Mistral AI Worldwide Hackathon - London 2026 - Ecotopia

Selected among 7,000+ global applicants for the Mistral AI Worldwide Hackathon in London (Feb 28 - Mar 1, 2026). Organized by Mistral AI and Iterate, sponsored by Weights & Biases, NVIDIA, Amazon Web Services (AWS), ElevenLabs and Hugging Face. Fine-Tuning Track (sponsored by W&B) — 48 hours to fine-tune open-source Mistral models and build a working application.

Selectionne parmi 7 000+ candidatures mondiales pour le Mistral AI Worldwide Hackathon a Londres (28 fev - 1er mars 2026). Organise par Mistral AI et Iterate, sponsorise par Weights & Biases, NVIDIA, Amazon Web Services (AWS), ElevenLabs et Hugging Face. Fine-Tuning Track (sponsorise par W&B) — 48 heures pour fine-tuner des modeles Mistral open-source et construire une application fonctionnelle.

Ecotopia is an interactive political simulation where the player is mayor of a city facing ecological collapse. Free-text speeches are analyzed by specialized fine-tuned models:

• Structured information extraction — Political promise NER, type categorization, contradiction detection
• Conditional text generation — Contextualized citizen reactions based on game state, citizen profiles, and trust history

Ecotopia est une simulation politique interactive ou le joueur incarne un maire face a l'effondrement ecologique. Les discours en texte libre sont analyses par des modeles specialises fine-tunes :

• Extraction d'information supervisee — NER de promesses politiques, categorisation, detection de contradictions
• Generation conditionnelle de texte — Reactions citoyennes contextualisees selon l'etat du jeu, les profils et l'historique de confiance

4 Mistral models fine-tuned via QLoRA (NF4 4-bit, LoRA r=16, alpha=32) on 690 synthetic examples generated via Amazon Bedrock, in under 10 minutes per model:

Task	Models	Training Examples
Promise Extraction	Ministral 8B, Nemo 12B	300 (3 difficulty tiers)
Citizen Reactions	Ministral 8B, Small 24B	390

4 modeles Mistral fine-tunes via QLoRA (NF4 4-bit, LoRA r=16, alpha=32) sur 690 exemples synthetiques generes via Amazon Bedrock, en moins de 10 minutes par modele :

Tache	Modeles	Exemples
Extraction de promesses	Ministral 8B, Nemo 12B	300 (3 niveaux de difficulte)
Reactions citoyennes	Ministral 8B, Small 24B	390

Our 8B fine-tuned SLMs outperform Mistral Large (base) across the entire structured output pipeline at 10x lower latency. Mistral Large scores 0% valid JSON on citizen reactions without fine-tuning. Specializing small models on precise tasks enables real-time applications where latency and output format reliability are hard constraints.

Nos SLMs 8B fine-tunes surpassent Mistral Large (base) sur toute la pipeline structuree avec une latence 10x inferieure. Mistral Large score 0% en JSON valide sur les reactions citoyennes sans fine-tuning. La specialisation de petits modeles sur des taches precises permet des applications temps reel ou la latence et la fiabilite du format de sortie sont des contraintes dures.