Edit model card

Chat-With-YourFAV-Character-Small

This model is a conversational agent fine-tuned on the game script of "The World Ends With You" for generating dialogues in the style of the character Joshua. It is based on the DialoGPT-small architecture and further trained to mimic the conversational patterns and style of Joshua, offering users the ability to interact with a simulated version of the character.

Model Details

Developed by

  • [Manan Raval- Virtual Height IT Services Pvt. Ltd.]

Model Description

This model was developed to explore the capabilities of language models in generating character-specific dialogues. It utilizes the transformer architecture for generating text that is coherent and contextually relevant to the inputs it receives.

  • Model type: DialoGPT-small

  • Language(s) (NLP): English

  • License: Apache-2.0

  • Sources

  • Finetuned from: microsoft/DialoGPT-small

Usage

Direct Use

This model is intended for direct interaction by users who want to engage in conversation with an AI that simulates the character Joshua from "The World Ends With You." It can be integrated into chat applications, games, or any platform where conversational AI can enhance user experience.

Out-of-Scope Use

This model is not designed for high-stakes decision-making processes, sensitive contexts, or as a replacement for human interaction in therapeutic or counseling settings.

Bias, Risks, and Limitations

The dialogues generated by this model are based on the character's portrayal in a specific video game and might not be appropriate for all contexts. Users should be aware of the potential for generating unexpected or offensive content.

Recommendations

It is recommended to monitor the interactions with the model and implement appropriate filtering mechanisms to prevent misuse or the generation of harmful content.

How to Get Started

from transformers import AutoModelForCausalLM, AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("mananvh/Chat-With-YourFAV-Character-Small")
model = AutoModelForCausalLM.from_pretrained("mananvh/Chat-With-YourFAV-Character-Small")

# For interactive conversation
input_text = "Your input text here"
input_ids = tokenizer.encode(input_text + tokenizer.eos_token, return_tensors='pt')
chat_history_ids = model.generate(input_ids, max_length=1000, pad_token_id=tokenizer.eos_token_id)
response = tokenizer.decode(chat_history_ids[:, input_ids.shape[-1]:][0], skip_special_tokens=True)

print(response)

Training and Evaluation

Training Data

The model was fine-tuned on a dataset extracted from the game script of "The World Ends With You," focusing specifically on dialogues involving the character Joshua.

Training Procedure

The model was trained using the Hugging Face Transformers library with the following hyperparameters: learning rate of 5e-5, batch size of 4, and a total of 4 epochs. The training was performed on a single GPU.

Evaluation

Due to the conversational nature of the model, its performance was primarily evaluated based on qualitative assessments of its ability to generate coherent and character-consistent dialogues.

Technical Specifications

Model Architecture

The model is based on the DialoGPT-small architecture, which is a variant of the GPT-2 model optimized for dialogue generation tasks.

Compute Infrastructure

The training was conducted on a single GPU setup, utilizing NVIDIA CUDA for acceleration. YourModelNameHere

Citation

@misc{chatprompt_ai_2024,
  title={ChatPrompt-AI: A Conversational Agent Simulating Joshua from TWEWY},
  author={Manan},
  year={2024},
  howpublished={\url{https://huggingface.co/mananvh/Chat-With-YourFAV-Character-Small}}
}

Contact

For any questions or feedback regarding this model, please contact [mananraval.vhits@gmail.com].

Downloads last month
2
Safetensors
Model size
124M params
Tensor type
F32
·