GaMS-9B-Instruct-Nemotron Overview

GaMS-9B-Instruct-Nemotron is a 9 billion parameter instruction-tuned language model, building upon the GaMS-9B-Instruct base. Developed by Timotej Petrič as part of a master's thesis, this model underwent supervised fine-tuning (SFT) using a specialized dataset.

Key Capabilities

• Multilingual Instruction Following: The model was fine-tuned on a curated subset of the nvidia/Nemotron-Post-Training-Dataset-v1, which included approximately 80,000 Slovenian instruction-response pairs and 20,000 English examples. This extensive training enables robust performance in both languages.
• Contextual Understanding: The Slovenian training data involved translations with additional modifications to adjust identity and context, enhancing the model's ability to understand and generate contextually relevant responses.

Good For

• Slovenian Language Applications: Excels in tasks requiring natural language understanding and generation in Slovenian, benefiting from its specialized training data.
• Bilingual (Slovenian-English) Use Cases: Suitable for applications that need to process or generate text in both Slovenian and English, leveraging its fine-tuning on a mixed dataset.
• Research and Development: Ideal for researchers and developers exploring instruction-tuned models with a focus on less-resourced languages like Slovenian, within a bilingual context.