File size: 1,499 Bytes
0292bfe 1c082de 0292bfe 1481eeb 0292bfe 1481eeb 0292bfe 1c082de 1481eeb 0292bfe 1481eeb 0292bfe 1481eeb 0292bfe |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 |
---
license: mit
library_name: pytorch
tags:
- bilstm
- lstm
- pytorch
- text-classification
- spam-detection
model_details:
parameters: 4403585
task_categories:
- text-classification
datasets:
- ucirvine/sms_spam
language:
- en
---
# BiLSTM Text Classifier
Simple BiLSTM model PyTorch trained for SPAM detection on SMS Spam Collection
(Almeida, Tiago and Jos Hidalgo. 2011. *SMS Spam Collection*.
UCI Machine Learning Repository. https://doi.org/10.24432/C5CC84).
## Important Notes
- The model returns **logits** as output; to obtain probabilities, apply `torch.sigmoid`.
- The model uses the `bert-base-uncased` tokenizer **only for tokenization** (the encoder is NOT BERT).
- Number of parameters: ~4.4M
## Files
- `BiLSTMClassifier.safetensors`: trained weights
- `BiLSTMClassifier.py`: model definition
- `config.json`: hyperparameters
## Usage
```python
import json
import torch
from transformers import BertTokenizer
from safetensors.torch import load_file
from BiLSTMClassifier import BiLSTMClassifier
with open("config.json") as f:
cfg = json.load(f)
model = BiLSTMClassifier(**cfg)
state_dict = load_file("BiLSTMClassifier.safetensors")
model.load_state_dict(state_dict)
model.eval()
sample_text = "URGENT HIRING! Earn $500/day working from home. No experience needed."
tokenizer = BertTokenizer.from_pretrained("bert-base-uncased")
tokens = tokenizer(sample_text, return_tensors="pt")
logits = model(tokens["input_ids"])
prob = torch.sigmoid(logits)
|