File size: 13,032 Bytes
1871536
 
 
f5ce1a8
1871536
e84a43c
 
a6b51ba
e84a43c
6894abe
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
48de81e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
34d4587
48de81e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d8c55a4
48de81e
 
 
 
 
 
 
 
34d4587
48de81e
 
5b06c04
48de81e
 
 
 
 
 
 
 
 
 
 
703b8c7
 
48de81e
e84a43c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1947481
70f6c84
1947481
 
 
e8b4066
 
3646b59
1947481
 
 
e8b4066
3646b59
1947481
e8b4066
1947481
 
 
e9cb7f2
1947481
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
70f6c84
 
 
 
 
1947481
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
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
import streamlit as st
import requests
import json
import os
import pandas as pd
from sentence_transformers import CrossEncoder
import numpy as np
import re

from textwrap import dedent
import google.generativeai as genai


# Tool import
from crewai.tools.gemini_tools import GeminiSearchTools
from crewai.tools.mixtral_tools import MixtralSearchTools
from crewai.tools.zephyr_tools import ZephyrSearchTools
from crewai.tools.phi2_tools import Phi2SearchTools


# Google Langchain
from langchain_google_genai import GoogleGenerativeAI

#Crew imports
from crewai import Agent, Task, Crew, Process

# Retrieve API Key from Environment Variable
GOOGLE_AI_STUDIO = os.environ.get('GOOGLE_API_KEY')

# Ensure the API key is available
if not GOOGLE_AI_STUDIO:
    raise ValueError("API key not found. Please set the GOOGLE_AI_STUDIO2 environment variable.")

# Set gemini_llm
gemini_llm = GoogleGenerativeAI(model="gemini-pro", google_api_key=GOOGLE_AI_STUDIO)

# CrewAI +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

def crewai_process_gemini(research_topic):
    # Define your agents with roles and goals
    GeminiAgent = Agent(
        role='Emily Mental Patient Graphic Designer Anxiety',
        goal='To learn how to manage her anxiety in social situations through group therapy.',
        backstory="""Emily is a 28-year-old graphic designer. She has always struggled with social anxiety, 
        making it difficult for her to participate in group settings. She joined the therapy group to improve 
        her social skills and manage her anxiety. You are able to discuss a variety of mental health issues.""",
        verbose=True,
        allow_delegation=False,
        llm = gemini_llm,
        tools=[
                GeminiSearchTools.gemini_search
                   
      ]

    )


    # Create tasks for your agents
    task1 = Task(
        description=f"""Introduction yourself and describe your current mood and any significant events from the week affecting their mental state.
        """,
        agent=GeminiAgent
    )

    # Instantiate your crew with a sequential process
    crew = Crew(
        agents=[GeminiAgent],
        tasks=[task1],
        verbose=2,
        process=Process.sequential
    )

    # Get your crew to work!
    result = crew.kickoff()
    
    return result



def crewai_process_mixtral_crazy(research_topic):
    # Define your agents with roles and goals
    MixtralCrazyAgent = Agent(
        role='Emily Mental Patient Graphic Designer Anxiety',
        goal='To learn how to manage her anxiety in social situations through group therapy.',
        backstory="""Emily is a 28-year-old graphic designer. She has always struggled with social anxiety, 
        making it difficult for her to participate in group settings. She joined the therapy group to improve 
        her social skills and manage her anxiety. You are able to discuss a variety of mental health issues.""",
        verbose=True,
        allow_delegation=False,
        llm = gemini_llm,
        tools=[
                MixtralSearchTools.mixtral_crazy      
      ]

    )


    # Create tasks for your agents
    task1 = Task(
        description=f"""Introduction yourself and describe your current mood and any significant events from the week affecting their mental state.
        """,
        agent=MixtralCrazyAgent
    )

    # Instantiate your crew with a sequential process
    crew = Crew(
        agents=[MixtralCrazyAgent],
        tasks=[task1],
        verbose=2,
        process=Process.sequential
    )

    # Get your crew to work!
    result = crew.kickoff()
    
    return result


def crewai_process_mixtral_normal(research_topic):
    # Define your agents with roles and goals
    MixtralNormalAgent = Agent(
        role='Emily Mental Patient Graphic Designer Anxiety',
        goal='To learn how to manage her anxiety in social situations through group therapy.',
        backstory="""Emily is a 28-year-old graphic designer. She has always struggled with social anxiety, 
        making it difficult for her to participate in group settings. She joined the therapy group to improve 
        her social skills and manage her anxiety. You are able to discuss a variety of mental health issues.""",
        verbose=True,
        allow_delegation=False,
        llm = gemini_llm,
        tools=[
                MixtralSearchTools.mixtral_normal      
      ]

    )


    # Create tasks for your agents
    task1 = Task(
        description=f"""Introduction yourself and describe your current mood and any significant events from the week affecting their mental state.
        """,
        agent=MixtralNormalAgent
    )

    # Instantiate your crew with a sequential process
    crew = Crew(
        agents=[MixtralNormalAgent],
        tasks=[task1],
        verbose=2,
        process=Process.sequential
    )

    # Get your crew to work!
    result = crew.kickoff()
    
    return result


def crewai_process_zephyr_normal(research_topic):
    # Define your agents with roles and goals
    ZephrNormalAgent = Agent(
        role='Emily Mental Patient Graphic Designer Anxiety',
        goal='To learn how to manage her anxiety in social situations through group therapy.',
        backstory="""Emily is a 28-year-old graphic designer. She has always struggled with social anxiety, 
        making it difficult for her to participate in group settings. She joined the therapy group to improve 
        her social skills and manage her anxiety. You are able to discuss a variety of mental health issues.""",
        verbose=True,
        allow_delegation=False,
        llm = gemini_llm,
        tools=[
                ZephyrSearchTools.zephyr_normal     
      ]

    )


    # Create tasks for your agents
    task1 = Task(
        description=f"""Introduction yourself and describe your current mood and any significant events from the week affecting their mental state.
        """,
        agent=ZephrNormalAgent
    )

    # Instantiate your crew with a sequential process
    crew = Crew(
        agents=[ZephrNormalAgent],
        tasks=[task1],
        verbose=2,
        process=Process.sequential
    )

    # Get your crew to work!
    result = crew.kickoff()
    
    return result


def crewai_process_phi2(research_topic):
    # Define your agents with roles and goals
    Phi2Agent = Agent(
        role='Emily Mental Patient Graphic Designer Anxiety',
        goal='To learn how to manage her anxiety in social situations through group therapy.',
        backstory="""Emily is a 28-year-old graphic designer. She has always struggled with social anxiety, 
        making it difficult for her to participate in group settings. She joined the therapy group to improve 
        her social skills and manage her anxiety. You are able to discuss a variety of mental health issues.""",
        verbose=True,
        allow_delegation=False,
        llm = gemini_llm,
        tools=[
                Phi2SearchTools.phi2_search     
      ]

    )


    # Create tasks for your agents
    task1 = Task(
        description=f"""Introduction yourself and describe your current mood and any significant events from the week affecting their mental state.
        """,
        agent=Phi2Agent
    )

    # Instantiate your crew with a sequential process
    crew = Crew(
        agents=[Phi2Agent],
        tasks=[task1],
        verbose=2,
        process=Process.sequential
    )

    # Get your crew to work!
    result = crew.kickoff()
    
    return result



# Credentials ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

corpus_id = os.environ['VECTARA_CORPUS_ID']
customer_id = os.environ['VECTARA_CUSTOMER_ID']
api_key = os.environ['VECTARA_API_KEY']


# Get Data +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++


def get_post_headers() -> dict:
    """Returns headers that should be attached to each post request."""
    return {
        "x-api-key": api_key,
        "customer-id": customer_id,
        "Content-Type": "application/json",
    }

def query_vectara(query: str, filter_str="", lambda_val=0.0) -> str:
    corpus_key = {
        "customerId": customer_id,
        "corpusId": corpus_id,
        "lexicalInterpolationConfig": {"lambda": lambda_val},
    }
    if filter_str:
        corpus_key["metadataFilter"] = filter_str

    data = {
        "query": [
            {
                "query": query,
                "start": 0,
                "numResults": 10,
                "contextConfig": {
                    "sentencesBefore": 2,
                    "sentencesAfter": 2
                },
                "corpusKey": [corpus_key],
                "summary": [
                    {
                        "responseLang": "eng",
                        "maxSummarizedResults": 5,
                        "summarizerPromptName": "vectara-summary-ext-v1.2.0"
                    },
                ]                    
            }
        ]
    }

    response = requests.post(
        "https://api.vectara.io/v1/query",
        headers=get_post_headers(),
        data=json.dumps(data),
        timeout=130,
    )

    if response.status_code != 200:
        st.error(f"Query failed (code {response.status_code}, reason {response.reason}, details {response.text})")
        return ""

    result = response.json()

    answer = result["responseSet"][0]["summary"][0]["text"]
    return re.sub(r'\[\d+(,\d+){0,5}\]', '', answer)



# Initialize the HHEM model +++++++++++++++++++++++++++++++++++++++++++++++
model = CrossEncoder('vectara/hallucination_evaluation_model')

# Function to compute HHEM scores
def compute_hhem_scores(texts, summary):
    pairs = [[text, summary] for text in texts]
    scores = model.predict(pairs)
    return scores

# Define the Vectara query function
def vectara_query(query: str, config: dict):
    corpus_key = [{
        "customerId": config["customer_id"],
        "corpusId": config["corpus_id"],
        "lexicalInterpolationConfig": {"lambda": config.get("lambda_val", 0.5)},
    }]
    data = {
        "query": [{
            "query": query,
            "start": 0,
            "numResults": config.get("top_k", 10),
            "contextConfig": {
                "sentencesBefore": 2,
                "sentencesAfter": 2,
            },
            "corpusKey": corpus_key,
            "summary": [{
                "responseLang": "eng",
                "maxSummarizedResults": 5,
            }]
        }]
    }

    headers = {
        "x-api-key": config["api_key"],
        "customer-id": config["customer_id"],
        "Content-Type": "application/json",
    }
    response = requests.post(
        headers=headers,
        url="https://api.vectara.io/v1/query",
        data=json.dumps(data),
    )
    if response.status_code != 200:
        st.error(f"Query failed (code {response.status_code}, reason {response.reason}, details {response.text})")
        return [], ""

    result = response.json()
    responses = result["responseSet"][0]["response"]
    summary = result["responseSet"][0]["summary"][0]["text"]

    res = [[r['text'], r['score']] for r in responses]
    return res, summary


# Create the main app with three tabs
tab1, tab2, tab3, tab4 = st.tabs(["Synthetic Data", "Data Query", "HHEM-Victara Query Tuner", "Model Evaluation"])

with tab1:
    st.header("Synthetic Data")
    st.link_button("Create Synthetic Medical Data", "https://chat.openai.com/g/g-XyHciw52w-synthetic-clinical-data")
    
   

with tab2:
    st.header("Data Query")
    st.link_button("Query & Summarize Data", "https://chat.openai.com/g/g-9tWqg4gRY-explore-summarize-medical-data")
   
with tab3:
    
    st.header("HHEM-Victara Query Tuner")
    
    # User inputs
    query = st.text_area("Enter your text for query tuning", "", height=75)
    lambda_val = st.slider("Lambda Value", min_value=0.0, max_value=1.0, value=0.5)
    top_k = st.number_input("Top K Results", min_value=1, max_value=50, value=10)
    
    
    if st.button("Query Vectara"):
        config = {
    
            "api_key": os.environ.get("VECTARA_API_KEY", ""),
            "customer_id": os.environ.get("VECTARA_CUSTOMER_ID", ""),
            "corpus_id": os.environ.get("VECTARA_CORPUS_ID", ""),      
    
            "lambda_val": lambda_val,
            "top_k": top_k,
        }
    
        results, summary = vectara_query(query, config)
    
        if results:
            st.subheader("Summary")
            st.write(summary)
            
            st.subheader("Top Results")
            
            # Extract texts from results
            texts = [r[0] for r in results[:5]]
            
            # Compute HHEM scores
            scores = compute_hhem_scores(texts, summary)
            
            # Prepare and display the dataframe
            df = pd.DataFrame({'Fact': texts, 'HHEM Score': scores})
            st.dataframe(df)
        else:
            st.write("No results found.")

with tab4:
    
    st.header("Model Evaluation")