| TENDRIL most-embed-de | |
| Copyright 2026 Institute for Public Sector Transformation, Bern University of Applied Sciences | |
| This model is an adapted version of malteos/most-embed-de | |
| (https://huggingface.co/malteos/most-embed-de). The fine-tuning contribution of most-embed-de is | |
| licensed under the Creative Commons Attribution-NonCommercial 4.0 International License | |
| (CC-BY-NC-4.0, https://creativecommons.org/licenses/by-nc/4.0/). | |
| most-embed-de is built on nvidia/Nemotron-3-Embed-1B-BF16 | |
| (https://huggingface.co/nvidia/Nemotron-3-Embed-1B-BF16), licensed under the OpenMDW License | |
| Agreement v1.1 (https://openmdw.ai/license/1-1/), which in turn was derived from | |
| mistralai/Ministral-3-3B-Instruct-2512, licensed under the Apache License, Version 2.0. | |
| The notices of origin of these models are retained. | |
| Modifications: all weights were further fine-tuned for German passage retrieval in Swiss public | |
| procurement tenders, using the LLM-judged TENDRIL-Train data. The tokenizer, architecture, pooling | |
| and prompts of most-embed-de are unchanged. | |
| The adapted model is distributed under CC-BY-NC-4.0 (see LICENSE). It may not be used for | |
| commercial purposes. | |