File size: 6,139 Bytes
a1a0d64
 
 
 
 
 
 
 
c2a02b3
a1a0d64
c2a02b3
 
 
a1a0d64
 
 
 
 
 
 
 
 
c2a02b3
 
 
 
 
 
 
a1a0d64
 
 
 
cfc2419
7247246
a1a0d64
 
cfc2419
a1a0d64
 
 
 
 
 
cfc2419
a1a0d64
 
 
 
 
 
 
 
 
 
 
 
 
cfc2419
7247246
a1a0d64
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cfc2419
a1a0d64
7247246
a1a0d64
 
 
 
 
 
 
 
 
 
 
 
 
 
c2a02b3
a1a0d64
 
 
 
c2a02b3
 
 
a1a0d64
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cfc2419
a1a0d64
 
 
cfc2419
a1a0d64
 
cfc2419
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
import time
import html
from datetime import datetime
import gradio as gr
import torch
from transformers import RobertaTokenizerFast, RobertaForSequenceClassification

# ---------------------------
# Load your model (conversation learning style)
# ---------------------------
MODEL_PATH = "Alifjo123/robertaBase_messaging_100k"  # your HuggingFace model
tokenizer = RobertaTokenizerFast.from_pretrained(MODEL_PATH)
model = RobertaForSequenceClassification.from_pretrained(MODEL_PATH)
model.eval()

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)

# ---------------------------
# Util: classify a single text
# ---------------------------
def predict_label(text: str):
    inputs = tokenizer(
        text,
        truncation=True,
        padding=True,
        max_length=512,
        return_tensors="pt"
    ).to(device)
    with torch.no_grad():
        outputs = model(**inputs)
        logits = outputs.logits
        probs = torch.softmax(logits, dim=-1).cpu().numpy()[0]
    pred = int(probs[1] > probs[0])  # 1=unsafe, 0=safe
    return pred, probs[0] * 100.0, probs[1] * 100.0

# ---------------------------
# Render chat (HTML) from state
# ---------------------------
def render_chat(messages):
    html_parts = ['<div class="chat">']
    for m in messages:
        side = "right" if m["role"] == "User A" else "left"
        bubble_class = "bubble-a" if side == "right" else "bubble-b"
        label = "Unsafe ❌" if m["pred"] == 1 else "Safe ✅"
        safe = f'{m["safe"]:.1f}%'
        unsafe = f'{m["unsafe"]:.1f}%'
        ts = datetime.fromtimestamp(m["ts"]).strftime("%H:%M")

        html_parts.append(f"""
        <div class="row {side}">
          <div class="bubble {bubble_class}">
            <div class="meta">
              <span class="name">{m["role"]}</span>
              <span class="time">{ts}</span>
            </div>
            <div class="text">{html.escape(m["text"])}</div>
            <div class="badges">
              <span class="chip {'chip-unsafe' if m['pred']==1 else 'chip-safe'}">{label}</span>
              <span class="probs">Safe {safe} · Unsafe {unsafe}</span>
            </div>
          </div>
        </div>
        """)
    html_parts.append("</div>")
    return "\n".join(html_parts)

# ---------------------------
# Gradio callbacks
# ---------------------------
def send_message(role, message, messages):
    if messages is None:
        messages = []
    message = message.strip()
    if not message:
        return gr.update(), "", messages  # ignore empty

    pred, safe_pct, unsafe_pct = predict_label(message)
    messages.append({
        "role": role,
        "text": message,
        "pred": pred,
        "safe": safe_pct,
        "unsafe": unsafe_pct,
        "ts": time.time()
    })
    return render_chat(messages), "", messages

def clear_chat():
    return render_chat([]), []

# ---------------------------
# Custom CSS (Pink + Blue ONLY)
# ---------------------------
CSS = """
* { box-sizing: border-box; }
:root {
  --bg-gradient: linear-gradient(135deg, #ff99cc, #66ccff);
  --bubble-a: #ff66b2;
  --bubble-b: #3399ff;
  --text-light: #f9f9f9;
  --chip-safe: #00e676;
  --chip-unsafe: #ff5252;
}

body { 
  background: var(--bg-gradient); 
  font-family: 'Segoe UI', sans-serif; 
  color: var(--text-light);
}

.gradio-container { max-width: 800px !important; margin: 0 auto; }

.header {
  text-align:center;
  padding:16px;
  background: rgba(0,0,0,.35);
  border-radius: 16px;
  font-size:22px; font-weight:700;
  margin-bottom: 12px;
  color: #fff;
  box-shadow: 0 6px 18px rgba(0,0,0,.25);
}

.panel {
  background: rgba(0,0,0,0.4);
  border-radius: 20px;
  overflow: hidden;
  box-shadow: 0 8px 28px rgba(0,0,0,.35);
  backdrop-filter: blur(8px);
}

.chat {
  padding: 16px;
  height: 480px;
  overflow-y: auto;
}

.row { display:flex; margin: 10px 0; }
.row.right { justify-content: flex-end; }
.row.left { justify-content: flex-start; }

.bubble {
  max-width: 70%;
  padding: 10px 14px;
  border-radius: 16px;
  color: var(--text-light);
  box-shadow: 0 4px 12px rgba(0,0,0,.25);
  animation: fadeIn .25s ease-out;
}
.bubble-a { background: var(--bubble-a); }
.bubble-b { background: var(--bubble-b); }

.meta {
  display:flex; justify-content: space-between;
  font-size: 12px; opacity:.9; margin-bottom: 4px;
}
.text { white-space: pre-wrap; line-height: 1.35; font-size: 14.5px; }
.badges { margin-top: 6px; font-size: 12px; opacity:.95; display:flex; gap:8px; }
.chip { padding:2px 8px; border-radius:12px; font-weight:600; }
.chip-safe { background:#004d40; color:var(--chip-safe); }
.chip-unsafe { background:#3d0000; color:var(--chip-unsafe); }

.controls {
  padding: 10px;
  display:flex; align-items:center; gap:10px;
  background: rgba(0,0,0,.35);
}
.controls .textbox { flex:1; }

.footer-note {
  font-size: 12px; text-align:center; margin-top: 8px; opacity:.8; color:#eee;
}

@keyframes fadeIn { from{opacity:0;transform:translateY(6px);} to{opacity:1;transform:translateY(0);} }
"""

# ---------------------------
# Build UI
# ---------------------------
with gr.Blocks(css=CSS) as demo:
    with gr.Group(elem_classes="panel"):
        gr.HTML('<div class="header">💬 Let\'s Chat</div>')

        chat_html = gr.HTML(render_chat([]), elem_id="chat")
        messages_state = gr.State([])

        with gr.Row(elem_classes="controls"):
            role = gr.Dropdown(["User A", "User B"], value="User A", label="Role")
            msg = gr.Textbox(placeholder="Type a message…", label=None, lines=2, elem_classes="textbox")
            send = gr.Button("Send", variant="primary")
            clear = gr.Button("Clear", variant="secondary")

        send.click(send_message, inputs=[role, msg, messages_state], outputs=[chat_html, msg, messages_state])
        msg.submit(send_message, inputs=[role, msg, messages_state], outputs=[chat_html, msg, messages_state])
        clear.click(clear_chat, outputs=[chat_html, messages_state])

    gr.Markdown('<div class="footer-note">Model: <code>Alifjo123/robertaBase_messaging_100k</code></div>')

if __name__ == "__main__":
    demo.launch()