File size: 6,684 Bytes
24af1c0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4114529
24af1c0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
import gradio as gr

# Available models
AVAILABLE_MODELS = [
    "(Select Model)",
    "mistralai/Mistral-7B-v0.1",
]

def create_model_config_section():
    """
    Creates the "Head to Head - Choose Models" section with two model configurations side by side.
    Returns the components needed for the main app.
    """
    with gr.Column() as model_config_container:
        gr.Markdown("## (C) Head to Head - Choose Models to evaluate against each other")
        
        with gr.Row():
            # Left column - Model 1 configuration
            with gr.Column(scale=1) as model1_column:
                with gr.Group(elem_classes=["config-box"]):
                    gr.Markdown("### Model 1")
                    
                    model1_dropdown = gr.Dropdown(
                        choices=AVAILABLE_MODELS,
                        value="(Select Model)",
                        label="Select Model 1",
                        info="Choose the first model for head-to-head comparison"
                    )
                    
                    model1_shots = gr.Slider(
                        minimum=0,
                        maximum=5,
                        value=5,
                        step=1,
                        label="Number of Few-shot Examples",
                        info="Number of examples to use for few-shot learning (0-5)"
                    )
                    
                    model1_regex = gr.Textbox(
                        label="Regex Pattern",
                        placeholder="Optional: Apply regex pattern to model outputs",
                        info="Leave empty for no regex pattern"
                    )
                    
                    model1_flash_attn = gr.Checkbox(
                        label="Use FlashAttention",
                        value=True,
                        info="Use FlashAttention for better performance (if supported by model)"
                    )
            
            # Divider in the middle
            with gr.Column(scale=0.1):
                gr.Markdown('<div style="border-left: 1px solid #ddd; height: 100%;"></div>', elem_classes=["center-divider"])
            
            # Right column - Model 2 configuration
            with gr.Column(scale=1) as model2_column:
                with gr.Group(elem_classes=["config-box"]):
                    gr.Markdown("### Model 2")
                    
                    model2_dropdown = gr.Dropdown(
                        choices=AVAILABLE_MODELS,
                        value="(Select Model)",
                        label="Select Model 2",
                        info="Choose the second model for head-to-head comparison"
                    )
                    
                    model2_shots = gr.Slider(
                        minimum=0,
                        maximum=5,
                        value=5,
                        step=1,
                        label="Number of Few-shot Examples",
                        info="Number of examples to use for few-shot learning (0-5)"
                    )
                    
                    model2_regex = gr.Textbox(
                        label="Regex Pattern",
                        placeholder="Optional: Apply regex pattern to model outputs",
                        info="Leave empty for no regex pattern"
                    )
                    
                    model2_flash_attn = gr.Checkbox(
                        label="Use FlashAttention",
                        value=True,
                        info="Use FlashAttention for better performance (if supported by model)"
                    )
        
        # Error message area - initially hidden
        model_config_error = gr.Markdown(
            visible=False,
            value="⚠️ **Error**: Both models and configurations are identical. Please select different models or configurations for comparison.",
            elem_classes=["error-message"]
        )
    
    return {
        'container': model_config_container,
        'model1_dropdown': model1_dropdown,
        'model1_shots': model1_shots,
        'model1_regex': model1_regex,
        'model1_flash_attn': model1_flash_attn,
        'model2_dropdown': model2_dropdown,
        'model2_shots': model2_shots,
        'model2_regex': model2_regex,
        'model2_flash_attn': model2_flash_attn,
        'error_message': model_config_error
    }

def validate_model_configs(model1, model1_shots, model1_regex, model1_flash, 
                          model2, model2_shots, model2_regex, model2_flash):
    """
    Validates that the two model configurations are not identical.
    Returns:
    - bool: Whether the configurations are valid (not identical)
    - str: Error message if invalid, otherwise empty string
    """
    if model1 == "(Select Model)" or model2 == "(Select Model)":
        return True, ""
    
    # Check if models and all configs are identical
    if (model1 == model2 and 
        model1_shots == model2_shots and 
        model1_regex == model2_regex and 
        model1_flash == model2_flash):
        return False, "⚠️ **Error**: Both configurations are identical. Please select different configurations (e.g., number of few-shot examples) for comparison."
    
    return True, ""

def update_eval_button_state(model1, model1_shots, model1_regex, model1_flash, 
                            model2, model2_shots, model2_regex, model2_flash):
    """
    Checks model configurations and updates the error message visibility and eval button state.
    """
    is_valid, error_msg = validate_model_configs(
        model1, model1_shots, model1_regex, model1_flash,
        model2, model2_shots, model2_regex, model2_flash
    )
    
    if model1 == "(Select Model)" or model2 == "(Select Model)":
        return gr.update(visible=False), gr.update(interactive=False)
    
    if not is_valid:
        return gr.update(visible=True, value=error_msg), gr.update(interactive=False)
    
    return gr.update(visible=False), gr.update(interactive=True)

def get_model_configs(model1, model1_shots, model1_regex, model1_flash,
                     model2, model2_shots, model2_regex, model2_flash):
    """
    Returns the model configurations as structured data for the evaluation function.
    """
    return {
        "model1": {
            "name": model1,
            "shots": model1_shots,
            "regex": model1_regex,
            "flash_attention": model1_flash
        },
        "model2": {
            "name": model2,
            "shots": model2_shots,
            "regex": model2_regex,
            "flash_attention": model2_flash
        }
    }