Rec_ Business_ Names

Introduction

The Rec_Business_Names model aims to assist users in generating viable business names by leveraging the creative potential of the T5 small pre-trained language model. By fine-tuning this model on a dataset scraped from approximately 350,000 websites, it can generate business names based on a summary or keywords related to a business idea. This approach surpasses traditional rule-based systems by considering context, resulting in more creative and relevant names.

Architecture

The model is built on the T5 small architecture, known for its ability to generate text with near-human quality by understanding the context of a given prefix. Fine-tuning the T5 model on domain names and meta contexts enhances its capability to relate domain names with website content, facilitating more contextually aware business name generation.

Training

The training process involved fine-tuning the T5 small model on a large and clean dataset, which is crucial for achieving high model quality. The dataset, compiled from various domain lists, is accessible upon request. The fine-tuning process aimed to improve the model's ability to understand the relationship between domain names and website content, thereby enhancing its text generation capabilities.

Guide: Running Locally

To run the Rec_Business_Names model locally, follow these steps:

1. Install Dependencies: Ensure you have Python and the Transformers library installed.

2. Load the Model:

   from transformers import AutoTokenizer, AutoModelForSeq2SeqLM 
   
   tokenizer = AutoTokenizer.from_pretrained("abdelhalim/Rec_Business_Names")
   model = AutoModelForSeq2SeqLM.from_pretrained("abdelhalim/Rec_Business_Names")
   

3. Generate Business Names:

   encoder_input_str = "fourniture and decor brand"
   number_of_business_names = 10
   
   input_ids = tokenizer(encoder_input_str, return_tensors="pt").input_ids
   
   outputs = model.generate(
       input_ids,
       num_beams=number_of_business_names,
       num_return_sequences=number_of_business_names,
       no_repeat_ngram_size=1,
       remove_invalid_values=True,
   )
   
   for i in range(len(outputs)):
     print(tokenizer.decode(outputs[i], skip_special_tokens=True))
   

4. Output: The code will output a list of potential business names based on the input.

For optimal performance, consider using cloud GPUs from providers such as AWS, Google Cloud, or Azure, which can handle the computational requirements efficiently.

License

The Rec_Business_Names model is released under the BSD-1 License. This license permits use, distribution, and modification, provided that proper attribution is given to the original author.