Hugging Face's logo

tcepi
/
prog_integridade_model

Text Classification
Transformers
Safetensors
PyTorch
Portuguese
modernbert
binary-classification
Eval Results (legacy)
text-embeddings-inference
Model card Files
xet
 Community
vic35get commited on
Commit
8339f70
·
verified ·
1 Parent(s): 0637243

Add detailed Model Card with metrics

Browse files
Files changed (1) hide show
  1. README.md +42 -42
README.md CHANGED
@@ -30,21 +30,21 @@ model-index:
30
  type: tcepi/prog_integridade_dataset
31
  split: test
32
  metrics:
33
- - type: accuracy
34
- value: 0.9878
35
- name: Accuracy
36
- - type: f1
37
- value: 0.9878
38
- name: F1
39
- - type: precision
40
- value: 0.988
41
- name: Precision
42
- - type: recall
43
- value: 0.9878
44
- name: Recall
45
- - type: roc_auc
46
- value: 0.997
47
- name: ROC-AUC
48
  ---
49
 
50
  # Programa de Integridade Classification Model
@@ -64,31 +64,31 @@ Este modelo é um fine-tune do [ModernBERT-base](https://huggingface.co/answerdo
64
 
65
  | Métrica | Valor |
66
  |---------|-------|
67
- | **Accuracy** | 0.9878 |
68
- | **F1-Score** | 0.9878 |
69
- | **Precision** | 0.9880 |
70
- | **Recall** | 0.9878 |
71
- | **ROC-AUC** | 0.9970 |
72
- | **Specificity** | 0.9791 |
73
 
74
  ### Matriz de Confusão
75
 
76
  | | Predito Negativo | Predito Positivo |
77
  |--|-----------------|-----------------|
78
- | **Real Negativo** | 516 (TN) | 11 (FP) |
79
- | **Real Positivo** | 2 (FN) | 540 (TP) |
80
 
81
  ### Relatório de Classificação
82
 
83
  ```
84
  precision recall f1-score support
85
 
86
- Negativo 0.9961 0.9791 0.9876 527
87
- Positivo 0.9800 0.9963 0.9881 542
88
 
89
- accuracy 0.9878 1069
90
- macro avg 0.9881 0.9877 0.9878 1069
91
- weighted avg 0.9880 0.9878 0.9878 1069
92
 
93
  ```
94
 
@@ -104,7 +104,7 @@ model = AutoModelForSequenceClassification.from_pretrained("tcepi/prog_integrida
104
 
105
  # Classificar texto
106
  text = "Seu texto aqui"
107
- inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=512)
108
 
109
  with torch.no_grad():
110
  outputs = model(**inputs)
@@ -129,24 +129,24 @@ print(f"Probabilidades: {predictions.tolist()}")
129
 
130
  ### Informações de Treinamento
131
 
132
- - **Tempo Total:** 1281.24 segundos
133
- - **Samples/segundo:** 30.01
134
- - **Loss Final:** 0.0905
135
 
136
  ### Tabela de Métricas por Época
137
 
138
  | epoch | eval_loss | eval_accuracy | eval_f1 | eval_roc_auc | eval_specificity |
139
  |--------:|------------:|----------------:|----------:|---------------:|-------------------:|
140
- | 1 | 0.151197 | 0.957944 | 0.957959 | 0.994956 | 0.965686 |
141
- | 2 | 0.0504656 | 0.985981 | 0.985969 | 0.998654 | 0.970588 |
142
- | 3 | 0.0681938 | 0.985981 | 0.985969 | 0.998588 | 0.970588 |
143
- | 4 | 0.0468312 | 0.990654 | 0.990649 | 0.998961 | 0.980392 |
144
- | 5 | 0.0373923 | 0.990654 | 0.990649 | 0.99907 | 0.980392 |
145
- | 6 | 0.0413175 | 0.990654 | 0.990649 | 0.999179 | 0.980392 |
146
- | 7 | 0.0415714 | 0.990654 | 0.990649 | 0.999311 | 0.980392 |
147
- | 8 | 0.0415429 | 0.988318 | 0.988316 | 0.999442 | 0.985294 |
148
- | 9 | 0.043584 | 0.990654 | 0.990652 | 0.999486 | 0.985294 |
149
- | 10 | 0.0377868 | 0.990654 | 0.990652 | 0.999551 | 0.985294 |
150
 
151
  ### Curvas de Treinamento
152
 
 
30
  type: tcepi/prog_integridade_dataset
31
  split: test
32
  metrics:
33
+ - name: Accuracy
34
+ type: accuracy
35
+ value: 0.9944
36
+ - name: F1
37
+ type: f1
38
+ value: 0.9944
39
+ - name: Precision
40
+ type: precision
41
+ value: 0.9944
42
+ - name: Recall
43
+ type: recall
44
+ value: 0.9944
45
+ - name: ROC-AUC
46
+ type: roc_auc
47
+ value: 0.9984
48
  ---
49
 
50
  # Programa de Integridade Classification Model
 
64
 
65
  | Métrica | Valor |
66
  |---------|-------|
67
+ | **Accuracy** | 0.9944 |
68
+ | **F1-Score** | 0.9944 |
69
+ | **Precision** | 0.9944 |
70
+ | **Recall** | 0.9944 |
71
+ | **ROC-AUC** | 0.9984 |
72
+ | **Specificity** | 0.9905 |
73
 
74
  ### Matriz de Confusão
75
 
76
  | | Predito Negativo | Predito Positivo |
77
  |--|-----------------|-----------------|
78
+ | **Real Negativo** | 522 (TN) | 5 (FP) |
79
+ | **Real Positivo** | 1 (FN) | 541 (TP) |
80
 
81
  ### Relatório de Classificação
82
 
83
  ```
84
  precision recall f1-score support
85
 
86
+ Negativo 0.9981 0.9905 0.9943 527
87
+ Positivo 0.9908 0.9982 0.9945 542
88
 
89
+ accuracy 0.9944 1069
90
+ macro avg 0.9945 0.9943 0.9944 1069
91
+ weighted avg 0.9944 0.9944 0.9944 1069
92
 
93
  ```
94
 
 
104
 
105
  # Classificar texto
106
  text = "Seu texto aqui"
107
+ inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=8192)
108
 
109
  with torch.no_grad():
110
  outputs = model(**inputs)
 
129
 
130
  ### Informações de Treinamento
131
 
132
+ - **Tempo Total:** 2731.90 segundos
133
+ - **Samples/segundo:** 14.07
134
+ - **Loss Final:** 0.0701
135
 
136
  ### Tabela de Métricas por Época
137
 
138
  | epoch | eval_loss | eval_accuracy | eval_f1 | eval_roc_auc | eval_specificity |
139
  |--------:|------------:|----------------:|----------:|---------------:|-------------------:|
140
+ | 1 | 0.188758 | 0.946262 | 0.946277 | 0.997855 | 0.985294 |
141
+ | 2 | 0.0218751 | 0.990654 | 0.990654 | 0.999912 | 0.990196 |
142
+ | 3 | 0.0222455 | 0.992991 | 0.992993 | 0.999869 | 1 |
143
+ | 4 | 0.0971301 | 0.983645 | 0.983653 | 0.999934 | 1 |
144
+ | 5 | 0.0337453 | 0.990654 | 0.990649 | 1 | 0.980392 |
145
+ | 6 | 0.0275761 | 0.997664 | 0.997664 | 0.999956 | 1 |
146
+ | 7 | 0.0167756 | 0.997664 | 0.997664 | 1 | 1 |
147
+ | 8 | 0.0292979 | 0.997664 | 0.997664 | 0.999912 | 1 |
148
+ | 9 | 0.0309485 | 0.997664 | 0.997664 | 0.999912 | 1 |
149
+ | 10 | 0.0248101 | 0.997664 | 0.997664 | 0.999956 | 1 |
150
 
151
  ### Curvas de Treinamento
152