mrshu/mistral-sk-7b-alpaca-slovak-it
mrshu/mistral-sk-7b-alpaca-slovak-it is a 7 billion parameter Mistral-based causal language model, instruction-tuned by mrshu for Slovak language tasks. It is derived from slovak-nlp/mistral-sk-7b and fine-tuned using LoRA on a Slovak Alpaca-style dataset. This model is optimized for Slovak instruction following, question answering, drafting, and general assistant workflows, offering a compact solution for lower memory environments.
Loading preview...
Model Overview
mrshu/mistral-sk-7b-alpaca-slovak-it is a 7 billion parameter instruction-tuned model based on the Mistral architecture, specifically derived from slovak-nlp/mistral-sk-7b. It was fine-tuned using a LoRA supervised fine-tuning approach on the saillab/alpaca-slovak-cleaned dataset, which consists of Slovak Alpaca-style instruction data. The LoRA adapter was subsequently merged into the base model, resulting in a full-weight checkpoint that can be loaded directly.
Key Capabilities
- Slovak Instruction Following: Excels at understanding and responding to instructions in Slovak.
- Slovak Question Answering: Designed for answering questions in the Slovak language.
- Text Generation: Capable of drafting, rewriting, and summarization-style prompts in Slovak.
- General Assistant Workflows: Suitable for various assistant tasks where Slovak is the primary language.
- Compact Size: A 7B parameter model intended for experiments where lower memory usage is prioritized.
Intended Use Cases
This model is primarily intended for applications requiring strong Slovak language capabilities. While it can handle English prompts and translation requests, its language control for non-Slovak tasks is noted as weaker compared to larger models. Users should implement additional application-level checks for strict formatting, policy compliance, or high-reliability translation. It is not safety-aligned or moderated for high-stakes decisions.