This study examines the effects of Fine-Tuning on LLMs, like LegalBERT, for adaptation to the legal text-domain. This work offers a case study on domain adaptation by training the model on specific tasks, including extracting legal entities (laws, sections, and actions) with a legal dataset. The paper intends to evaluate the efficacy of the customized model against generic models like SpaCy (model: en_core_web_lg) utilizing traditional quantitative metrics like Accuracy and F1-score, alongside language model-specific metrics like BLEU and ROUGE, to assess the quality of the extracted texts and their capacity to convey legal meanings effectively. The endeavor aims to achieve two principal objectives: to improve text classification inside legal texts by using the capacity of LLMs to understand context and accurately retrieve legal information. Secondly, create a legal dictionary using the extracted terminology to improve access to legal knowledge. This paper offers a thorough analysis that compares quantitative and qualitative approaches to evaluate the effectiveness of customized models with generic models, highlighting the benefits and challenges associated with the implementation of large language models in specialized legal settings.