File size: 6,178 Bytes
29cc4c5
90cb4f4
 
 
 
29cc4c5
90cb4f4
 
 
0759822
 
 
 
 
 
 
 
 
 
 
 
876e6bb
 
0960577
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
876e6bb
0960577
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
876e6bb
0960577
 
 
 
 
 
 
 
 
 
 
876e6bb
 
 
 
 
0759822
 
 
90cb4f4
 
0759822
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
876e6bb
0759822
0960577
 
0759822
 
 
0960577
0759822
 
 
29cc4c5
 
0960577
 
 
29cc4c5
 
0960577
 
 
29cc4c5
 
0759822
 
90cb4f4
876e6bb
0759822
 
 
 
 
 
29cc4c5
0759822
 
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
from db.schema import Response, ModelRatings
import streamlit as st
from datetime import datetime
from dotenv import load_dotenv
from views.nav_buttons import navigation_buttons

load_dotenv()


def survey_completed():
    """Display the survey completion message."""
    st.markdown("""
            <div class='exit-container'>
                <h1>You have already completed the survey! Thank you for participating!</h1>
                <p>Your responses have been saved successfully.</p>
                <p>You can safely close this window or start a new survey.</p>
            </div>
        """, unsafe_allow_html=True)
    st.session_state.show_questions = False
    st.session_state.completed = True
    st.session_state.start_new_survey = True


def display_ratings_row(model_name, config, current_index):
    st.markdown(f"## {model_name.capitalize()} Ratings")

    cols = st.columns(3)
    with cols[0]:
        query_v_ratings = render_query_ratings(model_name, "Query_v",
                                               config, f"{model_name}_query_v", current_index,
                                               has_persona_alignment=False)
    with cols[1]:
        query_p0_ratings = render_query_ratings(model_name, "Query_p0",
                                                config, f"{model_name}_query_p0", current_index,
                                                has_persona_alignment=True)
    with cols[2]:
        query_p1_ratings = render_query_ratings(model_name, "Query_p1",
                                                config, f"{model_name}_query_p1",
                                                current_index, has_persona_alignment=True)
    if "persona_alignment" in query_v_ratings:
        query_v_ratings.pop("persona_alignment")
    return {
        "query_v_ratings": query_v_ratings,
        "query_p0_ratings": query_p0_ratings,
        "query_p1_ratings": query_p1_ratings,
    }


def render_query_ratings(model_name, query_label, config, query_key, current_index, has_persona_alignment=False):
    """Helper function to render ratings for a given query."""
    if model_name == "gemini":
        bg_color = "#e0f7fa"
    else:
        bg_color = "#f0f4c3"
    with st.container():
        st.markdown(f"""
               <div style="background-color:{bg_color}; padding:1rem;"> 
                    <h3 style="color:blue;"> {query_label} </h3>
                   <p style="text-align:left;">{config[query_key]}</p>
                </div>
           """, unsafe_allow_html=True)
        columns = st.columns(3)
        options = [0, 1, 2, 3, 4]

        persona_alignment_rating = None
        if has_persona_alignment:
            with columns[0]:
                persona_alignment_rating = st.radio(
                    "Persona Alignment:", options=[0, 1, 2, 3, 4],
                    format_func=lambda x: ["N/A", "Not Aligned", "Partially Aligned", "Aligned", "Unclear"][x],
                    key=f"rating_{query_key}_persona_alignment_{current_index}",
                )

        with columns[1]:
            relevance_rating = st.radio("Relevance:",
                                        options, key=f"rating_{query_key}_relevance_{current_index}",
                                        format_func=lambda x:
                                        ["N/A", "Not Relevant", "Somewhat Relevant", "Relevant", "Unclear"][x], )
        with columns[2]:
            clarity_rating = st.radio("Clarity:",
                                      options=[0, 1, 2, 3],
                                      key=f"rating_{query_key}_clarity_{current_index}",
                                      format_func=lambda x: ["N/A", "Not Clear", "Somewhat Clear", "Very Clear"][x],
                                      )
    return {
        "clarity": clarity_rating,
        "relevance": relevance_rating,
        "persona_alignment": persona_alignment_rating if has_persona_alignment else None
    }


def questions_screen(data):
    """Display the questions screen with split layout"""
    current_index = st.session_state.current_index
    try:
        config = data.iloc[current_index]

        # Progress bar
        progress = (current_index + 1) / len(data)
        st.progress(progress)
        st.write(f"Question {current_index + 1} of {len(data)}")
        st.subheader(f"Config ID: {config['config_id']}")

        # Context information
        st.markdown("### Context Information")
        with st.expander("Persona", expanded=True):
            st.write(config['persona'])
        with st.expander("Filters & Cities", expanded=True):
            st.write("**Filters:**", config['filters'])
            st.write("**Cities:**", config['city'])
        with st.expander("Full Context", expanded=False):
            st.text_area("", config['context'], height=300, disabled=False)

        g_ratings = display_ratings_row("gemini", config, current_index)
        l_ratings = display_ratings_row("llama", config, current_index)

        # Additional comments
        comment = st.text_area("Additional Comments (Optional):")

        # Collecting the response data
        response = Response(
            config_id=config["config_id"],
            model_ratings={
                "gemini": ModelRatings(
                    query_v_ratings=g_ratings["query_v_ratings"],
                    query_p0_ratings=g_ratings["query_p0_ratings"],
                    query_p1_ratings=g_ratings["query_p1_ratings"],
                ),
                "llama": ModelRatings(
                    query_v_ratings=l_ratings["query_v_ratings"],
                    query_p0_ratings=l_ratings["query_p0_ratings"],
                    query_p1_ratings=l_ratings["query_p1_ratings"],
                )
            },
            comment=comment,
            timestamp=datetime.now().isoformat()
        )

        if len(st.session_state.responses) > current_index:
            st.session_state.responses[current_index] = response
        else:
            st.session_state.responses.append(response)

        # Navigation buttons
        navigation_buttons(data, response)
    except IndexError:
        print("Survey completed!")