Upload model_mf_v3 (epoch 5, 75u/194w, RMSE=1.1956) (model card)

README.md

@@ -18,14 +18,14 @@ real user-wine ratings from the Swirl platform.
 | Property | Value |
 |---|---|
 | Architecture | Biased MF with optional side-feature projection |
-| Latent dimension | 64 |
+| Latent dimension | 16 |
 | Side features | Yes (768-dim wine embeddings) |
-| Parameters | 123,654 |
-| Users | 73 |
-| Wines | 188 |
-| Best epoch | 7 |
-| Val RMSE | 0.8635 |
-| Val MAE | 0.606194019317627 |
+| Parameters | 29,710 |
+| Users | 75 |
+| Wines | 194 |
+| Best epoch | 5 |
+| Val RMSE | 1.1956 |
+| Val MAE | 0.7387912273406982 |
 | Version | mf-v1 |
 
 ## Scoring formula
@@ -44,7 +44,7 @@ Two-stage scoring for production:
 ```python
 from matrix_factorization.inference import MFScorer
 
-scorer = MFScorer.load("model_mf_v2.pt")
+scorer = MFScorer.load("model_mf_v3.pt")
 score = scorer.score(user_id="...", wine_id="...")
 ```