-
ICML 2026 Workshop on Weight-Space Symmetries2026Multi-task model merging combines separately trained expert models into a single model that handles all tasks without co-training. Standard practice merges experts at their optimal validation loss. We challenge this convention by systematically studying how training duration of domain experts affects the quality of the merged model. We fine-tune experts on five domains (Math, Code, Instruction Following
-
ACL 2026 Workshop on Sustainable and Efficient Language, Vision, and Action Models (SELVA)2026Optimizing Large Language Models (LLMs) for production AI agent deployment demands substantial computational resources and specialized human expertise (e.g., prompt engineering). Self-evolution offers a promising solution by enabling agents to autonomously enhance capabilities through structured feedback, improving performance without expensive manual optimization. However, most existing self-evolving agents
-
ICML 2026 Workshop on Resource-Adaptive Foundation Model Inference (AdaptFM)2026We establish a formal equivalence between the Quantized Johnson–Lindenstrauss (QJL) transform of the TurboQuant KV cache compression scheme and the classical 1-bit compressive sensing (1-bit CS) model of Boufounos and Baraniuk (2008), which lets us import 1-bit CS theory into QJL analysis. From it we derive three new consequences. First, reconstruction guarantees for QJL side-channel estimates in terms
-
LREC 20262026Large language models (LLMs) have been widely deployed and have achieved remarkable success in downstream tasks. However, their high latency continues to pose challenges for real-time applications that require fast inference, and the need to train and deploy distinct models for different hardware constraints increases both financial and computational costs. To address this, we propose Nested Matrix Learning
-
2026Clinical diagnosis is time-consuming, requiring intensive interactions between patients and medical professionals. While large language models (LLMs) could ease the pre-diagnostic workload, their limited domain knowledge hinders effective medical question generation. We introduce a Knowledge Graph-augmented LLM with active in-context learning to generate relevant and important follow-up questions, KG-Followup
Related content
-
August 24, 2018This year’s Interspeech — the largest conference in speech technology — will take place in Hyderabad, India, the first week of September. More than 40 Amazon researchers will be attending, including Björn Hoffmeister, the senior manager for machine learning in the Alexa Automatic Speech Recognition group. He took a few minutes to answer three questions about this year’s conference. -
August 23, 2018Here’s a fairly common interaction with Alexa: “Alexa, set volume to five”; “Alexa, play music”. Even though the queries come in quick succession, the customer needs to repeat the wake word “Alexa”. To allow for more natural interactions, the device could immediately re-enter its listening state after the first query, without wake-word repetition; but that would require it to detect whether a follow-up speech input is indeed a query intended for the device (“device-directed”) or just background speech (“non-device-directed”).
-
August 19, 2018At the annual meeting of the North American chapter of the Association for Computational Linguistics in June, researchers at Amazon and the University of Sheffield released a new dataset that can be used to train machine-learning systems to determine the veracity of factual assertions online. The dataset is called FEVER, for fact extraction and verification.
-
August 18, 2018"Perfect hashing" is among the techniques that reduce the memory footprints of machine learning models by 94%.
-
August 8, 2018New machine-learned multilingual named-entity transliteration system.
-
July 23, 2018Automatic speech recognition systems, which convert spoken words into text, are an important component of conversational agents such as Alexa. These systems generally comprise an acoustic model, a pronunciation model, and a statistical language model. The role of the statistical language model is to assign a probability to the next word in a sentence, given the previous ones. For instance, the phrases “Pulitzer Prize” and “pullet surprise” may have very similar acoustic profiles, but statistically, one is far more likely to conclude a question that begins “Alexa, what playwright just won a … ?”
