Interview with Loubna Ben Allal
summary
Describes building small LMs for phones that trade accuracy for much lower inference cost, outlines experiments on training and runtime efficiency, and explains choosing model size and languages based on end‑user constraints. Shows deployment work that fits models on laptops and mobile, demonstrates multilingual testing including Moroccan dialect behavior, and mentions using Cursor for coding and AI for slide drafting so viewers see concrete tools, tests, and deployment choices.
AI-written summary · about
related talks
21mBeyond the Chatbox: What Changes When AI Runs Where You Are
42mPost-Training Small Language Models: the adventures of a practitioner
36mPanel Discussion
31mLeveraging LLMs to build supervised datasets suitable for smaller models
30mLearning to wield large language models
27mAI Demystified Unveiling the Intelligence Behind the Machine
36mProduction-Ready AI Agents: From LLMs to Small Language Models
28mNavigating LLM deployment: Tips, tricks, and techniques