Model Card for SchnabelTim/t5-catify-de-en

Model Details

Model Name: SchnabelTim/t5-catify-de-en
Model Architecture: T5
Base Model: Einmalumdiewelt/T5-Base_GNAD
Language(s): German, English
License: MIT
Author: SchnabelTim

Model Description

SchnabelTim/t5-catify-de-en is a Transformer-based model fine-tuned from the T5 architecture. It is designed to transform person-related data into cat-related data, functioning effectively in both German and English. The model can take input sentences that are about people and convert them to be about cats, maintaining the original context and meaning as much as possible.

Training Data

The model was trained on a self-created dataset. The dataset includes sentences related to people and their corresponding cat-related transformations. This dataset was curated to ensure diverse and contextually rich examples for robust performance across various scenarios.

Training Procedure

Number of Epochs: 5
Batch Size: 12
Optimizer: AdamW
Learning Rate: 5e-5

Training was monitored using TensorBoard, and the following metrics were observed:

Training Loss: 0.098
Validation Loss: 0.096

Evaluation

The model was evaluated on a held-out test set from the same distribution as the training data. The following metrics were used to assess model performance:

Test Loss: 0.096

Usage

To use this model, you can load it using the Hugging Face Transformers library as follows:

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("SchnabelTim/t5-catify-de-en")
model = AutoModelForSeq2SeqLM.from_pretrained("SchnabelTim/t5-catify-de-en")

def catify_text(input_text):
    inputs = tokenizer(input_text, return_tensors="pt")
    outputs = model.generate(**inputs)
    return tokenizer.decode(outputs[0], skip_special_tokens=True)

input_text = "What is a Human?"
print(catify_text(input_text))  # Output: "What is a cat?"