DeepSeek-V3-689M trained on Shakespeare plays
SmolLM2-135M trained on Shakespeare plays
124M size model (similar to GPT2) overfitted on a small data
Marathi language tokenizer
