Update README.md

README.md

@@ -4,8 +4,21 @@ tags:
 - Audio
 - SSL
 - EAT
+- flexEAT
+- audio embeddings
+- audio features
 library_name: transformers
+language:
+- en
+base_model:
+- worstchan/EAT-base_epoch30_pretrain
+pipeline_tag: audio-classification
 ---
+  
+**⚠️ Codebase Update: Input Flexibility & Fine-Tuning Preparation**  
+
+This repository has been updated to support **dynamic input lengths**. The model now utilizes on-the-fly positional embeddings, removing the restriction on fixed audio durations found in previous versions.
+While the core pre-trained weights and architectural logic remain unchanged, we have introduced new infrastructure to facilitate **downstream fine-tuning** (including dynamic classification heads and normalization layers). Additional fine-tuning configurations and hyperparameters will be documented in future updates.  
 
 # EAT-base (Epoch 30, Pre-trained Checkpoint)