Spaces:
Sleeping
Sleeping
| import gradio as gr | |
| import gpt | |
| """ | |
| For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface | |
| """ | |
| demo = gr.Interface(fn=gpt.get_response, inputs=["textbox", | |
| gr.Slider(0, 100, value=50, step=1), | |
| gr.Slider(0.1, 2.0, value=1.0), | |
| gr.Dropdown( | |
| ["mike-chat", "mike-code", "mike-code-600m"], value="mike-chat"), | |
| ], outputs=gr.Markdown(line_breaks=True), title="Mike Chat", article=""" | |
| Notice: if you have a GPU, I would highly recommend cloning the space and running it locally. The CPU provided by spaces isn't very fast. | |
| Mike is a small GPT-style language model. It was trained for about 8 hrs on my PC using fineweb-edu and open orca datasets. While it hallucinates a lot, it seems to be about on par with other LMs of its size (about 160M params). Model details: | |
| block_size: 512 | |
| n_layers: 12 | |
| n_heads: 12 | |
| d_model: 768 | |
| (Same as gpt-2 but without weight tying) | |
| Architecture for Mike-Code-600m: | |
| block_size: 256 | |
| n_layers: 16 | |
| n_heads: 12 | |
| d_model: 1536""") | |
| if __name__ == "__main__": | |
| demo.launch() |