Adaptation of a Large Language Model for Sentiment and Emotion Classification using Parameter Efficient Fine-Tuning
Overview
This pipeline provides code and instructions on how to efficiently adapt a large language model for sentiment and emotion classification in user input. It is suitable for many domains such as customer feedback analysis, market research, and public opinion mining. The prerequisites are a modest amount of task-based data, i.e., sentiment- or emotion-annotated texts. The pipeline first helps users in preparing data for fine-tuning, and then provides the know-how and code for fine-tuning the model with the prepared data. Finally, it provides code for efficient inference using the fine-tuned model. A worked example of fine-tuning is provided. The pipeline uses Slovene large language models (LLMs) such as GaMS, but is adaptable to other state-of-the-art open multilingual LLMs, libraries that support LLM fine-tuning with LoRA and QLoRA, and vLLM for efficient inference.
Outcome
Documentation and code for domain-specific fine-tuning of models for sentiment and emotion classification.
Service Card
| Category | AI development and deployment |
| Domain | Natural Language Processing, Sentiment Analysis |
| Vertical | Cross-domain |
| Target users | SMEs, Enterprises, Public Sector, Scientific Users |
| Service type | Pipeline |
| Outputs | Pipeline for adaptation of models for sentiment and emotion classification, including data preparation scripts, fine-tuning code, inference code, and a worked example. |
| Requirements | Requires NVIDIA GPU node. |
| Limitations | |
| Availability | Q4/2026 |
| Licensing | MIT |
| Visual content | |
| Visual caption |