Edoardo Maria Ponti
Associate Professor
Leads projects on adaptive tokenization and memory in foundation models, studying efficient, long-context and multilingual large language models.
Edoardo Maria Ponti's lab at Imperial works on large language models, multilingual, translation & linguistics and generative & foundation models. 34 papers since 2023, 22 at top venues, mostly NeurIPS and NAACL. 10 PhD students. 3 alumni. Applications close .
At a glance
- papers, 2023–2026
- 9/ year
- at T1 venues
- 65%
- PhD students
- 10
- MS students
- 0
- alumni listed
- 3
Papers by yearfrom DBLP, 2023–2026
- 4 papers2026 · T1 0 · collecting
- 11 papers2025 · T1 10
- 10 papers2024 · T1 8
- 9 papers2023 · T1 4
T1 venuesOtherVenues: NeurIPS, NAACL, Findings of ACL, ACL, EACL
Research focuspaper titles that use each topic's words
- Large language models10 papers
- Generative & foundation models4 papers
- Multilingual, translation & linguistics2 papers
- multilingual2 papers
- long-context0 papers
Paper list
From DBLP.
20264 papers
- Findings of ACLCan VLMs Predict Future States? Bootstrapping World Models from Inverse Dynamics (opens in a new tab)
- EACLIs Information Density Uniform when Utterances are Grounded on Perception and Discourse? (opens in a new tab)
- CoNLLMeasuring the Effects of Visual Salience in Human and AI Descriptions with Image Editing (opens in a new tab)
- Findings of ACLThe Sparse Frontier: Sparse Attention Trade-offs in Transformer LLMs (opens in a new tab)
202511 papers
- NAACL (T1)A Grounded Typology of Word Classes (opens in a new tab)
- NAACL (T1)Cross-Lingual and Cross-Cultural Variation in Image Descriptions (opens in a new tab)
- NAACL (T1)Fine-Tuning Large Language Models with Sequential Instructions (opens in a new tab)
- NeurIPS (T1)Inference-Time Hyper-Scaling with KV Cache Compression (opens in a new tab)
- ACL (T1)Mixtures of In-Context Learners (opens in a new tab)
- NeurIPS (T1)MoE-CAP: Benchmarking Cost, Accuracy and Performance of Sparse Mixture-of-Experts Systems (opens in a new tab)
- NeurIPS (T1)Neurosymbolic Diffusion Models (opens in a new tab)
- NeSyNeurosymbolic Reasoning Shortcuts under the Independence Assumption (opens in a new tab)
- ICLR (T1)Post-hoc Reward Calibration: A Case Study on Length Bias (opens in a new tab)
- ACL (T1)The Cross-linguistic Role of Animacy in Grammar Structures (opens in a new tab)
- NeurIPS (T1)Universal Cross-Tokenizer Distillation via Approximate Likelihood Matching (opens in a new tab)
202410 papers
- NAACL (T1)Are Large Language Model Temporally Grounded? (opens in a new tab)
- J. Lang. Model.Corpus-based measures discriminate inflection and derivation cross-linguistically (opens in a new tab)
- ICML (T1)Dynamic Memory Compression: Retrofitting LLMs for Accelerated Inference (opens in a new tab)
- NAACL (T1)Elastic Weight Removal for Faithful and Abstractive Dialogue Generation (opens in a new tab)
- ICLR (T1)Model Merging by Uncertainty-Based Gradient Matching (opens in a new tab)
- ICML (T1)On the Independence Assumption in Neurosymbolic Learning (opens in a new tab)
- Findings of ACLProbing the Emergence of Cross-lingual Alignment during LLM Training (opens in a new tab)
- NeurIPS (T1)Spectral Editing of Activations for Large Language Model Alignment (opens in a new tab)
- ICML (T1)Towards Modular LLMs by Building and Reusing a Library of LoRAs (opens in a new tab)
- NeurIPS (T1)Zero-Shot Tokenizer Transfer (opens in a new tab)
20239 papers
- EACLCombining Parameter-efficient Modules for Task-level Generalisation (opens in a new tab)
- Trans. Assoc. Comput. LinguisticsCross-Lingual Dialogue Dataset Creation via Outline-Based Generation (opens in a new tab)
- EMNLP (T1)Detecting and Mitigating Hallucinations in Multilingual Summarisation (opens in a new tab)
- Findings of ACLDistilling Efficient Language-Specific Models for Cross-Lingual Transfer (opens in a new tab)
- ACL (T1)Efficient Transformers with Dynamic Token Pooling (opens in a new tab)
- Trans. Mach. Learn. Res.Modular Deep Learning (opens in a new tab)
- NeurIPS (T1)Multi-Head Adapter Routing for Cross-Task Generalization (opens in a new tab)
- EACLProbing Cross-Lingual Lexical Knowledge from Multilingual Sentence Encoders (opens in a new tab)
- EMNLP (T1)Unifying Cross-Lingual Transfer across Scenarios of Resource Scarcity (opens in a new tab)
Where alumni went3 listed
Apply to Imperial2027 (Apr/Oct)
Computing PhD
- Round
- Round 2 (1 Dec 2026)
- Opens
- —
- GRE
- See program page
- English
- See program page
- Fee
- none
Confirmed on the program page.
Similar labsnearest by research summary
- CNTsinghua
Ning Ding
- Large language models
- USUCSD
Jingbo Shang
- Knowledge, QA & retrieval
- USUSC
Jieyu Zhao
- Interpretability, evaluation & safety of LMs
- USUCLA
Kai-Wei Chang
- Large language models
- USUC Berkeley
Dan Klein
- Multilingual, translation & linguistics
- USMIT
Paul Liang
21 T1
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- DBLP
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
Imperialcloses