nvt_triton_client

# External dependencies
import os
from time import time

import cudf
import tritonclient.grpc as grpcclient
import nvtabular.inference.triton as nvt_triton

#!curl -i triton:8000/v2/health/ready

import tritonhttpclient
try:
    triton_client = tritonhttpclient.InferenceServerClient(url="localhost:8000", verbose=True)
    print("client created.")
except Exception as e:
    print("channel creation failed: " + str(e))
    
triton_client.is_server_live()

NUM_ROWS = 10000
inputs = {
    'user_session': np.random.randint(1, 10000, NUM_ROWS),
    'product_id': np.random.randint(1, 51996, NUM_ROWS),
    'category_id': np.random.randint(0, 332, NUM_ROWS),
    'event_time_ts': np.random.randint(1570373000, 1670373390, NUM_ROWS),
    'prod_first_event_time_ts' : np.random.randint(1570373000, 1570373382, NUM_ROWS),
    'price' : np.random.uniform(0, 2750, NUM_ROWS)
    }
df = cudf.DataFrame(inputs)

batch = df[['category_id',
 'event_time_ts',
 'user_session',
 'prod_first_event_time_ts',
 'price',
 'product_id']].iloc[:3, :]

print(batch)

inputs = nvt_triton.convert_df_to_triton_input(batch.columns, batch, grpcclient.InferInput)

output_names_org = ['user_session',
 'product_id-count',
 'product_id-list_seq',
 'et_dayofweek_sin-list_seq',
 'price_log_norm-list_seq',
 'relative_price_to_avg_categ_id-list_seq',
 'category_id-list_seq',
 'product_recency_days_log_norm-list_seq',
 'et_dayofweek_cos-list_seq',
 'day_index']

output_names = []

for name in output_names_org:
    if '-list' in name:
        output_names.append(name + '__values')
        output_names.append(name + '__nnzs')
    else:
        output_names.append(name)
    
outputs = []

MODEL_NAME_NVT = "model_nvt"

for col in output_names:
    outputs.append(grpcclient.InferRequestedOutput(col))

with grpcclient.InferenceServerClient("localhost:8001") as client:
    response = client.infer(MODEL_NAME_NVT, inputs, request_id="1", outputs=outputs)

for col in output_names:
    print(col, response.as_numpy(col), response.as_numpy(col).shape)