Spaces:
Sleeping
Sleeping
File size: 11,294 Bytes
28c5847 |
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 |
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# 📈 Stock Market BPE Tokenizer - Usage Examples\n",
"\n",
"This notebook demonstrates how to use the Stock Market BPE tokenizer.\n",
"\n",
"## 🎯 What You'll Learn\n",
"- How to load the trained tokenizer\n",
"- How to encode stock data\n",
"- How to decode tokens back to text\n",
"- How to calculate compression ratios\n",
"- Real-world examples with actual stock data"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 📦 Setup"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from tokenizer import StockBPE\n",
"import json"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 🔧 Load the Trained Tokenizer"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Initialize and load the trained tokenizer\n",
"tokenizer = StockBPE()\n",
"tokenizer.load(\"stock_bpe\")\n",
"\n",
"print(f\"✅ Tokenizer loaded!\")\n",
"print(f\"📚 Vocabulary size: {len(tokenizer.vocab):,}\")\n",
"print(f\"🔀 Number of merges: {len(tokenizer.merges):,}\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 📊 Example 1: Encode a Single Stock Record"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Sample stock data\n",
"stock_data = \"AAPL|2024-01-15|150.25|152.30|149.80|151.50|1000000\"\n",
"\n",
"print(\"📈 Original Stock Data:\")\n",
"print(stock_data)\n",
"print(f\"\\n📏 Length: {len(stock_data)} characters\")\n",
"\n",
"# Encode\n",
"tokens = tokenizer.encode(stock_data)\n",
"print(f\"\\n🔤 Encoded Tokens:\")\n",
"print(tokens)\n",
"print(f\"\\n📊 Token count: {len(tokens)}\")\n",
"\n",
"# Calculate compression\n",
"original_bytes = len(stock_data.encode('utf-8'))\n",
"compression_ratio = original_bytes / len(tokens)\n",
"print(f\"\\n🗜️ Compression ratio: {compression_ratio:.2f}x\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 🔄 Example 2: Decode Tokens Back to Text"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Decode the tokens back to original text\n",
"decoded = tokenizer.decode(tokens)\n",
"\n",
"print(\"🔓 Decoded Text:\")\n",
"print(decoded)\n",
"\n",
"# Verify it matches the original\n",
"print(f\"\\n✅ Match: {stock_data == decoded}\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 📊 Example 3: Multiple Stock Records"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Multiple stock records\n",
"multi_stock = \"\"\"AAPL|2024-01-15|150.25|152.30|149.80|151.50|1000000\n",
"MSFT|2024-01-15|380.50|385.20|379.00|384.75|850000\n",
"GOOGL|2024-01-15|140.10|142.50|139.80|141.90|920000\"\"\"\n",
"\n",
"print(\"📈 Multiple Stock Records:\")\n",
"print(multi_stock)\n",
"\n",
"# Encode\n",
"tokens = tokenizer.encode(multi_stock)\n",
"print(f\"\\n🔤 Total tokens: {len(tokens)}\")\n",
"\n",
"# Compression\n",
"ratio = tokenizer.calculate_compression_ratio(multi_stock)\n",
"print(f\"🗜️ Compression ratio: {ratio:.2f}x\")\n",
"\n",
"# Decode and verify\n",
"decoded = tokenizer.decode(tokens)\n",
"print(f\"\\n✅ Decoding successful: {multi_stock == decoded}\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 🎯 Example 4: Analyze Compression Patterns"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Test different types of stock data\n",
"test_cases = [\n",
" (\"Single record\", \"AAPL|2024-01-15|150.25|152.30|149.80|151.50|1000000\"),\n",
" (\"High price\", \"GOOGL|2024-01-15|2800.50|2850.20|2790.00|2845.75|500000\"),\n",
" (\"Low price\", \"F|2024-01-15|12.50|12.80|12.30|12.75|5000000\"),\n",
"]\n",
"\n",
"print(\"📊 Compression Analysis:\\n\")\n",
"for name, data in test_cases:\n",
" ratio = tokenizer.calculate_compression_ratio(data)\n",
" tokens = len(tokenizer.encode(data))\n",
" print(f\"{name:15} | Ratio: {ratio:.2f}x | Tokens: {tokens}\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 🔍 Example 5: Inspect Learned Patterns"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Show some learned merge patterns\n",
"print(\"🧠 Sample Learned Patterns:\\n\")\n",
"\n",
"# Get first 10 merges\n",
"for i, ((p0, p1), idx) in enumerate(list(tokenizer.merges.items())[:10]):\n",
" try:\n",
" pattern = tokenizer.vocab[p0].decode('utf-8', errors='ignore') + \\\n",
" tokenizer.vocab[p1].decode('utf-8', errors='ignore')\n",
" print(f\"Merge {i+1}: '{pattern}' -> Token {idx}\")\n",
" except:\n",
" print(f\"Merge {i+1}: Bytes ({p0}, {p1}) -> Token {idx}\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 📈 Example 6: Real-World Usage Simulation"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Simulate processing a day's worth of stock data\n",
"daily_data = \"\"\"AAPL|2024-01-15|150.25|152.30|149.80|151.50|1000000\n",
"AAPL|2024-01-16|151.60|153.20|151.00|152.80|1200000\n",
"AAPL|2024-01-17|152.90|154.50|152.00|153.75|980000\n",
"MSFT|2024-01-15|380.50|385.20|379.00|384.75|850000\n",
"MSFT|2024-01-16|385.00|388.50|384.00|387.25|920000\n",
"MSFT|2024-01-17|387.50|390.00|386.50|389.50|880000\"\"\"\n",
"\n",
"print(\"📊 Processing Daily Stock Data\\n\")\n",
"print(f\"Original size: {len(daily_data)} characters\")\n",
"\n",
"# Encode\n",
"tokens = tokenizer.encode(daily_data)\n",
"print(f\"Tokenized: {len(tokens)} tokens\")\n",
"\n",
"# Calculate savings\n",
"original_bytes = len(daily_data.encode('utf-8'))\n",
"token_bytes = len(tokens) * 2 # Assuming 2 bytes per token\n",
"savings = (1 - token_bytes / original_bytes) * 100\n",
"\n",
"print(f\"\\n💾 Storage Savings:\")\n",
"print(f\" Original: {original_bytes} bytes\")\n",
"print(f\" Tokenized: {token_bytes} bytes\")\n",
"print(f\" Savings: {savings:.1f}%\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 🎓 Summary\n",
"\n",
"### ✅ What We Learned\n",
"- How to load and use the Stock Market BPE tokenizer\n",
"- Encoding stock data into tokens\n",
"- Decoding tokens back to original format\n",
"- Calculating compression ratios\n",
"- Analyzing learned patterns\n",
"\n",
"### 📊 Key Metrics\n",
"- **Vocabulary Size:** 5,500+ tokens\n",
"- **Compression Ratio:** 3.5x average\n",
"- **Accuracy:** 100% (lossless encoding/decoding)\n",
"\n",
"### 🚀 Next Steps\n",
"- Use this tokenizer in ML models for stock prediction\n",
"- Compress large financial datasets\n",
"- Analyze learned patterns for market insights\n",
"\n",
"---\n",
"\n",
"**Happy tokenizing! 📈🤖**"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.8.0"
}
},
"nbformat": 4,
"nbformat_minor": 4
} |