VLM, multilingual, and specialized domains. • 44 items • Updated Aug 11 • 22

Nemotron-Personas

A collection of multilingual, region-specific synthetic persona datasets that support sovereign AI development across many countries and regions. • 10 items • Updated Aug 11 • 76

RADIO

A collection of Foundation Vision Models that combine multiple models (CLIP, DINOv2, SAM, etc.). • 19 items • Updated Aug 11 • 39

Parakeet ASR

NeMo Parakeet ASR Models attain strong speech recognition accuracy while being efficient for inference. Available in CTC and RNN-Transducer variants. • 16 items • Updated Aug 11 • 84

Nemotron Speech

Open, state-of-the-art, production‑ready enterprise speech models from the NVIDIA Speech research team for ASR, TTS, Speaker Diarization and S2S • 15 items • Updated 10 days ago • 71

upvoted an article 23 days ago

Article

How to Fine-Tune Nemotron 3.5 ASR for Your Language, Domain, or Accent

nvidia

•

Jun 4

• 79

upvoted 4 papers 23 days ago

Code as Worlds: Agentic Discovery of Executable World Representations for Physical Reasoning

Paper • 2608.27549 • Published about 1 month ago • 49

Agentic Artifact Creation: Systems, Evaluation, Principles, and Opportunities

Paper • 2608.28122 • Published 30 days ago • 66

Qwen-Drive-1.0: An Initial Step towards a Vision-Language Foundation Model for Autonomous Driving

Paper • 2609.00111 • Published 27 days ago • 314

Repo-To-Skill: Distilling GitHub Repositories Into AI4AI Skills

Paper • 2609.02749 • Published 25 days ago • 402

upvoted a paper 26 days ago

Open Source State-Of-the-Art Solution for Romanian Speech Recognition

Paper • 2511.03361 • Published Nov 5, 2025 • 1

upvoted 2 collections about 2 months ago

MOSS-TTS

14 items • Updated Jul 11 • 44

Draft-OPD

9 items • Updated 23 days ago • 3

upvoted an article about 2 months ago

Article

Build Low-Latency Multilingual Voice Agents: Open Weights & Full Deployment Control with NVIDIA Magpie TTS

nvidia

•

Aug 10

• 38

upvoted a collection about 2 months ago

Dream-Cubed

Collection for Dream-Cubed: Controllable Generative Modeling in Minecraft by Training on Billions of Cubes • 4 items • Updated May 7 • 4

upvoted a collection 3 months ago

RO-EN Surogate Models 🇷🇴🇺🇸

Romanian-first bilingual (RO/EN), dual-mode reasoner models • 3 items • Updated 1 day ago • 5

upvoted 2 papers 3 months ago

LFM2 Technical Report

Paper • 2511.23404 • Published Nov 28, 2025 • 73

VectraYX-Nano: A 42M-Parameter Spanish Cybersecurity Language Model with Curriculum Learning and Native Tool Use

Paper • 2605.13989 • Published May 13 • 1