Finetuning on Kaushik's Portfoliohttps://kauuu.github.io/topics/finetuning/Recent content in Finetuning on Kaushik's PortfolioHugoen-usThu, 25 Dec 2025 00:00:00 +0100Evaluating and Finetuning LLMs for Multilingual Legal Summarisationhttps://kauuu.github.io/projects/llm-legal-summarisation/Thu, 25 Dec 2025 00:00:00 +0100https://kauuu.github.io/projects/llm-legal-summarisation/Adapted Apertus 8B/70B for Swiss legal summarisation, showing full fine-tuning outperforms GPT-4o and Claude 3.5 Sonnet on BERTScore and ROUGE, while highlighting trade-offs between LoRA and full fine-tuning. Click to read more!