Upload Maaza-MLM-135M-JSON-v1 - v1.0.0 production release

README.md

@@ -65,7 +65,7 @@ Evaluated on 158 test cases across 24 schema types:
 | **Field F1** | 0.520 |
 | **Schema Compliance** | 41.1% |
 | **Latency (CPU)** | 18.5 tokens/sec |
-| **Training Time** | 48.7 seconds |
+| **Training Time** | <1 minute |
 
 ### By Complexity Level
 
@@ -89,7 +89,7 @@ Evaluated on 158 test cases across 24 schema types:
 ## Training Data
 
 ### Dataset: EdgeJSON v3
-- **Total Examples**: 787 (100% validated)
+- **Total Examples**: 787 (validated)
 - **Train Split**: 629 examples (80%)
 - **Test Split**: 158 examples (20%)
 - **Validation Rate**: 100% (all examples pass schema validation)
@@ -114,7 +114,7 @@ Output:
 
 ### Hardware
 - **GPU**: NVIDIA RTX 4080 SUPER (16GB)
-- **Training Time**: 48.7 seconds
+- **Training Time**: <1 minute
 - **Effective Batch Size**: 32 (4 per device × 8 gradient accumulation)
 
 ### Hyperparameters
@@ -286,7 +286,7 @@ If you use this model in your research, please cite:
 
 ### v1.0.0 (2025-11-20)
 - Initial release
-- Trained on EdgeJSON v3 dataset (100% validated)
+- Trained on EdgeJSON v3 dataset (validated)
 - 24.7% JSONExact, 0.520 Field F1
 - LoRA fine-tuning (r=16, alpha=32)
 - 48.7 second training time