Update README.md

README.md

@@ -56,7 +56,7 @@ library_name: transformers
 [![Tasks](https://img.shields.io/badge/Tasks-NLI%20%7C%20Intent--Detection%20%7C%20Sentiment%20Analysis-orange)](#)
 [![Inference Speed](https://img.shields.io/badge/Blazing%20Fast-Edge%20Devices-green)](#)
 
-Say hello to `NeuroBERT-Mini`, the **game-changing NLP model** that brings **world-class performance** to **low-resource devices**! Fine-tuned from the robust `google-bert/bert-base-uncased`, this **ultra-compact** model weighs in at just **~35MB** with **~11M parameters**, delivering an **outstanding ~95% accuracy** on tasks like masked language modeling, NER, and text classification. Perfect for **IoT devices**, **mobile apps**, **wearables**, and **edge AI systems**, NeuroBERT-Mini is your ticket to **fast, offline, and context-aware** NLP in 2025! 🌟
+Say hello to `NeuroBERT-Mini`, the **game-changing NLP model** that brings **world-class performance** to **low-resource devices**! Fine-tuned from the robust `google-bert/bert-base-uncased`, this **ultra-compact** model weighs in at just **~35MB** with **~10M parameters**, delivering an **outstanding ~95% accuracy** on tasks like masked language modeling, NER, and text classification. Perfect for **IoT devices**, **mobile apps**, **wearables**, and **edge AI systems**, NeuroBERT-Mini is your ticket to **fast, offline, and context-aware** NLP in 2025! 🌟
 
 ---
 
@@ -179,7 +179,7 @@ Input: The capital of France is [MASK].
 - **MNLI (MultiNLI)**: Built for natural language inference.
 - **All-NLI**: Enhanced with extra NLI data for smarter understanding.
 
-*Fine-Tuning Brilliance*: Starting from `google-bert/bert-base-uncased` (12 layers, 768 hidden, 110M parameters), NeuroBERT-Mini was fine-tuned to a streamlined 4 layers, 256 hidden, and ~11M parameters, creating a compact yet powerful NLP solution for edge AI! 🪄
+*Fine-Tuning Brilliance*: Starting from `google-bert/bert-base-uncased` (12 layers, 768 hidden, 110M parameters), NeuroBERT-Mini was fine-tuned to a streamlined 4 layers, 256 hidden, and ~10M parameters, creating a compact yet powerful NLP solution for edge AI! 🪄
 
 ---