Siddharth Dalmia
@siddalmia05
Followers
1K
Following
473
Media
8
Statuses
245
Audio LLMs @ Waveforms AI | #SpeechProc and #NLProc | Previously Research Scientist @GoogleDeepmind | PhD @LTIatCMU @SCSatCMU
New York
Joined September 2016
Happy new year! Personal updates from 2022,. -- I defended my Ph.D. @LTIatCMU! Thanks to my amazing committee - Florian Metze, Alan Black, @shinjiw_at_cmu, @AbdoMohamedML. -- I am now in MTV working as a Research Scientist @GoogleAI in the intersection of #SpeechProc and #NLProc
18
16
233
RT @lasha_nlp: Life update: I’m excited to share that I’ll be starting as faculty at the Max Planck Institute for Software Systems(@mpi_sws….
0
43
0
Started a new role at @WaveFormsAI, founded by @alex_conneau and Coralie Lemaitre (. I am excited to be working with a fantastic team of AI dreamers building the future of audio LLMs. Ready to give form to the coming wave of audio intelligence. 🔊🌊🧠
6
5
71
RT @lasha_nlp: Want to know what training data has been memorized by models like GPT-4?. We propose information-guided probes, a method to….
0
43
0
Interesting and thoughtful work on tracing LLM hallucinations to pretraining corpora!!!.
We are launching HALoGEN💡, a way to systematically study *when* and *why* LLMs still hallucinate. New work w/ @shrusti_ghela* @davidjwadden @YejinChoinka 💫. 🧵 [1/n]
0
2
6
RT @lasha_nlp: We are launching HALoGEN💡, a way to systematically study *when* and *why* LLMs still hallucinate. New work w/ @shrusti_ghel….
0
41
0
RT @lasha_nlp: ✨I’m on the faculty job market for 2024-2025! ✨ . My research focuses on advancing Responsible AI—enhancing factuality, robu….
0
47
0
RT @WiNLPWorkshop: 🌊 Sail with us at #WiNLP2024! 🌊. Join panel "Sailing the NLP Seas: Navigating Research in the Age of LLMs" on Nov 15, 11….
0
10
0
RT @kiranvodrahalli: Happy to share Michelangelo (, a long-context reasoning benchmark which measures performance b….
arxiv.org
We introduce Michelangelo: a minimal, synthetic, and unleaked long-context reasoning evaluation for large language models which is also easy to automatically score. This evaluation is derived via...
0
57
0
RT @lmsysorg: Exciting News from Chatbot Arena!. @GoogleDeepMind's new Gemini 1.5 Pro (Experimental 0801) has been tested in Arena for the….
0
407
0
RT @SaberaTalukder: 👀👀 poster session + @LiuQunMTtoDeath's talk!. @WilliamWangNLP @yayitsamyzhang +.Language Control Diffusion. @rach_it_ @….
0
4
0
I'm curious to see, if we can use these models help unlock paired speech-text ASR and speech translation data in long-tail languages!. Great effort by everyone in the team! @knarfamlap Ramon, Gustavo, Daniel and @yunhsuansung.
0
0
4
We see that we can match speech and text modalities in languages that have *no* paired speech-text data. Cross-lingual modality matching can also be enhanced with just utilizing paired text translation data.
1
0
5
Excited to see one of the things we worked on last year @GoogleDeepMind @GoogleAI finally out!. We show that we can convert a text based LLMs into Multilingual Multimodal Retrieval Models with minimal paired speech-text data.
Our new paper introduces our approach that matches speech and text across 102 languages by converting LLMs into multi-modal retrieval systems! 🔊.
1
6
43
RT @ShikharSSU: Is that speech English, Spanish, or Mandarin? 🕵️♂️. 🔊Super excited to share several works that improve language identifica….
0
11
0
RT @samridhishree: Super pumped to show what my incredible team in @amazon has been working on with @BMW at #CES2024 ! We leverage our in-….
0
2
0
RT @omarsar0: LLM Augmented LLMs. Yeah, you heard that right! This work explores composing existing foundation models with specific models….
0
113
0
RT @shinjiw_at_cmu: Happy New Year! Last year, our group published over 70 papers (ICASSPx29, Interspeechx17, ASRUx12, etc.)! I'm very happ….
0
11
0
RT @acl_sigel: ✨ComputEL-7 will be co-located with EACL 2024!✨ We are excited to see you in Malta!. 📢📢The submission deadline is December 1….
0
9
0
RT @shinjiw_at_cmu: Dear friends, please spread this tweet! If you're interested, please do not hesitate to contact me or David. We're look….
0
18
0