Update README.md

README.md

@@ -1,5 +1,43 @@
----
-license: other
-license_name: link-attribution
-license_link: https://dejanmarketing.com/link-attribution/
----
+---
+license: other
+license_name: link-attribution
+license_link: https://dejanmarketing.com/link-attribution/
+---
+
+# Model Card: dejanseo/ecommerce-taxonomy-classifier
+
+## Model Description
+**dejanseo/ecommerce-taxonomy-classifier** is a multi-level text classification model designed to categorize ecommerce product descriptions (or similar text) into a hierarchical taxonomy. It uses a pretrained ALBERT (albert-base-v2) backbone and a custom classification head that leverages parent-level one-hot encodings for deeper levels in the taxonomy.
+
+### Model Architecture
+- **Base Model**: [ALBERT (albert-base-v2)](https://huggingface.co/albert-base-v2)
+- **Classification Head**: A linear layer (or multi-layer head) on top of the ALBERT pooled output, concatenated with a parent-level one-hot vector representing the higher-level class.
+
+### Intended Use
+- **Primary Application**: Categorizing product descriptions in online retail or marketplace scenarios.
+- **Potential Use Cases**:
+  - E-commerce product listing
+  - Product categorization for inventory management
+  - Enriching product feeds for better search/discovery
+
+### How to Use
+1. **Installation**: Install `transformers`, `torch`, etc.  
+2. **Pipeline Example**:
+   ```python
+   from transformers import AlbertTokenizer, AlbertForSequenceClassification
+   import torch
+   
+   # Load tokenizer and model from the Hugging Face Hub
+   tokenizer = AlbertTokenizer.from_pretrained("dejanseo/ecommerce-taxonomy-classifier")
+   model = AlbertForSequenceClassification.from_pretrained("dejanseo/ecommerce-taxonomy-classifier")
+   model.eval()
+
+   text = "Experience the magic of music with the Clavinova CLP-800 series digital pianos."
+   inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True)
+
+   with torch.no_grad():
+       outputs = model(**inputs)
+       logits = outputs.logits
+       predicted_class = torch.argmax(logits, dim=-1).item()
+
+   print("Predicted Class:", predicted_class)