File size: 11,981 Bytes
7c04f3e
1fa6961
 
 
 
 
 
 
e85946a
7c04f3e
1fa6961
 
 
 
 
 
 
 
 
 
 
 
7c04f3e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1fa6961
b102339
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1fa6961
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b102339
1fa6961
 
 
 
 
 
7c04f3e
 
 
 
 
 
 
 
 
 
 
1fa6961
 
 
 
 
 
 
 
7c04f3e
 
 
 
 
 
 
 
 
 
 
1fa6961
 
 
 
 
 
 
 
7c04f3e
 
 
 
 
 
 
 
 
 
1fa6961
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b102339
1fa6961
b102339
1fa6961
 
 
 
 
b102339
 
1fa6961
 
 
b102339
 
 
 
 
1fa6961
 
b102339
 
 
 
 
 
 
 
1fa6961
 
 
b102339
1fa6961
b102339
1fa6961
 
 
 
b102339
7c04f3e
b102339
1fa6961
7c04f3e
 
 
 
b102339
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7c04f3e
 
 
 
b102339
7c04f3e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b102339
7c04f3e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1fa6961
7c04f3e
1fa6961
7c04f3e
 
 
1fa6961
 
b102339
7c04f3e
 
 
1fa6961
7c04f3e
 
 
 
 
 
b102339
7c04f3e
b102339
7c04f3e
 
 
 
 
 
b102339
7c04f3e
 
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
import os, json, time, random
from dotenv import load_dotenv

# Load environment variables
load_dotenv()

# Imports
from langchain_nvidia_ai_endpoints import NVIDIAEmbeddings
from langchain_groq import ChatGroq
from langchain_nvidia_ai_endpoints import ChatNVIDIA
from langchain_community.tools.tavily_search import TavilySearchResults
from langchain_community.document_loaders import WikipediaLoader
from langchain_community.document_loaders import ArxivLoader
from langchain_community.vectorstores import FAISS
from langchain_core.messages import SystemMessage, HumanMessage
from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
from langchain_core.tools import tool
from langchain.tools.retriever import create_retriever_tool
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain_community.document_loaders import JSONLoader
from langgraph.prebuilt import create_react_agent
from langgraph.checkpoint.memory import MemorySaver
from langchain_core.rate_limiters import InMemoryRateLimiter

# Rate limiters for different providers
groq_rate_limiter = InMemoryRateLimiter(
    requests_per_second=0.5,  # 30 requests per minute
    check_every_n_seconds=0.1,
    max_bucket_size=10
)

google_rate_limiter = InMemoryRateLimiter(
    requests_per_second=0.33,  # 20 requests per minute
    check_every_n_seconds=0.1,
    max_bucket_size=10
)

nvidia_rate_limiter = InMemoryRateLimiter(
    requests_per_second=0.25,  # 15 requests per minute
    check_every_n_seconds=0.1,
    max_bucket_size=10
)

# Initialize individual LLMs
groq_llm = ChatGroq(
    model="llama-3.3-70b-versatile",
    temperature=0,
    api_key=os.getenv("GROQ_API_KEY"),
    rate_limiter=groq_rate_limiter,
    max_retries=2,
    request_timeout=60
)

nvidia_llm = ChatNVIDIA(
    model="meta/llama-3.1-405b-instruct",
    temperature=0,
    api_key=os.getenv("NVIDIA_API_KEY"),
    rate_limiter=nvidia_rate_limiter,
    max_retries=2
)

# Create LLM tools that can be selected by the agent
@tool
def groq_reasoning_tool(query: str) -> str:
    """Use Groq's Llama model for fast reasoning, mathematical calculations, and logical problems.
    Best for: Math problems, logical reasoning, quick calculations, code generation.
    
    Args:
        query: The question or problem to solve
    """
    try:
        time.sleep(random.uniform(1, 2))  # Rate limiting
        response = groq_llm.invoke([HumanMessage(content=query)])
        return f"Groq Response: {response.content}"
    except Exception as e:
        return f"Groq tool failed: {str(e)}"


@tool
def nvidia_specialist_tool(query: str) -> str:
    """Use NVIDIA's large model for specialized tasks, technical questions, and domain expertise.
    Best for: Technical questions, specialized domains, scientific problems, detailed analysis.
    
    Args:
        query: The specialized question or technical problem
    """
    try:
        time.sleep(random.uniform(2, 4))  # Rate limiting
        response = nvidia_llm.invoke([HumanMessage(content=query)])
        return f"NVIDIA Response: {response.content}"
    except Exception as e:
        return f"NVIDIA tool failed: {str(e)}"

# Define calculation tools
@tool
def multiply(a: int | float, b: int | float) -> int | float:
    """Multiply two numbers.
    Args:
        a: first int | float
        b: second int | float
    """
    return a * b

@tool
def add(a: int | float, b: int | float) -> int | float:
    """Add two numbers.
    
    Args:
        a: first int | float
        b: second int | float
    """
    return a + b

@tool
def subtract(a: int | float , b: int | float) -> int | float:
    """Subtract two numbers.
    
    Args:
        a: first int | float
        b: second int | float
    """
    return a - b

@tool
def divide(a: int | float, b: int | float) -> int | float:
    """Divide two numbers.
    
    Args:
        a: first int | float
        b: second int | float
    """
    if b == 0:
        raise ValueError("Cannot divide by zero.")
    return a / b

@tool
def modulus(a: int | float, b: int | float) -> int | float:
    """Get the modulus of two numbers.
    
    Args:
        a: first int | float
        b: second int | float
    """
    return a % b

# Define search tools
@tool
def wiki_search(query: str) -> str:
    """Search the wikipedia for a query and return the first paragraph
    args:
        query: the query to search for
    """
    try:
        loader = WikipediaLoader(query=query, load_max_docs=1)
        data = loader.load()
        formatted_search_docs = "\n\n---\n\n".join(
            [
                f'\n{doc.page_content}\n'
                for doc in data
            ])
        return formatted_search_docs
    except Exception as e:
        return f"Wikipedia search failed: {str(e)}"

@tool
def web_search(query: str) -> str:
    """Search Tavily for a query and return maximum 3 results.
    
    Args:
        query: The search query.
    """
    try:
        time.sleep(random.uniform(1, 3))
        search_docs = TavilySearchResults(max_results=3).invoke(query=query)
        formatted_search_docs = "\n\n---\n\n".join(
            [
                f'\n{doc.get("content", "")}\n'
                for doc in search_docs
            ])
        return formatted_search_docs
    except Exception as e:
        return f"Web search failed: {str(e)}"

@tool
def arxiv_search(query: str) -> str:
    """Search Arxiv for a query and return maximum 3 result.
    
    Args:
        query: The search query.
    """
    try:
        search_docs = ArxivLoader(query=query, load_max_docs=3).load()
        formatted_search_docs = "\n\n---\n\n".join(
            [
                f'\n{doc.page_content[:1000]}\n'
                for doc in search_docs
            ])
        return formatted_search_docs
    except Exception as e:
        return f"ArXiv search failed: {str(e)}"

# Load and process your JSONL data
jq_schema = """
{
  page_content: .Question,
  metadata: {
    task_id: .task_id,
    Level: .Level,
    Final_answer: ."Final answer",
    file_name: .file_name,
    Steps: .["Annotator Metadata"].Steps,
    Number_of_steps: .["Annotator Metadata"]["Number of steps"],
    How_long: .["Annotator Metadata"]["How long did this take?"],
    Tools: .["Annotator Metadata"].Tools,
    Number_of_tools: .["Annotator Metadata"]["Number of tools"]
  }
}
"""

# Load documents and create vector database
json_loader = JSONLoader(file_path="metadata.jsonl", jq_schema=jq_schema, json_lines=True, text_content=False)
json_docs = json_loader.load()

# Split documents
text_splitter = RecursiveCharacterTextSplitter(chunk_size=512, chunk_overlap=200)
json_chunks = text_splitter.split_documents(json_docs)

# Create vector database
database = FAISS.from_documents(json_chunks, NVIDIAEmbeddings())

# Create retriever and retriever tool
retriever = database.as_retriever(search_type="similarity", search_kwargs={"k": 3})

retriever_tool = create_retriever_tool(
    retriever=retriever,
    name="question_search",
    description="Search for similar questions and their solutions from the knowledge base."
)

# Combine all tools including LLM tools
tools = [
    # Math tools
    multiply,
    add,
    subtract,
    divide,
    modulus,
    
    # Search tools
    wiki_search,
    web_search,
    arxiv_search,
    retriever_tool,
    
    # LLM tools - agent can choose which LLM to use
    groq_reasoning_tool,
    nvidia_specialist_tool
]

# Use a lightweight coordinator LLM (Groq for speed)
coordinator_llm = ChatGroq(
    model="llama-3.3-70b-versatile",
    temperature=0,
    api_key=os.getenv("GROQ_API_KEY"),
    rate_limiter=groq_rate_limiter
)

# Create memory for conversation
memory = MemorySaver()

# Create the agent with coordinator LLM
agent_executor = create_react_agent(
    model=coordinator_llm,
    tools=tools,
    checkpointer=memory
)

# Enhanced robust agent run
def robust_agent_run(query, thread_id="robust_conversation", max_retries=3):
    """Run agent with error handling, rate limiting, and LLM tool selection"""
    
    for attempt in range(max_retries):
        try:
            config = {"configurable": {"thread_id": f"{thread_id}_{attempt}"}}
            
            system_msg = SystemMessage(content='''You are a helpful assistant with access to multiple specialized LLM tools and other utilities. 

AVAILABLE LLM TOOLS:
- groq_reasoning_tool: Fast reasoning, math, calculations, code (use for quick logical problems)
- google_analysis_tool: Complex analysis, creative tasks, detailed explanations (use for comprehensive analysis)
- nvidia_specialist_tool: Technical questions, specialized domains, scientific problems (use for expert-level tasks)

TOOL SELECTION STRATEGY:
- For math/calculations: Use basic math tools (add, multiply, etc.) OR groq_reasoning_tool for complex math
- For factual questions: Use web_search, wiki_search, or arxiv_search first
- For analysis/reasoning: Choose the most appropriate LLM tool based on complexity
- For technical/scientific: Use nvidia_specialist_tool
- For creative/comprehensive: Use google_analysis_tool
- For quick logical problems: Use groq_reasoning_tool

Always finish with: FINAL ANSWER: [YOUR FINAL ANSWER]
Your answer should be a number OR few words OR comma separated list as appropriate.''')
            
            user_msg = HumanMessage(content=query)
            result = []
            
            print(f"Attempt {attempt + 1}: Processing query with multi-LLM agent...")
            
            for step in agent_executor.stream(
                {"messages": [system_msg, user_msg]}, 
                config, 
                stream_mode="values"
            ):
                result = step["messages"]
                
            final_response = result[-1].content if result else "No response generated"
            print(f"Query processed successfully on attempt {attempt + 1}")
            return final_response
            
        except Exception as e:
            error_msg = str(e).lower()
            
            if any(keyword in error_msg for keyword in ['rate limit', 'too many requests', '429', 'quota exceeded']):
                wait_time = (2 ** attempt) + random.uniform(1, 3)
                print(f"Rate limit hit on attempt {attempt + 1}. Waiting {wait_time:.2f} seconds...")
                time.sleep(wait_time)
                
                if attempt == max_retries - 1:
                    return f"Rate limit exceeded after {max_retries} attempts: {str(e)}"
                continue
                
            elif any(keyword in error_msg for keyword in ['api', 'connection', 'timeout', 'service unavailable']):
                wait_time = (2 ** attempt) + random.uniform(0.5, 1.5)
                print(f"API error on attempt {attempt + 1}. Retrying in {wait_time:.2f} seconds...")
                time.sleep(wait_time)
                
                if attempt == max_retries - 1:
                    return f"API error after {max_retries} attempts: {str(e)}"
                continue
                
            else:
                return f"Error occurred: {str(e)}"
    
    return "Maximum retries exceeded"

# Main function with request tracking
request_count = 0
last_request_time = time.time()

def main(query: str) -> str:
    """Main function to run the multi-LLM agent"""
    global request_count, last_request_time
    
    current_time = time.time()
    
    # Reset counter every minute
    if current_time - last_request_time > 60:
        request_count = 0
        last_request_time = current_time
    
    request_count += 1
    print(f"Processing request #{request_count} with multi-LLM agent")
    
    # Add delay between requests
    if request_count > 1:
        time.sleep(random.uniform(2, 5))
    
    return robust_agent_run(query)

if __name__ == "__main__":
    # Test the multi-LLM agent
    result = main("What are the names of the US presidents who were assassinated?")
    print(result)