File size: 5,451 Bytes
2c29579 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 | import os
import json
from datetime import datetime
# Optional imports
try:
import mlflow
except Exception:
mlflow = None
try:
import wandb
except Exception:
wandb = None
class MLTracking:
"""Handles integration with MLflow and Weights & Biases."""
@staticmethod
def init_mlflow(experiment_name="AutoML_Studio"):
# ✅ FIX 1: guard if mlflow not installed
if not mlflow:
return
try:
if os.getenv("MLFLOW_TRACKING_URI"):
mlflow.set_tracking_uri(os.getenv("MLFLOW_TRACKING_URI"))
mlflow.set_experiment(experiment_name)
except Exception as e:
print(f"MLflow init failed: {e}") # ✅ FIX 2
@staticmethod
def log_run(job_id, params, metrics, model=None, artifact_path=None):
"""Logs a single run to both MLflow and W&B if configured."""
# ✅ FIX 3: ensure safe defaults
params = params or {}
metrics = metrics or {}
flat_metrics = {
k: v for k, v in metrics.items()
if isinstance(v, (int, float))
}
# MLflow
if mlflow and os.getenv("MLFLOW_ENABLED", "false").lower() == "true":
try:
with mlflow.start_run(run_name=f"Job_{job_id}"):
if params:
mlflow.log_params(params)
if flat_metrics:
mlflow.log_metrics(flat_metrics)
# ✅ FIX 4: guard sklearn submodule
if model and hasattr(mlflow, "sklearn"):
try:
mlflow.sklearn.log_model(model, "model")
except Exception:
pass # don't break run
if artifact_path and os.path.exists(artifact_path):
mlflow.log_artifact(artifact_path)
except Exception as e:
print(f"MLflow logging failed: {e}")
# W&B
if wandb and os.getenv("WANDB_API_KEY"):
try:
wandb.init(
project="AutoML_Studio",
name=f"Job_{job_id}",
config=params,
reinit=True
)
if metrics:
wandb.log(flat_metrics)
# ✅ FIX 5: safe artifact handling
if artifact_path and os.path.exists(artifact_path):
try:
wandb.save(artifact_path)
except Exception:
pass
wandb.finish()
except Exception as e:
print(f"W&B logging failed: {e}")
# Local experiment tracking DB
try:
from infra.database import get_db, JobModel, ExperimentRun
with get_db() as db:
job = db.query(JobModel).filter(JobModel.id == job_id).first()
if job:
try:
job_params = json.loads(job.params_json) if job.params_json else {}
except Exception:
job_params = {}
score = (
metrics.get("score")
if metrics.get("score") is not None
else flat_metrics.get("test_score")
)
metric_name = (
metrics.get("metric_name")
or params.get("metric_name")
or job_params.get("eval_metric")
or "Score"
)
best_model = (
metrics.get("best_model")
or params.get("best_model")
or ""
)
run = ExperimentRun(
job_id=job_id,
dataset_id=job.dataset_id,
model_name=str(best_model) if best_model else None,
metric_name=str(metric_name),
score=str(score) if score is not None else None,
hyperparams_json=json.dumps(params),
metrics_json=json.dumps(metrics),
leaderboard_json=json.dumps(metrics.get("leaderboard", [])),
feature_count=str(len(metrics.get("feature_names") or [])),
row_count=str((metrics.get("eda_summary") or {}).get("rows_after_target_cleaning", "")),
task_type="classification" if metrics.get("is_classification") else "regression",
mode=str(params.get("mode") or job_params.get("mode") or ""),
goal=str(params.get("goal") or job_params.get("goal") or ""),
)
db.add(run)
except Exception as e:
print(f"Experiment tracking DB write failed: {e}")
class StructuredLogger:
@staticmethod
def log(event, **kwargs):
# ✅ FIX 6: safe timestamp + formatting
try:
timestamp = datetime.utcnow().isoformat()
except Exception:
timestamp = "unknown_time"
try:
data = ", ".join([f"{k}={v}" for k, v in kwargs.items()])
except Exception:
data = ""
print(f"[{timestamp}] EVENT={event} | {data}")
|