WOOGPT

🧙 WOOGPT

WOOGPT is a custom language model fine-tuned on Wizard of Oz books and movie scripts. It aims to generate whimsical, story-rich, and character-driven text in the tone of L. Frank Baum’s magical universe.

---

🧠 Model Details

• Base model: GPT-2 architecture
• Architecture:
  • Embedding dimension: 384
  • Attention heads: 8
  • Transformer layers: 8
  • Dropout: 0.2
• Sequence length (block size): 128 tokens
• Language: English
• License: MIT
• Trained by: Aksshat Govind

---

🏋️ Training Configuration

Note: Training was performed on Wizard of Oz text data including books and screenplay dialogue. Training used a causal language modeling objective with teacher forcing.

---

📚 Training Dataset

WOOGPT was trained on:

• Public domain books by L. Frank Baum
• Wizard of Oz film dialogue transcripts
• Stage play and derivative text adaptations

Training epochs: 3000
Tokenizer: GPT-2 tokenizer (byte-level BPE) Device: MPS and CUDA

---

🚀 Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("thisisakz/WOOGPT")
tokenizer = AutoTokenizer.from_pretrained("thisisakz/WOOGPT")

prompt = "Dorothy looked at the yellow brick road and said,"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=100)
print(tokenizer.decode(outputs[0]))