Renamed demo script and added initial pre-trained test python notebook
Browse files- test.py → demo.py +0 -0
- test_pretrained.ipynb +303 -0
test.py → demo.py
RENAMED
|
File without changes
|
test_pretrained.ipynb
ADDED
|
@@ -0,0 +1,303 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cells": [
|
| 3 |
+
{
|
| 4 |
+
"cell_type": "markdown",
|
| 5 |
+
"metadata": {},
|
| 6 |
+
"source": [
|
| 7 |
+
"# Run pre-trained DeepSeek Coder 1.3B Model on Chat-GPT 4o generated dataset"
|
| 8 |
+
]
|
| 9 |
+
},
|
| 10 |
+
{
|
| 11 |
+
"cell_type": "markdown",
|
| 12 |
+
"metadata": {},
|
| 13 |
+
"source": [
|
| 14 |
+
"## First load dataset into pandas dataframe"
|
| 15 |
+
]
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"cell_type": "code",
|
| 19 |
+
"execution_count": 83,
|
| 20 |
+
"metadata": {},
|
| 21 |
+
"outputs": [
|
| 22 |
+
{
|
| 23 |
+
"name": "stdout",
|
| 24 |
+
"output_type": "stream",
|
| 25 |
+
"text": [
|
| 26 |
+
"Total dataset examples: 1044\n",
|
| 27 |
+
"\n",
|
| 28 |
+
"\n",
|
| 29 |
+
"What is the highest number of assists recorded by the Indiana Pacers in a single home game?\n",
|
| 30 |
+
"SELECT MAX(ast_home) FROM game WHERE team_name_home = 'Indiana Pacers';\n",
|
| 31 |
+
"44.0\n"
|
| 32 |
+
]
|
| 33 |
+
}
|
| 34 |
+
],
|
| 35 |
+
"source": [
|
| 36 |
+
"import pandas as pd \n",
|
| 37 |
+
"\n",
|
| 38 |
+
"# Load dataset and check length\n",
|
| 39 |
+
"df = pd.read_csv(\"./train-data/sql_train.tsv\", sep='\\t')\n",
|
| 40 |
+
"print(\"Total dataset examples: \" + str(len(df)))\n",
|
| 41 |
+
"print(\"\\n\")\n",
|
| 42 |
+
"\n",
|
| 43 |
+
"# Test sampling\n",
|
| 44 |
+
"sample = df.sample(n=1)\n",
|
| 45 |
+
"print(sample[\"natural_query\"].values[0])\n",
|
| 46 |
+
"print(sample[\"sql_query\"].values[0])\n",
|
| 47 |
+
"print(sample[\"result\"].values[0])"
|
| 48 |
+
]
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"cell_type": "markdown",
|
| 52 |
+
"metadata": {},
|
| 53 |
+
"source": [
|
| 54 |
+
"## Load pre-trained DeepSeek model using transformers and pytorch packages"
|
| 55 |
+
]
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"cell_type": "code",
|
| 59 |
+
"execution_count": 84,
|
| 60 |
+
"metadata": {},
|
| 61 |
+
"outputs": [],
|
| 62 |
+
"source": [
|
| 63 |
+
"from transformers import AutoTokenizer, AutoModelForCausalLM\n",
|
| 64 |
+
"import torch\n",
|
| 65 |
+
"\n",
|
| 66 |
+
"# Set device to cuda if available, otherwise CPU\n",
|
| 67 |
+
"device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
|
| 68 |
+
"\n",
|
| 69 |
+
"# Load model and tokenizer\n",
|
| 70 |
+
"tokenizer = AutoTokenizer.from_pretrained(\"./deepseek-coder-1.3b-instruct\")\n",
|
| 71 |
+
"model = AutoModelForCausalLM.from_pretrained(\"./deepseek-coder-1.3b-instruct\", torch_dtype=torch.bfloat16, device_map=device) "
|
| 72 |
+
]
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"cell_type": "markdown",
|
| 76 |
+
"metadata": {},
|
| 77 |
+
"source": [
|
| 78 |
+
"## Create prompt to setup the model for better performance"
|
| 79 |
+
]
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"cell_type": "code",
|
| 83 |
+
"execution_count": 85,
|
| 84 |
+
"metadata": {},
|
| 85 |
+
"outputs": [],
|
| 86 |
+
"source": [
|
| 87 |
+
"input_text = \"\"\"You are an AI assistant that generates SQLite queries for an NBA database based on user questions. The database consists of two tables:\n",
|
| 88 |
+
"\n",
|
| 89 |
+
"1. `team` - Stores information about NBA teams.\n",
|
| 90 |
+
" - `id`: Unique team identifier.\n",
|
| 91 |
+
" - `full_name`: Full team name (e.g., \"Los Angeles Lakers\").\n",
|
| 92 |
+
" - `abbreviation`: 3-letter team code (e.g., \"LAL\").\n",
|
| 93 |
+
" - `city`, `state`: Location of the team.\n",
|
| 94 |
+
" - `year_founded`: The year the team was founded.\n",
|
| 95 |
+
"\n",
|
| 96 |
+
"2. `game` - Stores details of individual games.\n",
|
| 97 |
+
" - `game_date`: Date of the game.\n",
|
| 98 |
+
" - `team_id_home`, `team_id_away`: Unique IDs of home and away teams.\n",
|
| 99 |
+
" - `team_name_home`, `team_name_away`: Full names of the teams.\n",
|
| 100 |
+
" - `pts_home`, `pts_away`: Points scored by home and away teams.\n",
|
| 101 |
+
" - `wl_home`: \"W\" if the home team won, \"L\" if they lost.\n",
|
| 102 |
+
" - `reb_home`, `reb_away`: Total rebounds.\n",
|
| 103 |
+
" - `ast_home`, `ast_away`: Total assists.\n",
|
| 104 |
+
" - Other statistics include field goals (`fgm_home`, `fg_pct_home`), three-pointers (`fg3m_home`), free throws (`ftm_home`), and turnovers (`tov_home`).\n",
|
| 105 |
+
"\n",
|
| 106 |
+
"### Instructions:\n",
|
| 107 |
+
"- Generate a valid SQLite query to retrieve relevant data from the database.\n",
|
| 108 |
+
"- Use column names correctly based on the provided schema.\n",
|
| 109 |
+
"- Ensure the query is well-structured and avoids unnecessary joins.\n",
|
| 110 |
+
"- Format the query with proper indentation.\n",
|
| 111 |
+
"\n",
|
| 112 |
+
"### Example Queries:\n",
|
| 113 |
+
"User: \"What is the most points the Los Angeles Lakers have ever scored at home?\"\n",
|
| 114 |
+
"SQLite:\n",
|
| 115 |
+
"SELECT MAX(pts_home) \n",
|
| 116 |
+
"FROM game \n",
|
| 117 |
+
"WHERE team_name_home = 'Los Angeles Lakers';\n",
|
| 118 |
+
"\n",
|
| 119 |
+
"User: \"List all games where the Golden State Warriors scored more than 130 points.\" \n",
|
| 120 |
+
"SQLite:\n",
|
| 121 |
+
"SELECT game_date, team_name_home, pts_home, team_name_away, pts_away\n",
|
| 122 |
+
"FROM game\n",
|
| 123 |
+
"WHERE (team_name_home = 'Golden State Warriors' AND pts_home > 130)\n",
|
| 124 |
+
" OR (team_name_away = 'Golden State Warriors' AND pts_away > 130);\n",
|
| 125 |
+
" \n",
|
| 126 |
+
"Now, generate a SQL query based on the following user request: \"\"\""
|
| 127 |
+
]
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"cell_type": "markdown",
|
| 131 |
+
"metadata": {},
|
| 132 |
+
"source": [
|
| 133 |
+
"## Test model performance on a single example"
|
| 134 |
+
]
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"cell_type": "code",
|
| 138 |
+
"execution_count": 86,
|
| 139 |
+
"metadata": {},
|
| 140 |
+
"outputs": [
|
| 141 |
+
{
|
| 142 |
+
"name": "stderr",
|
| 143 |
+
"output_type": "stream",
|
| 144 |
+
"text": [
|
| 145 |
+
"c:\\Users\\Dean\\AppData\\Local\\Programs\\Python\\Python312\\Lib\\site-packages\\transformers\\generation\\configuration_utils.py:634: UserWarning: `do_sample` is set to `False`. However, `top_p` is set to `0.95` -- this flag is only used in sample-based generation modes. You should set `do_sample=True` or unset `top_p`.\n",
|
| 146 |
+
" warnings.warn(\n",
|
| 147 |
+
"The attention mask and the pad token id were not set. As a consequence, you may observe unexpected behavior. Please pass your input's `attention_mask` to obtain reliable results.\n",
|
| 148 |
+
"Setting `pad_token_id` to `eos_token_id`:32021 for open-end generation.\n"
|
| 149 |
+
]
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"name": "stdout",
|
| 153 |
+
"output_type": "stream",
|
| 154 |
+
"text": [
|
| 155 |
+
"SQLite:\n",
|
| 156 |
+
"SELECT MAX(ast_home) \n",
|
| 157 |
+
"FROM game \n",
|
| 158 |
+
"WHERE team_name_home = 'Indiana Pacers';\n",
|
| 159 |
+
"\n"
|
| 160 |
+
]
|
| 161 |
+
}
|
| 162 |
+
],
|
| 163 |
+
"source": [
|
| 164 |
+
"# Create message with sample query and run model\n",
|
| 165 |
+
"message=[{ 'role': 'user', 'content': input_text + sample[\"natural_query\"].values[0]}]\n",
|
| 166 |
+
"inputs = tokenizer.apply_chat_template(message, add_generation_prompt=True, return_tensors=\"pt\").to(model.device)\n",
|
| 167 |
+
"outputs = model.generate(inputs, max_new_tokens=512, do_sample=False, top_k=50, top_p=0.95, num_return_sequences=1, eos_token_id=tokenizer.eos_token_id)\n",
|
| 168 |
+
"\n",
|
| 169 |
+
"# Print output\n",
|
| 170 |
+
"query_output = tokenizer.decode(outputs[0][len(inputs[0]):], skip_special_tokens=True)\n",
|
| 171 |
+
"print(query_output)"
|
| 172 |
+
]
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"cell_type": "markdown",
|
| 176 |
+
"metadata": {},
|
| 177 |
+
"source": [
|
| 178 |
+
"# Test sample output on sqlite3 database"
|
| 179 |
+
]
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"cell_type": "code",
|
| 183 |
+
"execution_count": null,
|
| 184 |
+
"metadata": {},
|
| 185 |
+
"outputs": [
|
| 186 |
+
{
|
| 187 |
+
"name": "stdout",
|
| 188 |
+
"output_type": "stream",
|
| 189 |
+
"text": [
|
| 190 |
+
"cleaned\n",
|
| 191 |
+
"(44.0,)\n"
|
| 192 |
+
]
|
| 193 |
+
}
|
| 194 |
+
],
|
| 195 |
+
"source": [
|
| 196 |
+
"import sqlite3 as sql\n",
|
| 197 |
+
"\n",
|
| 198 |
+
"# Create connection to sqlite3 database\n",
|
| 199 |
+
"connection = sql.connect('./nba-data/nba.sqlite')\n",
|
| 200 |
+
"cursor = connection.cursor()\n",
|
| 201 |
+
"\n",
|
| 202 |
+
"# Execute query from model output and print result\n",
|
| 203 |
+
"if query_output[0:7] == \"SQLite:\":\n",
|
| 204 |
+
" print(\"cleaned\")\n",
|
| 205 |
+
" query = query_output[7:]\n",
|
| 206 |
+
"elif query_output[0:4] == \"SQL:\":\n",
|
| 207 |
+
" query = query_output[4:]\n",
|
| 208 |
+
"else:\n",
|
| 209 |
+
" query = query_output\n",
|
| 210 |
+
"cursor.execute(query)\n",
|
| 211 |
+
"rows = cursor.fetchall()\n",
|
| 212 |
+
"for row in rows:\n",
|
| 213 |
+
" print(row)"
|
| 214 |
+
]
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"cell_type": "markdown",
|
| 218 |
+
"metadata": {},
|
| 219 |
+
"source": [
|
| 220 |
+
"## Create function to compare output to ground truth result from examples"
|
| 221 |
+
]
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"cell_type": "code",
|
| 225 |
+
"execution_count": null,
|
| 226 |
+
"metadata": {},
|
| 227 |
+
"outputs": [
|
| 228 |
+
{
|
| 229 |
+
"name": "stdout",
|
| 230 |
+
"output_type": "stream",
|
| 231 |
+
"text": [
|
| 232 |
+
"cleaned\n",
|
| 233 |
+
"[(44.0,)]\n",
|
| 234 |
+
"\n",
|
| 235 |
+
"SELECT MAX(ast_home) \n",
|
| 236 |
+
"FROM game \n",
|
| 237 |
+
"WHERE team_name_home = 'Indiana Pacers';\n",
|
| 238 |
+
"\n",
|
| 239 |
+
"SELECT MAX(ast_home) FROM game WHERE team_name_home = 'Indiana Pacers';\n",
|
| 240 |
+
"44.0\n",
|
| 241 |
+
"44.0\n",
|
| 242 |
+
"SQL matched? True\n",
|
| 243 |
+
"Result matched? True\n"
|
| 244 |
+
]
|
| 245 |
+
}
|
| 246 |
+
],
|
| 247 |
+
"source": [
|
| 248 |
+
"def compare_result(sample_query, sample_result, query_output):\n",
|
| 249 |
+
" # Clean model output to only have the query output\n",
|
| 250 |
+
" if query_output[0:7] == \"SQLite:\":\n",
|
| 251 |
+
" query = query_output[7:]\n",
|
| 252 |
+
" elif query_output[0:4] == \"SQL:\":\n",
|
| 253 |
+
" query = query_output[4:]\n",
|
| 254 |
+
" else:\n",
|
| 255 |
+
" query = query_output\n",
|
| 256 |
+
" \n",
|
| 257 |
+
" # Try to execute query, if it fails, then this is a failure of the model\n",
|
| 258 |
+
" try:\n",
|
| 259 |
+
" # Execute query and obtain result\n",
|
| 260 |
+
" cursor.execute(query)\n",
|
| 261 |
+
" rows = cursor.fetchall()\n",
|
| 262 |
+
"\n",
|
| 263 |
+
" # Check if this is a multi-line query\n",
|
| 264 |
+
" if \"|\" in sample_result:\n",
|
| 265 |
+
" return True, True\n",
|
| 266 |
+
" else:\n",
|
| 267 |
+
" # Strip all whitespace before comparing queries since there may be differences in spacing, newlines, tabs, etc.\n",
|
| 268 |
+
" query = query.replace(\" \", \"\").replace(\"\\n\", \"\").replace(\"\\t\", \"\")\n",
|
| 269 |
+
" sample_query = sample_query.replace(\" \", \"\").replace(\"\\n\", \"\").replace(\"\\t\", \"\")\n",
|
| 270 |
+
"\n",
|
| 271 |
+
" # Compare results and return\n",
|
| 272 |
+
" return (query == sample_query), (str(rows[0][0]) == str(sample_result))\n",
|
| 273 |
+
" except:\n",
|
| 274 |
+
" return False, False\n",
|
| 275 |
+
"\n",
|
| 276 |
+
"result = compare_result(sample[\"sql_query\"].values[0], sample[\"result\"].values[0], query_output)\n",
|
| 277 |
+
"print(\"SQL matched? \" + str(result[0]))\n",
|
| 278 |
+
"print(\"Result matched? \" + str(result[1]))"
|
| 279 |
+
]
|
| 280 |
+
}
|
| 281 |
+
],
|
| 282 |
+
"metadata": {
|
| 283 |
+
"kernelspec": {
|
| 284 |
+
"display_name": "Python 3",
|
| 285 |
+
"language": "python",
|
| 286 |
+
"name": "python3"
|
| 287 |
+
},
|
| 288 |
+
"language_info": {
|
| 289 |
+
"codemirror_mode": {
|
| 290 |
+
"name": "ipython",
|
| 291 |
+
"version": 3
|
| 292 |
+
},
|
| 293 |
+
"file_extension": ".py",
|
| 294 |
+
"mimetype": "text/x-python",
|
| 295 |
+
"name": "python",
|
| 296 |
+
"nbconvert_exporter": "python",
|
| 297 |
+
"pygments_lexer": "ipython3",
|
| 298 |
+
"version": "3.12.6"
|
| 299 |
+
}
|
| 300 |
+
},
|
| 301 |
+
"nbformat": 4,
|
| 302 |
+
"nbformat_minor": 2
|
| 303 |
+
}
|