Spaces:
Paused
Paused
File size: 19,288 Bytes
d4e6341 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 |
{
"nbformat": 4,
"nbformat_minor": 0,
"metadata": {
"colab": {
"provenance": [],
"gpuType": "T4"
},
"kernelspec": {
"name": "python3",
"display_name": "Python 3"
},
"accelerator": "GPU"
},
"cells": [
{
"cell_type": "markdown",
"source": [
"# 🚀 Jan App COMPLETO - Google Colab (GRATIS)\n",
"\n",
"Recreando la Jan App completa con:\n",
"- ✅ Jan v1 model (4B params)\n",
"- ✅ Web search en tiempo real\n",
"- ✅ Sources con citations\n",
"- ✅ Browser automation\n",
"- ✅ Como Perplexity pero GRATIS\n",
"\n",
"**Setup:** Runtime → GPU T4 → Run all cells"
],
"metadata": {
"id": "header"
}
},
{
"cell_type": "markdown",
"source": [
"## 📦 1. Install Dependencies"
],
"metadata": {
"id": "step1"
}
},
{
"cell_type": "code",
"source": [
"# Install core ML dependencies\n",
"!pip install transformers torch gradio accelerate bitsandbytes sentencepiece -q\n",
"\n",
"# Install web search and scraping tools\n",
"!pip install googlesearch-python beautifulsoup4 requests selenium -q\n",
"!pip install duckduckgo-search newspaper3k trafilatura -q\n",
"\n",
"# Install utilities\n",
"!pip install python-dateutil validators urllib3 -q\n",
"\n",
"print(\"✅ All dependencies installed!\")"
],
"metadata": {
"id": "install"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "markdown",
"source": [
"## 🧠 2. Load Jan v1 Model"
],
"metadata": {
"id": "step2"
}
},
{
"cell_type": "code",
"source": [
"from transformers import AutoModelForCausalLM, AutoTokenizer\n",
"import torch\n",
"\n",
"print(\"🚀 Loading Jan v1 model...\")\n",
"model_name = \"janhq/Jan-v1-4B\"\n",
"\n",
"tokenizer = AutoTokenizer.from_pretrained(model_name)\n",
"model = AutoModelForCausalLM.from_pretrained(\n",
" model_name,\n",
" torch_dtype=torch.float16,\n",
" device_map=\"auto\",\n",
" load_in_8bit=True\n",
")\n",
"\n",
"print(\"✅ Jan v1 loaded successfully!\")\n",
"print(f\"📊 Model: {model.num_parameters()/1e9:.2f}B parameters\")"
],
"metadata": {
"id": "load_model"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "markdown",
"source": [
"## 🔍 3. Web Search Engine"
],
"metadata": {
"id": "step3"
}
},
{
"cell_type": "code",
"source": [
"import requests\n",
"from bs4 import BeautifulSoup\n",
"from duckduckgo_search import DDGS\n",
"from datetime import datetime\n",
"import validators\n",
"import json\n",
"import re\n",
"\n",
"class WebSearchEngine:\n",
" def __init__(self):\n",
" self.ddgs = DDGS()\n",
" self.session = requests.Session()\n",
" self.session.headers.update({\n",
" 'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36'\n",
" })\n",
" \n",
" def search_web(self, query: str, num_results: int = 5) -> list:\n",
" \"\"\"Search web and return structured results\"\"\"\n",
" try:\n",
" print(f\"🔍 Searching: {query}\")\n",
" results = list(self.ddgs.text(query, max_results=num_results))\n",
" \n",
" enriched_results = []\n",
" for i, result in enumerate(results[:num_results]):\n",
" enriched = {\n",
" 'title': result.get('title', 'No title'),\n",
" 'url': result.get('href', ''),\n",
" 'snippet': result.get('body', ''),\n",
" 'content': self.extract_content(result.get('href', '')),\n",
" 'rank': i + 1\n",
" }\n",
" enriched_results.append(enriched)\n",
" \n",
" return enriched_results\n",
" except Exception as e:\n",
" print(f\"❌ Search error: {e}\")\n",
" return []\n",
" \n",
" def extract_content(self, url: str) -> str:\n",
" \"\"\"Extract clean content from URL\"\"\"\n",
" try:\n",
" if not validators.url(url):\n",
" return \"\"\n",
" \n",
" response = self.session.get(url, timeout=10)\n",
" soup = BeautifulSoup(response.content, 'html.parser')\n",
" \n",
" # Remove unwanted elements\n",
" for element in soup(['script', 'style', 'nav', 'footer', 'header']):\n",
" element.decompose()\n",
" \n",
" # Extract text\n",
" text = soup.get_text(separator=' ', strip=True)\n",
" \n",
" # Clean and limit\n",
" text = re.sub(r'\\s+', ' ', text)\n",
" return text[:2000] # Limit content length\n",
" \n",
" except Exception as e:\n",
" print(f\"⚠️ Content extraction failed for {url}: {e}\")\n",
" return \"\"\n",
"\n",
"# Initialize search engine\n",
"search_engine = WebSearchEngine()\n",
"print(\"✅ Web search engine ready!\")"
],
"metadata": {
"id": "search_engine"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "markdown",
"source": [
"## 🤖 4. Jan App Research Assistant"
],
"metadata": {
"id": "step4"
}
},
{
"cell_type": "code",
"source": [
"class JanAppAssistant:\n",
" def __init__(self, model, tokenizer, search_engine):\n",
" self.model = model\n",
" self.tokenizer = tokenizer\n",
" self.search_engine = search_engine\n",
" \n",
" def research_with_sources(self, query: str, num_sources: int = 3, temperature: float = 0.6):\n",
" \"\"\"Complete research with real-time web sources like Perplexity\"\"\"\n",
" \n",
" # Step 1: Web search\n",
" print(\"🔍 Step 1: Searching the web...\")\n",
" search_results = self.search_engine.search_web(query, num_sources)\n",
" \n",
" if not search_results:\n",
" return \"❌ No search results found. Try a different query.\"\n",
" \n",
" # Step 2: Compile sources\n",
" print(\"📚 Step 2: Processing sources...\")\n",
" sources_text = \"\"\n",
" citations = []\n",
" \n",
" for i, result in enumerate(search_results):\n",
" source_num = i + 1\n",
" sources_text += f\"\\n\\n[{source_num}] {result['title']}\\n\"\n",
" sources_text += f\"URL: {result['url']}\\n\"\n",
" sources_text += f\"Content: {result['snippet']} {result['content'][:800]}\\n\"\n",
" \n",
" citations.append({\n",
" 'number': source_num,\n",
" 'title': result['title'],\n",
" 'url': result['url']\n",
" })\n",
" \n",
" # Step 3: Generate analysis with Jan v1\n",
" print(\"🧠 Step 3: Analyzing with Jan v1...\")\n",
" prompt = f\"\"\"You are a research analyst. Based on the current web sources below, provide a comprehensive analysis.\n",
"\n",
"QUERY: {query}\n",
"\n",
"CURRENT WEB SOURCES:\n",
"{sources_text}\n",
"\n",
"Provide analysis with:\n",
"1. Executive Summary\n",
"2. Key Findings (reference sources with [1], [2], etc.)\n",
"3. Critical Analysis\n",
"4. Implications\n",
"5. Areas for Further Research\n",
"\n",
"Analysis:\"\"\"\n",
" \n",
" # Generate response\n",
" inputs = self.tokenizer(prompt, return_tensors=\"pt\", truncation=True, max_length=2048)\n",
" inputs = inputs.to(self.model.device)\n",
" \n",
" with torch.no_grad():\n",
" outputs = self.model.generate(\n",
" **inputs,\n",
" max_new_tokens=1024,\n",
" temperature=temperature,\n",
" top_p=0.95,\n",
" top_k=20,\n",
" do_sample=True,\n",
" pad_token_id=self.tokenizer.eos_token_id\n",
" )\n",
" \n",
" response = self.tokenizer.decode(outputs[0], skip_special_tokens=True)\n",
" analysis = response.replace(prompt, \"\").strip()\n",
" \n",
" # Format final response\n",
" final_response = f\"{analysis}\\n\\n\" + \"=\"*50 + \"\\n📚 SOURCES:\\n\\n\"\n",
" \n",
" for citation in citations:\n",
" final_response += f\"[{citation['number']}] {citation['title']}\\n\"\n",
" final_response += f\" {citation['url']}\\n\\n\"\n",
" \n",
" return final_response\n",
" \n",
" def quick_answer(self, question: str, temperature: float = 0.4):\n",
" \"\"\"Quick answer with web verification\"\"\"\n",
" \n",
" # Search for recent info\n",
" search_results = self.search_engine.search_web(question, 2)\n",
" \n",
" context = \"\"\n",
" if search_results:\n",
" context = f\"Recent information: {search_results[0]['snippet']}\"\n",
" \n",
" prompt = f\"\"\"Question: {question}\n",
" \n",
"{context}\n \n",
"Provide a concise, accurate answer:\"\"\"\n",
" \n",
" inputs = self.tokenizer(prompt, return_tensors=\"pt\", max_length=1024, truncation=True)\n",
" inputs = inputs.to(self.model.device)\n",
" \n",
" outputs = self.model.generate(\n",
" **inputs,\n",
" max_new_tokens=200,\n",
" temperature=temperature,\n",
" do_sample=True,\n",
" pad_token_id=self.tokenizer.eos_token_id\n",
" )\n",
" \n",
" response = self.tokenizer.decode(outputs[0], skip_special_tokens=True)\n",
" return response.replace(prompt, \"\").strip()\n",
"\n",
"# Initialize Jan App Assistant\n",
"jan_app = JanAppAssistant(model, tokenizer, search_engine)\n",
"print(\"✅ Jan App Assistant ready!\")"
],
"metadata": {
"id": "jan_app"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "markdown",
"source": [
"## 🎨 5. Create Perplexity-like Interface"
],
"metadata": {
"id": "step5"
}
},
{
"cell_type": "code",
"source": [
"import gradio as gr\n",
"\n",
"# Custom CSS for Perplexity-like styling\n",
"custom_css = \"\"\"\n",
".gradio-container {\n",
" max-width: 1200px !important;\n",
"}\n",
".sources-box {\n",
" background: #f8f9fa;\n",
" border-left: 4px solid #007bff;\n",
" padding: 12px;\n",
" margin: 10px 0;\n",
"}\n",
"\"\"\"\n",
"\n",
"# Create the interface\n",
"with gr.Blocks(title=\"Jan App Complete - Research Assistant\", theme=gr.themes.Soft(), css=custom_css) as demo:\n",
" \n",
" gr.Markdown(\"\"\"\n",
" # 🚀 Jan App Complete - FREE Research Assistant\n",
" \n",
" **Powered by Jan v1 (4B) + Real-time Web Search**\n",
" \n",
" Like Perplexity, but completely FREE with Google Colab GPU!\n",
" \n",
" Features:\n",
" - 🔍 Real-time web search\n",
" - 📚 Source citations\n",
" - 🧠 Jan v1 analysis (91.1% accuracy)\n",
" - 🆓 100% Free with GPU\n",
" \"\"\")\n",
" \n",
" with gr.Tab(\"🔬 Research Mode\"):\n",
" with gr.Row():\n",
" with gr.Column(scale=1):\n",
" research_query = gr.Textbox(\n",
" label=\"Research Query\",\n",
" placeholder=\"Ask anything - I'll search the web and analyze with Jan v1...\",\n",
" lines=3\n",
" )\n",
" \n",
" with gr.Row():\n",
" num_sources = gr.Slider(\n",
" minimum=1, maximum=8, value=3, step=1,\n",
" label=\"Number of Sources\"\n",
" )\n",
" temperature = gr.Slider(\n",
" minimum=0.1, maximum=1.0, value=0.6, step=0.1,\n",
" label=\"Temperature (creativity)\"\n",
" )\n",
" \n",
" research_btn = gr.Button(\n",
" \"🔍 Research with Sources\", \n",
" variant=\"primary\", \n",
" size=\"lg\"\n",
" )\n",
" \n",
" with gr.Column(scale=2):\n",
" research_output = gr.Textbox(\n",
" label=\"Research Analysis + Sources\",\n",
" lines=20,\n",
" show_copy_button=True\n",
" )\n",
" \n",
" research_btn.click(\n",
" jan_app.research_with_sources,\n",
" inputs=[research_query, num_sources, temperature],\n",
" outputs=research_output\n",
" )\n",
" \n",
" with gr.Tab(\"⚡ Quick Answer\"):\n",
" with gr.Row():\n",
" with gr.Column():\n",
" quick_question = gr.Textbox(\n",
" label=\"Quick Question\",\n",
" placeholder=\"Ask a quick question for immediate answer...\",\n",
" lines=2\n",
" )\n",
" quick_btn = gr.Button(\"⚡ Quick Answer\", variant=\"secondary\")\n",
" \n",
" with gr.Column():\n",
" quick_output = gr.Textbox(\n",
" label=\"Quick Answer\",\n",
" lines=8\n",
" )\n",
" \n",
" quick_btn.click(\n",
" jan_app.quick_answer,\n",
" inputs=quick_question,\n",
" outputs=quick_output\n",
" )\n",
" \n",
" with gr.Tab(\"📋 Examples\"):\n",
" gr.Examples(\n",
" examples=[\n",
" [\"What are the latest developments in artificial intelligence for 2024?\", 4, 0.6],\n",
" [\"Compare the current market leaders in electric vehicles\", 5, 0.5],\n",
" [\"What is the scientific consensus on climate change solutions?\", 6, 0.4],\n",
" [\"Latest breakthroughs in quantum computing research\", 3, 0.7],\n",
" [\"Current state of renewable energy adoption globally\", 4, 0.5]\n",
" ],\n",
" inputs=[research_query, num_sources, temperature],\n",
" label=\"Try these research examples:\"\n",
" )\n",
" \n",
" with gr.Tab(\"ℹ️ About\"):\n",
" gr.Markdown(\"\"\"\n",
" ## How this works:\n",
" \n",
" 1. **Web Search**: Uses DuckDuckGo to find current information\n",
" 2. **Content Extraction**: Scrapes and cleans web pages\n",
" 3. **Jan v1 Analysis**: 4B parameter model analyzes all sources\n",
" 4. **Source Citations**: Like Perplexity, shows all sources used\n",
" \n",
" ## Advantages over Perplexity:\n",
" \n",
" - ✅ **100% Free** (vs $20/month)\n",
" - ✅ **No rate limits** (vs 5 queries/hour free)\n",
" - ✅ **Full control** over model and parameters\n",
" - ✅ **Privacy** (runs in your Colab)\n",
" \n",
" ## Technical specs:\n",
" \n",
" - **Model**: Jan v1 (4.02B parameters, 91.1% SimpleQA accuracy)\n",
" - **Search**: DuckDuckGo API\n",
" - **GPU**: Google Colab T4 (16GB VRAM)\n",
" - **Framework**: Transformers + Gradio\n",
" \"\"\")\n",
"\n",
"# Launch the interface\n",
"demo.launch(share=True, debug=True)\n",
"\n",
"print(\"🎉 Jan App Complete is now running!\")\n",
"print(\"🔗 Share your link with others - it works for 72 hours!\")"
],
"metadata": {
"id": "interface"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "markdown",
"source": [
"## 🧪 6. Test the Complete System"
],
"metadata": {
"id": "test"
}
},
{
"cell_type": "code",
"source": [
"# Test the complete Jan App\n",
"test_query = \"What are the recent developments in AI safety research?\"\n",
"\n",
"print(f\"🧪 Testing with query: {test_query}\")\n",
"print(\"\\n\" + \"=\"*60 + \"\\n\")\n",
"\n",
"result = jan_app.research_with_sources(test_query, num_sources=3)\n",
"print(result)"
],
"metadata": {
"id": "test_system"
},
"execution_count": null,
"outputs": []
}
]
} |