license: mit library_name: transformers language: en tags: - gpt2 - morse-code - text-to-morse - tars-humor - mini-llm - fine-tuned

Morsent-128T

Compact GPT-2 model for Morse code translation, decoding, and sci-fi humor — all in 128 tokens or less.

🧠 About the Model

Morsent-128T is a tiny fine-tuned language model based on GPT-2 that’s trained to:

🔡 Translate English → Morse Code
📻 Decode Morse Code → English
🤖 Respond with quirky sci-fi humor (inspired by TARS from Interstellar)
🧩 Work within just 128 tokens (perfect for small devices or fast inference)

It's meant to be small, weird, and a little bit wonderful.

💡 Key Highlights

Feature	Details
Base Model	GPT-2
Context Length	128 tokens
Tokenizer	Custom-trained
Fine-tuned Tasks	Morse translation, decoding, TARS-style humor
Intended Use	Educational, demo, fun!
License	MIT

🚀 Example Usage

from transformers import GPT2LMHeadModel, GPT2TokenizerFast

model = GPT2LMHeadModel.from_pretrained("SrihariV/Morsent-128T")
tokenizer = GPT2TokenizerFast.from_pretrained("SrihariV/Morsent-128T")

prompt = "Translate: emergency to Morse code [/INST]"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=50)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))

##🧪 Known Issues 🚫 Max length is short: 128 tokens only 🤯 Sometimes responds with hallucinated or surreal outputs 😅 TARS-style jokes may be too weird

#📦 Files pytorch_model.bin — fine-tuned weights tokenizer.json, tokenizer_config.json — custom tokenizer config.json — model architecture

📚 Inspiration

Morse code from historical datasets Instruction-tuning format inspired by [LLaMA-Instruct] TARS-style humor inspired by sci-fi language modeling

🤝 Citation

@misc{morsent2025,
  title = {Morsent-128T: A Tiny Morse & Humor LLM},
  author = {Srihari Venkatesan},
  year = 2025,
  howpublished = {\url{https://huggingface.co./SrihariV/Morsent-128T}},
}

🗣 Contact Questions? Bugs? Want to collab?

📫 Reach me at: [email protected]

“Sometimes the smallest minds deliver the weirdest wonders.” – Morsent

SrihariV
/

Morsent-128-mini