> For the complete documentation index, see [llms.txt](https://trust.memori.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://trust.memori.ai/en/legal-documents/gpai-general-purpose-ai-models/gpai-document-ncas.md). # GPAI Document: NCAs Document for the National Competent Authorities The NCAs (National Competent Authorities) document is intended for the National Competent Authorities of the EU Member States in the exercise of their supervisory functions under the AI Act. It contains the information needed to assess the compliance of high-risk AI systems built on MemoriNLP, in particular when the provider of the system is different from Memori s.r.l. ### What the NCAs document contains * **Provider identity and model authenticity:** company name, URL endpoint, release dates * **Architecture and design specifications:** description of the hybrid fastText + DistilBERT system and rationale for the design choices * **Model size (range):** 500M – 5B parameters * **Training data:** type, origin, number of tokens, scope and characteristics of the corpus * **Anti-bias and inappropriate-content measures:** systemic issues identified and countermeasures * **Computational resources (order of magnitude):** qualitative training time * **Energy consumption and inference benchmarks:** with measurement methodology * **Distribution channels and license:** all delivery models * **Acceptable Use Policy and intended uses** ### How to request the NCAs document The NCAs document is **not public** and is not distributed proactively. It is made available upon request by the AI Office (AIO), made on its own initiative or at the request of the National Competent Authority concerned. Requests must state the legal basis and specific purpose. * **Authorized party:** National Competent Authorities of the EU Member States, via the AIO * **Procedure:** formal request via the AIO, stating the legal basis and supervisory purpose * **Handling:** under Art. 78 of the AI Act (confidentiality and trade secrets) * **Memori contact:** — subject "NCAs Request – MemoriNLP" {% hint style="info" %} Unlike the AIO document, the NCAs document does not include precise computational values (e.g. exact number of tokens, detailed FLOP counts) or the specifications on how the training data was obtained and selected. This information remains confidential to the AI Office. {% endhint %}

Request NCAs