Upload modelo treinado - 20251205_140224_hf

README.md

@@ -19,15 +19,15 @@ model-index:
       type: legal_documents
     metrics:
     - type: accuracy
-      value: 0.927038626609442
+      value: 0.9124293785310734
     - type: f1
-      value: 0.9253745140607927
+      value: 0.9069008258581538
     - type: precision
-      value: 0.9279357771017352
+      value: 0.9052447065095824
     - type: recall
-      value: 0.927038626609442
+      value: 0.9124293785310734
     - type: f1_macro
-      value: 0.8587213814786654
+      value: 0.6832937813302449
 ---
 
 # Model Card for Model ID
@@ -51,7 +51,7 @@ Este modelo foi treinado para classificar documentos jurídicos brasileiros usan
 
 - **Base Model:** google-bert/bert-base-multilingual-cased
 - **Tipo:** Classificação Multiclasse
-- **Número de Classes:** 7
+- **Número de Classes:** 8
 - **Idioma:** Português (Brasil)
 - **Domínio:** Documentos Jurídicos
 
@@ -64,33 +64,35 @@ Este modelo foi treinado para classificar documentos jurídicos brasileiros usan
 4. Art.40
 5. Não
 6. Teste
+7. Teste 2
 
 ## Métricas de Avaliação
 
 As métricas abaixo foram calculadas no conjunto de validação:
 
-- **Accuracy:** 0.9270
-- **F1-Score (Weighted):** 0.9254
-- **Precision (Weighted):** 0.9279
-- **Recall (Weighted):** 0.9270
-- **F1-Score (Macro):** 0.8587
+- **Accuracy:** 0.9124
+- **F1-Score (Weighted):** 0.9069
+- **Precision (Weighted):** 0.9052
+- **Recall (Weighted):** 0.9124
+- **F1-Score (Macro):** 0.6833
 
 ## Classification Report Completo
 
 ```
               precision    recall  f1-score   support
 
-           0       0.67      0.29      0.40         7
-           1       0.88      1.00      0.94        46
-           2       1.00      0.80      0.89        56
-           3       0.93      0.95      0.94        39
-           4       0.87      0.94      0.90       120
-           5       0.94      0.94      0.94       421
-           6       1.00      1.00      1.00        10
-
-    accuracy                           0.93       699
-   macro avg       0.90      0.85      0.86       699
-weighted avg       0.93      0.93      0.93       699
+           0       0.00      0.00      0.00         7
+           1       0.94      0.98      0.96        48
+           2       0.95      0.72      0.82        57
+           3       0.82      0.95      0.88        42
+           4       0.88      0.95      0.91       132
+           5       0.94      0.95      0.94       408
+           6       0.40      0.33      0.36         6
+           7       0.56      0.62      0.59         8
+
+    accuracy                           0.91       708
+   macro avg       0.69      0.69      0.68       708
+weighted avg       0.91      0.91      0.91       708
 
 ```
 

config.json

@@ -16,7 +16,8 @@
     "3": "Cita\u00e7\u00e3o",
     "4": "Art.40",
     "5": "N\u00e3o",
-    "6": "Teste"
+    "6": "Teste",
+    "7": "Teste 2"
   },
   "initializer_range": 0.02,
   "intermediate_size": 3072,
@@ -27,7 +28,8 @@
     "N\u00e3o": 5,
     "Penhora/Arresto/Sequestro": 2,
     "Senten\u00e7a": 1,
-    "Teste": 6
+    "Teste": 6,
+    "Teste 2": 7
   },
   "layer_norm_eps": 1e-12,
   "max_position_embeddings": 512,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7ce861602c0a806e8a87b95156402241cf004d3dddfc086703769e85a1377620
-size 711458836
+oid sha256:efbed19f1962756e5e336210fcff4fc033dc4354158b0f808ec2cd744a2e0842
+size 711461912