app.py

import gradio as gr
import yaml

def generate_config(base_model, models, layer_range, merge_method):
    slices = []
    for model in models:
        slice_config = {
            "sources": [
                {
                    "model": model,
                    "layer_range": layer_range
                }
            ]
        }
        slices.append(slice_config)

    config = {
        "slices": slices,
        "merge_method": merge_method,
        "base_model": base_model,
        "parameters": {
            "t": [
                {
                    "filter": "self_attn",
                    "value": [0, 0.5, 0.3, 0.7, 1]
                },
                {
                    "filter": "mlp",
                    "value": [1, 0.5, 0.7, 0.3, 0]
                },
                {
                    "value": 0.5
                }
            ]
        },
        "dtype": "bfloat16"
    }

    return yaml.dump(config)

iface = gr.Interface(
    fn=generate_config,
    inputs="dropdown",
    inputs2="multiselect",
    inputs3=gr.inputs.NumberSlider(minimum=0, maximum=32, step=1),
    inputs4="dropdown",
    output_type="text",
    output_formats=["yaml"],
    title="Merge Config Builder",
    description="Build a merge config file for your models"
)

iface.launch()
'''
    We define a function generate_config that takes in the base model, a list of additional models, the layer range, and the merge method as inputs.
    We create a list of slices, where each slice contains the configuration for a single model.
    We create the final config dictionary, which includes the slices, merge method, base model, parameters, and data type.
    We use the yaml.dump function to convert the config dictionary to a YAML string.
    We create a Gradio interface using the gr.Interface function.
    We define the input and output components of the interface.
    We launch the interface using the launch method.
'''