--- title: "Martin Jaggi — author profile and papers" description: "Martin Jaggi, Electric Propulsion Laboratory (United States): 3 papers on Elamur Research (2018–2024), h-index 39 — affiliations, topics, co-authors and…" canonical: "https://research.elamur.ai/free/en/authors/martin-jaggi/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/authors/martin-jaggi/) [home](https://md.research.elamur.ai/free/en/) / [authors](https://md.research.elamur.ai/free/en/authors/) / Martin Jaggi # Martin Jaggi Martin Jaggi's papers here concern machine learning for natural language, with attention both to how text is represented and to how large models can be run more cheaply. The two works on this site span 2018 to 2024. They share a practical concern: simple, scalable methods that avoid labelled data or heavy calibration while keeping model quality intact. The approaches include training objectives borrowed from word embeddings and linear-algebra transformations applied to the weights of trained networks. The most cited of these papers is [Unsupervised Learning of Sentence Embeddings Using Compositional n-Gram Features](https://md.research.elamur.ai/free/en/papers/pagliardini-2018-unsupervised-learning-sentence-embeddings-compositional/) (2018), which introduced Sent2Vec. It widens the C-BOW objective from a small local window to an entire sentence, so that a sentence vector becomes the average of learned embeddings for its words and n-grams. Models were trained on book text, Wikipedia sentences and tweets. The second paper, [QuaRot](https://md.research.elamur.ai/free/en/papers/ashkboos-2024-quarot-outlier-free-4-bit/) (2024), tackles 4-bit quantization of large language models. Outlier channels in the activations make low precision costly. Instead of keeping those channels at higher precision, QuaRot multiplies the weights by random orthogonal Hadamard matrices, mostly fused into the weights themselves. This leaves the model's output unchanged while making the activations easier to quantize. The newer of the two, QuaRot, moves the work from representation learning towards efficient inference for large models, within the topic of [Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/). Jaggi is based at the [École Polytechnique Fédérale de Lausanne](https://md.research.elamur.ai/free/en/universities/ecole-polytechnique-federale-de-lausanne/), with earlier bylines at [ETH Zurich](https://md.research.elamur.ai/free/en/universities/eth-zurich/) and the [École Polytechnique](https://md.research.elamur.ai/free/en/universities/ecole-polytechnique/) in Palaiseau. Co-authors on the papers here include [Matteo Pagliardini](https://md.research.elamur.ai/free/en/authors/matteo-pagliardini/), [Prakhar Gupta](https://md.research.elamur.ai/free/en/authors/prakhar-gupta/), [Amirkeivan Mohtashami](https://md.research.elamur.ai/free/en/authors/amirkeivan-mohtashami/) and [Bo Li](https://md.research.elamur.ai/free/en/authors/bo-li/). 3 papers on this site · 2018–2024 current affiliation [Electric Propulsion Laboratory (United States)](https://ror.org/01eb6s760), Monument, United States; [EPF - École d'ingénieurs](https://ror.org/03dg0f179), Sceaux, France; [École Polytechnique Fédérale de Lausanne](https://md.research.elamur.ai/free/en/universities/ecole-polytechnique-federale-de-lausanne/), Lausanne, Switzerland Martin Jaggi has 3 papers on this site, in 2018–2024, most of them on [Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/); often with [Amirkeivan Mohtashami](https://md.research.elamur.ai/free/en/authors/amirkeivan-mohtashami/), [Bo Li](https://md.research.elamur.ai/free/en/authors/bo-li/), [Elie Bakouch](https://md.research.elamur.ai/free/en/authors/elie-bakouch/). [OpenAlex](https://openalex.org/A5073756389) · [ORCID](https://orcid.org/0000-0003-1579-5558) · [**h-index 39**](https://openalex.org/authors/A5073756389) · [228 works](https://openalex.org/works?filter=author.id:A5073756389) · [14 084 citations](https://openalex.org/works?filter=author.id:A5073756389&sort=cited_by_count:desc) OpenAlex, 2026-10-14 [3 works on this site](https://md.research.elamur.ai/free/en/authors/martin-jaggi/#papers) · [813 citations in our library](https://md.research.elamur.ai/free/en/authors/martin-jaggi/cited-by/) · 5 more works in our library · [⭐ Paper Star rank #1832](https://md.research.elamur.ai/free/en/paper-stars/page/19/#star-1832) The profile is the library's match of the name; two people with one name are not always told apart. Elsewhere: [Google Scholar](https://scholar.google.com/citations?view_op=search_authors&mauthors=Martin+Jaggi) · [Semantic Scholar](https://www.semanticscholar.org/search?q=Martin+Jaggi) · [dblp](https://dblp.org/search?q=Martin+Jaggi) · [ACL Anthology](https://aclanthology.org/search/?q=Martin+Jaggi) ## Affiliations - [École Polytechnique Fédérale de Lausanne](https://md.research.elamur.ai/free/en/universities/ecole-polytechnique-federale-de-lausanne/) · Lausanne, Switzerland —current · 2017–2026 on 152 papers · employment (Wikidata) - [EPF - École d'ingénieurs](https://ror.org/03dg0f179) · Sceaux, France —current · 2021–2026 on 2 papers - [Electric Propulsion Laboratory (United States)](https://ror.org/01eb6s760) · Monument, United States —current · 2021–2026 on 2 papers - [Helmholtz Center for Information Security](https://md.research.elamur.ai/free/en/labs/helmholtz-center-for-information-security/) · Saarbrücken, Germany —2021–2022 on 2 papers - [ETH Zurich](https://md.research.elamur.ai/free/en/universities/eth-zurich/) · Zurich, Switzerland —2009–2018 on 17 papers · education (Wikidata) - [École Polytechnique](https://md.research.elamur.ai/free/en/universities/ecole-polytechnique/) · Palaiseau, France —2009–2018 on 10 papers - [Centre de Mathématiques Appliquées de l'École polytechnique](https://ror.org/012e1xn46) · Palaiseau, France —2012–2018 on 3 papers From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 12 more institutions on single bylines are not listed. ## Topics [Language models & pretraining 1](https://md.research.elamur.ai/free/en/topics/language-models/) ## Works by year ### 2020s [**2024** 2 papers](https://md.research.elamur.ai/free/en/authors/martin-jaggi/year/2024/) ### 2010s [**2018** 1 paper](https://md.research.elamur.ai/free/en/papers/pagliardini-2018-unsupervised-learning-sentence-embeddings-compositional/) ## Papers The most cited first (OpenAlex and our library). 1. **[Unsupervised Learning of Sentence Embeddings Using Compositional n-Gram Features](https://md.research.elamur.ai/free/en/papers/pagliardini-2018-unsupervised-learning-sentence-embeddings-compositional/)** *(2018 · 840 citations)* The paper asks whether the simple, scalable recipe behind word embeddings can also produce strong general-purpose vectors for whole sentences without labeled data. 2. **[QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMs](https://md.research.elamur.ai/free/en/papers/ashkboos-2024-quarot-outlier-free-4-bit/)** *(2024 · 252 citations)* Quantizing large language models to 4 bits is hard because activations contain outliers: individual channels whose values are many times larger than the rest. 3. **[Scaling Laws and Compute-Optimal Training Beyond Fixed Training Durations](https://md.research.elamur.ai/free/en/papers/hagele-2024-scaling-laws-compute-optimal-training/)** *(2024 · 24 citations)* The paper questions the habit of training large language models with a cosine learning rate schedule. This schedule performs best solely when its cycle is set to equal the full training duration. As a result, the model has to be trained from scratch for every token budget, which makes scaling laws studies expensive.… ## Also in our library 5 more works of theirs are in our library without a page here yet; they will appear as we publish them. ## Co-authors here 7 co-authors with a page, the most papers together first. - [Amirkeivan Mohtashami](https://md.research.elamur.ai/free/en/authors/amirkeivan-mohtashami/) - [Bo Li](https://md.research.elamur.ai/free/en/authors/bo-li/) - [Elie Bakouch](https://md.research.elamur.ai/free/en/authors/elie-bakouch/) - [Leandro Von Werra](https://md.research.elamur.ai/free/en/authors/leandro-von-werra/) - [Loubna Ben Allal](https://md.research.elamur.ai/free/en/authors/loubna-ben-allal/) - [Matteo Pagliardini](https://md.research.elamur.ai/free/en/authors/matteo-pagliardini/) - [Prakhar Gupta](https://md.research.elamur.ai/free/en/authors/prakhar-gupta/) ## Cited by Papers of our library cite their works 813 times (of the 140 884 papers whose bibliographies we read); [47 citing papers with a page here →](https://md.research.elamur.ai/free/en/authors/martin-jaggi/cited-by/).