Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -171,8 +171,12 @@ The fine-tuned model shows consistent improvements across all metrics:
|
|
| 171 |
- **Training Strategy:** Contrastive learning with hard negative mining
|
| 172 |
- **Epochs:** 1
|
| 173 |
- **Batch Size:** 64
|
| 174 |
-
- **Learning Rate:** 2e-5
|
| 175 |
- **Training Samples:** 671,972 query-document pairs
|
|
|
|
|
|
|
|
|
|
|
|
|
| 176 |
- **Precision:** FP16
|
| 177 |
- **Hardware:** NVIDIA RTX A6000 (49GB VRAM)
|
| 178 |
|
|
|
|
| 171 |
- **Training Strategy:** Contrastive learning with hard negative mining
|
| 172 |
- **Epochs:** 1
|
| 173 |
- **Batch Size:** 64
|
| 174 |
+
- **Learning Rate:** 2e-5 (with 10% warmup)
|
| 175 |
- **Training Samples:** 671,972 query-document pairs
|
| 176 |
+
- **Total Steps:** 10,500
|
| 177 |
+
- **Training Duration:** 4 hours 6 minutes
|
| 178 |
+
- **Throughput:** 45.4 samples/second
|
| 179 |
+
- **Final Loss:** 2.245
|
| 180 |
- **Precision:** FP16
|
| 181 |
- **Hardware:** NVIDIA RTX A6000 (49GB VRAM)
|
| 182 |
|