Colin Raffel
Associate Professor
Studies large language models, collaborative and decentralized machine learning, and learning from limited data.
Colin Raffel's lab at Toronto works on large language models and human-AI collaboration & trust. 38 papers since 2023, 30 at top venues, mostly NeurIPS and ICML. 25 alumni. Applications close .
At a glance
- papers, 2023–2026
- 10/ year
- at T1 venues
- 79%
- PhD students
- 0
- MS students
- 0
- alumni listed
- 25
Papers by yearfrom the lab homepage, 2023–2026
- 0 papers2026 · T1 0 · collecting
- 10 papers2025 · T1 8
- 9 papers2024 · T1 7
- 19 papers2023 · T1 15
T1 venuesOtherVenues: NeurIPS, ICML, TMLR, ACL, EMNLP
Research focuspaper titles that use each topic's words
- Large language models21 papers
- Human-AI collaboration & trust0 papers
- limited0 papers
- decentralized0 papers
- studies large0 papers
Paper list
From the lab homepage.
202510 papers
- TMLR (T1)A Survey on Model MoErging: Recycling and Routing Among Specialized Experts for Collaborative Learning (opens in a new tab)
- ICLR (T1)AttriBoT: A Bag of Tricks for Efficiently Approximating Leave-One-Out Context Attribution (opens in a new tab)
- TMLR (T1)ComPEFT: Compression for Communicating Parameter Efficient Updates via Sparsification and Quantization (opens in a new tab)
- NeurIPS (T1)Enhancing Training Data Attribution with Representational Optimization (opens in a new tab)
- COLMFineWeb2: One Pipeline to Scale Them All -- Adapting Pre-Training Data Processing to Every Language (opens in a new tab)
- ICML (T1)Fishers for Free? Approximating the Fisher Information Matrix by Recycling the Squared Gradient Accumulator (opens in a new tab)
- COLMSmolLM2: When Smol Goes Big -- Data-Centric Training of a Small Language Model (opens in a new tab)
- ICML (T1)The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions (opens in a new tab)
- NeurIPS (T1)The Common Pile v0.1: An 8TB Dataset of Public Domain and Openly Licensed Text (opens in a new tab)
- ICML (T1)The Most Expensive Part of an LLM should be its Training Data (opens in a new tab)
20249 papers
- ICLR Blog Post TrackA New Alchemy: Language Model Development as a Subfield? (opens in a new tab)
- TMLR (T1)A Survey on Data Selection for Language Models (opens in a new tab)
- JMLR (T1)BLOOM: A 176B-Parameter Open-Access Multilingual Language Model (opens in a new tab)
- CACMCombining Machine Learning and Lifetime-based Resource Management for Memory Allocation and Beyond (opens in a new tab)
- ACL (T1)DataDreamer: A Tool for Synthetic Data Generation and Reproducible LLM Workflows (opens in a new tab)
- ICML (T1)Learning to Route Among Specialized Experts for Zero-Shot Generalization (opens in a new tab)
- TMLR (T1)Merging by Matching Models in Task Subspaces (opens in a new tab)
- TMLR (T1)Soft Merging of Experts with Adaptive Routing (opens in a new tab)
- NeurIPS (T1)The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scale (opens in a new tab)
202319 papers
- TMLR (T1)Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models (opens in a new tab)
- ICLR (T1)Bidirectional Language Models Are Also Few-shot Learners (opens in a new tab)
- CACMBuilding Machine Learning Models like Open-Source Software (opens in a new tab)
- ACL (T1)ColD Fusion: Collaborative Descent for Distributed Multitask Finetuning (opens in a new tab)
- NeurIPS Generative AI and Biology (GenBio) WorkshopConditional Generation of Antigen Specific T-cell Receptor Sequences (opens in a new tab)
- ACL (T1)Crosslingual Generalization through Multitask Finetuning (opens in a new tab)
- NeurIPS (T1)Distributed Inference and Fine-tuning of Large Language Models Over The Internet (opens in a new tab)
- TACL (T1)Efficient Methods for Natural Language Processing: A Survey (opens in a new tab)
- NeurIPS WorkshopEfficient Online Data Mixing For Language Model Pre-Training (opens in a new tab)
- ACL (T1)Evaluating the Factual Consistency of Large Language Models Through Summarization (opens in a new tab)
- ICML (T1)Git-Theta: A Git Extension for Collaborative Development of Machine Learning Models (opens in a new tab)
- NeurIPS (T1)Improving Few-Shot Generalization by Exploring and Exploiting Auxiliary Data (opens in a new tab)
Where alumni went25 listed
Apply to TorontoFall 2027
Computer Science MSc/PhD
- Round
- Single round (Dec 1)
- Opens
- —
- GRE
- optional
- English
- TOEFL/IELTS if applicable
- Fee
- CAD 135
Confirmed on the program page.
Similar labsnearest by research summary
- CNZhejiang
Jiwei Li
- Large language models
- CNTsinghua
Hua Xu
- Multimodal & grounded language
- CNZhejiang
Xin Du
- Distributed, cloud & edge systems
- CAToronto
Yang Xu
- Multilingual, translation & linguistics
- USNYU
Tal Linzen
- Multilingual, translation & linguistics
- USUMD
Douglas Oard
- Knowledge, QA & retrieval
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.
Torontocloses