Skip to content

Instantly share code, notes, and snippets.

@yamini

yamini/safe-synthetics-wandb-sweep.ipynb

Created August 14, 2025 15:57
Show Gist options
  • Save yamini/6b0933e7f34b774761cfa3ba1f7ce75c to your computer and use it in GitHub Desktop.
Save yamini/6b0933e7f34b774761cfa3ba1f7ce75c to your computer and use it in GitHub Desktop.
Download ZIP
Safe Synthetics hyperparam optimization using Weights & Biases
Display the source blob
Display the rendered blob
Raw
safe-synthetics-wandb-sweep.ipynb
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Safe Synthetics Hyperparameter Optimization with Weights & Biases\n",
"\n",
"This notebook demonstrates how to optimize Safe Synthetics model parameters using Weights & Biases (W&B) sweeps for automated hyperparameter tuning.\n",
"\n",
"## What this notebook does:\n",
"\n",
"**🎯 Automated Parameter Optimization**: Uses Bayesian optimization to find the best Safe Synthetics parameters for your dataset\n",
"\n",
"**📊 Built-in Evaluation**: Leverages Safe Synthetics' automatic data quality evaluation and reporting\n",
"\n",
"**🔍 Tracking**: Logs all experiments, parameters, and quality scores to W&B for easy comparison and analysis\n",
"\n",
"## How it works:\n",
"1. **Configure** your dataset and optimization parameters\n",
"2. **Launch** multiple Safe Synthetics training jobs with different parameter combinations \n",
"3. **Evaluate** synthetic data quality automatically using built-in metrics\n",
"4. **Optimize** parameters based on quality scores using Bayesian optimization\n",
"5. **Analyze** results through W&B's visualization dashboard\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 1) Install Dependencies\n",
"\n",
"Install the required Python packages for Safe Synthetics and Weights & Biases integration.\n"
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"%%capture\n",
"%pip install -U gretel-client wandb"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 2) Configure Gretel & Weights & Biases\n",
"\n",
"Set up authentication and project configuration for both Gretel and W&B services. The notebook will prompt for API keys if not already configured in your environment.\n",
"\n",
"**Configuration Options**:\n",
"- `MAX_TRIALS`: Number of different parameter combinations to test (default: 2) \n"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Logged in as [email protected] ✅\n",
"Using project: safe-syn-sweeps-ff62124b804335a\n",
"Project link: https://console.gretel.ai/proj_31FkDXkE3FFoXai92RsAAeoLW5l\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"\u001b[34m\u001b[1mwandb\u001b[0m: \u001b[33mWARNING\u001b[0m If you're specifying your api key in code, ensure this code is not shared publicly.\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \u001b[33mWARNING\u001b[0m Consider setting the WANDB_API_KEY environment variable, or running `wandb login` from the command line.\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: Appending key for api.wandb.ai to your netrc file: /Users/ykagal/.netrc\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: Currently logged in as: \u001b[33myamini_gretel\u001b[0m to \u001b[32mhttps://api.wandb.ai\u001b[0m. Use \u001b[1m`wandb login --relogin`\u001b[0m to force relogin\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"✅ Configured Gretel project: safe-syn-sweeps\n",
"✅ Configured W&B project: safe-syn-sweeps\n",
"✅ Entity: yamini_gretel\n"
]
}
],
"source": [
"import os, json, pathlib, uuid, time\n",
"from typing import Any, Dict, Optional\n",
"import pandas as pd\n",
"import concurrent.futures\n",
"import threading\n",
"from queue import Queue\n",
"\n",
"# Gretel Navigator client\n",
"from gretel_client.navigator_client import Gretel\n",
"\n",
"# --- Gretel config ---\n",
"DEFAULT_PROJECT_ID = os.getenv(\"GRETEL_PROJECT\", \"safe-syn-sweeps\")\n",
"gretel = Gretel(api_key=\"prompt\", default_project_id=DEFAULT_PROJECT_ID)\n",
"\n",
"# --- W&B config login ---\n",
"import wandb\n",
"from getpass import getpass\n",
"\n",
"WANDB_PROJECT = os.getenv(\"WANDB_PROJECT\", \"safe-syn-sweeps\")\n",
"WANDB_ENTITY = os.getenv(\"WANDB_ENTITY\", None) # optional\n",
"RUN_PREFIX = os.getenv(\"RUN_PREFIX\", \"safe-syn\")\n",
"MAX_TRIALS = int(os.getenv(\"MAX_TRIALS\", \"2\")) # Number of parameter combinations to test\n",
"\n",
"# Prompt for W&B API key if not already set\n",
"wandb_api_key = os.getenv(\"WANDB_API_KEY\")\n",
"if not wandb_api_key:\n",
" try:\n",
" wandb_api_key = getpass(\"Enter your Weights & Biases API key: \")\n",
" except Exception:\n",
" wandb_api_key = input(\"Enter your Weights & Biases API key: \")\n",
"wandb.login(key=wandb_api_key)\n",
"\n",
"# Resolve default entity if none provided\n",
"api = wandb.Api()\n",
"if WANDB_ENTITY is None:\n",
" try:\n",
" WANDB_ENTITY = api.default_entity\n",
" except Exception:\n",
" WANDB_ENTITY = None\n",
"\n",
"print(f\"✅ Configured Gretel project: {DEFAULT_PROJECT_ID}\")\n",
"print(f\"✅ Configured W&B project: {WANDB_PROJECT}\")\n",
"print(f\"✅ Entity: {WANDB_ENTITY}\")\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 3) Load Your Dataset\n",
"\n",
"Specify your training dataset for Safe Synthetics optimization. You can use the default public dataset or point to your own CSV/Parquet file by setting the `TRAINING_DATA_SOURCE` environment variable.\n"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Loaded training data with 500 rows and 8 columns\n",
"Columns: ['Email', 'Address', 'Avatar', 'Avg. Session Length', 'Time on App', 'Time on Website', 'Length of Membership', 'Yearly Amount Spent']\n"
]
},
{
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>Email</th>\n",
" <th>Address</th>\n",
" <th>Avatar</th>\n",
" <th>Avg. Session Length</th>\n",
" <th>Time on App</th>\n",
" <th>Time on Website</th>\n",
" <th>Length of Membership</th>\n",
" <th>Yearly Amount Spent</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>0</th>\n",
" <td>[email protected]</td>\n",
" <td>835 Frank Tunnel\\nWrightmouth, MI 82180-9605</td>\n",
" <td>Violet</td>\n",
" <td>34.497268</td>\n",
" <td>12.655651</td>\n",
" <td>39.577668</td>\n",
" <td>4.082621</td>\n",
" <td>587.951054</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>[email protected]</td>\n",
" <td>4547 Archer Common\\nDiazchester, CA 06566-8576</td>\n",
" <td>DarkGreen</td>\n",
" <td>31.926272</td>\n",
" <td>11.109461</td>\n",
" <td>37.268959</td>\n",
" <td>2.664034</td>\n",
" <td>392.204933</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>[email protected]</td>\n",
" <td>24645 Valerie Unions Suite 582\\nCobbborough, D...</td>\n",
" <td>Bisque</td>\n",
" <td>33.000915</td>\n",
" <td>11.330278</td>\n",
" <td>37.110597</td>\n",
" <td>4.104543</td>\n",
" <td>487.547505</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>[email protected]</td>\n",
" <td>1414 David Throughway\\nPort Jason, OH 22070-1220</td>\n",
" <td>SaddleBrown</td>\n",
" <td>34.305557</td>\n",
" <td>13.717514</td>\n",
" <td>36.721283</td>\n",
" <td>3.120179</td>\n",
" <td>581.852344</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>[email protected]</td>\n",
" <td>14023 Rodriguez Passage\\nPort Jacobville, PR 3...</td>\n",
" <td>MediumAquaMarine</td>\n",
" <td>33.330673</td>\n",
" <td>12.795189</td>\n",
" <td>37.536653</td>\n",
" <td>4.446308</td>\n",
" <td>599.406092</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"</div>"
],
"text/plain": [
" Email \\\n",
"0 [email protected] \n",
"1 [email protected] \n",
"2 [email protected] \n",
"3 [email protected] \n",
"4 [email protected] \n",
"\n",
" Address Avatar \\\n",
"0 835 Frank Tunnel\\nWrightmouth, MI 82180-9605 Violet \n",
"1 4547 Archer Common\\nDiazchester, CA 06566-8576 DarkGreen \n",
"2 24645 Valerie Unions Suite 582\\nCobbborough, D... Bisque \n",
"3 1414 David Throughway\\nPort Jason, OH 22070-1220 SaddleBrown \n",
"4 14023 Rodriguez Passage\\nPort Jacobville, PR 3... MediumAquaMarine \n",
"\n",
" Avg. Session Length Time on App Time on Website Length of Membership \\\n",
"0 34.497268 12.655651 39.577668 4.082621 \n",
"1 31.926272 11.109461 37.268959 2.664034 \n",
"2 33.000915 11.330278 37.110597 4.104543 \n",
"3 34.305557 13.717514 36.721283 3.120179 \n",
"4 33.330673 12.795189 37.536653 4.446308 \n",
"\n",
" Yearly Amount Spent \n",
"0 587.951054 \n",
"1 392.204933 \n",
"2 487.547505 \n",
"3 581.852344 \n",
"4 599.406092 "
]
},
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# Use the same public dataset or point to your own CSV/Parquet file\n",
"TRAINING_DATA_SOURCE = os.getenv(\n",
" \"TRAINING_DATA_SOURCE\",\n",
" \"https://gretel-datasets.s3.us-west-2.amazonaws.com/ecommerce_customers.csv\",\n",
")\n",
"\n",
"df = pd.read_csv(TRAINING_DATA_SOURCE)\n",
"print(f\"Loaded training data with {len(df):,} rows and {len(df.columns)} columns\")\n",
"print(f\"Columns: {list(df.columns)}\")\n",
"df.head()\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 4) Helper Functions\n",
"\n",
"Core functions that handle Safe Synthetics training and evaluation:\n",
"\n",
"- **`run_safe_syn_training_with_builtin_holdout()`**: Creates and runs a Safe Synthetics job with automatic 5% holdout for evaluation\n",
"- **`get_quality_score_from_job()`**: Extracts the overall quality score from Safe Synthetics' built-in evaluation report \n",
"- **`_extract_metric()`**: Utility function to safely extract nested metrics from evaluation reports\n"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"✅ Helper functions loaded:\n",
" - run_safe_syn_training_with_builtin_holdout: Creates Safe Synthetics jobs with automatic evaluation\n",
" - get_quality_score_from_job: Extracts quality scores from completed jobs\n",
" - _extract_metric: Utility for parsing evaluation reports\n",
"🚀 Ready for hyperparameter optimization!\n"
]
}
],
"source": [
"def run_safe_syn_training_with_builtin_holdout(params: Dict[str, Any], run_label: str) -> Any:\n",
" \"\"\"Create and run a Safe Synthetics job with the given parameters.\n",
" \n",
" Uses Safe Synthetics' built-in holdout (5%) and evaluation features.\n",
" \"\"\"\n",
" # Create Safe Synthetics workflow with automatic holdout\n",
" chain = (\n",
" gretel.safe_synthetic_dataset\n",
" .from_data_source(TRAINING_DATA_SOURCE, holdout=0.05) # 5% automatic holdout\n",
" .synthesize() # Generate synthetic data\n",
" )\n",
" \n",
" # Apply hyperparameters to the Safe Synthetics job\n",
" try:\n",
" chain = chain.with_parameters(params)\n",
" except Exception:\n",
" # Continue without parameters if not supported\n",
" pass\n",
"\n",
" try:\n",
" job = chain.create(run_name=run_label)\n",
" except TypeError:\n",
" job = chain.create()\n",
"\n",
" # Wait for job completion\n",
" try:\n",
" job.wait_until_done()\n",
" except Exception:\n",
" time.sleep(5)\n",
"\n",
" # Return the completed job (includes built-in evaluation)\n",
" return job\n",
"\n",
"\n",
"def get_quality_score_from_job(job: Any) -> dict:\n",
" \"\"\"Extract quality and privacy scores from a completed Safe Synthetics job.\"\"\"\n",
" try:\n",
" # For WorkflowRun objects, access the report directly\n",
" if hasattr(job, 'report') and hasattr(job.report, 'dict'):\n",
" report_dict = job.report.dict\n",
" \n",
" scores = {}\n",
" if 'synthetic_quality_score' in report_dict:\n",
" scores['quality_score'] = float(report_dict['synthetic_quality_score'])\n",
" if 'data_privacy_score' in report_dict:\n",
" scores['privacy_score'] = float(report_dict['data_privacy_score'])\n",
" \n",
" if scores:\n",
" return scores\n",
" \n",
" raise RuntimeError(\"Could not extract scores from job report\")\n",
" \n",
" except Exception as e:\n",
" print(f\"⚠️ Error extracting scores: {e}\")\n",
" raise\n",
"\n",
"def evaluate_synthetic_vs_holdout(synthetic_file: Any, holdout_file: Any) -> Dict[str, Any]:\n",
" wf = gretel.workflows.builder()\n",
" wf.add_step(gretel.tasks.EvaluateSafeSyntheticsDataset(), [synthetic_file.id, holdout_file.id], step_name=\"evaluate\")\n",
" results = wf.run(wait_until_done=True)\n",
"\n",
" # Extract the evaluation report in a flexible way\n",
" report = None\n",
" # Newer SDKs: results.report.to_json()\n",
" try:\n",
" report = results.report.to_json()\n",
" except Exception:\n",
" pass\n",
" # Named step output\n",
" if report is None:\n",
" try:\n",
" report = results.outputs[\"evaluate\"].report.to_json()\n",
" except Exception:\n",
" pass\n",
" # Fallback to raw dict if available\n",
" if report is None:\n",
" report = getattr(results, \"report\", None)\n",
" if report is None:\n",
" raise RuntimeError(\"Could not retrieve evaluation report. Inspect `results`.\")\n",
" return report\n",
"\n",
"\n",
"def _extract_metric(report: Dict[str, Any], dotted_key: str) -> float:\n",
" \"\"\"Extract a nested metric value from an evaluation report using dot notation.\"\"\"\n",
" cur: Any = report\n",
" for part in dotted_key.split(\".\"):\n",
" if not isinstance(cur, dict) or part not in cur:\n",
" raise KeyError(f\"Metric key '{dotted_key}' not found at '{part}'. Top-level keys: {list(cur.keys()) if isinstance(cur, dict) else cur}\")\n",
" cur = cur[part]\n",
" if not isinstance(cur, (int, float)):\n",
" raise TypeError(f\"Metric '{dotted_key}' is not numeric: {cur}\")\n",
" return float(cur)\n",
"\n",
"\n",
"print(\"✅ Helper functions loaded:\")\n",
"print(\" - run_safe_syn_training_with_builtin_holdout: Creates Safe Synthetics jobs with automatic evaluation\")\n",
"print(\" - get_quality_score_from_job: Extracts quality scores from completed jobs\") \n",
"print(\" - _extract_metric: Utility for parsing evaluation reports\")\n",
"print(\"🚀 Ready for hyperparameter optimization!\")\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 5) Hyperparameter Search Configuration\n",
"\n",
"Define the parameter space for Bayesian optimization using Weights & Biases sweeps. This configuration specifies which Safe Synthetics parameters to optimize and their ranges.\n",
"\n",
"**Optimization Strategy**: Bayesian optimization efficiently explores the parameter space to maximize synthetic data quality.\n"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"🎯 W&B Sweep Configuration:\n",
"{\n",
" \"name\": \"safe-syn-tabular-ft-sweep\",\n",
" \"method\": \"bayes\",\n",
" \"metric\": {\n",
" \"name\": \"quality_score\",\n",
" \"goal\": \"maximize\"\n",
" },\n",
" \"parameters\": {\n",
" \"num_records\": {\n",
" \"values\": [\n",
" 500,\n",
" 1000,\n",
" 2000\n",
" ]\n",
" },\n",
" \"learning_rate\": {\n",
" \"min\": 0.0001,\n",
" \"max\": 0.001,\n",
" \"distribution\": \"log_uniform_values\"\n",
" },\n",
" \"batch_size\": {\n",
" \"values\": [\n",
" 1,\n",
" 2,\n",
" 4\n",
" ]\n",
" },\n",
" \"num_train_epochs\": {\n",
" \"values\": [\n",
" 3,\n",
" 5,\n",
" 8\n",
" ]\n",
" },\n",
" \"gradient_accumulation_steps\": {\n",
" \"values\": [\n",
" 4,\n",
" 8,\n",
" 16\n",
" ]\n",
" },\n",
" \"weight_decay\": {\n",
" \"min\": 0.001,\n",
" \"max\": 0.1,\n",
" \"distribution\": \"log_uniform_values\"\n",
" },\n",
" \"warmup_ratio\": {\n",
" \"min\": 0.01,\n",
" \"max\": 0.1\n",
" },\n",
" \"lora_r\": {\n",
" \"values\": [\n",
" 16,\n",
" 32,\n",
" 64\n",
" ]\n",
" },\n",
" \"lora_alpha\": {\n",
" \"values\": [\n",
" 16,\n",
" 32,\n",
" 64,\n",
" 128\n",
" ]\n",
" },\n",
" \"lora_dropout\": {\n",
" \"min\": 0.01,\n",
" \"max\": 0.1\n",
" },\n",
" \"temperature\": {\n",
" \"min\": 0.7,\n",
" \"max\": 1.0\n",
" },\n",
" \"top_p\": {\n",
" \"min\": 0.9,\n",
" \"max\": 1.0\n",
" },\n",
" \"repetition_penalty\": {\n",
" \"min\": 1.0,\n",
" \"max\": 1.2\n",
" },\n",
" \"max_input_records\": {\n",
" \"values\": [\n",
" 5000,\n",
" 10000,\n",
" 25000\n",
" ]\n",
" }\n",
" },\n",
" \"early_terminate\": {\n",
" \"type\": \"hyperband\",\n",
" \"min_iter\": 1\n",
" }\n",
"}\n",
"\n",
"✅ Sweep configured for 2 trials\n",
" - Method: bayes\n",
" - Metric: quality_score (maximize)\n",
" - Parameters: 14 hyperparameters\n"
]
}
],
"source": [
"# W&B Sweep Configuration with Safe Synthetics + Tabular Fine-Tuning parameters\n",
"# Simplified flat structure for easier parameter passing\n",
"# Based on: https://docs.gretel.ai/create-synthetic-data/safe-synthetics/synthetics/gretel-tabular-fine-tuning\n",
"SWEEP_CONFIG = {\n",
" \"name\": \"safe-syn-tabular-ft-sweep\",\n",
" \"method\": \"bayes\", # Bayesian optimization for efficient search\n",
" \"metric\": {\"name\": \"quality_score\", \"goal\": \"maximize\"},\n",
" \"parameters\": {\n",
" # Safe Synthetics generation parameters\n",
" \"num_records\": {\n",
" \"values\": [500, 1000, 2000] # Number of synthetic records to generate\n",
" },\n",
" \n",
" # Core LoRA training parameters (flattened for easier passing)\n",
" \"learning_rate\": {\n",
" \"min\": 0.0001,\n",
" \"max\": 0.001,\n",
" \"distribution\": \"log_uniform_values\"\n",
" },\n",
" \"batch_size\": {\n",
" \"values\": [1, 2, 4] # Small batch sizes for tabular fine-tuning\n",
" },\n",
" \"num_train_epochs\": {\n",
" \"values\": [3, 5, 8] # Training epochs\n",
" },\n",
" \"gradient_accumulation_steps\": {\n",
" \"values\": [4, 8, 16]\n",
" },\n",
" \"weight_decay\": {\n",
" \"min\": 0.001,\n",
" \"max\": 0.1,\n",
" \"distribution\": \"log_uniform_values\"\n",
" },\n",
" \"warmup_ratio\": {\n",
" \"min\": 0.01,\n",
" \"max\": 0.1\n",
" },\n",
" \n",
" # LoRA-specific parameters\n",
" \"lora_r\": {\n",
" \"values\": [16, 32, 64] # LoRA rank - key parameter for adaptation\n",
" },\n",
" \"lora_alpha\": {\n",
" \"values\": [16, 32, 64, 128] # LoRA alpha parameter\n",
" },\n",
" \"lora_dropout\": {\n",
" \"min\": 0.01,\n",
" \"max\": 0.1\n",
" },\n",
" \n",
" # Generation parameters\n",
" \"temperature\": {\n",
" \"min\": 0.7,\n",
" \"max\": 1.0 # Controls randomness in generation\n",
" },\n",
" \"top_p\": {\n",
" \"min\": 0.9,\n",
" \"max\": 1.0 # Nucleus sampling parameter\n",
" },\n",
" \"repetition_penalty\": {\n",
" \"min\": 1.0,\n",
" \"max\": 1.2 # Prevents repetitive generation\n",
" },\n",
" \n",
" # Data sampling\n",
" \"max_input_records\": {\n",
" \"values\": [5000, 10000, 25000] # Input data sampling limit\n",
" }\n",
" },\n",
" \"early_terminate\": {\n",
" \"type\": \"hyperband\",\n",
" \"min_iter\": 1\n",
" }\n",
"}\n",
"\n",
"print(\"🎯 W&B Sweep Configuration:\")\n",
"print(json.dumps(SWEEP_CONFIG, indent=2))\n",
"print(f\"\\n✅ Sweep configured for {MAX_TRIALS} trials\")\n",
"print(f\" - Method: {SWEEP_CONFIG['method']}\")\n",
"print(f\" - Metric: {SWEEP_CONFIG['metric']['name']} ({SWEEP_CONFIG['metric']['goal']})\")\n",
"print(f\" - Parameters: {len(SWEEP_CONFIG['parameters'])} hyperparameters\")\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 6) Trial Execution Function\n",
"\n",
"This function defines how each optimization trial runs. Each trial tests one unique set of parameters suggested by W&B's Bayesian optimization, allowing proper comparison between different parameter combinations.\n"
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"✅ Trial function ready:\n",
" - Each trial tests one unique parameter combination\n",
" - Uses automatic holdout and evaluation\n",
" - Logs quality scores to Weights & Biases for comparison\n",
"🚀 Ready to start hyperparameter optimization!\n"
]
}
],
"source": [
"def trial_agent():\n",
" \"\"\"Execute a single optimization trial with one set of parameters.\"\"\"\n",
" run = wandb.init(project=WANDB_PROJECT, entity=WANDB_ENTITY, name=f\"{RUN_PREFIX}-{wandb.util.generate_id()}\")\n",
" params = dict(wandb.config)\n",
" \n",
" print(f\"🚀 Starting trial: {run.name}\")\n",
" print(f\"📊 Parameters: {params}\")\n",
" \n",
" try:\n",
" # Run a single Safe Synthetics job with this parameter set\n",
" job = run_safe_syn_training_with_builtin_holdout(params=params, run_label=run.name)\n",
" print(f\"✅ Training complete!\")\n",
" \n",
" # Extract quality score from built-in evaluation\n",
" scores = get_quality_score_from_job(job)\n",
" print(f\"🎯 Quality score: {scores['quality_score']}\")\n",
" print(f\"🔒 Privacy score: {scores['privacy_score']}\")\n",
"\n",
" # Log results to Weights & Biases\n",
" wandb.log({\n",
" \"quality_score\": scores['quality_score'],\n",
" \"privacy_score\": scores['privacy_score']\n",
" })\n",
" wandb.summary.update({\n",
" \"quality_score\": scores['quality_score'],\n",
" \"privacy_score\": scores['privacy_score']\n",
" })\n",
" \n",
" print(f\"✅ Trial complete → score={score}\")\n",
" \n",
" except Exception as e:\n",
" print(f\"❌ Trial failed: {e}\")\n",
" wandb.log({\"error\": str(e)})\n",
" raise\n",
" finally:\n",
" # Close the W&B run\n",
" wandb.finish()\n",
"\n",
"\n",
"print(\"✅ Trial function ready:\")\n",
"print(f\" - Each trial tests one unique parameter combination\")\n",
"print(f\" - Uses automatic holdout and evaluation\") \n",
"print(f\" - Logs quality scores to Weights & Biases for comparison\")\n",
"print(\"🚀 Ready to start hyperparameter optimization!\")\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 7) Configuration Summary\n",
"\n",
"Review your current configuration settings before running the optimization.\n"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"📋 Current Configuration:\n",
"📊 Dataset: https://gretel-datasets.s3.us-west-2.amazonaws.com/ecommerce_customers.csv\n",
"📈 Optimization trials: 2\n",
"🎯 Optimization method: bayes\n",
"📋 Parameters to optimize: 14\n",
"\n",
"🔗 W&B Project: safe-syn-sweeps\n",
"👤 W&B Entity: yamini_gretel\n",
"\n",
"🎛️ Key parameters being optimized:\n",
" - learning_rate: 0.0001 to 0.001\n",
" - batch_size: [1, 2, 4]\n",
" - temperature: 0.7 to 1.0\n",
" - lora_r: [16, 32, 64]\n",
" - num_records: [500, 1000, 2000]\n",
"\n",
"✅ Configuration ready for hyperparameter optimization!\n"
]
}
],
"source": [
"# Display current configuration for review\n",
"print(\"📋 Current Configuration:\")\n",
"\n",
"print(f\"📊 Dataset: {TRAINING_DATA_SOURCE}\")\n",
"print(f\"📈 Optimization trials: {MAX_TRIALS}\")\n",
"print(f\"🎯 Optimization method: {SWEEP_CONFIG['method']}\")\n",
"print(f\"📋 Parameters to optimize: {len(SWEEP_CONFIG['parameters'])}\")\n",
"\n",
"print(f\"\\n🔗 W&B Project: {WANDB_PROJECT}\")\n",
"print(f\"👤 W&B Entity: {WANDB_ENTITY}\")\n",
"\n",
"# Show key parameters being optimized\n",
"key_params = ['learning_rate', 'batch_size', 'temperature', 'lora_r', 'num_records']\n",
"print(f\"\\n🎛️ Key parameters being optimized:\")\n",
"for param in key_params:\n",
" if param in SWEEP_CONFIG['parameters']:\n",
" config = SWEEP_CONFIG['parameters'][param]\n",
" if 'values' in config:\n",
" print(f\" - {param}: {config['values']}\")\n",
" elif 'min' in config and 'max' in config:\n",
" print(f\" - {param}: {config['min']} to {config['max']}\")\n",
"\n",
"print(f\"\\n✅ Configuration ready for hyperparameter optimization!\")\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 8) Visualizing Results in W&B Dashboard\n",
"\n",
"Tips for creating effective visualizations of your hyperparameter optimization results in the Weights & Biases dashboard.\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"📊 W&B Dashboard Configuration Tips\n",
"==================================================\n",
"\n",
"🎨 Creating Effective W&B Visualizations:\n",
"1. Navigate to your W&B sweep dashboard\n",
"2. Add recommended chart types for comprehensive analysis\n",
"3. Use these panel configurations for best insights:\n",
"\n",
"📈 Recommended Dashboard Panels:\n",
"┌─────────────────────────────────────────────────────────┐\n",
"│ Panel Type │ Metric Name │ Description │\n",
"├─────────────────────────────────────────────────────────┤\n",
"│ Line Plot │ quality_score │ Main metric │\n",
"│ Scalar │ final_quality_score │ Final score │\n",
"│ Bar Chart │ trial_status │ Run status │\n",
"│ Histogram │ learning_rate │ LR dist. │\n",
"│ Scatter Plot │ temperature vs score │ Correlation │\n",
"│ Table │ All hyperparameters │ Best runs │\n",
"│ Parallel Coordinates│ All params + score │ Patterns │\n",
"└─────────────────────────────────────────────────────────┘\n",
"\n",
"🎯 Key Metrics to Track:\n",
" • quality_score - Primary optimization metric (logged during run)\n",
" • final_quality_score - Final score for each run (in summary)\n",
" • trial_status - completed/failed status (string, not boolean)\n",
" • learning_rate - Tabular FT learning rate\n",
" • batch_size - Training batch size\n",
" • temperature - Generation temperature\n",
" • lora_r - LoRA rank parameter\n",
" • records_generated - Number of synthetic records created\n",
" • error_message - Error details for failed runs (string, not media)\n",
"\n",
"🚀 Pro Tips for W&B Visualization:\n",
" 1. Use 'final_quality_score' for comparing final results\n",
" 2. Use 'quality_score' for tracking optimization progress\n",
" 3. Filter by 'trial_status' == 'completed' to exclude failed runs\n",
" 4. Create scatter plots: 'learning_rate' vs 'final_quality_score'\n",
" 5. Use parallel coordinates to find parameter patterns\n",
"\n",
"📋 Sample W&B Panel Configurations:\n",
"```\n",
"# Scatter Plot: Learning Rate vs Quality\n",
"X-axis: learning_rate\n",
"Y-axis: final_quality_score\n",
"Color: batch_size\n",
"Filter: trial_status == 'completed'\n",
"\n",
"# Parallel Coordinates: All Parameters\n",
"Include: learning_rate, batch_size, temperature, lora_r, final_quality_score\n",
"Color: final_quality_score\n",
"```\n",
"\n",
"🔗 Access your sweep dashboard: https://wandb.ai/yamini_gretel/safe-syn-sweeps\n",
"✅ Use these visualizations to identify optimal parameter combinations!\n"
]
}
],
"source": [
"# W&B Dashboard Configuration Guide\n",
"print(\"📊 W&B Dashboard Configuration Tips\")\n",
"print(\"=\" * 50)\n",
"\n",
"print(\"\\n🎨 Creating Effective W&B Visualizations:\")\n",
"print(\"1. Navigate to your W&B sweep dashboard\")\n",
"print(\"2. Add recommended chart types for comprehensive analysis\")\n",
"print(\"3. Use these panel configurations for best insights:\")\n",
"\n",
"print(\"\\n📈 Recommended Dashboard Panels:\")\n",
"print(\"┌─────────────────────────────────────────────────────────┐\")\n",
"print(\"│ Panel Type │ Metric Name │ Description │\")\n",
"print(\"├─────────────────────────────────────────────────────────┤\")\n",
"print(\"│ Line Plot │ quality_score │ Main metric │\")\n",
"print(\"│ Line Plot │ privacy_score │ Privacy metric │\")\n",
"print(\"│ Scalar │ final_quality_score │ Final score │\")\n",
"print(\"│ Bar Chart │ trial_status │ Run status │\")\n",
"print(\"│ Histogram │ learning_rate │ LR dist. │\")\n",
"print(\"│ Scatter Plot │ temperature vs score │ Correlation │\")\n",
"print(\"│ Table │ All hyperparameters │ Best runs │\")\n",
"print(\"│ Parallel Coordinates│ All params + score │ Patterns │\")\n",
"print(\"└─────────────────────────────────────────────────────────┘\")\n",
"\n",
"print(\"\\n🎯 Key Metrics to Track:\")\n",
"metrics_info = {\n",
" \"quality_score\": \"Primary optimization metric (logged during run)\",\n",
" \"privacy_score\": \"Privacy score (logged during run)\",\n",
" \"final_quality_score\": \"Final score for each run (in summary)\",\n",
" \"trial_status\": \"completed/failed status (string, not boolean)\",\n",
" \"learning_rate\": \"Tabular FT learning rate\",\n",
" \"batch_size\": \"Training batch size\",\n",
" \"temperature\": \"Generation temperature\",\n",
" \"lora_r\": \"LoRA rank parameter\",\n",
" \"records_generated\": \"Number of synthetic records created\",\n",
" \"error_message\": \"Error details for failed runs (string, not media)\"\n",
"}\n",
"\n",
"for metric, description in metrics_info.items():\n",
" print(f\" • {metric:20} - {description}\")\n",
"\n",
"print(\"\\n🚀 Pro Tips for W&B Visualization:\")\n",
"print(\" 1. Use 'final_quality_score' for comparing final results\")\n",
"print(\" 2. Use 'quality_score' for tracking optimization progress\")\n",
"print(\" 3. Filter by 'trial_status' == 'completed' to exclude failed runs\")\n",
"print(\" 4. Create scatter plots: 'learning_rate' vs 'final_quality_score'\")\n",
"print(\" 5. Use parallel coordinates to find parameter patterns\")\n",
"\n",
"print(\"\\n📋 Sample W&B Panel Configurations:\")\n",
"print(\"```\")\n",
"print(\"# Scatter Plot: Learning Rate vs Quality\")\n",
"print(\"X-axis: learning_rate\")\n",
"print(\"Y-axis: final_quality_score\")\n",
"print(\"Color: batch_size\")\n",
"print(\"Filter: trial_status == 'completed'\")\n",
"print(\"\")\n",
"print(\"# Parallel Coordinates: All Parameters\")\n",
"print(\"Include: learning_rate, batch_size, temperature, lora_r, final_quality_score\")\n",
"print(\"Color: final_quality_score\")\n",
"print(\"```\")\n",
"\n",
"print(f\"\\n🔗 Access your sweep dashboard: https://wandb.ai/{WANDB_ENTITY or 'your-entity'}/{WANDB_PROJECT}\")\n",
"print(\"✅ Use these visualizations to identify optimal parameter combinations!\")\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 9) Launch Hyperparameter Optimization\n",
"\n",
"Start the Weights & Biases sweep to automatically optimize Safe Synthetics parameters. The sweep will run multiple trials, each testing different parameter combinations to find the best settings for your dataset.\n"
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"🚀 Launching W&B Sweep...\n",
"📋 Configuration: 2 trials using bayes optimization\n",
"🎯 Optimizing: quality_score (maximize)\n",
"📊 Dataset: 500 rows, 8 columns\n",
"\n",
"📝 Creating sweep...\n",
"Create sweep with ID: e8o48dv0\n",
"Sweep URL: https://wandb.ai/yamini_gretel/safe-syn-sweeps/sweeps/e8o48dv0\n",
"✅ Sweep created with ID: e8o48dv0\n",
"🔗 View sweep at: https://wandb.ai/yamini_gretel/safe-syn-sweeps/sweeps/e8o48dv0\n",
"\n",
"🤖 Starting sweep agent (max 2 trials)...\n",
"================================================================================\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"\u001b[34m\u001b[1mwandb\u001b[0m: Agent Starting Run: dk0an880 with config:\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tbatch_size: 2\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tgradient_accumulation_steps: 4\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tlearning_rate: 0.0005028487963675808\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tlora_alpha: 16\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tlora_dropout: 0.07498826011840841\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tlora_r: 32\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tmax_input_records: 5000\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tnum_records: 1000\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tnum_train_epochs: 3\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \trepetition_penalty: 1.0997807762466294\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \ttemperature: 0.9576905174699012\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \ttop_p: 0.989535007770401\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \twarmup_ratio: 0.06641367922936457\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tweight_decay: 0.0013366680486922313\n"
]
},
{
"data": {
"text/html": [
"Ignoring project 'safe-syn-sweeps' when running a sweep."
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"Ignoring entity 'yamini_gretel' when running a sweep."
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"Tracking run with wandb version 0.21.1"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"Run data is saved locally in <code>/Users/ykagal/Documents/notebooks/wandb/run-20250813_192506-dk0an880</code>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"Syncing run <strong><a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps/runs/dk0an880' target=\"_blank\">safe-syn-lxmr1ys7</a></strong> to <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps' target=\"_blank\">Weights & Biases</a> (<a href='https://wandb.me/developer-guide' target=\"_blank\">docs</a>)<br>Sweep page: <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps/sweeps/e8o48dv0' target=\"_blank\">https://wandb.ai/yamini_gretel/safe-syn-sweeps/sweeps/e8o48dv0</a>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" View project at <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps' target=\"_blank\">https://wandb.ai/yamini_gretel/safe-syn-sweeps</a>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" View sweep at <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps/sweeps/e8o48dv0' target=\"_blank\">https://wandb.ai/yamini_gretel/safe-syn-sweeps/sweeps/e8o48dv0</a>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" View run at <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps/runs/dk0an880' target=\"_blank\">https://wandb.ai/yamini_gretel/safe-syn-sweeps/runs/dk0an880</a>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"🚀 Starting trial: safe-syn-lxmr1ys7\n",
"📊 Parameters: {'batch_size': 2, 'gradient_accumulation_steps': 4, 'learning_rate': 0.0005028487963675808, 'lora_alpha': 16, 'lora_dropout': 0.07498826011840841, 'lora_r': 32, 'max_input_records': 5000, 'num_records': 1000, 'num_train_epochs': 3, 'repetition_penalty': 1.0997807762466294, 'temperature': 0.9576905174699012, 'top_p': 0.989535007770401, 'warmup_ratio': 0.06641367922936457, 'weight_decay': 0.0013366680486922313}\n",
"Configuring generator for data source: https://gretel-datasets.s3.us-west-2.amazonaws.com/ecommerce_customers.csv\n",
"Configuring holdout: 0.05\n",
"Configuring synthetic data generation model: tabular_ft/default\n",
"▶️ Creating Workflow: w_31Fl4cur1ZouU3Yha3TgN4tKDj3\n",
"▶️ Created Workflow Run: wr_31Fl4deAXRL3OO8DIz7w82T2rXo\n",
"🔗 Workflow Run console link: https://console.gretel.ai/workflows/w_31Fl4cur1ZouU3Yha3TgN4tKDj3/runs/wr_31Fl4deAXRL3OO8DIz7w82T2rXo\n",
"Fetching task logs for workflow run wr_31Fl4deAXRL3OO8DIz7w82T2rXo\n",
"Workflow run is now in status: RUN_STATUS_CREATED\n",
"Got task wt_31Fl4gG53nK8zDMau097FvoxJau\n",
"Workflow run is now in status: RUN_STATUS_ACTIVE\n",
"[read-data-source] Task Status is now: RUN_STATUS_ACTIVE\n",
"[read-data-source] 2025-08-13 23:25:29.292469+00:00 Preparing step 'read-data-source'\n",
"[read-data-source] 2025-08-13 23:25:38.771531+00:00 Starting 'data_source' task execution\n",
"[read-data-source] 2025-08-13 23:25:40.716385+00:00 Task 'data_source' executed successfully\n",
"[read-data-source] 2025-08-13 23:25:40.716960+00:00 Task execution completed. Saving task outputs.\n",
"[read-data-source] 2025-08-13 23:25:41.485640+00:00 Task outputs saved.\n",
"[read-data-source] Task Status is now: RUN_STATUS_COMPLETED\n",
"Got task wt_31Fl4goPXJKlltr1P4FLL0UBnPn\n",
"[holdout] 2025-08-13 23:26:42.644205+00:00 Preparing step 'holdout'\n",
"[holdout] 2025-08-13 23:26:53.144817+00:00 Starting 'holdout' task execution\n",
"[holdout] 2025-08-13 23:26:53.147032+00:00 Dataset test holdout split configuration complete. Train dataset: 475 rows x 8 columns. Test dataset: 25 rows x 8 columns.\n",
"[holdout] 2025-08-13 23:26:53.156365+00:00 Task 'holdout' executed successfully\n",
"[holdout] 2025-08-13 23:26:53.156692+00:00 Task execution completed. Saving task outputs.\n",
"[holdout] 2025-08-13 23:26:54.029717+00:00 Task outputs saved.\n",
"[holdout] Task Status is now: RUN_STATUS_COMPLETED\n",
"Got task wt_31Fl4eong6zfqInjJ5xKR4hkIIK\n",
"[tabular-ft] Task Status is now: RUN_STATUS_ACTIVE\n",
"[tabular-ft] 2025-08-13 23:33:50.225124+00:00 Preparing step 'tabular-ft'\n",
"[tabular-ft] 2025-08-13 23:34:08.746790+00:00 Starting 'tabular_ft' task execution\n",
"[tabular-ft] 2025-08-13 23:34:25.345406+00:00 Analyzing input data and checking for auto-params...\n",
"[tabular-ft] 2025-08-13 23:34:25.369633+00:00 Parameter `rope_scaling_factor` was automatically set to 1 based on an estimated token count given the lengths of each training record and the column names.\n",
"[tabular-ft] 2025-08-13 23:34:25.369951+00:00 Found 3 auto-params that were set based on input data. - num_input_records_to_sample: 25000, use_unsloth: True, rope_scaling_factor: 1\n",
"[tabular-ft] 2025-08-13 23:34:25.370049+00:00 Using updated model config: \n",
"{\n",
" \"group_training_examples_by\": null,\n",
" \"order_training_examples_by\": null,\n",
" \"max_sequences_per_example\": null,\n",
" \"pretrained_model\": null,\n",
" \"params\": {\n",
" \"num_input_records_to_sample\": 25000,\n",
" \"batch_size\": 1,\n",
" \"gradient_accumulation_steps\": 8,\n",
" \"weight_decay\": 0.01,\n",
" \"warmup_ratio\": 0.05,\n",
" \"lr_scheduler\": \"cosine\",\n",
" \"learning_rate\": 0.0005,\n",
" \"lora_r\": 32,\n",
" \"lora_alpha_over_r\": 1.0,\n",
" \"lora_target_modules\": [\n",
" \"q_proj\",\n",
" \"k_proj\",\n",
" \"v_proj\",\n",
" \"o_proj\"\n",
" ],\n",
" \"use_unsloth\": true,\n",
" \"rope_scaling_factor\": 1,\n",
" \"validation_ratio\": 0.0,\n",
" \"validation_steps\": 15\n",
" },\n",
" \"privacy_params\": {\n",
" \"dp\": false,\n",
" \"epsilon\": 8.0,\n",
" \"delta\": \"auto\",\n",
" \"per_sample_max_grad_norm\": 1.0\n",
" },\n",
" \"data_config\": null\n",
"}\n",
"[tabular-ft] 2025-08-13 23:34:25.371157+00:00 << 🧭 Tabular FT >> Preparing for training\n",
"[tabular-ft] 2025-08-13 23:34:45.026544+00:00 << 🧭 Tabular FT >> Tokenizing records\n",
"[tabular-ft] 2025-08-13 23:34:45.165041+00:00 << 🧭 Tabular FT >> Number of unique train records: 475\n",
"[tabular-ft] 2025-08-13 23:34:45.165188+00:00 << 🧭 Tabular FT >> Assembling examples from 5263.2% of the input records\n",
"[tabular-ft] 2025-08-13 23:35:22.840856+00:00 << 🧭 Tabular FT >> Training Example Statistics:\n",
"[tabular-ft] 2025-08-13 23:35:22.841759+00:00 \n",
"╒════════╤═════════════════════╤══════════════════════╤═══════════════════════╕\n",
"│ │ Tokens per record │ Tokens per example │ Records per example │\n",
"╞════════╪═════════════════════╪══════════════════════╪═══════════════════════╡\n",
"│ min │ 136 │ 383 │ 2 │\n",
"├────────┼─────────────────────┼──────────────────────┼───────────────────────┤\n",
"│ max │ 164 │ 2048 │ 13 │\n",
"├────────┼─────────────────────┼──────────────────────┼───────────────────────┤\n",
"│ mean │ 151.034 │ 1974.62 │ 12.525 │\n",
"├────────┼─────────────────────┼──────────────────────┼───────────────────────┤\n",
"│ stddev │ 5.103 │ 72.027 │ 0.552 │\n",
"╘════════╧═════════════════════╧══════════════════════╧═══════════════════════╛\n",
"[tabular-ft] 2025-08-13 23:35:22.841870+00:00 << 🧭 Tabular FT >> Number of training examples: 1996\n",
"[tabular-ft] 2025-08-13 23:35:26.247635+00:00 << 🧭 Tabular FT >> Using PEFT - 9.01 million parameters are trainable\n",
"[tabular-ft] 2025-08-13 23:35:45.912722+00:00 Training 0.4% complete - step: 8, epoch: 0.004008016032064128, loss: 1.6175\n",
"[tabular-ft] 2025-08-13 23:35:48.508916+00:00 Training 0.8% complete - step: 16, epoch: 0.008016032064128256, loss: 1.6386\n",
"[tabular-ft] 2025-08-13 23:35:51.152695+00:00 Training 1.2% complete - step: 24, epoch: 0.012024048096192385, loss: 1.6077\n",
"[tabular-ft] 2025-08-13 23:35:54.241759+00:00 Training 1.6% complete - step: 32, epoch: 0.01603206412825651, loss: 1.5778\n",
"[tabular-ft] 2025-08-13 23:35:57.985694+00:00 Training 2.0% complete - step: 40, epoch: 0.02004008016032064, loss: 1.5495\n",
"[tabular-ft] 2025-08-13 23:36:00.644875+00:00 Training 2.4% complete - step: 48, epoch: 0.02404809619238477, loss: 1.5286\n",
"[tabular-ft] 2025-08-13 23:36:03.253112+00:00 Training 2.8% complete - step: 56, epoch: 0.028056112224448898, loss: 1.5051\n",
"[tabular-ft] 2025-08-13 23:36:05.908115+00:00 Training 3.2% complete - step: 64, epoch: 0.03206412825651302, loss: 1.4938\n",
"[tabular-ft] 2025-08-13 23:36:08.574781+00:00 Training 3.6% complete - step: 72, epoch: 0.036072144288577156, loss: 1.4738\n",
"[tabular-ft] 2025-08-13 23:36:11.258610+00:00 Training 4.0% complete - step: 80, epoch: 0.04008016032064128, loss: 1.4616\n",
"[tabular-ft] 2025-08-13 23:36:13.933793+00:00 Training 4.4% complete - step: 88, epoch: 0.04408817635270541, loss: 1.4524\n",
"[tabular-ft] 2025-08-13 23:36:16.551139+00:00 Training 4.8% complete - step: 96, epoch: 0.04809619238476954, loss: 1.445\n",
"[tabular-ft] 2025-08-13 23:36:19.161562+00:00 Training 5.2% complete - step: 104, epoch: 0.052104208416833664, loss: 1.4328\n",
"[tabular-ft] 2025-08-13 23:36:21.834688+00:00 Training 5.6% complete - step: 112, epoch: 0.056112224448897796, loss: 1.4251\n",
"[tabular-ft] 2025-08-13 23:36:24.556978+00:00 Training 6.0% complete - step: 120, epoch: 0.06012024048096192, loss: 1.4157\n",
"[tabular-ft] 2025-08-13 23:36:27.226823+00:00 Training 6.4% complete - step: 128, epoch: 0.06412825651302605, loss: 1.4197\n",
"[tabular-ft] 2025-08-13 23:36:29.909867+00:00 Training 6.8% complete - step: 136, epoch: 0.06813627254509018, loss: 1.416\n",
"[tabular-ft] 2025-08-13 23:36:32.567605+00:00 Training 7.2% complete - step: 144, epoch: 0.07214428857715431, loss: 1.4088\n",
"[tabular-ft] 2025-08-13 23:36:35.203375+00:00 Training 7.6% complete - step: 152, epoch: 0.07615230460921844, loss: 1.4068\n",
"[tabular-ft] 2025-08-13 23:36:37.842510+00:00 Training 8.0% complete - step: 160, epoch: 0.08016032064128256, loss: 1.3954\n",
"[tabular-ft] 2025-08-13 23:36:40.468999+00:00 Training 8.4% complete - step: 168, epoch: 0.0841683366733467, loss: 1.4023\n",
"[tabular-ft] 2025-08-13 23:36:43.124597+00:00 Training 8.8% complete - step: 176, epoch: 0.08817635270541083, loss: 1.3937\n",
"[tabular-ft] 2025-08-13 23:36:45.784133+00:00 Training 9.2% complete - step: 184, epoch: 0.09218436873747494, loss: 1.3896\n",
"[tabular-ft] 2025-08-13 23:36:48.461455+00:00 Training 9.6% complete - step: 192, epoch: 0.09619238476953908, loss: 1.3906\n",
"[tabular-ft] 2025-08-13 23:36:51.166096+00:00 Training 10.0% complete - step: 200, epoch: 0.10020040080160321, loss: 1.3905\n",
"[tabular-ft] 2025-08-13 23:36:53.800457+00:00 Training 10.4% complete - step: 208, epoch: 0.10420841683366733, loss: 1.3834\n",
"[tabular-ft] 2025-08-13 23:36:56.436637+00:00 Training 10.8% complete - step: 216, epoch: 0.10821643286573146, loss: 1.3858\n",
"[tabular-ft] 2025-08-13 23:36:59.114579+00:00 Training 11.2% complete - step: 224, epoch: 0.11222444889779559, loss: 1.3801\n",
"[tabular-ft] 2025-08-13 23:37:01.757610+00:00 Training 11.6% complete - step: 232, epoch: 0.11623246492985972, loss: 1.3765\n",
"[tabular-ft] 2025-08-13 23:37:04.436560+00:00 Training 12.0% complete - step: 240, epoch: 0.12024048096192384, loss: 1.3719\n",
"[tabular-ft] 2025-08-13 23:37:07.099163+00:00 Training 12.4% complete - step: 248, epoch: 0.12424849699398798, loss: 1.3682\n",
"[tabular-ft] 2025-08-13 23:37:09.761428+00:00 Training 12.9% complete - step: 256, epoch: 0.1282565130260521, loss: 1.3632\n",
"[tabular-ft] 2025-08-13 23:37:12.408489+00:00 Training 13.3% complete - step: 264, epoch: 0.13226452905811623, loss: 1.3736\n",
"[tabular-ft] 2025-08-13 23:37:15.068767+00:00 Training 13.7% complete - step: 272, epoch: 0.13627254509018036, loss: 1.3598\n",
"[tabular-ft] 2025-08-13 23:37:17.682682+00:00 Training 14.1% complete - step: 280, epoch: 0.1402805611222445, loss: 1.3747\n",
"[tabular-ft] 2025-08-13 23:37:20.323148+00:00 Training 14.5% complete - step: 288, epoch: 0.14428857715430862, loss: 1.3585\n",
"[tabular-ft] 2025-08-13 23:37:22.951873+00:00 Training 14.9% complete - step: 296, epoch: 0.14829659318637275, loss: 1.3603\n",
"[tabular-ft] 2025-08-13 23:37:25.589083+00:00 Training 15.3% complete - step: 304, epoch: 0.1523046092184369, loss: 1.3553\n",
"[tabular-ft] 2025-08-13 23:37:28.212147+00:00 Training 15.7% complete - step: 312, epoch: 0.156312625250501, loss: 1.3477\n",
"[tabular-ft] 2025-08-13 23:37:30.913382+00:00 Training 16.1% complete - step: 320, epoch: 0.16032064128256512, loss: 1.358\n",
"[tabular-ft] 2025-08-13 23:37:33.543990+00:00 Training 16.5% complete - step: 328, epoch: 0.16432865731462926, loss: 1.3469\n",
"[tabular-ft] 2025-08-13 23:37:36.245607+00:00 Training 16.9% complete - step: 336, epoch: 0.1683366733466934, loss: 1.3411\n",
"[tabular-ft] 2025-08-13 23:37:38.896248+00:00 Training 17.3% complete - step: 344, epoch: 0.17234468937875752, loss: 1.3438\n",
"[tabular-ft] 2025-08-13 23:37:41.555320+00:00 Training 17.7% complete - step: 352, epoch: 0.17635270541082165, loss: 1.3471\n",
"[tabular-ft] 2025-08-13 23:37:44.228521+00:00 Training 18.1% complete - step: 360, epoch: 0.18036072144288579, loss: 1.345\n",
"[tabular-ft] 2025-08-13 23:37:46.917886+00:00 Training 18.5% complete - step: 368, epoch: 0.1843687374749499, loss: 1.329\n",
"[tabular-ft] 2025-08-13 23:37:49.612344+00:00 Training 18.9% complete - step: 376, epoch: 0.18837675350701402, loss: 1.3304\n",
"[tabular-ft] 2025-08-13 23:37:52.240507+00:00 Training 19.3% complete - step: 384, epoch: 0.19238476953907815, loss: 1.3178\n",
"[tabular-ft] 2025-08-13 23:37:54.831890+00:00 Training 19.7% complete - step: 392, epoch: 0.1963927855711423, loss: 1.3135\n",
"[tabular-ft] 2025-08-13 23:37:57.466658+00:00 Training 20.1% complete - step: 400, epoch: 0.20040080160320642, loss: 1.3223\n",
"[tabular-ft] 2025-08-13 23:38:00.099881+00:00 Training 20.5% complete - step: 408, epoch: 0.20440881763527055, loss: 1.3265\n",
"[tabular-ft] 2025-08-13 23:38:02.756201+00:00 Training 20.9% complete - step: 416, epoch: 0.20841683366733466, loss: 1.301\n",
"[tabular-ft] 2025-08-13 23:38:05.427058+00:00 Training 21.3% complete - step: 424, epoch: 0.2124248496993988, loss: 1.3079\n",
"[tabular-ft] 2025-08-13 23:38:08.109885+00:00 Training 21.7% complete - step: 432, epoch: 0.21643286573146292, loss: 1.3128\n",
"[tabular-ft] 2025-08-13 23:38:10.771521+00:00 Training 22.1% complete - step: 440, epoch: 0.22044088176352705, loss: 1.2931\n",
"[tabular-ft] 2025-08-13 23:38:13.433392+00:00 Training 22.5% complete - step: 448, epoch: 0.22444889779559118, loss: 1.2967\n",
"[tabular-ft] 2025-08-13 23:38:16.036406+00:00 Training 22.9% complete - step: 456, epoch: 0.22845691382765532, loss: 1.28\n",
"[tabular-ft] 2025-08-13 23:38:18.708530+00:00 Training 23.3% complete - step: 464, epoch: 0.23246492985971945, loss: 1.2749\n",
"[tabular-ft] 2025-08-13 23:38:21.337210+00:00 Training 23.7% complete - step: 472, epoch: 0.23647294589178355, loss: 1.2857\n",
"[tabular-ft] 2025-08-13 23:38:23.928470+00:00 Training 24.1% complete - step: 480, epoch: 0.24048096192384769, loss: 1.2786\n",
"[tabular-ft] 2025-08-13 23:38:26.497299+00:00 Training 24.5% complete - step: 488, epoch: 0.24448897795591182, loss: 1.2641\n",
"[tabular-ft] 2025-08-13 23:38:29.155486+00:00 Training 24.9% complete - step: 496, epoch: 0.24849699398797595, loss: 1.2651\n",
"[tabular-ft] 2025-08-13 23:38:31.790154+00:00 Training 25.3% complete - step: 504, epoch: 0.25250501002004005, loss: 1.2516\n",
"[tabular-ft] 2025-08-13 23:38:34.422756+00:00 Training 25.7% complete - step: 512, epoch: 0.2565130260521042, loss: 1.2599\n",
"[tabular-ft] 2025-08-13 23:38:37.112427+00:00 Training 26.1% complete - step: 520, epoch: 0.2605210420841683, loss: 1.2573\n",
"[tabular-ft] 2025-08-13 23:38:39.755347+00:00 Training 26.5% complete - step: 528, epoch: 0.26452905811623245, loss: 1.2278\n",
"[tabular-ft] 2025-08-13 23:38:42.391372+00:00 Training 26.9% complete - step: 536, epoch: 0.2685370741482966, loss: 1.2342\n",
"[tabular-ft] 2025-08-13 23:38:45.028056+00:00 Training 27.3% complete - step: 544, epoch: 0.2725450901803607, loss: 1.227\n",
"[tabular-ft] 2025-08-13 23:38:47.668950+00:00 Training 27.7% complete - step: 552, epoch: 0.27655310621242485, loss: 1.2238\n",
"[tabular-ft] 2025-08-13 23:38:50.263201+00:00 Training 28.1% complete - step: 560, epoch: 0.280561122244489, loss: 1.2118\n",
"[tabular-ft] 2025-08-13 23:38:52.968271+00:00 Training 28.5% complete - step: 568, epoch: 0.2845691382765531, loss: 1.2066\n",
"[tabular-ft] 2025-08-13 23:38:55.668479+00:00 Training 28.9% complete - step: 576, epoch: 0.28857715430861725, loss: 1.1836\n",
"[tabular-ft] 2025-08-13 23:38:58.320688+00:00 Training 29.3% complete - step: 584, epoch: 0.2925851703406814, loss: 1.1853\n",
"[tabular-ft] 2025-08-13 23:39:00.977971+00:00 Training 29.7% complete - step: 592, epoch: 0.2965931863727455, loss: 1.1919\n",
"[tabular-ft] 2025-08-13 23:39:03.589475+00:00 Training 30.1% complete - step: 600, epoch: 0.30060120240480964, loss: 1.1829\n",
"[tabular-ft] 2025-08-13 23:39:06.270146+00:00 Training 30.5% complete - step: 608, epoch: 0.3046092184368738, loss: 1.1642\n",
"[tabular-ft] 2025-08-13 23:39:08.974597+00:00 Training 30.9% complete - step: 616, epoch: 0.30861723446893785, loss: 1.1509\n",
"[tabular-ft] 2025-08-13 23:39:11.610935+00:00 Training 31.3% complete - step: 624, epoch: 0.312625250501002, loss: 1.1644\n",
"[tabular-ft] 2025-08-13 23:39:14.192375+00:00 Training 31.7% complete - step: 632, epoch: 0.3166332665330661, loss: 1.1503\n",
"[tabular-ft] 2025-08-13 23:39:16.829465+00:00 Training 32.1% complete - step: 640, epoch: 0.32064128256513025, loss: 1.1271\n",
"[tabular-ft] 2025-08-13 23:39:19.486353+00:00 Training 32.5% complete - step: 648, epoch: 0.3246492985971944, loss: 1.1067\n",
"[tabular-ft] 2025-08-13 23:39:22.179017+00:00 Training 32.9% complete - step: 656, epoch: 0.3286573146292585, loss: 1.1193\n",
"[tabular-ft] 2025-08-13 23:39:24.799312+00:00 Training 33.3% complete - step: 664, epoch: 0.33266533066132264, loss: 1.11\n",
"[tabular-ft] 2025-08-13 23:39:27.489235+00:00 Training 33.7% complete - step: 672, epoch: 0.3366733466933868, loss: 1.1034\n",
"[tabular-ft] 2025-08-13 23:39:30.144459+00:00 Training 34.1% complete - step: 680, epoch: 0.3406813627254509, loss: 1.0859\n",
"[tabular-ft] 2025-08-13 23:39:32.798694+00:00 Training 34.5% complete - step: 688, epoch: 0.34468937875751504, loss: 1.087\n",
"[tabular-ft] 2025-08-13 23:39:35.452054+00:00 Training 34.9% complete - step: 696, epoch: 0.3486973947895792, loss: 1.0793\n",
"[tabular-ft] 2025-08-13 23:39:38.099167+00:00 Training 35.3% complete - step: 704, epoch: 0.3527054108216433, loss: 1.0802\n",
"[tabular-ft] 2025-08-13 23:39:40.733369+00:00 Training 35.7% complete - step: 712, epoch: 0.35671342685370744, loss: 1.0563\n",
"[tabular-ft] 2025-08-13 23:39:43.320315+00:00 Training 36.1% complete - step: 720, epoch: 0.36072144288577157, loss: 1.0537\n",
"[tabular-ft] 2025-08-13 23:39:45.955766+00:00 Training 36.5% complete - step: 728, epoch: 0.36472945891783565, loss: 1.049\n",
"[tabular-ft] 2025-08-13 23:39:48.643796+00:00 Training 36.9% complete - step: 736, epoch: 0.3687374749498998, loss: 1.0484\n",
"[tabular-ft] 2025-08-13 23:39:51.283195+00:00 Training 37.3% complete - step: 744, epoch: 0.3727454909819639, loss: 1.0402\n",
"[tabular-ft] 2025-08-13 23:39:53.919615+00:00 Training 37.8% complete - step: 752, epoch: 0.37675350701402804, loss: 1.0126\n",
"[tabular-ft] 2025-08-13 23:39:56.566854+00:00 Training 38.2% complete - step: 760, epoch: 0.3807615230460922, loss: 1.0146\n",
"[tabular-ft] 2025-08-13 23:39:59.204664+00:00 Training 38.6% complete - step: 768, epoch: 0.3847695390781563, loss: 1.0084\n",
"[tabular-ft] 2025-08-13 23:40:01.882853+00:00 Training 39.0% complete - step: 776, epoch: 0.38877755511022044, loss: 0.9901\n",
"[tabular-ft] 2025-08-13 23:40:04.541881+00:00 Training 39.4% complete - step: 784, epoch: 0.3927855711422846, loss: 0.9999\n",
"[tabular-ft] 2025-08-13 23:40:07.216623+00:00 Training 39.8% complete - step: 792, epoch: 0.3967935871743487, loss: 0.9839\n",
"[tabular-ft] 2025-08-13 23:40:09.855703+00:00 Training 40.2% complete - step: 800, epoch: 0.40080160320641284, loss: 0.9735\n",
"[tabular-ft] 2025-08-13 23:40:12.516338+00:00 Training 40.6% complete - step: 808, epoch: 0.40480961923847697, loss: 0.975\n",
"[tabular-ft] 2025-08-13 23:40:15.148256+00:00 Training 41.0% complete - step: 816, epoch: 0.4088176352705411, loss: 0.9496\n",
"[tabular-ft] 2025-08-13 23:40:17.817683+00:00 Training 41.4% complete - step: 824, epoch: 0.41282565130260523, loss: 0.9297\n",
"[tabular-ft] 2025-08-13 23:40:20.454690+00:00 Training 41.8% complete - step: 832, epoch: 0.4168336673346693, loss: 0.9317\n",
"[tabular-ft] 2025-08-13 23:40:23.114357+00:00 Training 42.2% complete - step: 840, epoch: 0.42084168336673344, loss: 0.9285\n",
"[tabular-ft] 2025-08-13 23:40:25.745680+00:00 Training 42.6% complete - step: 848, epoch: 0.4248496993987976, loss: 0.9209\n",
"[tabular-ft] 2025-08-13 23:40:28.424971+00:00 Training 43.0% complete - step: 856, epoch: 0.4288577154308617, loss: 0.9228\n",
"[tabular-ft] 2025-08-13 23:40:31.078355+00:00 Training 43.4% complete - step: 864, epoch: 0.43286573146292584, loss: 0.8968\n",
"[tabular-ft] 2025-08-13 23:40:33.692527+00:00 Training 43.8% complete - step: 872, epoch: 0.43687374749499, loss: 0.8939\n",
"[tabular-ft] 2025-08-13 23:40:36.334084+00:00 Training 44.2% complete - step: 880, epoch: 0.4408817635270541, loss: 0.8961\n",
"[tabular-ft] 2025-08-13 23:40:38.974663+00:00 Training 44.6% complete - step: 888, epoch: 0.44488977955911824, loss: 0.8737\n",
"[tabular-ft] 2025-08-13 23:40:41.629365+00:00 Training 45.0% complete - step: 896, epoch: 0.44889779559118237, loss: 0.8519\n",
"[tabular-ft] 2025-08-13 23:40:44.307689+00:00 Training 45.4% complete - step: 904, epoch: 0.4529058116232465, loss: 0.8555\n",
"[tabular-ft] 2025-08-13 23:40:46.948226+00:00 Training 45.8% complete - step: 912, epoch: 0.45691382765531063, loss: 0.844\n",
"[tabular-ft] 2025-08-13 23:40:49.629630+00:00 Training 46.2% complete - step: 920, epoch: 0.46092184368737477, loss: 0.8391\n",
"[tabular-ft] 2025-08-13 23:40:52.246282+00:00 Training 46.6% complete - step: 928, epoch: 0.4649298597194389, loss: 0.8291\n",
"[tabular-ft] 2025-08-13 23:40:54.858042+00:00 Training 47.0% complete - step: 936, epoch: 0.46893787575150303, loss: 0.8152\n",
"[tabular-ft] 2025-08-13 23:40:57.553515+00:00 Training 47.4% complete - step: 944, epoch: 0.4729458917835671, loss: 0.8257\n",
"[tabular-ft] 2025-08-13 23:41:00.242550+00:00 Training 47.8% complete - step: 952, epoch: 0.47695390781563124, loss: 0.8058\n",
"[tabular-ft] 2025-08-13 23:41:02.878694+00:00 Training 48.2% complete - step: 960, epoch: 0.48096192384769537, loss: 0.8047\n",
"[tabular-ft] 2025-08-13 23:41:05.517341+00:00 Training 48.6% complete - step: 968, epoch: 0.4849699398797595, loss: 0.7807\n",
"[tabular-ft] 2025-08-13 23:41:08.132868+00:00 Training 49.0% complete - step: 976, epoch: 0.48897795591182364, loss: 0.7651\n",
"[tabular-ft] 2025-08-13 23:41:10.804449+00:00 Training 49.4% complete - step: 984, epoch: 0.49298597194388777, loss: 0.763\n",
"[tabular-ft] 2025-08-13 23:41:13.441438+00:00 Training 49.8% complete - step: 992, epoch: 0.4969939879759519, loss: 0.7662\n",
"[tabular-ft] 2025-08-13 23:41:16.129124+00:00 Training 50.2% complete - step: 1000, epoch: 0.501002004008016, loss: 0.7454\n",
"[tabular-ft] 2025-08-13 23:41:18.791720+00:00 Training 50.6% complete - step: 1008, epoch: 0.5050100200400801, loss: 0.7214\n",
"[tabular-ft] 2025-08-13 23:41:21.465662+00:00 Training 51.0% complete - step: 1016, epoch: 0.5090180360721442, loss: 0.7398\n",
"[tabular-ft] 2025-08-13 23:41:24.115508+00:00 Training 51.4% complete - step: 1024, epoch: 0.5130260521042084, loss: 0.7026\n",
"[tabular-ft] 2025-08-13 23:41:26.794303+00:00 Training 51.8% complete - step: 1032, epoch: 0.5170340681362725, loss: 0.7156\n",
"[tabular-ft] 2025-08-13 23:41:29.519791+00:00 Training 52.2% complete - step: 1040, epoch: 0.5210420841683366, loss: 0.7076\n",
"[tabular-ft] 2025-08-13 23:41:32.157547+00:00 Training 52.6% complete - step: 1048, epoch: 0.5250501002004008, loss: 0.6837\n",
"[tabular-ft] 2025-08-13 23:41:34.799579+00:00 Training 53.0% complete - step: 1056, epoch: 0.5290581162324649, loss: 0.6686\n",
"[tabular-ft] 2025-08-13 23:41:37.442255+00:00 Training 53.4% complete - step: 1064, epoch: 0.533066132264529, loss: 0.6623\n",
"[tabular-ft] 2025-08-13 23:41:40.126922+00:00 Training 53.8% complete - step: 1072, epoch: 0.5370741482965932, loss: 0.6513\n",
"[tabular-ft] 2025-08-13 23:41:42.810276+00:00 Training 54.2% complete - step: 1080, epoch: 0.5410821643286573, loss: 0.6323\n",
"[tabular-ft] 2025-08-13 23:41:45.445236+00:00 Training 54.6% complete - step: 1088, epoch: 0.5450901803607214, loss: 0.6166\n",
"[tabular-ft] 2025-08-13 23:41:48.108551+00:00 Training 55.0% complete - step: 1096, epoch: 0.5490981963927856, loss: 0.6189\n",
"[tabular-ft] 2025-08-13 23:41:50.791149+00:00 Training 55.4% complete - step: 1104, epoch: 0.5531062124248497, loss: 0.6012\n",
"[tabular-ft] 2025-08-13 23:41:53.427541+00:00 Training 55.8% complete - step: 1112, epoch: 0.5571142284569138, loss: 0.6168\n",
"[tabular-ft] 2025-08-13 23:41:56.092489+00:00 Training 56.2% complete - step: 1120, epoch: 0.561122244488978, loss: 0.59\n",
"[tabular-ft] 2025-08-13 23:41:58.719855+00:00 Training 56.6% complete - step: 1128, epoch: 0.5651302605210421, loss: 0.5641\n",
"[tabular-ft] 2025-08-13 23:42:01.387299+00:00 Training 57.0% complete - step: 1136, epoch: 0.5691382765531062, loss: 0.5664\n",
"[tabular-ft] 2025-08-13 23:42:04.066789+00:00 Training 57.4% complete - step: 1144, epoch: 0.5731462925851704, loss: 0.5675\n",
"[tabular-ft] 2025-08-13 23:42:06.721857+00:00 Training 57.8% complete - step: 1152, epoch: 0.5771543086172345, loss: 0.5428\n",
"[tabular-ft] 2025-08-13 23:42:09.428256+00:00 Training 58.2% complete - step: 1160, epoch: 0.5811623246492986, loss: 0.5549\n",
"[tabular-ft] 2025-08-13 23:42:12.106990+00:00 Training 58.6% complete - step: 1168, epoch: 0.5851703406813628, loss: 0.5367\n",
"[tabular-ft] 2025-08-13 23:42:14.763185+00:00 Training 59.0% complete - step: 1176, epoch: 0.5891783567134269, loss: 0.5166\n",
"[tabular-ft] 2025-08-13 23:42:17.414028+00:00 Training 59.4% complete - step: 1184, epoch: 0.593186372745491, loss: 0.5113\n",
"[tabular-ft] 2025-08-13 23:42:20.095374+00:00 Training 59.8% complete - step: 1192, epoch: 0.5971943887775552, loss: 0.5088\n",
"[tabular-ft] 2025-08-13 23:42:22.781641+00:00 Training 60.2% complete - step: 1200, epoch: 0.6012024048096193, loss: 0.4955\n",
"[tabular-ft] 2025-08-13 23:42:25.415974+00:00 Training 60.6% complete - step: 1208, epoch: 0.6052104208416834, loss: 0.4828\n",
"[tabular-ft] 2025-08-13 23:42:28.025311+00:00 Training 61.0% complete - step: 1216, epoch: 0.6092184368737475, loss: 0.4813\n",
"[tabular-ft] 2025-08-13 23:42:30.707710+00:00 Training 61.4% complete - step: 1224, epoch: 0.6132264529058116, loss: 0.467\n",
"[tabular-ft] 2025-08-13 23:42:33.339939+00:00 Training 61.8% complete - step: 1232, epoch: 0.6172344689378757, loss: 0.45\n",
"[tabular-ft] 2025-08-13 23:42:35.994136+00:00 Training 62.2% complete - step: 1240, epoch: 0.6212424849699398, loss: 0.4484\n",
"[tabular-ft] 2025-08-13 23:42:38.609587+00:00 Training 62.7% complete - step: 1248, epoch: 0.625250501002004, loss: 0.4433\n",
"[tabular-ft] 2025-08-13 23:42:41.311730+00:00 Training 63.1% complete - step: 1256, epoch: 0.6292585170340681, loss: 0.4252\n",
"[tabular-ft] 2025-08-13 23:42:43.944683+00:00 Training 63.5% complete - step: 1264, epoch: 0.6332665330661322, loss: 0.4144\n",
"[tabular-ft] 2025-08-13 23:42:46.603808+00:00 Training 63.9% complete - step: 1272, epoch: 0.6372745490981964, loss: 0.4147\n",
"[tabular-ft] 2025-08-13 23:42:49.258769+00:00 Training 64.3% complete - step: 1280, epoch: 0.6412825651302605, loss: 0.4186\n",
"[tabular-ft] 2025-08-13 23:42:51.919066+00:00 Training 64.7% complete - step: 1288, epoch: 0.6452905811623246, loss: 0.3976\n",
"[tabular-ft] 2025-08-13 23:42:54.576172+00:00 Training 65.1% complete - step: 1296, epoch: 0.6492985971943888, loss: 0.3845\n",
"[tabular-ft] 2025-08-13 23:42:57.244006+00:00 Training 65.5% complete - step: 1304, epoch: 0.6533066132264529, loss: 0.3897\n",
"[tabular-ft] 2025-08-13 23:42:59.905507+00:00 Training 65.9% complete - step: 1312, epoch: 0.657314629258517, loss: 0.3878\n",
"[tabular-ft] 2025-08-13 23:43:02.547869+00:00 Training 66.3% complete - step: 1320, epoch: 0.6613226452905812, loss: 0.3707\n",
"[tabular-ft] 2025-08-13 23:43:05.186095+00:00 Training 66.7% complete - step: 1328, epoch: 0.6653306613226453, loss: 0.3612\n",
"[tabular-ft] 2025-08-13 23:43:07.866421+00:00 Training 67.1% complete - step: 1336, epoch: 0.6693386773547094, loss: 0.3533\n",
"[tabular-ft] 2025-08-13 23:43:10.552125+00:00 Training 67.5% complete - step: 1344, epoch: 0.6733466933867736, loss: 0.3517\n",
"[tabular-ft] 2025-08-13 23:43:13.200301+00:00 Training 67.9% complete - step: 1352, epoch: 0.6773547094188377, loss: 0.3398\n",
"[tabular-ft] 2025-08-13 23:43:15.835834+00:00 Training 68.3% complete - step: 1360, epoch: 0.6813627254509018, loss: 0.3411\n",
"[tabular-ft] 2025-08-13 23:43:18.446423+00:00 Training 68.7% complete - step: 1368, epoch: 0.685370741482966, loss: 0.3268\n",
"[tabular-ft] 2025-08-13 23:43:21.122529+00:00 Training 69.1% complete - step: 1376, epoch: 0.6893787575150301, loss: 0.3222\n",
"[tabular-ft] 2025-08-13 23:43:23.720156+00:00 Training 69.5% complete - step: 1384, epoch: 0.6933867735470942, loss: 0.3316\n",
"[tabular-ft] 2025-08-13 23:43:26.440709+00:00 Training 69.9% complete - step: 1392, epoch: 0.6973947895791583, loss: 0.3145\n",
"[tabular-ft] 2025-08-13 23:43:29.077983+00:00 Training 70.3% complete - step: 1400, epoch: 0.7014028056112225, loss: 0.314\n",
"[tabular-ft] 2025-08-13 23:43:31.761547+00:00 Training 70.7% complete - step: 1408, epoch: 0.7054108216432866, loss: 0.3126\n",
"[tabular-ft] 2025-08-13 23:43:34.462737+00:00 Training 71.1% complete - step: 1416, epoch: 0.7094188376753507, loss: 0.3068\n",
"[tabular-ft] 2025-08-13 23:43:37.117111+00:00 Training 71.5% complete - step: 1424, epoch: 0.7134268537074149, loss: 0.2908\n",
"[tabular-ft] 2025-08-13 23:43:39.800065+00:00 Training 71.9% complete - step: 1432, epoch: 0.717434869739479, loss: 0.2985\n",
"[tabular-ft] 2025-08-13 23:43:42.458903+00:00 Training 72.3% complete - step: 1440, epoch: 0.7214428857715431, loss: 0.2839\n",
"[tabular-ft] 2025-08-13 23:43:45.070781+00:00 Training 72.7% complete - step: 1448, epoch: 0.7254509018036072, loss: 0.2771\n",
"[tabular-ft] 2025-08-13 23:43:47.793546+00:00 Training 73.1% complete - step: 1456, epoch: 0.7294589178356713, loss: 0.2693\n",
"[tabular-ft] 2025-08-13 23:43:50.467123+00:00 Training 73.5% complete - step: 1464, epoch: 0.7334669338677354, loss: 0.2639\n",
"[tabular-ft] 2025-08-13 23:43:53.130724+00:00 Training 73.9% complete - step: 1472, epoch: 0.7374749498997996, loss: 0.2678\n",
"[tabular-ft] 2025-08-13 23:43:55.815900+00:00 Training 74.3% complete - step: 1480, epoch: 0.7414829659318637, loss: 0.2553\n",
"[tabular-ft] 2025-08-13 23:43:58.480200+00:00 Training 74.7% complete - step: 1488, epoch: 0.7454909819639278, loss: 0.2505\n",
"[tabular-ft] 2025-08-13 23:44:01.180398+00:00 Training 75.1% complete - step: 1496, epoch: 0.749498997995992, loss: 0.2539\n",
"[tabular-ft] 2025-08-13 23:44:03.840363+00:00 Training 75.5% complete - step: 1504, epoch: 0.7535070140280561, loss: 0.2422\n",
"[tabular-ft] 2025-08-13 23:44:06.525943+00:00 Training 75.9% complete - step: 1512, epoch: 0.7575150300601202, loss: 0.2444\n",
"[tabular-ft] 2025-08-13 23:44:09.190984+00:00 Training 76.3% complete - step: 1520, epoch: 0.7615230460921844, loss: 0.2339\n",
"[tabular-ft] 2025-08-13 23:44:11.844365+00:00 Training 76.7% complete - step: 1528, epoch: 0.7655310621242485, loss: 0.233\n",
"[tabular-ft] 2025-08-13 23:44:14.531036+00:00 Training 77.1% complete - step: 1536, epoch: 0.7695390781563126, loss: 0.2251\n",
"[tabular-ft] 2025-08-13 23:44:17.170154+00:00 Training 77.5% complete - step: 1544, epoch: 0.7735470941883767, loss: 0.2266\n",
"[tabular-ft] 2025-08-13 23:44:19.805560+00:00 Training 77.9% complete - step: 1552, epoch: 0.7775551102204409, loss: 0.2161\n",
"[tabular-ft] 2025-08-13 23:44:22.469081+00:00 Training 78.3% complete - step: 1560, epoch: 0.781563126252505, loss: 0.2191\n",
"[tabular-ft] 2025-08-13 23:44:25.147737+00:00 Training 78.7% complete - step: 1568, epoch: 0.7855711422845691, loss: 0.2218\n",
"[tabular-ft] 2025-08-13 23:44:27.744403+00:00 Training 79.1% complete - step: 1576, epoch: 0.7895791583166333, loss: 0.2144\n",
"[tabular-ft] 2025-08-13 23:44:30.371847+00:00 Training 79.5% complete - step: 1584, epoch: 0.7935871743486974, loss: 0.2135\n",
"[tabular-ft] 2025-08-13 23:44:33.002447+00:00 Training 79.9% complete - step: 1592, epoch: 0.7975951903807615, loss: 0.2049\n",
"[tabular-ft] 2025-08-13 23:44:35.636489+00:00 Training 80.3% complete - step: 1600, epoch: 0.8016032064128257, loss: 0.204\n",
"[tabular-ft] 2025-08-13 23:44:38.295312+00:00 Training 80.7% complete - step: 1608, epoch: 0.8056112224448898, loss: 0.1986\n",
"[tabular-ft] 2025-08-13 23:44:41.014251+00:00 Training 81.1% complete - step: 1616, epoch: 0.8096192384769539, loss: 0.2044\n",
"[tabular-ft] 2025-08-13 23:44:43.628455+00:00 Training 81.5% complete - step: 1624, epoch: 0.8136272545090181, loss: 0.1934\n",
"[tabular-ft] 2025-08-13 23:44:46.332196+00:00 Training 81.9% complete - step: 1632, epoch: 0.8176352705410822, loss: 0.1963\n",
"[tabular-ft] 2025-08-13 23:44:49.006131+00:00 Training 82.3% complete - step: 1640, epoch: 0.8216432865731463, loss: 0.1795\n",
"[tabular-ft] 2025-08-13 23:44:51.653568+00:00 Training 82.7% complete - step: 1648, epoch: 0.8256513026052105, loss: 0.1807\n",
"[tabular-ft] 2025-08-13 23:44:54.371145+00:00 Training 83.1% complete - step: 1656, epoch: 0.8296593186372746, loss: 0.1821\n",
"[tabular-ft] 2025-08-13 23:44:57.025214+00:00 Training 83.5% complete - step: 1664, epoch: 0.8336673346693386, loss: 0.1831\n",
"[tabular-ft] 2025-08-13 23:44:59.672162+00:00 Training 83.9% complete - step: 1672, epoch: 0.8376753507014028, loss: 0.1805\n",
"[tabular-ft] 2025-08-13 23:45:02.307011+00:00 Training 84.3% complete - step: 1680, epoch: 0.8416833667334669, loss: 0.1724\n",
"[tabular-ft] 2025-08-13 23:45:04.958236+00:00 Training 84.7% complete - step: 1688, epoch: 0.845691382765531, loss: 0.1783\n",
"[tabular-ft] 2025-08-13 23:45:07.620960+00:00 Training 85.1% complete - step: 1696, epoch: 0.8496993987975952, loss: 0.1705\n",
"[tabular-ft] 2025-08-13 23:45:10.229429+00:00 Training 85.5% complete - step: 1704, epoch: 0.8537074148296593, loss: 0.1726\n",
"[tabular-ft] 2025-08-13 23:45:12.902175+00:00 Training 85.9% complete - step: 1712, epoch: 0.8577154308617234, loss: 0.1631\n",
"[tabular-ft] 2025-08-13 23:45:15.573886+00:00 Training 86.3% complete - step: 1720, epoch: 0.8617234468937875, loss: 0.167\n",
"[tabular-ft] 2025-08-13 23:45:18.206394+00:00 Training 86.7% complete - step: 1728, epoch: 0.8657314629258517, loss: 0.1659\n",
"[tabular-ft] 2025-08-13 23:45:20.858470+00:00 Training 87.1% complete - step: 1736, epoch: 0.8697394789579158, loss: 0.1639\n",
"[tabular-ft] 2025-08-13 23:45:23.525020+00:00 Training 87.6% complete - step: 1744, epoch: 0.87374749498998, loss: 0.1552\n",
"[tabular-ft] 2025-08-13 23:45:26.125000+00:00 Training 88.0% complete - step: 1752, epoch: 0.8777555110220441, loss: 0.1643\n",
"[tabular-ft] 2025-08-13 23:45:28.811200+00:00 Training 88.4% complete - step: 1760, epoch: 0.8817635270541082, loss: 0.1581\n",
"[tabular-ft] 2025-08-13 23:45:31.422186+00:00 Training 88.8% complete - step: 1768, epoch: 0.8857715430861723, loss: 0.157\n",
"[tabular-ft] 2025-08-13 23:45:34.008210+00:00 Training 89.2% complete - step: 1776, epoch: 0.8897795591182365, loss: 0.1601\n",
"[tabular-ft] 2025-08-13 23:45:36.621176+00:00 Training 89.6% complete - step: 1784, epoch: 0.8937875751503006, loss: 0.1552\n",
"[tabular-ft] 2025-08-13 23:45:39.233644+00:00 Training 90.0% complete - step: 1792, epoch: 0.8977955911823647, loss: 0.1602\n",
"[tabular-ft] 2025-08-13 23:45:41.825551+00:00 Training 90.4% complete - step: 1800, epoch: 0.9018036072144289, loss: 0.1544\n",
"[tabular-ft] 2025-08-13 23:45:44.526956+00:00 Training 90.8% complete - step: 1808, epoch: 0.905811623246493, loss: 0.1502\n",
"[tabular-ft] 2025-08-13 23:45:47.208540+00:00 Training 91.2% complete - step: 1816, epoch: 0.9098196392785571, loss: 0.1516\n",
"[tabular-ft] 2025-08-13 23:45:49.864248+00:00 Training 91.6% complete - step: 1824, epoch: 0.9138276553106213, loss: 0.1548\n",
"[tabular-ft] 2025-08-13 23:45:52.517566+00:00 Training 92.0% complete - step: 1832, epoch: 0.9178356713426854, loss: 0.1498\n",
"[tabular-ft] 2025-08-13 23:45:55.175873+00:00 Training 92.4% complete - step: 1840, epoch: 0.9218436873747495, loss: 0.1511\n",
"[tabular-ft] 2025-08-13 23:45:57.856332+00:00 Training 92.8% complete - step: 1848, epoch: 0.9258517034068137, loss: 0.1484\n",
"[tabular-ft] 2025-08-13 23:46:00.446309+00:00 Training 93.2% complete - step: 1856, epoch: 0.9298597194388778, loss: 0.1521\n",
"[tabular-ft] 2025-08-13 23:46:03.088862+00:00 Training 93.6% complete - step: 1864, epoch: 0.9338677354709419, loss: 0.1487\n",
"[tabular-ft] 2025-08-13 23:46:05.702522+00:00 Training 94.0% complete - step: 1872, epoch: 0.9378757515030061, loss: 0.1484\n",
"[tabular-ft] 2025-08-13 23:46:08.396645+00:00 Training 94.4% complete - step: 1880, epoch: 0.9418837675350702, loss: 0.1428\n",
"[tabular-ft] 2025-08-13 23:46:11.032617+00:00 Training 94.8% complete - step: 1888, epoch: 0.9458917835671342, loss: 0.1451\n",
"[tabular-ft] 2025-08-13 23:46:13.686205+00:00 Training 95.2% complete - step: 1896, epoch: 0.9498997995991983, loss: 0.1476\n",
"[tabular-ft] 2025-08-13 23:46:16.313181+00:00 Training 95.6% complete - step: 1904, epoch: 0.9539078156312625, loss: 0.1438\n",
"[tabular-ft] 2025-08-13 23:46:18.972540+00:00 Training 96.0% complete - step: 1912, epoch: 0.9579158316633266, loss: 0.1462\n",
"[tabular-ft] 2025-08-13 23:46:21.632784+00:00 Training 96.4% complete - step: 1920, epoch: 0.9619238476953907, loss: 0.1432\n",
"[tabular-ft] 2025-08-13 23:46:24.307284+00:00 Training 96.8% complete - step: 1928, epoch: 0.9659318637274549, loss: 0.1455\n",
"[tabular-ft] 2025-08-13 23:46:26.964089+00:00 Training 97.2% complete - step: 1936, epoch: 0.969939879759519, loss: 0.1486\n",
"[tabular-ft] 2025-08-13 23:46:29.573306+00:00 Training 97.6% complete - step: 1944, epoch: 0.9739478957915831, loss: 0.1465\n",
"[tabular-ft] 2025-08-13 23:46:32.232521+00:00 Training 98.0% complete - step: 1952, epoch: 0.9779559118236473, loss: 0.1442\n",
"[tabular-ft] 2025-08-13 23:46:34.820749+00:00 Training 98.4% complete - step: 1960, epoch: 0.9819639278557114, loss: 0.1426\n",
"[tabular-ft] 2025-08-13 23:46:37.480681+00:00 Training 98.8% complete - step: 1968, epoch: 0.9859719438877755, loss: 0.1446\n",
"[tabular-ft] 2025-08-13 23:46:40.134961+00:00 Training 99.2% complete - step: 1976, epoch: 0.9899799599198397, loss: 0.1491\n",
"[tabular-ft] 2025-08-13 23:46:42.835872+00:00 Training 99.6% complete - step: 1984, epoch: 0.9939879759519038, loss: 0.1402\n",
"[tabular-ft] 2025-08-13 23:46:45.869078+00:00 Training 100.0% complete - step: 1992, epoch: 0.9979959919839679, loss: 0.1465\n",
"[tabular-ft] 2025-08-13 23:46:46.089261+00:00 << 🧭 Tabular FT >> Saving LoRA adapter\n",
"[tabular-ft] 2025-08-13 23:46:46.726043+00:00 🙌 Training Completed\n",
"[tabular-ft] 2025-08-13 23:46:47.205597+00:00 Task 'train_tabular_ft' executed successfully\n",
"[tabular-ft] 2025-08-13 23:46:47.206483+00:00 << 🧭 Tabular FT >> 👀 Heads up -> Generation stopping is enabled:\n",
"patience: 3\n",
"invalid_fraction_threshold: 0.8\n",
"[tabular-ft] 2025-08-13 23:48:38.896533+00:00 Batch Generation Summary 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"🦜 Number of prompts submitted: 100\n",
"👍 Number of valid records generated: 182\n",
"👎 Number of invalid records generated: 363\n",
"📊 Percentage of records that are valid: 33.39%\n",
"🔬 Top invalid record errors:\n",
"╒═════════════════════════════════════════════════════════════╤══════════════╕\n",
"│ Error Category │ Percentage │\n",
"╞═════════════════════════════════════════════════════════════╪══════════════╡\n",
"│ Invalid JSON │ 72.7% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Missing required field │ 24.8% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be at most a given length │ 1.1% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be greater than or equal to a given number │ 0.6% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Invalid field type │ 0.6% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be less than or equal to a given number │ 0.3% │\n",
"╘═════════════════════════════════════════════════════════════╧══════════════╛\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-13 23:48:38.896736+00:00 Batch timing and progress 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"⏱️ Generation time: 33.9 seconds\n",
"🐢 Generation speed: 5.4 records per second.\n",
"⏳ Progress: 182 out of 1000 records generated.\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-13 23:49:14.011784+00:00 Batch Generation Summary 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"🦜 Number of prompts submitted: 100\n",
"👍 Number of valid records generated: 167\n",
"👎 Number of invalid records generated: 397\n",
"📊 Percentage of records that are valid: 29.61%\n",
"🔬 Top invalid record errors:\n",
"╒══════════════════════════════════════════════════════════╤══════════════╕\n",
"│ Error Category │ Percentage │\n",
"╞══════════════════════════════════════════════════════════╪══════════════╡\n",
"│ Invalid JSON │ 65.7% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Missing required field │ 31.7% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be less than or equal to a given number │ 1.5% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be at most a given length │ 0.8% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be at least a given length │ 0.3% │\n",
"╘══════════════════════════════════════════════════════════╧══════════════╛\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-13 23:49:14.012036+00:00 Batch timing and progress 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"⏱️ Generation time: 35.1 seconds\n",
"🐢 Generation speed: 4.8 records per second.\n",
"⏳ Progress: 349 out of 1000 records generated.\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-13 23:49:45.723007+00:00 Batch Generation Summary 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"🦜 Number of prompts submitted: 100\n",
"👍 Number of valid records generated: 183\n",
"👎 Number of invalid records generated: 333\n",
"📊 Percentage of records that are valid: 35.47%\n",
"🔬 Top invalid record errors:\n",
"╒═════════════════════════════════════════════════════════════╤══════════════╕\n",
"│ Error Category │ Percentage │\n",
"╞═════════════════════════════════════════════════════════════╪══════════════╡\n",
"│ Invalid JSON │ 78.7% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Missing required field │ 20.1% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be less than or equal to a given number │ 0.9% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be greater than or equal to a given number │ 0.3% │\n",
"╘═════════════════════════════════════════════════════════════╧══════════════╛\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-13 23:49:45.723294+00:00 Batch timing and progress 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"⏱️ Generation time: 31.7 seconds\n",
"🐢 Generation speed: 5.8 records per second.\n",
"⏳ Progress: 532 out of 1000 records generated.\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-13 23:50:17.967263+00:00 Batch Generation Summary 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"🦜 Number of prompts submitted: 100\n",
"👍 Number of valid records generated: 179\n",
"👎 Number of invalid records generated: 347\n",
"📊 Percentage of records that are valid: 34.03%\n",
"🔬 Top invalid record errors:\n",
"╒═════════════════════════════════════════════════════════════╤══════════════╕\n",
"│ Error Category │ Percentage │\n",
"╞═════════════════════════════════════════════════════════════╪══════════════╡\n",
"│ Invalid JSON │ 73.5% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Missing required field │ 23.9% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be less than or equal to a given number │ 1.4% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be greater than or equal to a given number │ 1.2% │\n",
"╘═════════════════════════════════════════════════════════════╧══════════════╛\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-13 23:50:17.967602+00:00 Batch timing and progress 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"⏱️ Generation time: 32.2 seconds\n",
"🐢 Generation speed: 5.6 records per second.\n",
"⏳ Progress: 711 out of 1000 records generated.\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-13 23:50:48.239439+00:00 Batch Generation Summary 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"🦜 Number of prompts submitted: 100\n",
"👍 Number of valid records generated: 145\n",
"👎 Number of invalid records generated: 314\n",
"📊 Percentage of records that are valid: 31.59%\n",
"🔬 Top invalid record errors:\n",
"╒═════════════════════════════════════════════════════════════╤══════════════╕\n",
"│ Error Category │ Percentage │\n",
"╞═════════════════════════════════════════════════════════════╪══════════════╡\n",
"│ Invalid JSON │ 75.5% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Missing required field │ 22.6% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be less than or equal to a given number │ 1.0% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be greater than or equal to a given number │ 0.6% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be at most a given length │ 0.3% │\n",
"╘═════════════════════════════════════════════════════════════╧══════════════╛\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-13 23:50:48.239830+00:00 Batch timing and progress 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"⏱️ Generation time: 30.3 seconds\n",
"🐢 Generation speed: 4.8 records per second.\n",
"⏳ Progress: 856 out of 1000 records generated.\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-13 23:51:20.552155+00:00 Batch Generation Summary 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"🦜 Number of prompts submitted: 94\n",
"👍 Number of valid records generated: 187\n",
"👎 Number of invalid records generated: 352\n",
"📊 Percentage of records that are valid: 34.69%\n",
"🔬 Top invalid record errors:\n",
"╒═════════════════════════════════════════════════════════════╤══════════════╕\n",
"│ Error Category │ Percentage │\n",
"╞═════════════════════════════════════════════════════════════╪══════════════╡\n",
"│ Invalid JSON │ 79.5% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Missing required field │ 18.2% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be at most a given length │ 1.1% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be greater than or equal to a given number │ 0.9% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be less than or equal to a given number │ 0.3% │\n",
"╘═════════════════════════════════════════════════════════════╧══════════════╛\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-13 23:51:20.552597+00:00 Batch timing and progress 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"⏱️ Generation time: 32.3 seconds\n",
"🐢 Generation speed: 5.8 records per second.\n",
"⌛️ Progress: 1043 out of 1000 records generated.\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-13 23:51:20.552771+00:00 🎉 Generation complete 🎉\n",
"[tabular-ft] 2025-08-13 23:51:21.957350+00:00 🙌 Successfully generated 1000 records\n",
"[tabular-ft] 2025-08-13 23:51:22.087923+00:00 Task 'generate_from_tabular_ft' executed successfully\n",
"[tabular-ft] 2025-08-13 23:51:22.283463+00:00 Task 'tabular_ft' executed successfully\n",
"[tabular-ft] 2025-08-13 23:51:22.283874+00:00 Task execution completed. Saving task outputs.\n",
"[tabular-ft] 2025-08-13 23:51:22.885338+00:00 Task outputs saved.\n",
"[tabular-ft] Task Status is now: RUN_STATUS_COMPLETED\n",
"Got task wt_31Fl4jQkgvGU2KPBHS0XTAZcQv1\n",
"[evaluate-safe-synthetics-dataset] Task Status is now: RUN_STATUS_ACTIVE\n",
"[evaluate-safe-synthetics-dataset] 2025-08-13 23:52:13.564574+00:00 Preparing step 'evaluate-safe-synthetics-dataset'\n",
"[evaluate-safe-synthetics-dataset] 2025-08-13 23:52:30.786570+00:00 Starting 'evaluate_safe_synthetics_dataset' task execution\n",
"[evaluate-safe-synthetics-dataset] 2025-08-13 23:52:47.489056+00:00 LLM column classification took 2.572293884999908 seconds.\n",
"[evaluate-safe-synthetics-dataset] 2025-08-13 23:53:21.223743+00:00 Task 'evaluate_safe_synthetics_dataset' executed successfully\n",
"[evaluate-safe-synthetics-dataset] 2025-08-13 23:53:21.224976+00:00 Task execution completed. Saving task outputs.\n",
"[evaluate-safe-synthetics-dataset] 2025-08-13 23:53:21.821792+00:00 Task outputs saved.\n",
"[evaluate-safe-synthetics-dataset] Task Status is now: RUN_STATUS_COMPLETED\n",
"Workflow run is now in status: RUN_STATUS_COMPLETED\n",
"✅ Training complete!\n",
"🎯 Quality score: 8.4\n",
"🔒 Privacy score: 5.8\n",
"❌ Trial failed: name 'score' is not defined\n"
]
},
{
"data": {
"text/html": [],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"<br> <style><br> .wandb-row {<br> display: flex;<br> flex-direction: row;<br> flex-wrap: wrap;<br> justify-content: flex-start;<br> width: 100%;<br> }<br> .wandb-col {<br> display: flex;<br> flex-direction: column;<br> flex-basis: 100%;<br> flex: 1;<br> padding: 10px;<br> }<br> </style><br><div class=\"wandb-row\"><div class=\"wandb-col\"><h3>Run history:</h3><br/><table class=\"wandb\"><tr><td>privacy_score</td><td>▁</td></tr><tr><td>quality_score</td><td>▁</td></tr></table><br/></div><div class=\"wandb-col\"><h3>Run summary:</h3><br/><table class=\"wandb\"><tr><td>error</td><td>name 'score' is not ...</td></tr><tr><td>privacy_score</td><td>5.8</td></tr><tr><td>quality_score</td><td>8.4</td></tr></table><br/></div></div>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" View run <strong style=\"color:#cdcd00\">safe-syn-lxmr1ys7</strong> at: <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps/runs/dk0an880' target=\"_blank\">https://wandb.ai/yamini_gretel/safe-syn-sweeps/runs/dk0an880</a><br> View project at: <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps' target=\"_blank\">https://wandb.ai/yamini_gretel/safe-syn-sweeps</a><br>Synced 5 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s)"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"Find logs at: <code>./wandb/run-20250813_192506-dk0an880/logs</code>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Traceback (most recent call last):\n",
" File \"/Users/ykagal/.venvs/jupyter/lib/python3.13/site-packages/wandb/agents/pyagent.py\", line 297, in _run_job\n",
" self._function()\n",
" ~~~~~~~~~~~~~~^^\n",
" File \"/var/folders/kx/_q607d_9603cg0389qczz_f00000gp/T/ipykernel_66388/2497270407.py\", line 29, in trial_agent\n",
" print(f\"✅ Trial complete → score={score}\")\n",
" ^^^^^\n",
"NameError: name 'score' is not defined. Did you mean: 'scores'?\n",
"\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \u001b[32m\u001b[41mERROR\u001b[0m Run dk0an880 errored: name 'score' is not defined\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: Agent Starting Run: 9hww7mum with config:\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tbatch_size: 2\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tgradient_accumulation_steps: 16\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tlearning_rate: 0.00018847766722608427\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tlora_alpha: 64\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tlora_dropout: 0.08576453159743103\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tlora_r: 32\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tmax_input_records: 5000\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tnum_records: 2000\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tnum_train_epochs: 5\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \trepetition_penalty: 1.1397895049956188\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \ttemperature: 0.8948162775702677\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \ttop_p: 0.9971495213568616\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \twarmup_ratio: 0.05901020017604965\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \tweight_decay: 0.0018779079890300656\n"
]
},
{
"data": {
"text/html": [
"Ignoring project 'safe-syn-sweeps' when running a sweep."
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"Ignoring entity 'yamini_gretel' when running a sweep."
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"Tracking run with wandb version 0.21.1"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"Run data is saved locally in <code>/Users/ykagal/Documents/notebooks/wandb/run-20250813_195359-9hww7mum</code>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"Syncing run <strong><a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps/runs/9hww7mum' target=\"_blank\">safe-syn-pd0lrdry</a></strong> to <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps' target=\"_blank\">Weights & Biases</a> (<a href='https://wandb.me/developer-guide' target=\"_blank\">docs</a>)<br>Sweep page: <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps/sweeps/e8o48dv0' target=\"_blank\">https://wandb.ai/yamini_gretel/safe-syn-sweeps/sweeps/e8o48dv0</a>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" View project at <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps' target=\"_blank\">https://wandb.ai/yamini_gretel/safe-syn-sweeps</a>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" View sweep at <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps/sweeps/e8o48dv0' target=\"_blank\">https://wandb.ai/yamini_gretel/safe-syn-sweeps/sweeps/e8o48dv0</a>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" View run at <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps/runs/9hww7mum' target=\"_blank\">https://wandb.ai/yamini_gretel/safe-syn-sweeps/runs/9hww7mum</a>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"🚀 Starting trial: safe-syn-pd0lrdry\n",
"📊 Parameters: {'batch_size': 2, 'gradient_accumulation_steps': 16, 'learning_rate': 0.00018847766722608427, 'lora_alpha': 64, 'lora_dropout': 0.08576453159743103, 'lora_r': 32, 'max_input_records': 5000, 'num_records': 2000, 'num_train_epochs': 5, 'repetition_penalty': 1.1397895049956188, 'temperature': 0.8948162775702677, 'top_p': 0.9971495213568616, 'warmup_ratio': 0.05901020017604965, 'weight_decay': 0.0018779079890300656}\n",
"Configuring generator for data source: https://gretel-datasets.s3.us-west-2.amazonaws.com/ecommerce_customers.csv\n",
"Configuring holdout: 0.05\n",
"Configuring synthetic data generation model: tabular_ft/default\n",
"▶️ Using Workflow: w_31Fl4cur1ZouU3Yha3TgN4tKDj3\n",
"▶️ Created Workflow Run: wr_31FoaHOmzhLPJ1bS3slOOUGpwlA\n",
"🔗 Workflow Run console link: https://console.gretel.ai/workflows/w_31Fl4cur1ZouU3Yha3TgN4tKDj3/runs/wr_31FoaHOmzhLPJ1bS3slOOUGpwlA\n",
"Fetching task logs for workflow run wr_31FoaHOmzhLPJ1bS3slOOUGpwlA\n",
"Workflow run is now in status: RUN_STATUS_CREATED\n",
"Got task wt_31FoaINK0nMC9fSMcYoQBpZK8Se\n",
"Workflow run is now in status: RUN_STATUS_ACTIVE\n",
"[read-data-source] 2025-08-13 23:54:20.998430+00:00 Preparing step 'read-data-source'\n",
"[read-data-source] 2025-08-13 23:54:30.464182+00:00 Starting 'data_source' task execution\n",
"[read-data-source] 2025-08-13 23:54:32.418878+00:00 Task 'data_source' executed successfully\n",
"[read-data-source] 2025-08-13 23:54:32.419393+00:00 Task execution completed. Saving task outputs.\n",
"[read-data-source] 2025-08-13 23:54:32.968810+00:00 Task outputs saved.\n",
"[read-data-source] Task Status is now: RUN_STATUS_COMPLETED\n",
"Got task wt_31FoaJHfPAjMLGahpHlMJF7ApJF\n",
"[holdout] Task Status is now: RUN_STATUS_ACTIVE\n",
"[holdout] 2025-08-13 23:55:30.102324+00:00 Preparing step 'holdout'\n",
"[holdout] 2025-08-13 23:55:40.589184+00:00 Starting 'holdout' task execution\n",
"[holdout] 2025-08-13 23:55:40.591490+00:00 Dataset test holdout split configuration complete. Train dataset: 475 rows x 8 columns. Test dataset: 25 rows x 8 columns.\n",
"[holdout] 2025-08-13 23:55:40.600746+00:00 Task 'holdout' executed successfully\n",
"[holdout] 2025-08-13 23:55:40.601161+00:00 Task execution completed. Saving task outputs.\n",
"[holdout] 2025-08-13 23:55:41.472792+00:00 Task outputs saved.\n",
"[holdout] Task Status is now: RUN_STATUS_COMPLETED\n",
"Got task wt_31FoaK1kdYpJujYd1XkeCphmCG1\n",
"[tabular-ft] Task Status is now: RUN_STATUS_ACTIVE\n",
"[tabular-ft] 2025-08-13 23:56:38.835883+00:00 Preparing step 'tabular-ft'\n",
"[tabular-ft] 2025-08-13 23:56:53.226971+00:00 Starting 'tabular_ft' task execution\n",
"[tabular-ft] 2025-08-13 23:57:08.768922+00:00 Analyzing input data and checking for auto-params...\n",
"[tabular-ft] 2025-08-13 23:57:08.789925+00:00 Parameter `rope_scaling_factor` was automatically set to 1 based on an estimated token count given the lengths of each training record and the column names.\n",
"[tabular-ft] 2025-08-13 23:57:08.790235+00:00 Found 3 auto-params that were set based on input data. - num_input_records_to_sample: 25000, use_unsloth: True, rope_scaling_factor: 1\n",
"[tabular-ft] 2025-08-13 23:57:08.790312+00:00 Using updated model config: \n",
"{\n",
" \"group_training_examples_by\": null,\n",
" \"order_training_examples_by\": null,\n",
" \"max_sequences_per_example\": null,\n",
" \"pretrained_model\": null,\n",
" \"params\": {\n",
" \"num_input_records_to_sample\": 25000,\n",
" \"batch_size\": 1,\n",
" \"gradient_accumulation_steps\": 8,\n",
" \"weight_decay\": 0.01,\n",
" \"warmup_ratio\": 0.05,\n",
" \"lr_scheduler\": \"cosine\",\n",
" \"learning_rate\": 0.0005,\n",
" \"lora_r\": 32,\n",
" \"lora_alpha_over_r\": 1.0,\n",
" \"lora_target_modules\": [\n",
" \"q_proj\",\n",
" \"k_proj\",\n",
" \"v_proj\",\n",
" \"o_proj\"\n",
" ],\n",
" \"use_unsloth\": true,\n",
" \"rope_scaling_factor\": 1,\n",
" \"validation_ratio\": 0.0,\n",
" \"validation_steps\": 15\n",
" },\n",
" \"privacy_params\": {\n",
" \"dp\": false,\n",
" \"epsilon\": 8.0,\n",
" \"delta\": \"auto\",\n",
" \"per_sample_max_grad_norm\": 1.0\n",
" },\n",
" \"data_config\": null\n",
"}\n",
"[tabular-ft] 2025-08-13 23:57:08.791350+00:00 << 🧭 Tabular FT >> Preparing for training\n",
"[tabular-ft] 2025-08-13 23:57:21.479266+00:00 << 🧭 Tabular FT >> Tokenizing records\n",
"[tabular-ft] 2025-08-13 23:57:21.620225+00:00 << 🧭 Tabular FT >> Number of unique train records: 475\n",
"[tabular-ft] 2025-08-13 23:57:21.620380+00:00 << 🧭 Tabular FT >> Assembling examples from 5263.2% of the input records\n",
"[tabular-ft] 2025-08-13 23:57:59.604105+00:00 << 🧭 Tabular FT >> Training Example Statistics:\n",
"[tabular-ft] 2025-08-13 23:57:59.605000+00:00 \n",
"╒════════╤═════════════════════╤══════════════════════╤═══════════════════════╕\n",
"│ │ Tokens per record │ Tokens per example │ Records per example │\n",
"╞════════╪═════════════════════╪══════════════════════╪═══════════════════════╡\n",
"│ min │ 136 │ 1590 │ 10 │\n",
"├────────┼─────────────────────┼──────────────────────┼───────────────────────┤\n",
"│ max │ 164 │ 2048 │ 13 │\n",
"├────────┼─────────────────────┼──────────────────────┼───────────────────────┤\n",
"│ mean │ 150.914 │ 1978.85 │ 12.562 │\n",
"├────────┼─────────────────────┼──────────────────────┼───────────────────────┤\n",
"│ stddev │ 5.152 │ 62.155 │ 0.499 │\n",
"╘════════╧═════════════════════╧══════════════════════╧═══════════════════════╛\n",
"[tabular-ft] 2025-08-13 23:57:59.605120+00:00 << 🧭 Tabular FT >> Number of training examples: 1990\n",
"[tabular-ft] 2025-08-13 23:58:03.045625+00:00 << 🧭 Tabular FT >> Using PEFT - 9.01 million parameters are trainable\n",
"[tabular-ft] 2025-08-13 23:58:18.590311+00:00 Training 0.4% complete - step: 8, epoch: 0.004020100502512563, loss: 1.633\n",
"[tabular-ft] 2025-08-13 23:58:21.228039+00:00 Training 0.8% complete - step: 16, epoch: 0.008040201005025126, loss: 1.6197\n",
"[tabular-ft] 2025-08-13 23:58:28.172734+00:00 Training 1.2% complete - step: 24, epoch: 0.012060301507537688, loss: 1.6065\n",
"[tabular-ft] 2025-08-13 23:58:30.739267+00:00 Training 1.6% complete - step: 32, epoch: 0.016080402010050253, loss: 1.5955\n",
"[tabular-ft] 2025-08-13 23:58:33.421629+00:00 Training 2.0% complete - step: 40, epoch: 0.020100502512562814, loss: 1.5539\n",
"[tabular-ft] 2025-08-13 23:58:36.063270+00:00 Training 2.4% complete - step: 48, epoch: 0.024120603015075376, loss: 1.5345\n",
"[tabular-ft] 2025-08-13 23:58:38.702425+00:00 Training 2.8% complete - step: 56, epoch: 0.02814070351758794, loss: 1.5071\n",
"[tabular-ft] 2025-08-13 23:58:41.393836+00:00 Training 3.2% complete - step: 64, epoch: 0.032160804020100506, loss: 1.4942\n",
"[tabular-ft] 2025-08-13 23:58:44.046447+00:00 Training 3.6% complete - step: 72, epoch: 0.036180904522613064, loss: 1.468\n",
"[tabular-ft] 2025-08-13 23:58:47.759247+00:00 Training 4.0% complete - step: 80, epoch: 0.04020100502512563, loss: 1.4623\n",
"[tabular-ft] 2025-08-13 23:58:50.440364+00:00 Training 4.4% complete - step: 88, epoch: 0.044221105527638194, loss: 1.4489\n",
"[tabular-ft] 2025-08-13 23:58:53.075880+00:00 Training 4.8% complete - step: 96, epoch: 0.04824120603015075, loss: 1.4488\n",
"[tabular-ft] 2025-08-13 23:58:55.713004+00:00 Training 5.2% complete - step: 104, epoch: 0.05226130653266332, loss: 1.4317\n",
"[tabular-ft] 2025-08-13 23:58:58.387937+00:00 Training 5.6% complete - step: 112, epoch: 0.05628140703517588, loss: 1.4284\n",
"[tabular-ft] 2025-08-13 23:59:01.010496+00:00 Training 6.0% complete - step: 120, epoch: 0.06030150753768844, loss: 1.4332\n",
"[tabular-ft] 2025-08-13 23:59:03.710113+00:00 Training 6.5% complete - step: 128, epoch: 0.06432160804020101, loss: 1.4236\n",
"[tabular-ft] 2025-08-13 23:59:06.396060+00:00 Training 6.9% complete - step: 136, epoch: 0.06834170854271357, loss: 1.4243\n",
"[tabular-ft] 2025-08-13 23:59:09.081333+00:00 Training 7.3% complete - step: 144, epoch: 0.07236180904522613, loss: 1.4142\n",
"[tabular-ft] 2025-08-13 23:59:11.720685+00:00 Training 7.7% complete - step: 152, epoch: 0.0763819095477387, loss: 1.4194\n",
"[tabular-ft] 2025-08-13 23:59:14.397093+00:00 Training 8.1% complete - step: 160, epoch: 0.08040201005025126, loss: 1.4113\n",
"[tabular-ft] 2025-08-13 23:59:17.078511+00:00 Training 8.5% complete - step: 168, epoch: 0.08442211055276382, loss: 1.402\n",
"[tabular-ft] 2025-08-13 23:59:19.713230+00:00 Training 8.9% complete - step: 176, epoch: 0.08844221105527639, loss: 1.3929\n",
"[tabular-ft] 2025-08-13 23:59:22.378771+00:00 Training 9.3% complete - step: 184, epoch: 0.09246231155778895, loss: 1.3911\n",
"[tabular-ft] 2025-08-13 23:59:25.086124+00:00 Training 9.7% complete - step: 192, epoch: 0.0964824120603015, loss: 1.3846\n",
"[tabular-ft] 2025-08-13 23:59:27.717608+00:00 Training 10.1% complete - step: 200, epoch: 0.10050251256281408, loss: 1.3923\n",
"[tabular-ft] 2025-08-13 23:59:30.333708+00:00 Training 10.5% complete - step: 208, epoch: 0.10452261306532663, loss: 1.3902\n",
"[tabular-ft] 2025-08-13 23:59:32.992370+00:00 Training 10.9% complete - step: 216, epoch: 0.10854271356783919, loss: 1.3907\n",
"[tabular-ft] 2025-08-13 23:59:35.650509+00:00 Training 11.3% complete - step: 224, epoch: 0.11256281407035176, loss: 1.3953\n",
"[tabular-ft] 2025-08-13 23:59:38.303336+00:00 Training 11.7% complete - step: 232, epoch: 0.11658291457286432, loss: 1.3665\n",
"[tabular-ft] 2025-08-13 23:59:40.980526+00:00 Training 12.1% complete - step: 240, epoch: 0.12060301507537688, loss: 1.3854\n",
"[tabular-ft] 2025-08-13 23:59:43.544087+00:00 Training 12.5% complete - step: 248, epoch: 0.12462311557788945, loss: 1.3709\n",
"[tabular-ft] 2025-08-13 23:59:46.172393+00:00 Training 12.9% complete - step: 256, epoch: 0.12864321608040202, loss: 1.3676\n",
"[tabular-ft] 2025-08-13 23:59:48.770151+00:00 Training 13.3% complete - step: 264, epoch: 0.13266331658291458, loss: 1.366\n",
"[tabular-ft] 2025-08-13 23:59:51.449635+00:00 Training 13.7% complete - step: 272, epoch: 0.13668341708542714, loss: 1.3794\n",
"[tabular-ft] 2025-08-13 23:59:54.112774+00:00 Training 14.1% complete - step: 280, epoch: 0.1407035175879397, loss: 1.3677\n",
"[tabular-ft] 2025-08-13 23:59:56.793204+00:00 Training 14.5% complete - step: 288, epoch: 0.14472361809045226, loss: 1.3608\n",
"[tabular-ft] 2025-08-13 23:59:59.390068+00:00 Training 14.9% complete - step: 296, epoch: 0.1487437185929648, loss: 1.3606\n",
"[tabular-ft] 2025-08-14 00:00:02.048901+00:00 Training 15.3% complete - step: 304, epoch: 0.1527638190954774, loss: 1.3676\n",
"[tabular-ft] 2025-08-14 00:00:04.725207+00:00 Training 15.7% complete - step: 312, epoch: 0.15678391959798996, loss: 1.3513\n",
"[tabular-ft] 2025-08-14 00:00:07.345552+00:00 Training 16.1% complete - step: 320, epoch: 0.16080402010050251, loss: 1.3443\n",
"[tabular-ft] 2025-08-14 00:00:09.979691+00:00 Training 16.5% complete - step: 328, epoch: 0.16482412060301507, loss: 1.3432\n",
"[tabular-ft] 2025-08-14 00:00:12.637681+00:00 Training 16.9% complete - step: 336, epoch: 0.16884422110552763, loss: 1.3382\n",
"[tabular-ft] 2025-08-14 00:00:15.344534+00:00 Training 17.3% complete - step: 344, epoch: 0.1728643216080402, loss: 1.3306\n",
"[tabular-ft] 2025-08-14 00:00:17.984500+00:00 Training 17.7% complete - step: 352, epoch: 0.17688442211055277, loss: 1.3477\n",
"[tabular-ft] 2025-08-14 00:00:20.708916+00:00 Training 18.1% complete - step: 360, epoch: 0.18090452261306533, loss: 1.342\n",
"[tabular-ft] 2025-08-14 00:00:23.411978+00:00 Training 18.5% complete - step: 368, epoch: 0.1849246231155779, loss: 1.3387\n",
"[tabular-ft] 2025-08-14 00:00:26.100577+00:00 Training 19.0% complete - step: 376, epoch: 0.18894472361809045, loss: 1.3343\n",
"[tabular-ft] 2025-08-14 00:00:28.774657+00:00 Training 19.4% complete - step: 384, epoch: 0.192964824120603, loss: 1.3254\n",
"[tabular-ft] 2025-08-14 00:00:31.407847+00:00 Training 19.8% complete - step: 392, epoch: 0.19698492462311556, loss: 1.3258\n",
"[tabular-ft] 2025-08-14 00:00:34.081079+00:00 Training 20.2% complete - step: 400, epoch: 0.20100502512562815, loss: 1.3348\n",
"[tabular-ft] 2025-08-14 00:00:36.688643+00:00 Training 20.6% complete - step: 408, epoch: 0.2050251256281407, loss: 1.3224\n",
"[tabular-ft] 2025-08-14 00:00:39.329309+00:00 Training 21.0% complete - step: 416, epoch: 0.20904522613065327, loss: 1.3164\n",
"[tabular-ft] 2025-08-14 00:00:41.987675+00:00 Training 21.4% complete - step: 424, epoch: 0.21306532663316582, loss: 1.3123\n",
"[tabular-ft] 2025-08-14 00:00:44.691948+00:00 Training 21.8% complete - step: 432, epoch: 0.21708542713567838, loss: 1.3012\n",
"[tabular-ft] 2025-08-14 00:00:47.350993+00:00 Training 22.2% complete - step: 440, epoch: 0.22110552763819097, loss: 1.2989\n",
"[tabular-ft] 2025-08-14 00:00:50.016618+00:00 Training 22.6% complete - step: 448, epoch: 0.22512562814070353, loss: 1.2987\n",
"[tabular-ft] 2025-08-14 00:00:52.614711+00:00 Training 23.0% complete - step: 456, epoch: 0.22914572864321608, loss: 1.2974\n",
"[tabular-ft] 2025-08-14 00:00:55.308223+00:00 Training 23.4% complete - step: 464, epoch: 0.23316582914572864, loss: 1.292\n",
"[tabular-ft] 2025-08-14 00:00:57.992890+00:00 Training 23.8% complete - step: 472, epoch: 0.2371859296482412, loss: 1.2895\n",
"[tabular-ft] 2025-08-14 00:01:00.655568+00:00 Training 24.2% complete - step: 480, epoch: 0.24120603015075376, loss: 1.2802\n",
"[tabular-ft] 2025-08-14 00:01:03.294888+00:00 Training 24.6% complete - step: 488, epoch: 0.24522613065326634, loss: 1.2779\n",
"[tabular-ft] 2025-08-14 00:01:05.981622+00:00 Training 25.0% complete - step: 496, epoch: 0.2492462311557789, loss: 1.2653\n",
"[tabular-ft] 2025-08-14 00:01:08.653750+00:00 Training 25.4% complete - step: 504, epoch: 0.25326633165829143, loss: 1.2488\n",
"[tabular-ft] 2025-08-14 00:01:11.291172+00:00 Training 25.8% complete - step: 512, epoch: 0.25728643216080405, loss: 1.2589\n",
"[tabular-ft] 2025-08-14 00:01:13.949220+00:00 Training 26.2% complete - step: 520, epoch: 0.2613065326633166, loss: 1.2604\n",
"[tabular-ft] 2025-08-14 00:01:16.630986+00:00 Training 26.6% complete - step: 528, epoch: 0.26532663316582916, loss: 1.2472\n",
"[tabular-ft] 2025-08-14 00:01:19.319122+00:00 Training 27.0% complete - step: 536, epoch: 0.2693467336683417, loss: 1.2565\n",
"[tabular-ft] 2025-08-14 00:01:22.002064+00:00 Training 27.4% complete - step: 544, epoch: 0.2733668341708543, loss: 1.2267\n",
"[tabular-ft] 2025-08-14 00:01:24.666359+00:00 Training 27.8% complete - step: 552, epoch: 0.27738693467336684, loss: 1.2268\n",
"[tabular-ft] 2025-08-14 00:01:27.369163+00:00 Training 28.2% complete - step: 560, epoch: 0.2814070351758794, loss: 1.2101\n",
"[tabular-ft] 2025-08-14 00:01:30.053101+00:00 Training 28.6% complete - step: 568, epoch: 0.28542713567839195, loss: 1.2261\n",
"[tabular-ft] 2025-08-14 00:01:32.665892+00:00 Training 29.0% complete - step: 576, epoch: 0.2894472361809045, loss: 1.2108\n",
"[tabular-ft] 2025-08-14 00:01:35.329623+00:00 Training 29.4% complete - step: 584, epoch: 0.29346733668341707, loss: 1.1933\n",
"[tabular-ft] 2025-08-14 00:01:37.981897+00:00 Training 29.8% complete - step: 592, epoch: 0.2974874371859296, loss: 1.1803\n",
"[tabular-ft] 2025-08-14 00:01:40.680849+00:00 Training 30.2% complete - step: 600, epoch: 0.3015075376884422, loss: 1.173\n",
"[tabular-ft] 2025-08-14 00:01:43.368352+00:00 Training 30.6% complete - step: 608, epoch: 0.3055276381909548, loss: 1.1609\n",
"[tabular-ft] 2025-08-14 00:01:46.060427+00:00 Training 31.0% complete - step: 616, epoch: 0.30954773869346736, loss: 1.1576\n",
"[tabular-ft] 2025-08-14 00:01:48.772524+00:00 Training 31.5% complete - step: 624, epoch: 0.3135678391959799, loss: 1.1625\n",
"[tabular-ft] 2025-08-14 00:01:51.433086+00:00 Training 31.9% complete - step: 632, epoch: 0.31758793969849247, loss: 1.1303\n",
"[tabular-ft] 2025-08-14 00:01:54.140214+00:00 Training 32.3% complete - step: 640, epoch: 0.32160804020100503, loss: 1.1455\n",
"[tabular-ft] 2025-08-14 00:01:56.821410+00:00 Training 32.7% complete - step: 648, epoch: 0.3256281407035176, loss: 1.1377\n",
"[tabular-ft] 2025-08-14 00:01:59.481492+00:00 Training 33.1% complete - step: 656, epoch: 0.32964824120603015, loss: 1.1273\n",
"[tabular-ft] 2025-08-14 00:02:02.108796+00:00 Training 33.5% complete - step: 664, epoch: 0.3336683417085427, loss: 1.099\n",
"[tabular-ft] 2025-08-14 00:02:04.766346+00:00 Training 33.9% complete - step: 672, epoch: 0.33768844221105526, loss: 1.1064\n",
"[tabular-ft] 2025-08-14 00:02:07.457081+00:00 Training 34.3% complete - step: 680, epoch: 0.3417085427135678, loss: 1.0869\n",
"[tabular-ft] 2025-08-14 00:02:10.163930+00:00 Training 34.7% complete - step: 688, epoch: 0.3457286432160804, loss: 1.1107\n",
"[tabular-ft] 2025-08-14 00:02:12.817256+00:00 Training 35.1% complete - step: 696, epoch: 0.349748743718593, loss: 1.0925\n",
"[tabular-ft] 2025-08-14 00:02:15.456478+00:00 Training 35.5% complete - step: 704, epoch: 0.35376884422110555, loss: 1.0706\n",
"[tabular-ft] 2025-08-14 00:02:18.134856+00:00 Training 35.9% complete - step: 712, epoch: 0.3577889447236181, loss: 1.0585\n",
"[tabular-ft] 2025-08-14 00:02:20.792100+00:00 Training 36.3% complete - step: 720, epoch: 0.36180904522613067, loss: 1.0573\n",
"[tabular-ft] 2025-08-14 00:02:23.430179+00:00 Training 36.7% complete - step: 728, epoch: 0.3658291457286432, loss: 1.0517\n",
"[tabular-ft] 2025-08-14 00:02:26.098421+00:00 Training 37.1% complete - step: 736, epoch: 0.3698492462311558, loss: 1.0501\n",
"[tabular-ft] 2025-08-14 00:02:28.741282+00:00 Training 37.5% complete - step: 744, epoch: 0.37386934673366834, loss: 1.0214\n",
"[tabular-ft] 2025-08-14 00:02:31.447596+00:00 Training 37.9% complete - step: 752, epoch: 0.3778894472361809, loss: 1.0298\n",
"[tabular-ft] 2025-08-14 00:02:34.130542+00:00 Training 38.3% complete - step: 760, epoch: 0.38190954773869346, loss: 0.995\n",
"[tabular-ft] 2025-08-14 00:02:36.788075+00:00 Training 38.7% complete - step: 768, epoch: 0.385929648241206, loss: 0.9956\n",
"[tabular-ft] 2025-08-14 00:02:39.467909+00:00 Training 39.1% complete - step: 776, epoch: 0.38994974874371857, loss: 0.9755\n",
"[tabular-ft] 2025-08-14 00:02:42.127553+00:00 Training 39.5% complete - step: 784, epoch: 0.39396984924623113, loss: 0.9677\n",
"[tabular-ft] 2025-08-14 00:02:44.778824+00:00 Training 39.9% complete - step: 792, epoch: 0.39798994974874374, loss: 0.9492\n",
"[tabular-ft] 2025-08-14 00:02:47.463296+00:00 Training 40.3% complete - step: 800, epoch: 0.4020100502512563, loss: 0.9501\n",
"[tabular-ft] 2025-08-14 00:02:50.120189+00:00 Training 40.7% complete - step: 808, epoch: 0.40603015075376886, loss: 0.9394\n",
"[tabular-ft] 2025-08-14 00:02:52.798973+00:00 Training 41.1% complete - step: 816, epoch: 0.4100502512562814, loss: 0.9201\n",
"[tabular-ft] 2025-08-14 00:02:55.461076+00:00 Training 41.5% complete - step: 824, epoch: 0.414070351758794, loss: 0.9045\n",
"[tabular-ft] 2025-08-14 00:02:58.116370+00:00 Training 41.9% complete - step: 832, epoch: 0.41809045226130653, loss: 0.9081\n",
"[tabular-ft] 2025-08-14 00:03:00.732577+00:00 Training 42.3% complete - step: 840, epoch: 0.4221105527638191, loss: 0.8765\n",
"[tabular-ft] 2025-08-14 00:03:03.411549+00:00 Training 42.7% complete - step: 848, epoch: 0.42613065326633165, loss: 0.8752\n",
"[tabular-ft] 2025-08-14 00:03:06.133046+00:00 Training 43.1% complete - step: 856, epoch: 0.4301507537688442, loss: 0.8801\n",
"[tabular-ft] 2025-08-14 00:03:08.818906+00:00 Training 43.5% complete - step: 864, epoch: 0.43417085427135677, loss: 0.8499\n",
"[tabular-ft] 2025-08-14 00:03:11.432584+00:00 Training 44.0% complete - step: 872, epoch: 0.4381909547738693, loss: 0.8575\n",
"[tabular-ft] 2025-08-14 00:03:14.116474+00:00 Training 44.4% complete - step: 880, epoch: 0.44221105527638194, loss: 0.8382\n",
"[tabular-ft] 2025-08-14 00:03:16.775495+00:00 Training 44.8% complete - step: 888, epoch: 0.4462311557788945, loss: 0.8438\n",
"[tabular-ft] 2025-08-14 00:03:19.458713+00:00 Training 45.2% complete - step: 896, epoch: 0.45025125628140705, loss: 0.7934\n",
"[tabular-ft] 2025-08-14 00:03:22.113246+00:00 Training 45.6% complete - step: 904, epoch: 0.4542713567839196, loss: 0.8099\n",
"[tabular-ft] 2025-08-14 00:03:24.754937+00:00 Training 46.0% complete - step: 912, epoch: 0.45829145728643217, loss: 0.8012\n",
"[tabular-ft] 2025-08-14 00:03:27.398932+00:00 Training 46.4% complete - step: 920, epoch: 0.4623115577889447, loss: 0.785\n",
"[tabular-ft] 2025-08-14 00:03:30.103460+00:00 Training 46.8% complete - step: 928, epoch: 0.4663316582914573, loss: 0.7576\n",
"[tabular-ft] 2025-08-14 00:03:32.773311+00:00 Training 47.2% complete - step: 936, epoch: 0.47035175879396984, loss: 0.7415\n",
"[tabular-ft] 2025-08-14 00:03:35.385502+00:00 Training 47.6% complete - step: 944, epoch: 0.4743718592964824, loss: 0.7515\n",
"[tabular-ft] 2025-08-14 00:03:38.068243+00:00 Training 48.0% complete - step: 952, epoch: 0.47839195979899496, loss: 0.7319\n",
"[tabular-ft] 2025-08-14 00:03:40.745770+00:00 Training 48.4% complete - step: 960, epoch: 0.4824120603015075, loss: 0.7374\n",
"[tabular-ft] 2025-08-14 00:03:43.386019+00:00 Training 48.8% complete - step: 968, epoch: 0.4864321608040201, loss: 0.7192\n",
"[tabular-ft] 2025-08-14 00:03:46.049509+00:00 Training 49.2% complete - step: 976, epoch: 0.4904522613065327, loss: 0.7137\n",
"[tabular-ft] 2025-08-14 00:03:51.100852+00:00 Training 49.6% complete - step: 984, epoch: 0.49447236180904525, loss: 0.6921\n",
"[tabular-ft] 2025-08-14 00:03:53.733492+00:00 Training 50.0% complete - step: 992, epoch: 0.4984924623115578, loss: 0.6771\n",
"[tabular-ft] 2025-08-14 00:03:56.375992+00:00 Training 50.4% complete - step: 1000, epoch: 0.5025125628140703, loss: 0.6391\n",
"[tabular-ft] 2025-08-14 00:03:59.009373+00:00 Training 50.8% complete - step: 1008, epoch: 0.5065326633165829, loss: 0.6554\n",
"[tabular-ft] 2025-08-14 00:04:01.665045+00:00 Training 51.2% complete - step: 1016, epoch: 0.5105527638190954, loss: 0.6129\n",
"[tabular-ft] 2025-08-14 00:04:04.310134+00:00 Training 51.6% complete - step: 1024, epoch: 0.5145728643216081, loss: 0.5918\n",
"[tabular-ft] 2025-08-14 00:04:06.969717+00:00 Training 52.0% complete - step: 1032, epoch: 0.5185929648241207, loss: 0.6113\n",
"[tabular-ft] 2025-08-14 00:04:09.680346+00:00 Training 52.4% complete - step: 1040, epoch: 0.5226130653266332, loss: 0.5906\n",
"[tabular-ft] 2025-08-14 00:04:12.406993+00:00 Training 52.8% complete - step: 1048, epoch: 0.5266331658291458, loss: 0.6095\n",
"[tabular-ft] 2025-08-14 00:04:15.047004+00:00 Training 53.2% complete - step: 1056, epoch: 0.5306532663316583, loss: 0.5846\n",
"[tabular-ft] 2025-08-14 00:04:17.674958+00:00 Training 53.6% complete - step: 1064, epoch: 0.5346733668341709, loss: 0.5646\n",
"[tabular-ft] 2025-08-14 00:04:20.315000+00:00 Training 54.0% complete - step: 1072, epoch: 0.5386934673366834, loss: 0.5494\n",
"[tabular-ft] 2025-08-14 00:04:22.988863+00:00 Training 54.4% complete - step: 1080, epoch: 0.542713567839196, loss: 0.5672\n",
"[tabular-ft] 2025-08-14 00:04:25.707657+00:00 Training 54.8% complete - step: 1088, epoch: 0.5467336683417086, loss: 0.5319\n",
"[tabular-ft] 2025-08-14 00:04:28.351650+00:00 Training 55.2% complete - step: 1096, epoch: 0.5507537688442211, loss: 0.5082\n",
"[tabular-ft] 2025-08-14 00:04:31.044928+00:00 Training 55.6% complete - step: 1104, epoch: 0.5547738693467337, loss: 0.5097\n",
"[tabular-ft] 2025-08-14 00:04:33.700255+00:00 Training 56.0% complete - step: 1112, epoch: 0.5587939698492462, loss: 0.4972\n",
"[tabular-ft] 2025-08-14 00:04:36.415593+00:00 Training 56.5% complete - step: 1120, epoch: 0.5628140703517588, loss: 0.4741\n",
"[tabular-ft] 2025-08-14 00:04:39.102164+00:00 Training 56.9% complete - step: 1128, epoch: 0.5668341708542713, loss: 0.4808\n",
"[tabular-ft] 2025-08-14 00:04:41.748762+00:00 Training 57.3% complete - step: 1136, epoch: 0.5708542713567839, loss: 0.4628\n",
"[tabular-ft] 2025-08-14 00:04:44.443490+00:00 Training 57.7% complete - step: 1144, epoch: 0.5748743718592965, loss: 0.4434\n",
"[tabular-ft] 2025-08-14 00:04:47.074340+00:00 Training 58.1% complete - step: 1152, epoch: 0.578894472361809, loss: 0.4236\n",
"[tabular-ft] 2025-08-14 00:04:49.780398+00:00 Training 58.5% complete - step: 1160, epoch: 0.5829145728643216, loss: 0.4328\n",
"[tabular-ft] 2025-08-14 00:04:52.494602+00:00 Training 58.9% complete - step: 1168, epoch: 0.5869346733668341, loss: 0.4052\n",
"[tabular-ft] 2025-08-14 00:04:55.162327+00:00 Training 59.3% complete - step: 1176, epoch: 0.5909547738693467, loss: 0.3918\n",
"[tabular-ft] 2025-08-14 00:04:57.877481+00:00 Training 59.7% complete - step: 1184, epoch: 0.5949748743718593, loss: 0.3962\n",
"[tabular-ft] 2025-08-14 00:05:00.559312+00:00 Training 60.1% complete - step: 1192, epoch: 0.5989949748743718, loss: 0.3801\n",
"[tabular-ft] 2025-08-14 00:05:03.250555+00:00 Training 60.5% complete - step: 1200, epoch: 0.6030150753768844, loss: 0.3752\n",
"[tabular-ft] 2025-08-14 00:05:05.873034+00:00 Training 60.9% complete - step: 1208, epoch: 0.607035175879397, loss: 0.3702\n",
"[tabular-ft] 2025-08-14 00:05:08.582364+00:00 Training 61.3% complete - step: 1216, epoch: 0.6110552763819096, loss: 0.3542\n",
"[tabular-ft] 2025-08-14 00:05:11.241851+00:00 Training 61.7% complete - step: 1224, epoch: 0.6150753768844222, loss: 0.3444\n",
"[tabular-ft] 2025-08-14 00:05:13.930729+00:00 Training 62.1% complete - step: 1232, epoch: 0.6190954773869347, loss: 0.3323\n",
"[tabular-ft] 2025-08-14 00:05:16.526277+00:00 Training 62.5% complete - step: 1240, epoch: 0.6231155778894473, loss: 0.3229\n",
"[tabular-ft] 2025-08-14 00:05:19.182807+00:00 Training 62.9% complete - step: 1248, epoch: 0.6271356783919598, loss: 0.3278\n",
"[tabular-ft] 2025-08-14 00:05:21.866731+00:00 Training 63.3% complete - step: 1256, epoch: 0.6311557788944724, loss: 0.321\n",
"[tabular-ft] 2025-08-14 00:05:24.500348+00:00 Training 63.7% complete - step: 1264, epoch: 0.6351758793969849, loss: 0.3083\n",
"[tabular-ft] 2025-08-14 00:05:27.163538+00:00 Training 64.1% complete - step: 1272, epoch: 0.6391959798994975, loss: 0.3061\n",
"[tabular-ft] 2025-08-14 00:05:29.827679+00:00 Training 64.5% complete - step: 1280, epoch: 0.6432160804020101, loss: 0.2902\n",
"[tabular-ft] 2025-08-14 00:05:32.505879+00:00 Training 64.9% complete - step: 1288, epoch: 0.6472361809045226, loss: 0.2884\n",
"[tabular-ft] 2025-08-14 00:05:35.142863+00:00 Training 65.3% complete - step: 1296, epoch: 0.6512562814070352, loss: 0.2884\n",
"[tabular-ft] 2025-08-14 00:05:37.783562+00:00 Training 65.7% complete - step: 1304, epoch: 0.6552763819095477, loss: 0.2779\n",
"[tabular-ft] 2025-08-14 00:05:40.448528+00:00 Training 66.1% complete - step: 1312, epoch: 0.6592964824120603, loss: 0.2685\n",
"[tabular-ft] 2025-08-14 00:05:43.108146+00:00 Training 66.5% complete - step: 1320, epoch: 0.6633165829145728, loss: 0.2635\n",
"[tabular-ft] 2025-08-14 00:05:45.771333+00:00 Training 66.9% complete - step: 1328, epoch: 0.6673366834170854, loss: 0.2495\n",
"[tabular-ft] 2025-08-14 00:05:48.404372+00:00 Training 67.3% complete - step: 1336, epoch: 0.671356783919598, loss: 0.2595\n",
"[tabular-ft] 2025-08-14 00:05:51.019163+00:00 Training 67.7% complete - step: 1344, epoch: 0.6753768844221105, loss: 0.2443\n",
"[tabular-ft] 2025-08-14 00:05:53.662589+00:00 Training 68.1% complete - step: 1352, epoch: 0.6793969849246231, loss: 0.2503\n",
"[tabular-ft] 2025-08-14 00:05:56.320767+00:00 Training 68.5% complete - step: 1360, epoch: 0.6834170854271356, loss: 0.2328\n",
"[tabular-ft] 2025-08-14 00:05:59.045336+00:00 Training 69.0% complete - step: 1368, epoch: 0.6874371859296482, loss: 0.2389\n",
"[tabular-ft] 2025-08-14 00:06:01.745431+00:00 Training 69.4% complete - step: 1376, epoch: 0.6914572864321608, loss: 0.2336\n",
"[tabular-ft] 2025-08-14 00:06:04.428580+00:00 Training 69.8% complete - step: 1384, epoch: 0.6954773869346733, loss: 0.2286\n",
"[tabular-ft] 2025-08-14 00:06:07.090899+00:00 Training 70.2% complete - step: 1392, epoch: 0.699497487437186, loss: 0.2274\n",
"[tabular-ft] 2025-08-14 00:06:09.755484+00:00 Training 70.6% complete - step: 1400, epoch: 0.7035175879396985, loss: 0.2114\n",
"[tabular-ft] 2025-08-14 00:06:12.411448+00:00 Training 71.0% complete - step: 1408, epoch: 0.7075376884422111, loss: 0.2143\n",
"[tabular-ft] 2025-08-14 00:06:15.069768+00:00 Training 71.4% complete - step: 1416, epoch: 0.7115577889447237, loss: 0.2096\n",
"[tabular-ft] 2025-08-14 00:06:17.705640+00:00 Training 71.8% complete - step: 1424, epoch: 0.7155778894472362, loss: 0.2082\n",
"[tabular-ft] 2025-08-14 00:06:20.390014+00:00 Training 72.2% complete - step: 1432, epoch: 0.7195979899497488, loss: 0.2018\n",
"[tabular-ft] 2025-08-14 00:06:23.045184+00:00 Training 72.6% complete - step: 1440, epoch: 0.7236180904522613, loss: 0.1987\n",
"[tabular-ft] 2025-08-14 00:06:25.717732+00:00 Training 73.0% complete - step: 1448, epoch: 0.7276381909547739, loss: 0.1899\n",
"[tabular-ft] 2025-08-14 00:06:28.408834+00:00 Training 73.4% complete - step: 1456, epoch: 0.7316582914572864, loss: 0.1865\n",
"[tabular-ft] 2025-08-14 00:06:31.046710+00:00 Training 73.8% complete - step: 1464, epoch: 0.735678391959799, loss: 0.1804\n",
"[tabular-ft] 2025-08-14 00:06:33.728106+00:00 Training 74.2% complete - step: 1472, epoch: 0.7396984924623116, loss: 0.1822\n",
"[tabular-ft] 2025-08-14 00:06:36.364570+00:00 Training 74.6% complete - step: 1480, epoch: 0.7437185929648241, loss: 0.1794\n",
"[tabular-ft] 2025-08-14 00:06:39.046295+00:00 Training 75.0% complete - step: 1488, epoch: 0.7477386934673367, loss: 0.1692\n",
"[tabular-ft] 2025-08-14 00:06:41.723194+00:00 Training 75.4% complete - step: 1496, epoch: 0.7517587939698492, loss: 0.1696\n",
"[tabular-ft] 2025-08-14 00:06:44.398150+00:00 Training 75.8% complete - step: 1504, epoch: 0.7557788944723618, loss: 0.1645\n",
"[tabular-ft] 2025-08-14 00:06:47.106497+00:00 Training 76.2% complete - step: 1512, epoch: 0.7597989949748744, loss: 0.1666\n",
"[tabular-ft] 2025-08-14 00:06:49.811871+00:00 Training 76.6% complete - step: 1520, epoch: 0.7638190954773869, loss: 0.1615\n",
"[tabular-ft] 2025-08-14 00:06:52.502170+00:00 Training 77.0% complete - step: 1528, epoch: 0.7678391959798995, loss: 0.1643\n",
"[tabular-ft] 2025-08-14 00:06:55.164214+00:00 Training 77.4% complete - step: 1536, epoch: 0.771859296482412, loss: 0.1632\n",
"[tabular-ft] 2025-08-14 00:06:57.842394+00:00 Training 77.8% complete - step: 1544, epoch: 0.7758793969849246, loss: 0.155\n",
"[tabular-ft] 2025-08-14 00:07:00.524177+00:00 Training 78.2% complete - step: 1552, epoch: 0.7798994974874371, loss: 0.1514\n",
"[tabular-ft] 2025-08-14 00:07:03.206482+00:00 Training 78.6% complete - step: 1560, epoch: 0.7839195979899497, loss: 0.152\n",
"[tabular-ft] 2025-08-14 00:07:05.913836+00:00 Training 79.0% complete - step: 1568, epoch: 0.7879396984924623, loss: 0.1437\n",
"[tabular-ft] 2025-08-14 00:07:08.568771+00:00 Training 79.4% complete - step: 1576, epoch: 0.7919597989949749, loss: 0.1432\n",
"[tabular-ft] 2025-08-14 00:07:11.208158+00:00 Training 79.8% complete - step: 1584, epoch: 0.7959798994974875, loss: 0.139\n",
"[tabular-ft] 2025-08-14 00:07:13.842799+00:00 Training 80.2% complete - step: 1592, epoch: 0.8, loss: 0.1388\n",
"[tabular-ft] 2025-08-14 00:07:16.502885+00:00 Training 80.6% complete - step: 1600, epoch: 0.8040201005025126, loss: 0.1349\n",
"[tabular-ft] 2025-08-14 00:07:19.099059+00:00 Training 81.0% complete - step: 1608, epoch: 0.8080402010050252, loss: 0.1366\n",
"[tabular-ft] 2025-08-14 00:07:21.699008+00:00 Training 81.5% complete - step: 1616, epoch: 0.8120603015075377, loss: 0.1309\n",
"[tabular-ft] 2025-08-14 00:07:24.355240+00:00 Training 81.9% complete - step: 1624, epoch: 0.8160804020100503, loss: 0.1281\n",
"[tabular-ft] 2025-08-14 00:07:26.990732+00:00 Training 82.3% complete - step: 1632, epoch: 0.8201005025125628, loss: 0.1257\n",
"[tabular-ft] 2025-08-14 00:07:29.621230+00:00 Training 82.7% complete - step: 1640, epoch: 0.8241206030150754, loss: 0.1231\n",
"[tabular-ft] 2025-08-14 00:07:32.263316+00:00 Training 83.1% complete - step: 1648, epoch: 0.828140703517588, loss: 0.1204\n",
"[tabular-ft] 2025-08-14 00:07:34.895720+00:00 Training 83.5% complete - step: 1656, epoch: 0.8321608040201005, loss: 0.1194\n",
"[tabular-ft] 2025-08-14 00:07:37.578019+00:00 Training 83.9% complete - step: 1664, epoch: 0.8361809045226131, loss: 0.1212\n",
"[tabular-ft] 2025-08-14 00:07:40.255171+00:00 Training 84.3% complete - step: 1672, epoch: 0.8402010050251256, loss: 0.1174\n",
"[tabular-ft] 2025-08-14 00:07:42.934041+00:00 Training 84.7% complete - step: 1680, epoch: 0.8442211055276382, loss: 0.1217\n",
"[tabular-ft] 2025-08-14 00:07:45.616359+00:00 Training 85.1% complete - step: 1688, epoch: 0.8482412060301507, loss: 0.1169\n",
"[tabular-ft] 2025-08-14 00:07:48.283406+00:00 Training 85.5% complete - step: 1696, epoch: 0.8522613065326633, loss: 0.1162\n",
"[tabular-ft] 2025-08-14 00:07:50.958988+00:00 Training 85.9% complete - step: 1704, epoch: 0.8562814070351759, loss: 0.1156\n",
"[tabular-ft] 2025-08-14 00:07:53.612795+00:00 Training 86.3% complete - step: 1712, epoch: 0.8603015075376884, loss: 0.1174\n",
"[tabular-ft] 2025-08-14 00:07:56.256406+00:00 Training 86.7% complete - step: 1720, epoch: 0.864321608040201, loss: 0.1137\n",
"[tabular-ft] 2025-08-14 00:07:58.917484+00:00 Training 87.1% complete - step: 1728, epoch: 0.8683417085427135, loss: 0.109\n",
"[tabular-ft] 2025-08-14 00:08:01.508900+00:00 Training 87.5% complete - step: 1736, epoch: 0.8723618090452261, loss: 0.1124\n",
"[tabular-ft] 2025-08-14 00:08:04.201114+00:00 Training 87.9% complete - step: 1744, epoch: 0.8763819095477386, loss: 0.1067\n",
"[tabular-ft] 2025-08-14 00:08:06.837567+00:00 Training 88.3% complete - step: 1752, epoch: 0.8804020100502512, loss: 0.1054\n",
"[tabular-ft] 2025-08-14 00:08:09.505592+00:00 Training 88.7% complete - step: 1760, epoch: 0.8844221105527639, loss: 0.1085\n",
"[tabular-ft] 2025-08-14 00:08:12.124297+00:00 Training 89.1% complete - step: 1768, epoch: 0.8884422110552764, loss: 0.1071\n",
"[tabular-ft] 2025-08-14 00:08:14.758059+00:00 Training 89.5% complete - step: 1776, epoch: 0.892462311557789, loss: 0.1074\n",
"[tabular-ft] 2025-08-14 00:08:17.385061+00:00 Training 89.9% complete - step: 1784, epoch: 0.8964824120603015, loss: 0.106\n",
"[tabular-ft] 2025-08-14 00:08:20.082565+00:00 Training 90.3% complete - step: 1792, epoch: 0.9005025125628141, loss: 0.1067\n",
"[tabular-ft] 2025-08-14 00:08:22.740840+00:00 Training 90.7% complete - step: 1800, epoch: 0.9045226130653267, loss: 0.1024\n",
"[tabular-ft] 2025-08-14 00:08:25.396468+00:00 Training 91.1% complete - step: 1808, epoch: 0.9085427135678392, loss: 0.1012\n",
"[tabular-ft] 2025-08-14 00:08:28.055845+00:00 Training 91.5% complete - step: 1816, epoch: 0.9125628140703518, loss: 0.1001\n",
"[tabular-ft] 2025-08-14 00:08:30.688997+00:00 Training 91.9% complete - step: 1824, epoch: 0.9165829145728643, loss: 0.1009\n",
"[tabular-ft] 2025-08-14 00:08:33.345058+00:00 Training 92.3% complete - step: 1832, epoch: 0.9206030150753769, loss: 0.1006\n",
"[tabular-ft] 2025-08-14 00:08:36.019629+00:00 Training 92.7% complete - step: 1840, epoch: 0.9246231155778895, loss: 0.099\n",
"[tabular-ft] 2025-08-14 00:08:38.697779+00:00 Training 93.1% complete - step: 1848, epoch: 0.928643216080402, loss: 0.1001\n",
"[tabular-ft] 2025-08-14 00:08:41.348676+00:00 Training 93.5% complete - step: 1856, epoch: 0.9326633165829146, loss: 0.1009\n",
"[tabular-ft] 2025-08-14 00:08:44.009193+00:00 Training 94.0% complete - step: 1864, epoch: 0.9366834170854271, loss: 0.0995\n",
"[tabular-ft] 2025-08-14 00:08:46.699939+00:00 Training 94.4% complete - step: 1872, epoch: 0.9407035175879397, loss: 0.1011\n",
"[tabular-ft] 2025-08-14 00:08:49.400558+00:00 Training 94.8% complete - step: 1880, epoch: 0.9447236180904522, loss: 0.0984\n",
"[tabular-ft] 2025-08-14 00:08:52.033506+00:00 Training 95.2% complete - step: 1888, epoch: 0.9487437185929648, loss: 0.0983\n",
"[tabular-ft] 2025-08-14 00:08:54.694544+00:00 Training 95.6% complete - step: 1896, epoch: 0.9527638190954774, loss: 0.1004\n",
"[tabular-ft] 2025-08-14 00:08:57.332808+00:00 Training 96.0% complete - step: 1904, epoch: 0.9567839195979899, loss: 0.0946\n",
"[tabular-ft] 2025-08-14 00:08:59.964268+00:00 Training 96.4% complete - step: 1912, epoch: 0.9608040201005025, loss: 0.094\n",
"[tabular-ft] 2025-08-14 00:09:02.661328+00:00 Training 96.8% complete - step: 1920, epoch: 0.964824120603015, loss: 0.0983\n",
"[tabular-ft] 2025-08-14 00:09:05.318804+00:00 Training 97.2% complete - step: 1928, epoch: 0.9688442211055276, loss: 0.0984\n",
"[tabular-ft] 2025-08-14 00:09:07.982879+00:00 Training 97.6% complete - step: 1936, epoch: 0.9728643216080402, loss: 0.1003\n",
"[tabular-ft] 2025-08-14 00:09:10.600919+00:00 Training 98.0% complete - step: 1944, epoch: 0.9768844221105528, loss: 0.0987\n",
"[tabular-ft] 2025-08-14 00:09:13.213365+00:00 Training 98.4% complete - step: 1952, epoch: 0.9809045226130654, loss: 0.0973\n",
"[tabular-ft] 2025-08-14 00:09:15.849383+00:00 Training 98.8% complete - step: 1960, epoch: 0.9849246231155779, loss: 0.0961\n",
"[tabular-ft] 2025-08-14 00:09:18.578649+00:00 Training 99.2% complete - step: 1968, epoch: 0.9889447236180905, loss: 0.1\n",
"[tabular-ft] 2025-08-14 00:09:21.257909+00:00 Training 99.6% complete - step: 1976, epoch: 0.992964824120603, loss: 0.0982\n",
"[tabular-ft] 2025-08-14 00:09:23.906181+00:00 Training 100.0% complete - step: 1984, epoch: 0.9969849246231156, loss: 0.0938\n",
"[tabular-ft] 2025-08-14 00:09:24.105317+00:00 << 🧭 Tabular FT >> Saving LoRA adapter\n",
"[tabular-ft] 2025-08-14 00:09:24.732421+00:00 🙌 Training Completed\n",
"[tabular-ft] 2025-08-14 00:09:25.206280+00:00 Task 'train_tabular_ft' executed successfully\n",
"[tabular-ft] 2025-08-14 00:09:25.207186+00:00 << 🧭 Tabular FT >> 👀 Heads up -> Generation stopping is enabled:\n",
"patience: 3\n",
"invalid_fraction_threshold: 0.8\n",
"[tabular-ft] 2025-08-14 00:11:07.599932+00:00 Batch Generation Summary 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"🦜 Number of prompts submitted: 100\n",
"👍 Number of valid records generated: 200\n",
"👎 Number of invalid records generated: 524\n",
"📊 Percentage of records that are valid: 27.62%\n",
"🔬 Top invalid record errors:\n",
"╒══════════════════════════════════════════════════════════╤══════════════╕\n",
"│ Error Category │ Percentage │\n",
"╞══════════════════════════════════════════════════════════╪══════════════╡\n",
"│ Invalid JSON │ 77.1% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Missing required field │ 21.8% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be less than or equal to a given number │ 0.8% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be at most a given length │ 0.4% │\n",
"╘══════════════════════════════════════════════════════════╧══════════════╛\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:11:07.600142+00:00 Batch timing and progress 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"⏱️ Generation time: 35.7 seconds\n",
"🐢 Generation speed: 5.6 records per second.\n",
"⏳ Progress: 200 out of 1000 records generated.\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:11:43.252790+00:00 Batch Generation Summary 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"🦜 Number of prompts submitted: 100\n",
"👍 Number of valid records generated: 140\n",
"👎 Number of invalid records generated: 559\n",
"📊 Percentage of records that are valid: 20.03%\n",
"🔬 Top invalid record errors:\n",
"╒════════════════════════════════════════════╤══════════════╕\n",
"│ Error Category │ Percentage │\n",
"╞════════════════════════════════════════════╪══════════════╡\n",
"│ Invalid JSON │ 74.8% │\n",
"├────────────────────────────────────────────┼──────────────┤\n",
"│ Missing required field │ 25.0% │\n",
"├────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be at most a given length │ 0.2% │\n",
"╘════════════════════════════════════════════╧══════════════╛\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:11:43.253030+00:00 Batch timing and progress 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"⏱️ Generation time: 35.6 seconds\n",
"🐢 Generation speed: 3.9 records per second.\n",
"⏳ Progress: 340 out of 1000 records generated.\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:12:16.562078+00:00 Batch Generation Summary 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"🦜 Number of prompts submitted: 100\n",
"👍 Number of valid records generated: 123\n",
"👎 Number of invalid records generated: 486\n",
"📊 Percentage of records that are valid: 20.20%\n",
"🔬 Top invalid record errors:\n",
"╒══════════════════════════════════════════════════════════╤══════════════╕\n",
"│ Error Category │ Percentage │\n",
"╞══════════════════════════════════════════════════════════╪══════════════╡\n",
"│ Invalid JSON │ 77.0% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Missing required field │ 22.6% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be less than or equal to a given number │ 0.4% │\n",
"╘══════════════════════════════════════════════════════════╧══════════════╛\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:12:16.562365+00:00 Batch timing and progress 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"⏱️ Generation time: 33.3 seconds\n",
"🐢 Generation speed: 3.7 records per second.\n",
"⏳ Progress: 463 out of 1000 records generated.\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:12:50.395771+00:00 Batch Generation Summary 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"🦜 Number of prompts submitted: 100\n",
"👍 Number of valid records generated: 116\n",
"👎 Number of invalid records generated: 546\n",
"📊 Percentage of records that are valid: 17.52%\n",
"🔬 Top invalid record errors:\n",
"╒══════════════════════════════════════════════════════════╤══════════════╕\n",
"│ Error Category │ Percentage │\n",
"╞══════════════════════════════════════════════════════════╪══════════════╡\n",
"│ Invalid JSON │ 69.8% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Missing required field │ 29.9% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be less than or equal to a given number │ 0.4% │\n",
"╘══════════════════════════════════════════════════════════╧══════════════╛\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:12:50.396118+00:00 Batch timing and progress 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"⏱️ Generation time: 33.8 seconds\n",
"🐢 Generation speed: 3.4 records per second.\n",
"⏳ Progress: 579 out of 1000 records generated.\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:13:26.667953+00:00 Batch Generation Summary 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"🦜 Number of prompts submitted: 100\n",
"👍 Number of valid records generated: 169\n",
"👎 Number of invalid records generated: 601\n",
"📊 Percentage of records that are valid: 21.95%\n",
"🔬 Top invalid record errors:\n",
"╒══════════════════════════════════════════════════════════╤══════════════╕\n",
"│ Error Category │ Percentage │\n",
"╞══════════════════════════════════════════════════════════╪══════════════╡\n",
"│ Invalid JSON │ 75.9% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Missing required field │ 23.6% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be at most a given length │ 0.3% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be less than or equal to a given number │ 0.2% │\n",
"╘══════════════════════════════════════════════════════════╧══════════════╛\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:13:26.668350+00:00 Batch timing and progress 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"⏱️ Generation time: 36.3 seconds\n",
"🐢 Generation speed: 4.7 records per second.\n",
"⏳ Progress: 748 out of 1000 records generated.\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:14:02.293920+00:00 Batch Generation Summary 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"🦜 Number of prompts submitted: 100\n",
"👍 Number of valid records generated: 153\n",
"👎 Number of invalid records generated: 611\n",
"📊 Percentage of records that are valid: 20.03%\n",
"🔬 Top invalid record errors:\n",
"╒═════════════════════════════════════════════════════════════╤══════════════╕\n",
"│ Error Category │ Percentage │\n",
"╞═════════════════════════════════════════════════════════════╪══════════════╡\n",
"│ Invalid JSON │ 74.1% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Missing required field │ 25.0% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be greater than or equal to a given number │ 0.5% │\n",
"├─────────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be less than or equal to a given number │ 0.3% │\n",
"╘═════════════════════════════════════════════════════════════╧══════════════╛\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:14:02.294334+00:00 Batch timing and progress 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"⏱️ Generation time: 35.6 seconds\n",
"🐢 Generation speed: 4.3 records per second.\n",
"⏳ Progress: 901 out of 1000 records generated.\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:14:30.566530+00:00 Batch Generation Summary 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"🦜 Number of prompts submitted: 76\n",
"👍 Number of valid records generated: 113\n",
"👎 Number of invalid records generated: 351\n",
"📊 Percentage of records that are valid: 24.35%\n",
"🔬 Top invalid record errors:\n",
"╒══════════════════════════════════════════════════════════╤══════════════╕\n",
"│ Error Category │ Percentage │\n",
"╞══════════════════════════════════════════════════════════╪══════════════╡\n",
"│ Invalid JSON │ 80.1% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Missing required field │ 19.4% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be at most a given length │ 0.3% │\n",
"├──────────────────────────────────────────────────────────┼──────────────┤\n",
"│ Field value must be less than or equal to a given number │ 0.3% │\n",
"╘══════════════════════════════════════════════════════════╧══════════════╛\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:14:30.567033+00:00 Batch timing and progress 👇\n",
"----------------------------------------------------------------------------------------------------\n",
"⏱️ Generation time: 28.3 seconds\n",
"🐢 Generation speed: 4.0 records per second.\n",
"⌛️ Progress: 1014 out of 1000 records generated.\n",
"----------------------------------------------------------------------------------------------------\n",
"[tabular-ft] 2025-08-14 00:14:30.567239+00:00 🎉 Generation complete 🎉\n",
"[tabular-ft] 2025-08-14 00:14:31.964482+00:00 🙌 Successfully generated 1000 records\n",
"[tabular-ft] 2025-08-14 00:14:32.098007+00:00 Task 'generate_from_tabular_ft' executed successfully\n",
"[tabular-ft] 2025-08-14 00:14:32.294104+00:00 Task 'tabular_ft' executed successfully\n",
"[tabular-ft] 2025-08-14 00:14:32.294515+00:00 Task execution completed. Saving task outputs.\n",
"[tabular-ft] 2025-08-14 00:14:33.012482+00:00 Task outputs saved.\n",
"[tabular-ft] Task Status is now: RUN_STATUS_COMPLETED\n",
"Got task wt_31FoaOsYjXxCuQwq01sy5O7tZrT\n",
"[evaluate-safe-synthetics-dataset] Task Status is now: RUN_STATUS_ACTIVE\n",
"[evaluate-safe-synthetics-dataset] 2025-08-14 00:15:20.045708+00:00 Preparing step 'evaluate-safe-synthetics-dataset'\n",
"[evaluate-safe-synthetics-dataset] 2025-08-14 00:15:34.705589+00:00 Starting 'evaluate_safe_synthetics_dataset' task execution\n",
"[evaluate-safe-synthetics-dataset] 2025-08-14 00:15:52.799623+00:00 LLM column classification took 2.4958897550000074 seconds.\n",
"[evaluate-safe-synthetics-dataset] 2025-08-14 00:16:26.829393+00:00 Task 'evaluate_safe_synthetics_dataset' executed successfully\n",
"[evaluate-safe-synthetics-dataset] 2025-08-14 00:16:26.831304+00:00 Task execution completed. Saving task outputs.\n",
"[evaluate-safe-synthetics-dataset] 2025-08-14 00:16:27.455332+00:00 Task outputs saved.\n",
"[evaluate-safe-synthetics-dataset] Task Status is now: RUN_STATUS_COMPLETED\n",
"Workflow run is now in status: RUN_STATUS_COMPLETED\n",
"✅ Training complete!\n",
"🎯 Quality score: 8.3\n",
"🔒 Privacy score: 5.9\n",
"❌ Trial failed: name 'score' is not defined\n"
]
},
{
"data": {
"text/html": [],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"<br> <style><br> .wandb-row {<br> display: flex;<br> flex-direction: row;<br> flex-wrap: wrap;<br> justify-content: flex-start;<br> width: 100%;<br> }<br> .wandb-col {<br> display: flex;<br> flex-direction: column;<br> flex-basis: 100%;<br> flex: 1;<br> padding: 10px;<br> }<br> </style><br><div class=\"wandb-row\"><div class=\"wandb-col\"><h3>Run history:</h3><br/><table class=\"wandb\"><tr><td>privacy_score</td><td>▁</td></tr><tr><td>quality_score</td><td>▁</td></tr></table><br/></div><div class=\"wandb-col\"><h3>Run summary:</h3><br/><table class=\"wandb\"><tr><td>error</td><td>name 'score' is not ...</td></tr><tr><td>privacy_score</td><td>5.9</td></tr><tr><td>quality_score</td><td>8.3</td></tr></table><br/></div></div>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" View run <strong style=\"color:#cdcd00\">safe-syn-pd0lrdry</strong> at: <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps/runs/9hww7mum' target=\"_blank\">https://wandb.ai/yamini_gretel/safe-syn-sweeps/runs/9hww7mum</a><br> View project at: <a href='https://wandb.ai/yamini_gretel/safe-syn-sweeps' target=\"_blank\">https://wandb.ai/yamini_gretel/safe-syn-sweeps</a><br>Synced 5 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s)"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"Find logs at: <code>./wandb/run-20250813_195359-9hww7mum/logs</code>"
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Traceback (most recent call last):\n",
" File \"/Users/ykagal/.venvs/jupyter/lib/python3.13/site-packages/wandb/agents/pyagent.py\", line 297, in _run_job\n",
" self._function()\n",
" ~~~~~~~~~~~~~~^^\n",
" File \"/var/folders/kx/_q607d_9603cg0389qczz_f00000gp/T/ipykernel_66388/2497270407.py\", line 29, in trial_agent\n",
" print(f\"✅ Trial complete → score={score}\")\n",
" ^^^^^\n",
"NameError: name 'score' is not defined. Did you mean: 'scores'?\n",
"\n",
"\u001b[34m\u001b[1mwandb\u001b[0m: \u001b[32m\u001b[41mERROR\u001b[0m Run 9hww7mum errored: name 'score' is not defined\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"================================================================================\n",
"🏁 Sweep completed!\n",
"⏱️ Total execution time: 52.0 minutes\n",
"📈 Average time per trial: 26.0 minutes\n",
"🔗 Final results: https://wandb.ai/yamini_gretel/safe-syn-sweeps/sweeps/e8o48dv0\n"
]
}
],
"source": [
"# Launch the Weights & Biases sweep\n",
"print(\"🚀 Launching W&B Sweep...\")\n",
"print(f\"📋 Configuration: {MAX_TRIALS} trials using {SWEEP_CONFIG['method']} optimization\")\n",
"print(f\"🎯 Optimizing: {SWEEP_CONFIG['metric']['name']} ({SWEEP_CONFIG['metric']['goal']})\")\n",
"print(f\"📊 Dataset: {len(df)} rows, {len(df.columns)} columns\")\n",
"\n",
"try:\n",
" # Create the sweep\n",
" print(\"\\n📝 Creating sweep...\")\n",
" sweep_id = wandb.sweep(SWEEP_CONFIG, project=WANDB_PROJECT, entity=WANDB_ENTITY)\n",
" print(f\"✅ Sweep created with ID: {sweep_id}\")\n",
" \n",
" # Show sweep URL\n",
" sweep_url = f\"https://wandb.ai/{WANDB_ENTITY or 'your-entity'}/{WANDB_PROJECT}/sweeps/{sweep_id}\"\n",
" print(f\"🔗 View sweep at: {sweep_url}\")\n",
" \n",
" # Launch the agent\n",
" print(f\"\\n🤖 Starting sweep agent (max {MAX_TRIALS} trials)...\")\n",
" print(\"=\" * 80)\n",
" \n",
" start_time = time.time()\n",
" wandb.agent(sweep_id, function=trial_agent, count=MAX_TRIALS)\n",
" end_time = time.time()\n",
" \n",
" execution_time = end_time - start_time\n",
" print(\"=\" * 80)\n",
" print(\"🏁 Sweep completed!\")\n",
" print(f\"⏱️ Total execution time: {execution_time/60:.1f} minutes\")\n",
" print(f\"📈 Average time per trial: {execution_time/MAX_TRIALS/60:.1f} minutes\")\n",
" print(f\"🔗 Final results: {sweep_url}\")\n",
" \n",
"except Exception as e:\n",
" print(f\"❌ Sweep failed: {e}\")\n",
" print(\"Check the error details above and ensure all previous cells ran successfully.\")\n",
" raise"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Tests for Debugging"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Check what variables are in memory from the recent trial runs\n",
"print(\"🔍 Checking available variables...\")\n",
"\n",
"# List all variables in the global namespace\n",
"variables = [var for var in globals().keys() if not var.startswith('_')]\n",
"print(f\"📋 Available variables: {variables}\")\n",
"\n",
"# Look for any job-related variables\n",
"job_vars = [var for var in variables if 'job' in var.lower()]\n",
"print(f\"🎯 Job-related variables: {job_vars}\")\n",
"\n",
"# Check if there are any gretel-related objects\n",
"gretel_vars = [var for var in variables if any(keyword in str(type(globals()[var])).lower() \n",
" for keyword in ['gretel', 'synthetic', 'workflow'])]\n",
"print(f\"🔧 Gretel-related variables: {gretel_vars}\")\n",
"\n",
"# If you see any promising variables, inspect them:\n",
"for var in gretel_vars[:3]: # Check first 3\n",
" try:\n",
" obj = globals()[var]\n",
" print(f\"\\n📊 {var}: {type(obj)}\")\n",
" if hasattr(obj, 'report'):\n",
" print(f\" ✅ Has report attribute\")\n",
" if hasattr(obj, 'id'):\n",
" print(f\" 🆔 ID: {obj.id}\")\n",
" except:\n",
" pass"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Test to extract quality and privacy scores\n",
"print(f\"🔍 Testing score extraction for: {job_id}\")\n",
"\n",
"try:\n",
" workflow_run = gretel.workflows.get_workflow_run(job_id)\n",
" report_dict = workflow_run.report.dict\n",
" \n",
" # Extract quality score\n",
" quality_score = report_dict['synthetic_quality_score']\n",
" print(f\"🎯 Quality Score: {quality_score}\")\n",
" \n",
" # Extract privacy score \n",
" privacy_score = report_dict['data_privacy_score']\n",
" print(f\"🔒 Privacy Score: {privacy_score}\")\n",
" \n",
" # Show other available scores for reference\n",
" print(f\"\\n📊 Other available scores:\")\n",
" score_keys = [k for k in report_dict.keys() if 'score' in k.lower()]\n",
" for key in score_keys:\n",
" print(f\" - {key}: {report_dict[key]}\")\n",
" \n",
" print(f\"\\n✅ SUCCESS! Both scores extracted successfully.\")\n",
" print(f\" Quality: {quality_score}, Privacy: {privacy_score}\")\n",
" \n",
"except Exception as e:\n",
" print(f\"❌ Error: {e}\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 10) Results Summary\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Optional: Analyze sweep results programmatically\n",
"print(\"📊 Analyzing sweep results...\")\n",
"\n",
"try:\n",
" # Get sweep results from W&B API\n",
" sweep = api.sweep(f\"{WANDB_ENTITY}/{WANDB_PROJECT}/{sweep_id}\")\n",
" runs = list(sweep.runs)\n",
" \n",
" print(f\"\\n✅ Sweep completed with {len(runs)} runs\")\n",
" \n",
" # Find best run\n",
" completed_runs = [run for run in runs if run.state == \"finished\"]\n",
" \n",
" if completed_runs:\n",
" best_run = max(completed_runs, key=lambda r: r.summary.get(\"quality_score\", 0))\n",
" \n",
" print(f\"\\n🏆 Best run: {best_run.name}\")\n",
" print(f\" - Quality Score: {best_run.summary.get('quality_score', 'N/A')}\")\n",
" print(f\" - Learning Rate: {best_run.config.get('learning_rate', 'N/A')}\")\n",
" print(f\" - Batch Size: {best_run.config.get('batch_size', 'N/A')}\")\n",
" print(f\" - Temperature: {best_run.config.get('temperature', 'N/A')}\")\n",
" print(f\" - LoRA R: {best_run.config.get('lora_r', 'N/A')}\")\n",
" print(f\" - Worker ID: {best_run.summary.get('worker_id', 'N/A')}\")\n",
" print(f\" - Run URL: {best_run.url}\")\n",
" \n",
" # Summary statistics\n",
" quality_scores = [run.summary.get(\"quality_score\", 0) for run in completed_runs]\n",
" worker_ids = [run.summary.get(\"worker_id\", 0) for run in completed_runs]\n",
" \n",
" print(f\"\\n📈 Quality Score Statistics:\")\n",
" print(f\" - Best: {max(quality_scores):.4f}\")\n",
" print(f\" - Worst: {min(quality_scores):.4f}\")\n",
" print(f\" - Average: {sum(quality_scores)/len(quality_scores):.4f}\")\n",
" print(f\" - Standard Deviation: {(sum((x - sum(quality_scores)/len(quality_scores))**2 for x in quality_scores) / len(quality_scores))**0.5:.4f}\")\n",
" \n",
" failed_runs = [run for run in runs if run.state == \"failed\"]\n",
" if failed_runs:\n",
" print(f\"\\n⚠️ {len(failed_runs)} run(s) failed\")\n",
" for run in failed_runs:\n",
" print(f\" - {run.name}: {run.summary.get('error_message', 'Unknown error')}\")\n",
"\n",
"except Exception as e:\n",
" print(f\"⚠️ Could not analyze results: {e}\")\n",
" print(f\"You can view results manually at: https://wandb.ai/{WANDB_ENTITY}/{WANDB_PROJECT}\")\n",
"\n",
"print(\"\\n🎉 Safe Synthetics sweep completed!\")\n",
"print(\"\\nKey improvements in this version:\")\n",
"print(\"✅ Removed slow holdout/transform steps\")\n",
"print(\"✅ Direct Safe Synthetics workflow\")\n",
"print(\"✅ Tabular Fine-Tuning parameter optimization\")\n",
"print(\"✅ Enhanced W&B logging and error handling\")\n",
"print(\"✅ Comprehensive quality score extraction\")\n",
"print(\"\\n📝 Configuration options:\")\n",
"print(\" - Set MAX_TRIALS=N to change number of trials\")\n"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "jupyter",
"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": "3.13.3"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment