metrics_comparison.txt

Model Performance Metrics Comparison
====================================

Baseline Performance
--------------------
Metric       │  768  │  512  │  256  │  128  │   64 
─────────────┼───────┼───────┼───────┼───────┼──────
ndcg@10      │ 0.548 │ 0.537 │ 0.513 │ 0.469 │ 0.379
mrr@10       │ 0.503 │ 0.490 │ 0.469 │ 0.423 │ 0.337
map@100      │ 0.512 │ 0.499 │ 0.478 │ 0.431 │ 0.347
accuracy@1   │ 0.416 │ 0.403 │ 0.384 │ 0.336 │ 0.262
accuracy@3   │ 0.561 │ 0.551 │ 0.517 │ 0.471 │ 0.384
accuracy@5   │ 0.618 │ 0.608 │ 0.576 │ 0.532 │ 0.432
accuracy@10  │ 0.690 │ 0.686 │ 0.654 │ 0.616 │ 0.515
precision@1  │ 0.416 │ 0.403 │ 0.384 │ 0.336 │ 0.262
precision@3  │ 0.187 │ 0.184 │ 0.172 │ 0.157 │ 0.128
precision@5  │ 0.124 │ 0.122 │ 0.115 │ 0.106 │ 0.086
precision@10 │ 0.069 │ 0.069 │ 0.065 │ 0.062 │ 0.051
recall@1     │ 0.416 │ 0.403 │ 0.384 │ 0.336 │ 0.262
recall@3     │ 0.561 │ 0.551 │ 0.517 │ 0.471 │ 0.384
recall@5     │ 0.618 │ 0.608 │ 0.576 │ 0.532 │ 0.432
recall@10    │ 0.690 │ 0.686 │ 0.654 │ 0.616 │ 0.515

Fine-Tuned Performance
----------------------
Metric       │  768  │  512  │  256  │  128  │   64 
─────────────┼───────┼───────┼───────┼───────┼──────
ndcg@10      │ 0.732 │ 0.730 │ 0.707 │ 0.657 │ 0.548
mrr@10       │ 0.682 │ 0.681 │ 0.658 │ 0.605 │ 0.494
map@100      │ 0.686 │ 0.685 │ 0.663 │ 0.610 │ 0.502
accuracy@1   │ 0.576 │ 0.577 │ 0.555 │ 0.499 │ 0.397
accuracy@3   │ 0.760 │ 0.760 │ 0.735 │ 0.673 │ 0.552
accuracy@5   │ 0.820 │ 0.820 │ 0.795 │ 0.737 │ 0.626
accuracy@10  │ 0.887 │ 0.882 │ 0.860 │ 0.823 │ 0.721
precision@1  │ 0.576 │ 0.577 │ 0.555 │ 0.499 │ 0.397
precision@3  │ 0.253 │ 0.253 │ 0.245 │ 0.224 │ 0.184
precision@5  │ 0.164 │ 0.164 │ 0.159 │ 0.147 │ 0.125
precision@10 │ 0.089 │ 0.088 │ 0.086 │ 0.082 │ 0.072
recall@1     │ 0.576 │ 0.577 │ 0.555 │ 0.499 │ 0.397
recall@3     │ 0.760 │ 0.760 │ 0.735 │ 0.673 │ 0.552
recall@5     │ 0.820 │ 0.820 │ 0.795 │ 0.737 │ 0.626
recall@10    │ 0.887 │ 0.882 │ 0.860 │ 0.823 │ 0.721

Absolute Changes (Δ)
--------------------
Metric       │  768   │  512   │  256   │  128   │   64  
─────────────┼────────┼────────┼────────┼────────┼───────
ndcg@10      │ +0.184 │ +0.193 │ +0.194 │ +0.188 │ +0.168
mrr@10       │ +0.179 │ +0.191 │ +0.190 │ +0.182 │ +0.156
map@100      │ +0.174 │ +0.187 │ +0.185 │ +0.179 │ +0.155
accuracy@1   │ +0.160 │ +0.174 │ +0.172 │ +0.163 │ +0.135
accuracy@3   │ +0.199 │ +0.209 │ +0.218 │ +0.202 │ +0.169
accuracy@5   │ +0.202 │ +0.212 │ +0.219 │ +0.205 │ +0.195
accuracy@10  │ +0.196 │ +0.196 │ +0.206 │ +0.206 │ +0.206
precision@1  │ +0.160 │ +0.174 │ +0.172 │ +0.163 │ +0.135
precision@3  │ +0.066 │ +0.070 │ +0.073 │ +0.067 │ +0.056
precision@5  │ +0.040 │ +0.042 │ +0.044 │ +0.041 │ +0.039
precision@10 │ +0.020 │ +0.020 │ +0.021 │ +0.021 │ +0.021
recall@1     │ +0.160 │ +0.174 │ +0.172 │ +0.163 │ +0.135
recall@3     │ +0.199 │ +0.209 │ +0.218 │ +0.202 │ +0.169
recall@5     │ +0.202 │ +0.212 │ +0.219 │ +0.205 │ +0.195
recall@10    │ +0.196 │ +0.196 │ +0.206 │ +0.206 │ +0.206

Percentage Changes
------------------
Metric       │  768   │  512   │  256   │  128   │   64  
─────────────┼────────┼────────┼────────┼────────┼───────
ndcg@10      │ +33.5% │ +35.9% │ +37.9% │ +40.2% │ +44.4%
mrr@10       │ +35.5% │ +39.0% │ +40.5% │ +43.1% │ +46.3%
map@100      │ +34.0% │ +37.5% │ +38.8% │ +41.5% │ +44.6%
accuracy@1   │ +38.5% │ +43.3% │ +44.7% │ +48.5% │ +51.7%
accuracy@3   │ +35.5% │ +38.0% │ +42.1% │ +42.9% │ +43.9%
accuracy@5   │ +32.7% │ +34.9% │ +38.1% │ +38.5% │ +45.1%
accuracy@10  │ +28.4% │ +28.6% │ +31.6% │ +33.5% │ +40.1%
precision@1  │ +38.5% │ +43.3% │ +44.7% │ +48.5% │ +51.7%
precision@3  │ +35.5% │ +38.0% │ +42.1% │ +42.9% │ +43.9%
precision@5  │ +32.7% │ +34.9% │ +38.1% │ +38.5% │ +45.1%
precision@10 │ +28.4% │ +28.6% │ +31.6% │ +33.5% │ +40.1%
recall@1     │ +38.5% │ +43.3% │ +44.7% │ +48.5% │ +51.7%
recall@3     │ +35.5% │ +38.0% │ +42.1% │ +42.9% │ +43.9%
recall@5     │ +32.7% │ +34.9% │ +38.1% │ +38.5% │ +45.1%
recall@10    │ +28.4% │ +28.6% │ +31.6% │ +33.5% │ +40.1%