masahi commented on a change in pull request #4714: Mxnet parser for Qnn dialect
URL: https://github.com/apache/incubator-tvm/pull/4714#discussion_r374951699
##########
File path: tests/python/frontend/mxnet/test_forward.py
##########
@@ -28,68 +28,173 @@
import model_zoo
+def get_gluon_output(name, x, dtype):
+ net = vision.get_model(name)
+ net.collect_params().initialize(mx.init.Xavier())
+ net_sym = gluon.nn.SymbolBlock(outputs=net(mx.sym.var('data')),
+ inputs=mx.sym.var('data'),
+ params=net.collect_params())
+ out = net_sym(mx.nd.array(x.astype(dtype))).asnumpy()
+ return out, net_sym
+
+
+def get_mxnet_output(symbol, x, dtype='float32', arg_params=None,
aux_params=None):
+ from collections import namedtuple
+ Batch = namedtuple('Batch', ['data'])
+ mod = mx.mod.Module(symbol, label_names=None)
+ mod.bind(data_shapes=[('data', x.shape)], for_training=False)
+ if aux_params is not None and arg_params is not None:
+ mod.set_params(arg_params, aux_params)
+ else:
+ mod.init_params()
+ mod.forward(Batch([mx.nd.array(x.astype(dtype))]))
+ out = mod.get_outputs()[0].asnumpy()
+ args, auxs = mod.get_params()
+ return out, args, auxs
+
+
+def get_tvm_output(symbol, x, args, auxs, target, ctx, out_shape,
dtype='float32', gluon_impl=False):
+ shape_dict = {"data": x.shape}
+ if gluon_impl:
+ mod, params = relay.frontend.from_mxnet(symbol, shape_dict)
+ else:
+ mod, params = relay.frontend.from_mxnet(symbol,
+ shape_dict,
+ arg_params=args,
+ aux_params=auxs)
+ with relay.build_config(opt_level=3):
+ graph, lib, params = relay.build(mod, target, params=params)
+ m = graph_runtime.create(graph, lib, ctx)
+ # set inputs
+ m.set_input("data", tvm.nd.array(x.astype(dtype)))
+ m.set_input(**params)
+ m.run()
+ # get outputs
+ out = m.get_output(0, tvm.nd.empty(out_shape, dtype))
+ return out.asnumpy()
+
+
def verify_mxnet_frontend_impl(mx_symbol,
data_shape=(1, 3, 224, 224),
out_shape=(1, 1000),
gluon_impl=False,
name=None,
dtype='float32'):
"""Use name different from test to avoid pytest picking it up"""
- if gluon_impl:
- def get_gluon_output(name, x):
- net = vision.get_model(name)
- net.collect_params().initialize(mx.init.Xavier())
- net_sym = gluon.nn.SymbolBlock(outputs=net(mx.sym.var('data')),
- inputs=mx.sym.var('data'),
- params=net.collect_params())
- out = net_sym(mx.nd.array(x.astype(dtype))).asnumpy()
- return out, net_sym
- else:
- def get_mxnet_output(symbol, x, dtype='float32'):
- from collections import namedtuple
- Batch = namedtuple('Batch', ['data'])
- mod = mx.mod.Module(symbol, label_names=None)
- mod.bind(data_shapes=[('data', x.shape)], for_training=False)
- mod.init_params()
- mod.forward(Batch([mx.nd.array(x.astype(dtype))]))
- out = mod.get_outputs()[0].asnumpy()
- args, auxs = mod.get_params()
- return out, args, auxs
-
- def get_tvm_output(symbol, x, args, auxs, target, ctx, dtype='float32'):
- shape_dict = {"data": x.shape}
- if gluon_impl:
- mod, params = relay.frontend.from_mxnet(symbol, shape_dict)
- else:
- mod, params = relay.frontend.from_mxnet(symbol,
- shape_dict,
- arg_params=args,
- aux_params=auxs)
- with relay.build_config(opt_level=3):
- graph, lib, params = relay.build(mod, target, params=params)
- m = graph_runtime.create(graph, lib, ctx)
- # set inputs
- m.set_input("data", tvm.nd.array(x.astype(dtype)))
- m.set_input(**params)
- m.run()
- # get outputs
- out = m.get_output(0, tvm.nd.empty(out_shape, dtype))
- return out.asnumpy()
-
# random input
x = np.random.uniform(size=data_shape)
if gluon_impl:
- gluon_out, gluon_sym = get_gluon_output(name, x)
+ gluon_out, gluon_sym = get_gluon_output(name, x, dtype)
for target, ctx in ctx_list():
tvm_out = get_tvm_output(gluon_sym, x, None, None, target, ctx,
dtype)
tvm.testing.assert_allclose(gluon_out, tvm_out, rtol=1e-5,
atol=1e-5)
else:
mx_out, args, auxs = get_mxnet_output(mx_symbol, x, dtype)
assert "data" not in args
for target, ctx in ctx_list():
- tvm_out = get_tvm_output(mx_symbol, x, args, auxs, target, ctx,
dtype)
+ tvm_out = get_tvm_output(mx_symbol, x, args, auxs, target, ctx,
out_shape, dtype)
tvm.testing.assert_allclose(mx_out, tvm_out, rtol=1e-5, atol=1e-5)
+
+def verify_mxnet_quantized_forward_impl(mx_symbol,
+ data_shape=(1, 3, 224, 224),
+ out_shape=(1, 1000),
+ name=None,
+ dtype='float32'):
+
+ def use_old_mxnet_quantization_api():
+ mx_version = float(mx.__version__[:3])
+ return mx_version <= 1.5
+
+ def quantize_with_old_api(sym, _arg_params, _aux_params, quantized_dtype,
_data_shape, logger):
+ """Calibrate and quantize the network with dummy data"""
+ ctx = mx.cpu(0)
+ batch_size = 1
+ num_calib_batches = 5
+ num_calib_examples = num_calib_batches * batch_size
+ calibration_data = [mx.random.uniform(-1.0, 1.0, shape=_data_shape,
ctx=ctx, dtype='float32')]
+ data = mx.io.NDArrayIter(data=calibration_data, label=[],
batch_size=batch_size,
+ shuffle=True, last_batch_handle='discard')
+ qsym = sym.get_backend_symbol('MKLDNN_QUANTIZE')
+ qsym, qarg_params, qaux_params =
mx.contrib.quantization.quantize_model(sym=qsym,
+
arg_params=_arg_params,
+
aux_params=_aux_params,
+
ctx=ctx,
+
excluded_sym_names=[],
+
calib_data=data,
+
num_calib_examples=num_calib_examples,
+
calib_mode='none',
+
quantized_dtype=quantized_dtype,
+
label_names=None,
+
logger=logger)
+ qsym = sym.get_backend_symbol('MKLDNN_QUANTIZE')
+ return qsym, qarg_params, qaux_params
+
+ def quantized_with_new_api(sym, _arg_params, _aux_params, quantized_dtype,
_data_shape, logger):
+ from mxnet.contrib.quantization import *
+
+ sym = sym.get_backend_symbol('MKLDNN_QUANTIZE')
+ # set exclude layers
+ excluded_names = []
+ # set calib mode.
+ calib_mode = 'naive'
+ # set calib_layer
+ calib_layer = None
+ # set quantized_dtype
+ ctx = mx.cpu(0)
+ cqsym, cqarg_params, _aux_params, collector = \
+ quantize_graph(sym=sym, arg_params=_arg_params,
aux_params=_aux_params,
+ excluded_sym_names=excluded_names,
+ calib_mode=calib_mode, calib_layer=calib_layer,
+ quantized_dtype=quantized_dtype, logger=logger)
+ batch_size = 16
+ calibration_data = [mx.random.uniform(-1.0, 1.0, shape=_data_shape,
ctx=ctx, dtype='float32')]
+
+ data = mx.io.NDArrayIter(data=calibration_data, label=[],
batch_size=batch_size,
+ shuffle=True, last_batch_handle='discard')
+ # create module
+ mod = mx.mod.Module(symbol=sym, label_names=None, context=mx.cpu())
+ mod.bind(for_training=False, data_shapes=data.provide_data,
label_shapes=None)
+ mod.set_params(_arg_params, _aux_params)
+ # calibration configs
+ # set num_calib_batches
+ num_calib_batches = 5
+ max_num_examples = num_calib_batches * batch_size
+ # monitor FP32 Inference
+ mod._exec_group.execs[0].set_monitor_callback(collector.collect,
monitor_all=True)
+ num_batches = 0
+ num_examples = 0
+ for batch in data:
+ mod.forward(data_batch=batch, is_train=False)
+ num_batches += 1
+ num_examples += batch_size
+ if num_examples >= max_num_examples:
+ break
+ if logger is not None:
+ logger.info("Collected statistics from %d batches with
batch_size=%d"
+ % (num_batches, batch_size))
+ cqsym, cqarg_params, _aux_params = calib_graph(qsym=cqsym,
+ arg_params=_arg_params,
+ aux_params=_aux_params,
+ collector=collector,
+ calib_mode=calib_mode,
+
quantized_dtype=quantized_dtype,
+ logger=logger)
+ return cqsym, cqarg_params, _aux_params
+
+ ctx = mx.cpu(0)
+ x = mx.random.uniform(-1.0, 1.0, shape=data_shape, ctx=ctx,
dtype='float32')
+ _, arg_params, aux_params = get_mxnet_output(mx_symbol, x)
+ quantize_api = quantize_with_old_api if use_old_mxnet_quantization_api()
else quantized_with_new_api
+
+
Review comment:
remove this white spaces
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
