Initial upload of BERT-Tiny AMD classifier

README.md

@@ -0,0 +1,141 @@
+---
+license: mit
+language:
+- en
+tags:
+- text-classification
+- answering-machine-detection
+- bert-tiny
+- binary-classification
+- call-center
+- voice-processing
+pipeline_tag: text-classification
+---
+
+# BERT-Tiny AMD Classifier
+
+A lightweight BERT-Tiny model fine-tuned for Answering Machine Detection (AMD) in call center environments.
+
+## Model Description
+
+This model is based on `prajjwal1/bert-tiny` and fine-tuned to classify phone call transcripts as either human or machine (answering machine/voicemail) responses. It's designed for real-time call center applications where quick and accurate detection of answering machines is crucial.
+
+## Model Architecture
+
+- **Base Model**: `prajjwal1/bert-tiny` (2 layers, 128 hidden size, 2 attention heads)
+- **Total Parameters**: ~4.4M (lightweight and efficient)
+- **Input**: User transcript text (max 128 tokens)
+- **Output**: Single logit with sigmoid activation for binary classification
+- **Loss Function**: BCEWithLogitsLoss with positive weight for class imbalance
+
+## Performance
+
+- **Validation Accuracy**: 97.75%
+- **Precision**: 95.79%
+- **Recall**: 95.79%
+- **F1-Score**: 95.79%
+- **Agreement with Rule-based System**: 97.75%
+
+## Training Data
+
+- **Total Samples**: 3,548 phone call transcripts
+- **Training Set**: 2,838 samples
+- **Validation Set**: 710 samples
+- **Class Distribution**: 26.8% machine calls, 73.2% human calls
+- **Source**: ElevateNow call center data
+
+## Usage
+
+### Basic Inference
+
+```python
+from transformers import AutoTokenizer, AutoModelForSequenceClassification
+import torch
+
+# Load model and tokenizer
+model = AutoModelForSequenceClassification.from_pretrained("your-username/bert-tiny-amd")
+tokenizer = AutoTokenizer.from_pretrained("your-username/bert-tiny-amd")
+
+# Prepare input
+text = "Hello, this is John speaking"
+inputs = tokenizer(text, return_tensors="pt", max_length=128, truncation=True, padding=True)
+
+# Make prediction
+with torch.no_grad():
+    outputs = model(**inputs)
+    logits = outputs.logits.squeeze(-1)
+    probability = torch.sigmoid(logits).item()
+    is_machine = probability >= 0.5
+
+print(f"Prediction: {'Machine' if is_machine else 'Human'}")
+print(f"Confidence: {probability:.4f}")
+```
+
+### Production Usage
+
+```python
+from transformers import AutoTokenizer, AutoModelForSequenceClassification
+import torch
+
+class AMDClassifier:
+    def __init__(self, model_name="your-username/bert-tiny-amd"):
+        self.model = AutoModelForSequenceClassification.from_pretrained(model_name)
+        self.tokenizer = AutoTokenizer.from_pretrained(model_name)
+        self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+        self.model.to(self.device)
+        self.model.eval()
+    
+    def predict(self, transcript_text, threshold=0.5):
+        """Predict if transcript is from answering machine"""
+        inputs = self.tokenizer(
+            transcript_text,
+            return_tensors="pt",
+            max_length=128,
+            truncation=True,
+            padding=True
+        ).to(self.device)
+        
+        with torch.no_grad():
+            outputs = self.model(**inputs)
+            logits = outputs.logits.squeeze(-1)
+            probability = torch.sigmoid(logits).item()
+            is_machine = probability >= threshold
+        
+        return is_machine, probability
+
+# Usage
+classifier = AMDClassifier()
+is_machine, confidence = classifier.predict("Hello, this is John speaking")
+```
+
+## Training Details
+
+- **Optimizer**: AdamW with weight decay (0.01)
+- **Learning Rate**: 3e-5 with linear scheduling
+- **Batch Size**: 32
+- **Epochs**: 12 (with early stopping)
+- **Early Stopping**: Patience of 3 epochs
+- **Class Imbalance**: Handled with positive weight (2.729)
+
+## Limitations
+
+- Trained on English phone call transcripts
+- May not generalize well to other languages or domains
+- Performance may vary with different transcription quality
+- Designed for short utterances (max 128 tokens)
+
+## Citation
+
+```bibtex
+@misc{bert-tiny-amd,
+  title={BERT-Tiny AMD Classifier for Answering Machine Detection},
+  author={Your Name},
+  year={2025},
+  publisher={Hugging Face},
+  howpublished={\url{https://huggingface.co/your-username/bert-tiny-amd}}
+}
+```
+
+## License
+
+MIT License - see LICENSE file for details.

config.json

@@ -0,0 +1,30 @@
+{
+  "architectures": [
+    "BertForSequenceClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "classifier_dropout": null,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 128,
+  "id2label": {
+    "0": "LABEL_0"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 512,
+  "label2id": {
+    "LABEL_0": 0
+  },
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 2,
+  "num_hidden_layers": 2,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "torch_dtype": "float32",
+  "transformers_version": "4.54.0",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 30522
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:079d2a82939195cf53e63521c9efc0cb4133e012d790625e511544f649069651
+size 17548796

special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "cls_token": "[CLS]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
+}

tokenizer_config.json

@@ -0,0 +1,58 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "101": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "102": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "103": {
+      "content": "[MASK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "do_basic_tokenize": true,
+  "do_lower_case": true,
+  "extra_special_tokens": {},
+  "mask_token": "[MASK]",
+  "model_max_length": 1000000000000000019884624838656,
+  "never_split": null,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "BertTokenizer",
+  "unk_token": "[UNK]"
+}

training_metadata.json

@@ -0,0 +1,76 @@
+{
+  "training_config": {
+    "model_name": "prajjwal1/bert-tiny",
+    "max_length": 128,
+    "batch_size": 32,
+    "learning_rate": 3e-05,
+    "num_epochs": 15,
+    "patience": 3,
+    "test_size": 0.2,
+    "device": "cpu",
+    "csv_file": "all_EN_calls.csv",
+    "s3_bucket": "voicex-call-recordings"
+  },
+  "final_metrics": {
+    "accuracy": 0.9774647887323944,
+    "precision": 0.9578947368421052,
+    "recall": 0.9578947368421052,
+    "f1": 0.9578947368421052,
+    "confusion_matrix": [
+      [
+        512,
+        8
+      ],
+      [
+        8,
+        182
+      ]
+    ]
+  },
+  "pos_weight": 2.729303547963206,
+  "threshold": 0.5,
+  "training_history": {
+    "train_losses": [
+      0.9435733710781912,
+      0.6628189873829317,
+      0.40206739406907155,
+      0.28053958831208475,
+      0.21479346770583913,
+      0.180794070108553,
+      0.14911148521337617,
+      0.13325696530636777,
+      0.12835281459468134,
+      0.11012767288792,
+      0.10539512767383222,
+      0.09656323011169272
+    ],
+    "val_losses": [
+      0.8112381230229917,
+      0.4864982029666071,
+      0.34563232180864917,
+      0.26932784072730853,
+      0.24466017180162927,
+      0.2034845212879388,
+      0.1938699973018273,
+      0.19390630900211955,
+      0.1721272283922071,
+      0.17268858526064002,
+      0.17224800457125125,
+      0.18237287065257196
+    ],
+    "val_accuracies": [
+      0.9042253521126761,
+      0.9690140845070423,
+      0.9704225352112676,
+      0.971830985915493,
+      0.9690140845070423,
+      0.976056338028169,
+      0.9746478873239437,
+      0.9746478873239437,
+      0.9774647887323944,
+      0.9732394366197183,
+      0.9746478873239437,
+      0.9774647887323944
+    ]
+  }
+}