Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
# ------------------------------
|
2 |
-
# Enhanced NeuroResearch AI System
|
3 |
# ------------------------------
|
4 |
import logging
|
5 |
import os
|
@@ -7,10 +7,12 @@ import re
|
|
7 |
import hashlib
|
8 |
import json
|
9 |
import time
|
|
|
10 |
from datetime import datetime
|
|
|
11 |
from concurrent.futures import ThreadPoolExecutor, as_completed
|
12 |
from typing import List, Dict, Any, Optional, Sequence
|
13 |
-
|
14 |
import requests
|
15 |
import streamlit as st
|
16 |
|
@@ -25,11 +27,14 @@ from langgraph.graph.message import add_messages
|
|
25 |
from typing_extensions import TypedDict, Annotated
|
26 |
from langchain.tools.retriever import create_retriever_tool
|
27 |
|
|
|
|
|
|
|
28 |
# ------------------------------
|
29 |
# Logging Configuration
|
30 |
# ------------------------------
|
31 |
logging.basicConfig(
|
32 |
-
level=logging.INFO,
|
33 |
format="%(asctime)s [%(levelname)s] %(message)s"
|
34 |
)
|
35 |
logger = logging.getLogger(__name__)
|
@@ -53,11 +58,11 @@ class ResearchConfig:
|
|
53 |
MAX_CONCURRENT_REQUESTS = 5
|
54 |
EMBEDDING_DIMENSIONS = 1536
|
55 |
DOCUMENT_MAP = {
|
56 |
-
"Research Report: Results of a New AI Model Improving Image Recognition Accuracy to 98%":
|
57 |
"CV-Transformer Hybrid Architecture",
|
58 |
-
"Academic Paper Summary: Why Transformers Became the Mainstream Architecture in Natural Language Processing":
|
59 |
"Transformer Architecture Analysis",
|
60 |
-
"Latest Trends in Machine Learning Methods Using Quantum Computing":
|
61 |
"Quantum ML Frontiers"
|
62 |
}
|
63 |
ANALYSIS_TEMPLATE = (
|
@@ -85,7 +90,7 @@ if not ResearchConfig.DEEPSEEK_API_KEY:
|
|
85 |
# ------------------------------
|
86 |
class QuantumDocumentManager:
|
87 |
"""
|
88 |
-
Manages
|
89 |
"""
|
90 |
def __init__(self) -> None:
|
91 |
try:
|
@@ -197,7 +202,7 @@ class CognitiveProcessor:
|
|
197 |
|
198 |
def process_query(self, prompt: str) -> Dict:
|
199 |
"""
|
200 |
-
|
201 |
"""
|
202 |
futures = []
|
203 |
for _ in range(3): # Triple redundancy for reliability
|
@@ -254,7 +259,6 @@ class CognitiveProcessor:
|
|
254 |
if not valid_results:
|
255 |
logger.error("All API requests failed.")
|
256 |
return {"error": "All API requests failed"}
|
257 |
-
# Choose the response with the longest content
|
258 |
return max(valid_results, key=lambda x: len(x.get('choices', [{}])[0].get('message', {}).get('content', '')))
|
259 |
|
260 |
# ------------------------------
|
@@ -291,14 +295,16 @@ class ResearchWorkflow:
|
|
291 |
|
292 |
def ingest_query(self, state: AgentState) -> Dict:
|
293 |
"""
|
294 |
-
Ingests the research query.
|
295 |
"""
|
296 |
try:
|
297 |
query = state["messages"][-1].content
|
|
|
|
|
298 |
logger.info("Query ingested.")
|
299 |
return {
|
300 |
"messages": [AIMessage(content="Query ingested successfully")],
|
301 |
-
"context":
|
302 |
"metadata": {"timestamp": datetime.now().isoformat()}
|
303 |
}
|
304 |
except Exception as e:
|
@@ -314,7 +320,7 @@ class ResearchWorkflow:
|
|
314 |
logger.info(f"Retrieved {len(docs)} documents for query.")
|
315 |
return {
|
316 |
"messages": [AIMessage(content=f"Retrieved {len(docs)} documents")],
|
317 |
-
"context": {"documents": docs, "retrieval_time": time.time()}
|
318 |
}
|
319 |
except Exception as e:
|
320 |
return self._error_state(f"Retrieval Error: {str(e)}")
|
@@ -333,7 +339,7 @@ class ResearchWorkflow:
|
|
333 |
logger.info("Content analysis completed.")
|
334 |
return {
|
335 |
"messages": [AIMessage(content=response.get('choices', [{}])[0].get('message', {}).get('content', ''))],
|
336 |
-
"context": {"analysis": response}
|
337 |
}
|
338 |
except Exception as e:
|
339 |
return self._error_state(f"Analysis Error: {str(e)}")
|
@@ -351,18 +357,18 @@ class ResearchWorkflow:
|
|
351 |
response = self.processor.process_query(validation_prompt)
|
352 |
logger.info("Output validation completed.")
|
353 |
return {
|
354 |
-
"messages": [
|
355 |
-
AIMessage(
|
356 |
-
content=analysis +
|
357 |
-
f"\n\nValidation: {response.get('choices', [{}])[0].get('message', {}).get('content', '')}"
|
358 |
-
)
|
359 |
-
]
|
360 |
}
|
361 |
|
362 |
def refine_results(self, state: AgentState) -> Dict:
|
363 |
"""
|
364 |
Refines the analysis report if validation fails.
|
|
|
365 |
"""
|
|
|
|
|
|
|
|
|
366 |
refinement_prompt = (
|
367 |
f"Refine this analysis:\n{state['messages'][-1].content}\n\n"
|
368 |
"Improve:\n1. Technical precision\n2. Empirical grounding\n3. Theoretical coherence"
|
@@ -370,18 +376,19 @@ class ResearchWorkflow:
|
|
370 |
response = self.processor.process_query(refinement_prompt)
|
371 |
logger.info("Refinement completed.")
|
372 |
return {
|
373 |
-
"messages": [
|
374 |
-
AIMessage(
|
375 |
-
content=response.get('choices', [{}])[0].get('message', {}).get('content', '')
|
376 |
-
)
|
377 |
-
],
|
378 |
"context": state["context"]
|
379 |
}
|
380 |
|
381 |
def _quality_check(self, state: AgentState) -> str:
|
382 |
"""
|
383 |
Checks whether the analysis report is valid.
|
|
|
384 |
"""
|
|
|
|
|
|
|
|
|
385 |
content = state["messages"][-1].content
|
386 |
quality = "valid" if "VALID" in content else "invalid"
|
387 |
logger.info(f"Quality check returned: {quality}")
|
|
|
1 |
# ------------------------------
|
2 |
+
# Enhanced NeuroResearch AI System with Refinement Counter
|
3 |
# ------------------------------
|
4 |
import logging
|
5 |
import os
|
|
|
7 |
import hashlib
|
8 |
import json
|
9 |
import time
|
10 |
+
import sys
|
11 |
from datetime import datetime
|
12 |
+
import chromdb
|
13 |
from concurrent.futures import ThreadPoolExecutor, as_completed
|
14 |
from typing import List, Dict, Any, Optional, Sequence
|
15 |
+
|
16 |
import requests
|
17 |
import streamlit as st
|
18 |
|
|
|
27 |
from typing_extensions import TypedDict, Annotated
|
28 |
from langchain.tools.retriever import create_retriever_tool
|
29 |
|
30 |
+
# Optionally increase the recursion limit if needed
|
31 |
+
sys.setrecursionlimit(100)
|
32 |
+
|
33 |
# ------------------------------
|
34 |
# Logging Configuration
|
35 |
# ------------------------------
|
36 |
logging.basicConfig(
|
37 |
+
level=logging.INFO,
|
38 |
format="%(asctime)s [%(levelname)s] %(message)s"
|
39 |
)
|
40 |
logger = logging.getLogger(__name__)
|
|
|
58 |
MAX_CONCURRENT_REQUESTS = 5
|
59 |
EMBEDDING_DIMENSIONS = 1536
|
60 |
DOCUMENT_MAP = {
|
61 |
+
"Research Report: Results of a New AI Model Improving Image Recognition Accuracy to 98%":
|
62 |
"CV-Transformer Hybrid Architecture",
|
63 |
+
"Academic Paper Summary: Why Transformers Became the Mainstream Architecture in Natural Language Processing":
|
64 |
"Transformer Architecture Analysis",
|
65 |
+
"Latest Trends in Machine Learning Methods Using Quantum Computing":
|
66 |
"Quantum ML Frontiers"
|
67 |
}
|
68 |
ANALYSIS_TEMPLATE = (
|
|
|
90 |
# ------------------------------
|
91 |
class QuantumDocumentManager:
|
92 |
"""
|
93 |
+
Manages creation of Chroma collections from raw document texts.
|
94 |
"""
|
95 |
def __init__(self) -> None:
|
96 |
try:
|
|
|
202 |
|
203 |
def process_query(self, prompt: str) -> Dict:
|
204 |
"""
|
205 |
+
Processes a query by sending multiple API requests in parallel.
|
206 |
"""
|
207 |
futures = []
|
208 |
for _ in range(3): # Triple redundancy for reliability
|
|
|
259 |
if not valid_results:
|
260 |
logger.error("All API requests failed.")
|
261 |
return {"error": "All API requests failed"}
|
|
|
262 |
return max(valid_results, key=lambda x: len(x.get('choices', [{}])[0].get('message', {}).get('content', '')))
|
263 |
|
264 |
# ------------------------------
|
|
|
295 |
|
296 |
def ingest_query(self, state: AgentState) -> Dict:
|
297 |
"""
|
298 |
+
Ingests the research query and initializes the refinement counter.
|
299 |
"""
|
300 |
try:
|
301 |
query = state["messages"][-1].content
|
302 |
+
# Initialize context with raw query and refinement counter
|
303 |
+
new_context = {"raw_query": query, "refine_count": 0}
|
304 |
logger.info("Query ingested.")
|
305 |
return {
|
306 |
"messages": [AIMessage(content="Query ingested successfully")],
|
307 |
+
"context": new_context,
|
308 |
"metadata": {"timestamp": datetime.now().isoformat()}
|
309 |
}
|
310 |
except Exception as e:
|
|
|
320 |
logger.info(f"Retrieved {len(docs)} documents for query.")
|
321 |
return {
|
322 |
"messages": [AIMessage(content=f"Retrieved {len(docs)} documents")],
|
323 |
+
"context": {"documents": docs, "retrieval_time": time.time(), "refine_count": state["context"].get("refine_count", 0)}
|
324 |
}
|
325 |
except Exception as e:
|
326 |
return self._error_state(f"Retrieval Error: {str(e)}")
|
|
|
339 |
logger.info("Content analysis completed.")
|
340 |
return {
|
341 |
"messages": [AIMessage(content=response.get('choices', [{}])[0].get('message', {}).get('content', ''))],
|
342 |
+
"context": {"analysis": response, "refine_count": state["context"].get("refine_count", 0)}
|
343 |
}
|
344 |
except Exception as e:
|
345 |
return self._error_state(f"Analysis Error: {str(e)}")
|
|
|
357 |
response = self.processor.process_query(validation_prompt)
|
358 |
logger.info("Output validation completed.")
|
359 |
return {
|
360 |
+
"messages": [AIMessage(content=analysis + f"\n\nValidation: {response.get('choices', [{}])[0].get('message', {}).get('content', '')}")]
|
|
|
|
|
|
|
|
|
|
|
361 |
}
|
362 |
|
363 |
def refine_results(self, state: AgentState) -> Dict:
|
364 |
"""
|
365 |
Refines the analysis report if validation fails.
|
366 |
+
Increments the refinement counter to limit infinite loops.
|
367 |
"""
|
368 |
+
current_count = state["context"].get("refine_count", 0)
|
369 |
+
state["context"]["refine_count"] = current_count + 1
|
370 |
+
logger.info(f"Refinement iteration: {state['context']['refine_count']}")
|
371 |
+
|
372 |
refinement_prompt = (
|
373 |
f"Refine this analysis:\n{state['messages'][-1].content}\n\n"
|
374 |
"Improve:\n1. Technical precision\n2. Empirical grounding\n3. Theoretical coherence"
|
|
|
376 |
response = self.processor.process_query(refinement_prompt)
|
377 |
logger.info("Refinement completed.")
|
378 |
return {
|
379 |
+
"messages": [AIMessage(content=response.get('choices', [{}])[0].get('message', {}).get('content', ''))],
|
|
|
|
|
|
|
|
|
380 |
"context": state["context"]
|
381 |
}
|
382 |
|
383 |
def _quality_check(self, state: AgentState) -> str:
|
384 |
"""
|
385 |
Checks whether the analysis report is valid.
|
386 |
+
Forces a valid state if the refinement count exceeds a threshold.
|
387 |
"""
|
388 |
+
refine_count = state["context"].get("refine_count", 0)
|
389 |
+
if refine_count >= 3:
|
390 |
+
logger.warning("Refinement limit reached. Forcing valid outcome to prevent infinite recursion.")
|
391 |
+
return "valid"
|
392 |
content = state["messages"][-1].content
|
393 |
quality = "valid" if "VALID" in content else "invalid"
|
394 |
logger.info(f"Quality check returned: {quality}")
|