Update README.md

README.md

@@ -1,39 +1,29 @@
 ---
 license: mit
 language:
-- ar
-- en
+  - ar
+  - en
 library_name: transformers
 tags:
-- arabic
-- text-generation
-- detoxification
-- ensemble
-- bloom
-- nlp
+  - arabic
+  - text-generation
+  - detoxification
+  - ensemble
+  - bloom
 pipeline_tag: text-generation
-base_model:
-- bigscience/bloom-1b7
-datasets:
-- custom
-metrics:
-- accuracy
 model-index:
-- name: arabic-detox-ensemble
-  results:
-  - task:
-      type: text-generation
-      name: Text Detoxification
-    metrics:
-    - type: j-score
-      value: 0.7129
-      name: J-Score
-    - type: accuracy
-      value: 0.95
-      name: Style Transfer Accuracy
-    - type: similarity
-      value: 0.9995
-      name: Reference Similarity
+  - name: arab-detoxification-isp
+    results:
+    - task:
+        type: text-generation
+        name: Text Generation
+      dataset:
+        type: custom
+        name: Arabic Detox Dataset
+      metrics:
+      - type: accuracy
+        value: 0.95
+        name: STA
 ---
 
 <div align="center">
@@ -49,7 +39,7 @@ model-index:
 
 **Transform toxic Arabic text into polite, neutral alternatives while preserving meaning**
 
-[Model Demo](#usage) | [Paper](#methodology) | [Dataset](#dataset) | [Results](#evaluation-results)
+[Model Demo](#-quick-start) | [Architecture](#-architecture-overview) | [Dataset](https://huggingface.co/datasets/ispromashka/arabic-detox-dataset) | [Results](#-evaluation-results)
 
 </div>
 
@@ -218,6 +208,9 @@ Where:
 
 ## 📁 Dataset
 
+Dataset used for training and evaluation:  
+[**ispromashka/arabic-detox-dataset**](https://huggingface.co/datasets/ispromashka/arabic-detox-dataset)
+
 ### Composition
 
 | Category | Examples | Description |
@@ -248,9 +241,9 @@ Where:
 |-----------|-------------|-------------|
 | Hardware | NVIDIA A100 40GB | NVIDIA A100 40GB |
 | Precision | BF16 | BF16 |
-| Batch Size | 8-16 | 8 |
-| Learning Rate | 2e-5 - 3e-5 | 1.5e-5 |
-| Epochs | 20-25 | 15 |
+| Batch Size | 8–16 | 8 |
+| Learning Rate | 2e-5 – 3e-5 | 1.5e-5 |
+| Epochs | 20–25 | 15 |
 | Optimizer | AdamW | AdamW |
 | Scheduler | Cosine | Cosine |
 | Warmup | 10% | 10% |
@@ -267,16 +260,6 @@ Where:
 
 ---
 
-## 🔮 Future Work
-
-- Expand to Arabic dialects (Egyptian, Gulf, Levantine)
-- Add toxicity detection classifier
-- Multi-turn conversation support
-- Larger model variants (3B, 7B)
-- Arabic-English code-switching support
-
----
-
 ## 📖 Citation
 
 ```bibtex
@@ -291,15 +274,6 @@ Where:
 
 ---
 
-## 🙏 Acknowledgments
-
-- [BigScience](https://bigscience.huggingface.co/) for BLOOM models
-- [AUB MIND Lab](https://mind.aub.edu.lb/) for AraGPT2
-- [SBERT](https://www.sbert.net/) for multilingual embeddings
-- [Hugging Face](https://huggingface.co/) for model hosting and Transformers library
-
----
-
 ## 📄 License
 
 This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.