---
library_name: transformers
tags: []
---

# Model Card for Super Tiny Bert
This is a super tiny Bert model for testing purposes.

## Model Details
This model has been generated using:

```
from transformers import BertTokenizer, BertModel, BertConfig

# Define a tiny BERT configuration
config = BertConfig(
    vocab_size=30,
    hidden_size=8,
    num_hidden_layers=2,
    num_attention_heads=2,
    intermediate_size=8,
    max_position_embeddings=8,
)

# Initialize a tiny BERT model with the custom configuration
model = BertModel(config)

# Create a custom vocabulary
vocab = {
    "[PAD]": 0,
    "[UNK]": 1,
    "[CLS]": 2,
    "[SEP]": 3,
    "[MASK]": 4,
    "hello": 5,
    "how": 6,
    "are": 7,
    "you": 8,
    "?": 9,
    "i": 10,
    "am": 11,
    "fine": 12,
    "thanks": 13,
    "and": 14,
    "good": 15,
    "morning": 16,
    "evening": 17,
    "night": 18,
    "yes": 19,
    "no": 20,
    "please": 21,
    "thank": 22,
    "welcome": 23,
    "sorry": 24,
    "bye": 25,
    "see": 26,
    "later": 27,
    "take": 28,
    "care": 29,
}

# Save the vocabulary to a file
vocab_file = "vocab.txt"
with open(vocab_file, "w") as f:
    for token, index in sorted(vocab.items(), key=lambda item: item[1]):
        f.write(f"{token}\n")

# Initialize the tokenizer with the custom vocabulary
tokenizer = BertTokenizer(vocab_file=vocab_file)

# Example usage: Tokenize input text
text = "Hello, how are you?"
inputs = tokenizer(text, return_tensors="pt")

# Forward pass through the model
outputs = model(**inputs)

# Extract the last hidden states
last_hidden_states = outputs.last_hidden_state

print("Last hidden states shape:", last_hidden_states.shape)

# Save the tokenizer and model to the Hugging Face Hub
model_name = "flexsystems/flex-e2e-super-tiny-bert-model"
tokenizer.push_to_hub(model_name, private=False)
model.push_to_hub(model_name, private=False)

print(f"Tiny BERT model and tokenizer saved to the Hugging Face Hub as '{model_name}'.")

```