Skip to content

Commit db4f890

Browse files
committed
add model pydoc
1 parent 650b7bc commit db4f890

File tree

2 files changed

+19
-0
lines changed
  • qa/python_models

2 files changed

+19
-0
lines changed

qa/python_models/response_parameters_bls/model.py

Lines changed: 11 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -31,6 +31,17 @@
3131

3232

3333
class TritonPythonModel:
34+
"""
35+
This model (A) is designed to test sending back response parameters when using BLS.
36+
It takes one input tensor, which is the RESPONSE_PARAMETERS and uses BLS to
37+
call response_parameters model (B). Model B would set RESPONSE_PARAMETERS (with a bit
38+
of data massage) as its response parameters. In the end, model A would also set its
39+
response parameters from model B's response parameters.
40+
41+
With above model set up, we can easily test whether the real response parameters are
42+
the same as the input response parameters.
43+
"""
44+
3445
def execute(self, requests):
3546
responses = []
3647

qa/python_models/response_parameters_bls_decoupled/model.py

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -31,6 +31,14 @@
3131

3232

3333
class TritonPythonModel:
34+
"""
35+
This model (A) is designed to test sending back response parameters when using BLS
36+
with decoupled model transaction policy.
37+
38+
The only difference vs. response_parameters_bls model is this model turns on decoupled
39+
model transaction policy. For more details, please check response_parameters_bls.
40+
"""
41+
3442
def execute(self, requests):
3543
for request in requests:
3644
res_params_tensor = pb_utils.get_input_tensor_by_name(

0 commit comments

Comments
 (0)