auth=svc_pr) # ### Enable SynapseML predict # Set the spark conf spark.synapse.ml.predict.enabled as true to enable the library. # In[5]: spark.conf.set("spark.synapse.ml.predict.enabled", "true") # ### Bind Model # In[6]: model = pcontext.bind_model(RETURN_TYPES, "mlflow", "ONNX_linear_regression", AML_MODEL_URI_ONNX, aml_workspace=ws).register() # ### Load Data # In[7]: df = spark.read.format("csv").option("header", "true").csv(DATA_FILE, inferSchema=True) df = df.select(df.columns[:9]) df.createOrReplaceTempView('data') df.show(10) df # In[8]:
RETURN_TYPES = "INT" # ### Enable SynapseML predict # Set the spark conf spark.synapse.ml.predict.enabled as true to enable the library. # In[6]: spark.conf.set("spark.synapse.ml.predict.enabled", "true") # ### Bind Model # In[7]: model = pcontext.bind_model(RETURN_TYPES, "mlflow", "sklearn_linear_regression", AML_MODEL_URI_SKLEARN, aml_workspace=ws).register() # ### Load Data # In[8]: df = spark.read.format("csv").option("header", "true").csv(DATA_FILE, inferSchema=True) df = df.select(df.columns[:9]) df.createOrReplaceTempView('data') df.show(10) df # In[9]:
# ### Enable SynapseML predict # Set the spark conf spark.synapse.ml.predict.enabled as true to enable the library. # In[5]: spark.conf.set("spark.synapse.ml.predict.enabled", "true") # ### Bind Model # In[6]: model = pcontext.bind_model(return_types=RETURN_TYPES, runtime="mlflow", model_alias="tensorflow_linear_regression", model_uri=AML_MODEL_URI_TENSORFLOW, meta_data={ 'meta_graph': ['serve'], 'signature_def_key': 'serving_default' }, aml_workspace=ws).register() # ### Load Data # In[7]: df = spark.read.format("csv").option("header", "true").csv(DATA_FILE, inferSchema=True) df = df.select(df.columns[:4]) df.createOrReplaceTempView('data') df.show(10) df
ADLS_MODEL_URI_XGBOOST = "abfss://[email protected]/predict/models/mlflow/xgboost/model_nf10/" RETURN_TYPES = "float" # ### Enable SynapseML predict # Set the spark conf spark.synapse.ml.predict.enabled as true to enable the library. # In[3]: spark.conf.set("spark.synapse.ml.predict.enabled", "true") # ### Bind Model # In[4]: model = pcontext.bind_model(return_types=RETURN_TYPES, runtime="mlflow", model_alias="xgboost_model", model_uri=ADLS_MODEL_URI_XGBOOST).register() # ### Load Data # In[5]: data = np.random.rand(5, 10) df = spark.createDataFrame(pd.DataFrame(data)) df.createOrReplaceTempView("data") df.show() # ### Model Prediction using SPARK_SQL # In[6]:
# ### Enable SynapseML predict # Set the spark conf spark.synapse.ml.predict.enabled as true to enable the library. # In[7]: spark.conf.set("spark.synapse.ml.predict.enabled","true") # ### Bind Model # In[8]: model = pcontext.bind_model(RETURN_TYPES, "mlflow","ONNX_linear_regression", ADLS_MODEL_URI_SKLEARN).register() # ### Load Data # In[9]: df = spark.read .format("csv") .option("header", "true") .csv(DATA_FILE, inferSchema=True) df = df.select(df.columns[:9]) df.createOrReplaceTempView('data') df.show(10) df
loader_module='mlflow.spark') # In[9]: MODEL_URI = './sparkml_pyfunc_model_path' RETURN_TYPES = 'float' # In[10]: model = pcontext.bind_model( return_types = RETURN_TYPES, runtime = 'mlflow', model_alias = 'sparkml_model', model_uri = MODEL_URI,).register() # In[11]: type(model) # In[12]: data = pd.DataFrame([(Vectors.dense(1, 2, 3, 4, 5, 6, 7, 8, 9, 10),)], columns=["features"]) model.predict(data)
ADLS_MODEL_URI_PYTORCH = "abfss://[email protected]/predict/models/mlflow/pytorch/linear_regression/" RETURN_TYPES = "INT" # ### Enable SynapseML predict # Set the spark conf spark.synapse.ml.predict.enabled as true to enable the library. # In[3]: spark.conf.set("spark.synapse.ml.predict.enabled", "true") # ### Bind Model # In[4]: model = pcontext.bind_model(RETURN_TYPES, "mlflow", "pytorch_linear_regression", ADLS_MODEL_URI_PYTORCH).register() # ### Load Data # In[5]: df = spark.read.format("csv").option("header", "true").csv(DATA_FILE, inferSchema=True) df = df.select(df.columns[:9]) df.createOrReplaceTempView('data') df.show(10) df # In[6]:
auth=svc_pr) # ### Enable SynapseML predict # Set the spark conf spark.synapse.ml.predict.enabled as true to enable the library. # In[11]: spark.conf.set("spark.synapse.ml.predict.enabled", "true") # ### Bind Model # In[12]: model = pcontext.bind_model(RETURN_TYPES, "mlflow", "pytorch_linear_regression", AML_MODEL_URI_PYTORCH, aml_workspace=ws).register() # ### Load Data # In[13]: df = spark.read.format("csv").option("header", "true").csv(DATA_FILE, inferSchema=True) df = df.select(df.columns[:9]) df.createOrReplaceTempView('data') df.show(10) df # In[14]: