Taylor Berg-Kirkpatrick
Associate Professor
Builds machine learning models for structured human data such as language, music and historical documents, with work on unsupervised NLP and language models.
Taylor Berg-Kirkpatrick's lab at UCSD works on large language models. 35 papers since 2023, 18 at top venues, mostly ICLR and ICASSP. 9 PhD students and 1 master's student. 23 alumni, most often to Google DeepMind and Adobe. Applications close .
At a glance
- papers, 2023–2026
- 9/ year
- at T1 venues
- 51%
- PhD students
- 9
- MS students
- 1
- alumni listed
- 23
Papers by yearfrom the lab homepage, 2023–2026
- 1 papers2026 · T1 1 · collecting
- 7 papers2025 · T1 4
- 14 papers2024 · T1 8
- 13 papers2023 · T1 5
T1 venuesOtherVenues: ICLR, ICASSP, ICDAR, ICML, AAAI
Research focuspaper titles that use each topic's words
- Large language models12 papers
- music8 papers
- builds machine0 papers
- unsupervised0 papers
- structured0 papers
Paper list
From the lab homepage.
20261 paper
- ICASSP (T1)MuseTok: Symbolic Music Tokenization for Generation and Semantic Understanding
20257 papers
- ICLR (T1)ClimaQA: An Automated Evaluation Framework for Climate Foundation Models
- WASPAAFast Text-to-Audio Generation with Adversarial Post-Training
- ISMIRGenerating symbolic music from natural language prompts using an llm-enhanced dataset
- ICASSP (T1)PDMX: A Large-Scale Public Domain MusicXML Dataset for Symbolic Music Processing
- ICLR (T1)Presto! distilling steps and layers for accelerating music generation
- COLMReadability != Learnability: Rethinking the Role of Simplicity in Training Small Language Models
- ICLR (T1)TEASERGEN: GENERATING TEASERS FOR LONG DOCUMENTARIES
202414 papers
- ICML (T1)Adapting While Learning: Grounding LLMs for Scientific Problems with Intelligent Tool Usage Adaptation
- ML4ALClassification of Paleographic Artifacts at Scale: Mitigating Confounds and Distribution Shift in Cuneiform Tablet Dating
- ICDARClustering Running Titles to Understand the Printing of Early Modern Books
- ISMIRDITTO-2: Distilled Diffusion Inference-Time T-Optimization for Music Generation
- ICML (T1)DITTO: Diffusion Inference-Time T-Optimization for Music Generation Authors
- NeurIPS (T1)Grammar-aligned decoding
- paperImproving Generalization of Speech Separation in Real-World Scenarios: Strategies in Simulation, Optimization, and Evaluation
- ICLR (T1)Is attention required for ICL? Exploring the Relationship Between Model Architecture and In-Context Learning Ability
- ACL (T1)LogogramNLP: Comparing Visual and Textual Representations of Ancient Logographic Writing Systems for NLP
- RecSys (T1)MAWI Rec: Leveraging Severe Weather Data in Recommendation
- Workshop on Computational PaleographyMONSTERMASH: multidirectional, overlapping, nested, spiral text extraction for recognition models of Arabic-script handwriting
- ICASSP (T1)MusicLDM: Enhancing Novelty in Text-to-Music Generation Using Beat-Synchronous Mixup Strategies
202313 papers
- CoNLLA Block Metropolis-Hastings Sampler for Controllable Energy-based Text Generation
- ICLR (T1)CLIPSep: Learning Text-queried Sound Separation with Noisy Unlabeled Videos
- WASPAACLIPSonic: Text-to-Audio Synthesis with Unlabeled Videos and Pretrained Language-Vision Models
- AAAI (T1)Contrastive Attention Networks for Attribution of Early Modern Print
- Journal of Open Humanities DataCuneiML: A Cuneiform Dataset for Machine Learning
- ICDAREEBO-Verse: Sifting for Poetry in Large Early Modern Corpora Using Visual Features
- World Wide Web (WWW), Industry TrackJointly modeling products and resource pages for task-oriented recommendation
- ICASSP (T1)Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation
- Findings of ACLMembership Inference Attacks against Language Models via Neighbourhood Comparison
- ICASSP (T1)Multitrack Music Transformer: Learning Long-Term Dependencies in Music with Diverse Instruments
- EMNLP (T1)Simple Temporal Adaptation to Changing Label Sets: Hashtag Prediction via Dense KNN
- ISMIRTowards Improving Harmonic Sensitivity and Prediction Stability for Singing Melody Extraction
Where alumni went23 listed, 47% to industry
Industry 7Academia 8
Most often to
- Google DeepMind
- Adobe
- Apple
Also to
- Carnegie Mellon University
- Georgia Tech
- Hong Kong University of Science and Technology
Apply to UCSDFall 2027
UCSD CSE Computer Science PhD
- Round
- Single round
- Opens
- —
- GRE
- See program page
- English
- See program page
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- CNPeking
Huishuai Zhang
- Large language models
- CAToronto
Zhijing Jin
- Large language models
- USUW
Noah A. Smith
- Large language models
- USUSC
Robin Jia
- Interpretability, evaluation & safety of LMs
- USUSC
Swabha Swayamdipta
- Interpretability, evaluation & safety of LMs
- CNPeking
Ge Li
- Large language models
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
UCSDcloses