File size: 19,980 Bytes
5092552
4efaf9c
 
5092552
 
d4557ee
 
 
ca98093
41f9740
 
1fa6961
4efaf9c
41f9740
ca98093
41f9740
 
 
ca98093
a35ea13
4efaf9c
 
 
 
 
 
 
 
 
 
 
 
 
 
41f9740
 
 
 
ca98093
41f9740
ca98093
4efaf9c
41f9740
4efaf9c
41f9740
 
 
 
 
 
 
 
 
 
 
ca98093
41f9740
f4505e9
86c8869
41f9740
f0e66e7
f4505e9
 
86c8869
41f9740
f0e66e7
1fa6961
f4505e9
86c8869
41f9740
f0e66e7
cc467c2
f4505e9
86c8869
41f9740
86c8869
f4505e9
 
 
 
86c8869
41f9740
f0e66e7
a55679f
f4505e9
41f9740
4efaf9c
25c1140
4efaf9c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
25c1140
d4557ee
0f81d99
f4505e9
41f9740
4efaf9c
25c1140
41f9740
 
 
 
 
 
25c1140
41f9740
a55679f
4efaf9c
 
 
ca98093
41f9740
4efaf9c
 
 
 
 
41f9740
4efaf9c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
41f9740
4efaf9c
41f9740
4efaf9c
 
 
 
 
 
 
 
 
 
 
41f9740
4efaf9c
 
 
 
 
 
 
 
 
 
 
 
 
41f9740
4efaf9c
 
41f9740
4efaf9c
 
 
 
 
 
 
ca98093
4efaf9c
41f9740
4efaf9c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0c69489
41f9740
 
 
 
 
 
 
 
 
 
4efaf9c
41f9740
4efaf9c
 
41f9740
4efaf9c
41f9740
 
4efaf9c
 
41f9740
 
 
 
 
4efaf9c
41f9740
4efaf9c
 
41f9740
4efaf9c
41f9740
ca98093
41f9740
 
4efaf9c
 
41f9740
4efaf9c
 
 
 
 
41f9740
4efaf9c
 
41f9740
4efaf9c
 
ca98093
4efaf9c
 
 
 
 
 
 
 
 
41f9740
4efaf9c
 
 
41f9740
 
 
 
 
5092552
41f9740
 
 
 
 
 
 
 
5092552
41f9740
 
 
 
 
 
 
 
5092552
4efaf9c
41f9740
 
4efaf9c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
41f9740
 
 
4efaf9c
 
 
 
 
 
 
 
 
 
 
 
41f9740
 
 
4efaf9c
41f9740
4efaf9c
 
 
41f9740
 
 
4efaf9c
 
 
 
 
41f9740
5092552
4efaf9c
41f9740
 
 
4efaf9c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0f81d99
41f9740
4efaf9c
41f9740
 
 
 
 
 
 
 
4efaf9c
72c7dbb
4efaf9c
41f9740
 
 
 
 
 
 
 
 
 
 
 
 
4efaf9c
 
 
 
 
 
 
 
 
 
 
5092552
4efaf9c
 
 
 
0c69489
4efaf9c
41f9740
4efaf9c
 
41f9740
4efaf9c
 
 
 
 
 
 
5092552
41f9740
5092552
 
4efaf9c
 
 
 
5092552
 
4efaf9c
 
 
 
 
 
 
41f9740
 
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
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
"""
Open-Source Multi-LLM Agent System
Uses only free and open-source models - no paid APIs required
"""

import os
import time
import random
import operator
from typing import List, Dict, Any, TypedDict, Annotated, Optional
from dotenv import load_dotenv

# Core LangChain imports
from langchain_core.tools import tool
from langchain_community.tools.tavily_search import TavilySearchResults
from langchain_community.document_loaders import WikipediaLoader
from langgraph.graph import StateGraph, END
from langgraph.checkpoint.memory import MemorySaver
from langchain_core.messages import SystemMessage, HumanMessage, AIMessage

# Open-source model integrations
from langchain_groq import ChatGroq  # Free tier available
from langchain_community.llms import Ollama
from langchain_community.chat_models import ChatOllama

# Hugging Face integration for open-source models
try:
    from langchain_huggingface import HuggingFacePipeline
    from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
    HF_AVAILABLE = True
except ImportError:
    HF_AVAILABLE = False

# Vector database imports
import faiss
import numpy as np
from sentence_transformers import SentenceTransformer
import json

load_dotenv()

# Enhanced system prompt
ENHANCED_SYSTEM_PROMPT = (
    "You are a helpful assistant tasked with answering questions using available tools. "
    "You must provide accurate, comprehensive answers based on available information. "
    "When answering questions, follow these guidelines:\n"
    "1. Use available tools to gather information when needed\n"
    "2. Provide precise, factual answers\n"
    "3. For numbers: don't use commas or units unless specified\n"
    "4. For strings: don't use articles or abbreviations, write digits in plain text\n"
    "5. For lists: apply above rules based on element type\n"
    "6. Always end with 'FINAL ANSWER: [YOUR ANSWER]'\n"
    "7. Be concise but thorough in your reasoning\n"
    "8. If you cannot find the answer, state that clearly"
)

# ---- Tool Definitions ----
@tool
def multiply(a: int, b: int) -> int:
    """Multiply two integers and return the product."""
    return a * b

@tool
def add(a: int, b: int) -> int:
    """Add two integers and return the sum."""
    return a + b

@tool
def subtract(a: int, b: int) -> int:
    """Subtract the second integer from the first and return the difference."""
    return a - b

@tool
def divide(a: int, b: int) -> float:
    """Divide the first integer by the second and return the quotient."""
    if b == 0:
        raise ValueError("Cannot divide by zero.")
    return a / b

@tool
def modulus(a: int, b: int) -> int:
    """Return the remainder when dividing the first integer by the second."""
    return a % b

@tool
def optimized_web_search(query: str) -> str:
    """Perform web search using free DuckDuckGo (fallback if Tavily not available)."""
    try:
        # Try Tavily first (free tier)
        if os.getenv("TAVILY_API_KEY"):
            time.sleep(random.uniform(0.7, 1.5))
            search_tool = TavilySearchResults(max_results=3)
            docs = search_tool.invoke({"query": query})
            return "\n\n---\n\n".join(
                f"<Doc url='{d.get('url','')}'>{d.get('content','')[:800]}</Doc>"
                for d in docs
            )
        else:
            # Fallback to DuckDuckGo (completely free)
            try:
                from duckduckgo_search import DDGS
                with DDGS() as ddgs:
                    results = list(ddgs.text(query, max_results=3))
                    return "\n\n---\n\n".join(
                        f"<Doc url='{r.get('href','')}'>{r.get('body','')[:800]}</Doc>"
                        for r in results
                    )
            except ImportError:
                return "Web search not available - install duckduckgo-search for free web search"
    except Exception as e:
        return f"Web search failed: {e}"

@tool
def optimized_wiki_search(query: str) -> str:
    """Perform Wikipedia search - completely free."""
    try:
        time.sleep(random.uniform(0.3, 1))
        docs = WikipediaLoader(query=query, load_max_docs=2).load()
        return "\n\n---\n\n".join(
            f"<Doc src='{d.metadata.get('source','Wikipedia')}'>{d.page_content[:1000]}</Doc>"
            for d in docs
        )
    except Exception as e:
        return f"Wikipedia search failed: {e}"

# ---- Open-Source Model Manager ----
class OpenSourceModelManager:
    """Manages only open-source and free models"""
    
    def __init__(self):
        self.available_models = {}
        self._initialize_models()
    
    def _initialize_models(self):
        """Initialize only open-source models"""
        
        # 1. Groq (Free tier with open-source models)
        if os.getenv("GROQ_API_KEY"):
            try:
                self.available_models['groq_llama3_70b'] = ChatGroq(
                    model="llama3-70b-8192", 
                    temperature=0, 
                    api_key=os.getenv("GROQ_API_KEY")
                )
                self.available_models['groq_llama3_8b'] = ChatGroq(
                    model="llama3-8b-8192", 
                    temperature=0, 
                    api_key=os.getenv("GROQ_API_KEY")
                )
                self.available_models['groq_mixtral'] = ChatGroq(
                    model="mixtral-8x7b-32768", 
                    temperature=0, 
                    api_key=os.getenv("GROQ_API_KEY")
                )
                self.available_models['groq_gemma'] = ChatGroq(
                    model="gemma-7b-it", 
                    temperature=0, 
                    api_key=os.getenv("GROQ_API_KEY")
                )
                print("Groq models initialized (free tier)")
            except Exception as e:
                print(f"Groq models not available: {e}")
        
        # 2. Ollama (Completely free local models)
        try:
            # Test if Ollama is running
            test_model = ChatOllama(model="llama3", base_url="http://localhost:11434")
            # If no error, add Ollama models
            self.available_models['ollama_llama3'] = ChatOllama(model="llama3")
            self.available_models['ollama_llama3_70b'] = ChatOllama(model="llama3:70b")
            self.available_models['ollama_mistral'] = ChatOllama(model="mistral")
            self.available_models['ollama_phi3'] = ChatOllama(model="phi3")
            self.available_models['ollama_codellama'] = ChatOllama(model="codellama")
            self.available_models['ollama_gemma'] = ChatOllama(model="gemma")
            self.available_models['ollama_qwen'] = ChatOllama(model="qwen")
            print("Ollama models initialized (local)")
        except Exception as e:
            print(f"Ollama not available: {e}")
        
        # 3. Hugging Face Transformers (Completely free)
        if HF_AVAILABLE:
            try:
                # Small models that can run on CPU
                self.available_models['hf_gpt2'] = self._create_hf_model("gpt2")
                self.available_models['hf_distilgpt2'] = self._create_hf_model("distilgpt2")
                print("Hugging Face models initialized (local)")
            except Exception as e:
                print(f"Hugging Face models not available: {e}")
        
        print(f"Total available open-source models: {len(self.available_models)}")
    
    def _create_hf_model(self, model_name: str):
        """Create Hugging Face pipeline model"""
        try:
            pipe = pipeline(
                "text-generation",
                model=model_name,
                max_length=512,
                do_sample=True,
                temperature=0.7,
                pad_token_id=50256
            )
            return HuggingFacePipeline(pipeline=pipe)
        except Exception as e:
            print(f"Failed to create HF model {model_name}: {e}")
            return None
    
    def get_model(self, model_name: str):
        """Get a specific model by name"""
        return self.available_models.get(model_name)
    
    def list_available_models(self) -> List[str]:
        """List all available model names"""
        return list(self.available_models.keys())
    
    def get_best_model_for_task(self, task_type: str):
        """Get the best available model for a specific task type"""
        if task_type == "reasoning":
            # Prefer larger models for reasoning
            for model_name in ['groq_llama3_70b', 'ollama_llama3_70b', 'groq_mixtral', 'ollama_llama3']:
                if model_name in self.available_models:
                    return self.available_models[model_name]
        
        elif task_type == "coding":
            # Prefer code-specialized models
            for model_name in ['ollama_codellama', 'groq_llama3_70b', 'ollama_llama3']:
                if model_name in self.available_models:
                    return self.available_models[model_name]
        
        elif task_type == "fast":
            # Prefer fast, smaller models
            for model_name in ['groq_llama3_8b', 'groq_gemma', 'ollama_phi3', 'hf_distilgpt2']:
                if model_name in self.available_models:
                    return self.available_models[model_name]
        
        # Default fallback to first available
        if self.available_models:
            return list(self.available_models.values())[0]
        return None

# ---- Enhanced Agent State ----
class EnhancedAgentState(TypedDict):
    """State structure for the enhanced multi-LLM agent system."""
    messages: Annotated[List[HumanMessage | AIMessage], operator.add]
    query: str
    agent_type: str
    final_answer: str
    perf: Dict[str, Any]
    tools_used: List[str]
    reasoning: str
    model_used: str

# ---- Open-Source Multi-LLM System ----
class OpenSourceMultiLLMSystem:
    """
    Multi-LLM system using only open-source and free models
    """
    
    def __init__(self):
        self.model_manager = OpenSourceModelManager()
        self.tools = [
            multiply, add, subtract, divide, modulus,
            optimized_web_search, optimized_wiki_search
        ]
        self.graph = self._build_graph()
    
    def _build_graph(self) -> StateGraph:
        """Build the LangGraph state machine with open-source models."""
        
        def router(st: EnhancedAgentState) -> EnhancedAgentState:
            """Route queries to appropriate model based on complexity and content analysis."""
            q = st["query"].lower()
            
            # Enhanced routing logic
            if any(keyword in q for keyword in ["calculate", "compute", "math", "multiply", "add", "subtract", "divide"]):
                model_type = "reasoning"
                agent_type = "math"
            elif any(keyword in q for keyword in ["search", "find", "lookup", "wikipedia", "information about"]):
                model_type = "fast"
                agent_type = "search_enhanced"
            elif any(keyword in q for keyword in ["code", "programming", "function", "algorithm"]):
                model_type = "coding"
                agent_type = "coding"
            elif len(q.split()) > 20:  # Complex queries
                model_type = "reasoning"
                agent_type = "complex"
            else:
                model_type = "fast"
                agent_type = "simple"
            
            # Get the best model for this task
            selected_model = self.model_manager.get_best_model_for_task(model_type)
            model_name = "unknown"
            for name, model in self.model_manager.available_models.items():
                if model == selected_model:
                    model_name = name
                    break
            
            return {**st, "agent_type": agent_type, "tools_used": [], "reasoning": "", "model_used": model_name}

        def math_node(st: EnhancedAgentState) -> EnhancedAgentState:
            """Process mathematical queries."""
            return self._process_with_model(st, "reasoning", "Mathematical calculation using open-source model")

        def search_enhanced_node(st: EnhancedAgentState) -> EnhancedAgentState:
            """Process query with search enhancement."""
            t0 = time.time()
            tools_used = []
            
            try:
                # Determine search strategy
                query = st["query"]
                search_results = ""
                
                if any(keyword in query.lower() for keyword in ["wikipedia", "wiki"]):
                    search_results = optimized_wiki_search.invoke({"query": query})
                    tools_used.append("wikipedia_search")
                else:
                    search_results = optimized_web_search.invoke({"query": query})
                    tools_used.append("web_search")
                
                enhanced_query = f"""
                Original Question: {query}
                
                Search Results:
                {search_results}
                
                Based on the search results above, provide a direct answer to the original question.
                """
                
                # Use fast model for search-enhanced queries
                model = self.model_manager.get_best_model_for_task("fast")
                if model:
                    sys = SystemMessage(content=ENHANCED_SYSTEM_PROMPT)
                    res = model.invoke([sys, HumanMessage(content=enhanced_query)])
                    
                    answer = res.content.strip() if hasattr(res, 'content') else str(res).strip()
                    if "FINAL ANSWER:" in answer:
                        answer = answer.split("FINAL ANSWER:")[-1].strip()
                    
                    return {**st,
                            "final_answer": answer,
                            "tools_used": tools_used,
                            "reasoning": "Used search enhancement with open-source model",
                            "perf": {"time": time.time() - t0, "prov": "Search-Enhanced"}}
                else:
                    return {**st, "final_answer": "No models available", "perf": {"error": "No models"}}
            except Exception as e:
                return {**st, "final_answer": f"Error: {e}", "perf": {"error": str(e)}}

        def coding_node(st: EnhancedAgentState) -> EnhancedAgentState:
            """Process coding-related queries."""
            return self._process_with_model(st, "coding", "Code generation using open-source model")

        def complex_node(st: EnhancedAgentState) -> EnhancedAgentState:
            """Process complex queries."""
            return self._process_with_model(st, "reasoning", "Complex reasoning using open-source model")

        def simple_node(st: EnhancedAgentState) -> EnhancedAgentState:
            """Process simple queries."""
            return self._process_with_model(st, "fast", "Simple query using fast open-source model")

        # Build graph
        g = StateGraph(EnhancedAgentState)
        g.add_node("router", router)
        g.add_node("math", math_node)
        g.add_node("search_enhanced", search_enhanced_node)
        g.add_node("coding", coding_node)
        g.add_node("complex", complex_node)
        g.add_node("simple", simple_node)
        
        g.set_entry_point("router")
        g.add_conditional_edges("router", lambda s: s["agent_type"], {
            "math": "math",
            "search_enhanced": "search_enhanced",
            "coding": "coding",
            "complex": "complex",
            "simple": "simple"
        })
        
        for node in ["math", "search_enhanced", "coding", "complex", "simple"]:
            g.add_edge(node, END)
            
        return g.compile(checkpointer=MemorySaver())
    
    def _process_with_model(self, st: EnhancedAgentState, model_type: str, reasoning: str) -> EnhancedAgentState:
        """Process query with specified model type"""
        t0 = time.time()
        try:
            model = self.model_manager.get_best_model_for_task(model_type)
            if not model:
                return {**st, "final_answer": "No suitable model available", "perf": {"error": "No model"}}
            
            enhanced_query = f"""
            Question: {st["query"]}
            
            Please provide a direct, accurate answer to this question.
            """
            
            sys = SystemMessage(content=ENHANCED_SYSTEM_PROMPT)
            res = model.invoke([sys, HumanMessage(content=enhanced_query)])
            
            answer = res.content.strip() if hasattr(res, 'content') else str(res).strip()
            if "FINAL ANSWER:" in answer:
                answer = answer.split("FINAL ANSWER:")[-1].strip()
            
            return {**st,
                    "final_answer": answer,
                    "reasoning": reasoning,
                    "perf": {"time": time.time() - t0, "prov": f"OpenSource-{model_type}"}}
        except Exception as e:
            return {**st, "final_answer": f"Error: {e}", "perf": {"error": str(e)}}

    def process_query(self, q: str) -> str:
        """Process a query through the open-source multi-LLM system."""
        state = {
            "messages": [HumanMessage(content=q)],
            "query": q,
            "agent_type": "",
            "final_answer": "",
            "perf": {},
            "tools_used": [],
            "reasoning": "",
            "model_used": ""
        }
        cfg = {"configurable": {"thread_id": f"opensource_qa_{hash(q)}"}}
        
        try:
            out = self.graph.invoke(state, cfg)
            answer = out.get("final_answer", "").strip()
            
            # Ensure we don't return the question as the answer
            if answer == q or answer.startswith(q):
                return "Information not available"
            
            return answer if answer else "No answer generated"
        except Exception as e:
            return f"Error processing query: {e}"
    
    def get_system_info(self) -> Dict[str, Any]:
        """Get information about available open-source models"""
        return {
            "available_models": self.model_manager.list_available_models(),
            "total_models": len(self.model_manager.available_models),
            "model_types": {
                "groq_free_tier": [m for m in self.model_manager.list_available_models() if m.startswith("groq_")],
                "ollama_local": [m for m in self.model_manager.list_available_models() if m.startswith("ollama_")],
                "huggingface_local": [m for m in self.model_manager.list_available_models() if m.startswith("hf_")]
            }
        }

# ---- Build Graph Function (for compatibility) ----
def build_graph(provider: str = "opensource"):
    """Build graph using only open-source models"""
    return OpenSourceMultiLLMSystem().graph

# ---- Main execution ----
if __name__ == "__main__":
    # Initialize the open-source system
    system = OpenSourceMultiLLMSystem()
    
    # Print system information
    info = system.get_system_info()
    print("Open-Source System Information:")
    print(f"Total Models Available: {info['total_models']}")
    for category, models in info['model_types'].items():
        if models:
            print(f"  {category}: {models}")
    
    # Test queries
    test_questions = [
        "What is 25 multiplied by 17?",
        "Find information about Mercedes Sosa albums between 2000-2009",
        "Write a simple Python function to calculate factorial",
        "Explain quantum computing in simple terms",
        "What is the capital of France?"
    ]
    
    print("\n" + "="*60)
    print("Testing Open-Source Multi-LLM System")
    print("="*60)
    
    for i, question in enumerate(test_questions, 1):
        print(f"\nQuestion {i}: {question}")
        print("-" * 50)
        answer = system.process_query(question)
        print(f"Answer: {answer}")