nesar2004/message-generator-gpt2

Message Generator GPT-2

This model is fine-tuned on GPT-2 for generating contextual messages across different categories like dating, flirty, love, and cheesy messages.

Model Description

• Model Type: GPT-2
• Language: English
• Training Data: Custom message dataset
• Input: Message category token (dating, flirty, love, cheesy)
• Output: Generated contextual message

Usage

from transformers import GPT2LMHeadModel, GPT2Tokenizer

# Load model and tokenizer
model = GPT2LMHeadModel.from_pretrained("nesar2004/message-generator-gpt2")
tokenizer = GPT2Tokenizer.from_pretrained("nesar2004/message-generator-gpt2")

# Generate message
category = "dating"  # Can be: dating, flirty, love, cheesy
input_text = f"<|{category}|><|message|>"
input_ids = tokenizer.encode(input_text, return_tensors='pt')

# Generate
output = model.generate(
    input_ids,
    max_length=60,
    num_return_sequences=1,
    temperature=0.9,
    top_p=0.95,
    do_sample=True,
    pad_token_id=tokenizer.pad_token_id,
    eos_token_id=tokenizer.eos_token_id
)

# Decode and clean up the message
message = tokenizer.decode(output[0], skip_special_tokens=True)
message = message.replace("<|message|>", "").strip()
message = message.replace(f"<|{category}|>", "").strip()
print(message)

Training Details

The model was fine-tuned on a custom dataset of messages across different categories. It uses special tokens for category identification and message generation.

Limitations

• The model may sometimes generate repetitive messages
• Output quality varies with generation parameters
• May require adjustment of temperature and top_p for optimal results

License

This model is released under the MIT License.