# Phoenix :rocket:

A tiny demo of **madcatter** rendering.

## Loss

Minimizes $L = -\sum_i y_i \log p_i$ per batch.

## Code

```python
def greet(name: str) -> str:
    return f"hello, {name}"
```

## Scores

| model | accuracy |
| ----- | -------: |
| base  |    0.812 |
| tuned |    0.947 |

More in the [docs](README.md).