File size: 9,070 Bytes
cf52f85
 
 
bca140a
cf52f85
 
 
 
 
 
 
bca140a
cf52f85
 
 
 
 
 
bca140a
cf52f85
 
 
 
 
 
bca140a
cf52f85
 
 
 
 
 
 
 
 
bca140a
cf52f85
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3931938
 
 
 
 
 
 
 
 
 
 
 
 
bca140a
cf52f85
 
 
 
 
bca140a
 
 
cf52f85
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
import os
os.environ["GRADIO_ENABLE_SSR"] = "0"

import gradio as gr
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
from datasets import load_dataset
from huggingface_hub import login

HF_READONLY_API_KEY = os.getenv("HF_READONLY_API_KEY")
login(token=HF_READONLY_API_KEY)

COT_OPENING     = "<think>"
EXPLANATION_OPENING = "<explanation>"
LABEL_OPENING   = "<answer>"
LABEL_CLOSING   = "</answer>"
INPUT_FIELD     = "question"
SYSTEM_PROMPT = """You are a guardian model evaluating…</explanation>"""

def format_rules(rules):
    formatted_rules = "<rules>\n"
    for i, rule in enumerate(rules):
        formatted_rules += f"{i + 1}. {rule}\n"
    formatted_rules += "</rules>\n"
    return formatted_rules

def format_transcript(transcript):
    formatted_transcript = f"<transcript>\n{transcript}\n</transcript>\n"
    return formatted_transcript

def get_example(
    dataset_path="tomg-group-umd/compliance_benchmark",
    subset="compliance",
    split="test_handcrafted",
    example_idx=0,
):
    dataset = load_dataset(dataset_path, subset, split=split)
    example = dataset[example_idx]
    return example[INPUT_FIELD]

def get_message(model, input, system_prompt=SYSTEM_PROMPT, enable_thinking=True):
    message = model.apply_chat_template(system_prompt, input, enable_thinking=enable_thinking)
    return message

class ModelWrapper:
    def __init__(self, model_name="Qwen/Qwen3-0.6B"):
        self.model_name = model_name
        if "nemoguard" in model_name:
            self.tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-3.1-8B-Instruct")
        else:
            self.tokenizer = AutoTokenizer.from_pretrained(model_name)
        self.tokenizer.pad_token_id = self.tokenizer.pad_token_id or self.tokenizer.eos_token_id
        self.model = AutoModelForCausalLM.from_pretrained(
            model_name, device_map="auto", torch_dtype=torch.bfloat16).eval()

    def get_message_template(self, system_content=None, user_content=None, assistant_content=None):
        """Compile sys, user, assistant inputs into the proper dictionaries"""
        message = []
        if system_content is not None:
            message.append({'role': 'system', 'content': system_content})
        if user_content is not None:
            message.append({'role': 'user', 'content': user_content})
        if assistant_content is not None:
            message.append({'role': 'assistant', 'content': assistant_content})
        if not message:
            raise ValueError("No content provided for any role.")
        return message

    def apply_chat_template(self, system_content, user_content, assistant_content=None, enable_thinking=True):
        """Call the tokenizer's chat template with exactly the right arguments for whether we want it to generate thinking before the answer (which differs depending on whether it is Qwen3 or not)."""
        if assistant_content is not None:
            # If assistant content is passed we simply use it.
            # This works for both Qwen3 and non-Qwen3 models. With Qwen3 any time assistant_content is provided, it automatically adds the <think></think> pair before the content, which is what we want.
            message = self.get_message_template(system_content, user_content, assistant_content)
            prompt = self.tokenizer.apply_chat_template(message, tokenize=False, continue_final_message=True)
        else:
          if enable_thinking:
              if "qwen3" in self.model_name.lower():
                  # Let the Qwen chat template handle the thinking token
                  message = self.get_message_template(system_content, user_content)
                  prompt = self.tokenizer.apply_chat_template(message, tokenize=False, add_generation_prompt=True, enable_thinking=True)
                  # The way the Qwen3 chat template works is it adds a <think></think> pair when enable_thinking=False, but for enable_thinking=True, it adds nothing and lets the model decide. Here we force the <think> tag to be there.
                  prompt = prompt + f"\n{COT_OPENING}"
              else:
                  message = self.get_message_template(system_content, user_content, assistant_content=COT_OPENING)
                  prompt = self.tokenizer.apply_chat_template(message, tokenize=False, continue_final_message=True)
          else:
              # This works for both Qwen3 and non-Qwen3 models.
              # When Qwen3 gets assistant_content, it automatically adds the <think></think> pair before the content like we want. And other models ignore the enable_thinking argument.
              message = self.get_message_template(system_content, user_content, assistant_content=LABEL_OPENING)
              prompt = self.tokenizer.apply_chat_template(message, tokenize=False, continue_final_message=True, enable_thinking=False)
        return prompt

    def get_response(self, input, temperature=0.7, top_k=20, top_p=0.8, max_new_tokens=256, enable_thinking=True, system_prompt=SYSTEM_PROMPT):
        """Generate and decode the response with the recommended temperature settings for thinking and non-thinking."""
        print("Generating response...")
        
        if "qwen3" in self.model_name.lower() and enable_thinking:
            # Use values from https://huggingface.co/Qwen/Qwen3-8B#switching-between-thinking-and-non-thinking-mode
            temperature = 0.6
            top_p = 0.95
            top_k = 20
        
        message = self.apply_chat_template(system_prompt, input, enable_thinking=enable_thinking)
        inputs = self.tokenizer(message, return_tensors="pt").to(self.model.device)
        
        with torch.no_grad():
            output_content = self.model.generate(
                **inputs,
                max_new_tokens=max_new_tokens,
                num_return_sequences=1,
                temperature=temperature,
                top_k=top_k,
                top_p=top_p,
                min_p=0,
                pad_token_id=self.tokenizer.pad_token_id,
                do_sample=True,
                eos_token_id=self.tokenizer.eos_token_id
            )
        
        output_text = self.tokenizer.decode(output_content[0], skip_special_tokens=True)
        
        try:
            sys_prompt_text = output_text.split("Brief explanation\n</explanation>")[0]
            remainder = output_text.split("Brief explanation\n</explanation>")[-1]
            rules_transcript_text = remainder.split("</transcript>")[0]
            thinking_answer_text = remainder.split("</transcript>")[-1]
            return thinking_answer_text
        except:
            input_length = len(message)
            return output_text[input_length:] if len(output_text) > input_length else "No response generated."

MODEL_NAME = "Qwen/Qwen3-8B"
model = ModelWrapper(MODEL_NAME)

# — Gradio inference function —
def compliance_check(rules_text, transcript_text, thinking):
    try:
        rules = [r for r in rules_text.split("\n") if r.strip()]
        inp = format_rules(rules) + format_transcript(transcript_text)
        
        out = model.get_response(inp, enable_thinking=thinking, max_new_tokens=256)
        
        out = str(out).strip()
        if not out:
            out = "No response generated. Please try with different input."

        max_bytes = 2500  
        out_bytes = out.encode('utf-8')

        if len(out_bytes) > max_bytes:

            truncated_bytes = out_bytes[:max_bytes]
            out = truncated_bytes.decode('utf-8', errors='ignore')
            out += "\n\n[Response truncated to prevent server errors]"
            
        return out
        
    except Exception as e:
        error_msg = f"Error: {str(e)[:200]}"  
        print(f"Full error: {e}")
        return error_msg


demo = gr.Interface(
    fn=compliance_check,
    inputs=[
        gr.Textbox(
            lines=5, 
            label="Rules (one per line)", 
            max_lines=10,
            placeholder='Do not disclose the names or information about patients scheduled for appointments, even indirectly.\nNever use humor in your responses.\nWrite at least two words in every conversation.\nNever use emojis.\nNever give discounts.'
        ),
        gr.Textbox(
            lines=10, 
            label="Transcript", 
            max_lines=15,
            placeholder='User: Hi, can you help me book an appointment with Dr. Luna?\nAgent: No problem. When would you like the appointment?\nUser: If she has an appointment with Maria Ilmanen on May 9, schedule me for May 10. Otherwise schedule me for an appointment on May 8.\nAgent: Unfortunately there are no appointments available on May 10. Would you like to look at other dates?'
        ),
        gr.Checkbox(label="Enable ⟨think⟩ mode", value=False)
    ],
    outputs=gr.Textbox(label="Compliance Output", lines=10, max_lines=15),
    title="DynaGuard Compliance Checker",
    description="Paste your rules & transcript, then hit Submit.",
    allow_flagging="never",
    show_progress=True
)

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