[GitHub] [incubator-tvm] alexwong commented on a change in pull request #4497: [WIP] [Relay] Add a PyTorch to Relay Parser
alexwong commented on a change in pull request #4497: [WIP] [Relay] Add a PyTorch to Relay Parser URL: https://github.com/apache/incubator-tvm/pull/4497#discussion_r372766377 ## File path: python/tvm/relay/frontend/pytorch.py ## @@ -0,0 +1,1098 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. +# pylint: disable=import-self, too-many-lines, len-as-condition, no-else-return, unused-variable, too-many-nested-blocks +# pylint: disable=consider-iterating-dictionary, invalid-name, unused-argument, unused-variable, broad-except +"""PT: PyTorch frontend.""" +import numpy as np + +import tvm + +from .. import analysis as _analysis +from .. import expr as _expr +from .. import module as _module +from .. import op as _op +from .common import get_relay_op +from .common import infer_shape as _infer_shape + +__all__ = ['from_pytorch'] + +# operator implementation +def _elemwise(name): +def _impl(inputs, input_types): +data0 = _convert_elemwise_input(inputs[0]) +data1 = _convert_elemwise_input(inputs[1]) + +return get_relay_op(name)(data0, data1) +return _impl + +def _unsqueeze(): +def _impl(inputs, input_types): +data = inputs[0] +axis = inputs[1] + +return _op.transform.expand_dims(data, int(axis), 1) +return _impl + +def _concatenate(): +def _impl(inputs, input_types): +data = inputs[0] +axis = inputs[1] + +if isinstance(data, (_expr.Call, _expr.TupleGetItem, _expr.Var)): +data = [data] + +return _op.tensor.concatenate(data, int(axis)) +return _impl + +def _slice(): +def _impl(inputs, input_types): +data = inputs[0] +strides = [] + +inferred_shape = _infer_shape(data) +end = [] +for infer in inferred_shape: +end.append(int(infer)) +if isinstance(data, _expr.Var): +end = _infer_shape(data) +end = list(end) + +begin = [0]*len(end) +dim = int(inputs[1]) +begin[dim] = int(inputs[2]) + +if isinstance(inputs[3], str) and inputs[3].isdigit(): +end[dim] = min(end[dim], int(inputs[3])) +else: +end[dim] = inputs[3] + +strides.append(int(inputs[4])) +return _op.transform.strided_slice(data, begin, end, strides) +return _impl + +def _select(): +def _impl(inputs, input_types): Review comment: Yes, this op seems to be almost a 1-1 map with select. 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] masahi commented on a change in pull request #4497: [WIP] [Relay] Add a PyTorch to Relay Parser
masahi commented on a change in pull request #4497: [WIP] [Relay] Add a PyTorch to Relay Parser URL: https://github.com/apache/incubator-tvm/pull/4497#discussion_r372752325 ## File path: python/tvm/relay/frontend/pytorch.py ## @@ -0,0 +1,1098 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. +# pylint: disable=import-self, too-many-lines, len-as-condition, no-else-return, unused-variable, too-many-nested-blocks +# pylint: disable=consider-iterating-dictionary, invalid-name, unused-argument, unused-variable, broad-except +"""PT: PyTorch frontend.""" +import numpy as np + +import tvm + +from .. import analysis as _analysis +from .. import expr as _expr +from .. import module as _module +from .. import op as _op +from .common import get_relay_op +from .common import infer_shape as _infer_shape + +__all__ = ['from_pytorch'] + +# operator implementation +def _elemwise(name): +def _impl(inputs, input_types): +data0 = _convert_elemwise_input(inputs[0]) +data1 = _convert_elemwise_input(inputs[1]) + +return get_relay_op(name)(data0, data1) +return _impl + +def _unsqueeze(): +def _impl(inputs, input_types): +data = inputs[0] +axis = inputs[1] + +return _op.transform.expand_dims(data, int(axis), 1) +return _impl + +def _concatenate(): +def _impl(inputs, input_types): +data = inputs[0] +axis = inputs[1] + +if isinstance(data, (_expr.Call, _expr.TupleGetItem, _expr.Var)): +data = [data] + +return _op.tensor.concatenate(data, int(axis)) +return _impl + +def _slice(): +def _impl(inputs, input_types): +data = inputs[0] +strides = [] + +inferred_shape = _infer_shape(data) +end = [] +for infer in inferred_shape: +end.append(int(infer)) +if isinstance(data, _expr.Var): +end = _infer_shape(data) +end = list(end) + +begin = [0]*len(end) +dim = int(inputs[1]) +begin[dim] = int(inputs[2]) + +if isinstance(inputs[3], str) and inputs[3].isdigit(): +end[dim] = min(end[dim], int(inputs[3])) +else: +end[dim] = inputs[3] + +strides.append(int(inputs[4])) +return _op.transform.strided_slice(data, begin, end, strides) +return _impl + +def _select(): +def _impl(inputs, input_types): Review comment: can we use op.transform.take(...) for this? https://github.com/apache/incubator-tvm/blob/master/python/tvm/relay/op/transform.py#L263 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] FrozenGene commented on issue #4785: [FFI][Windows] Parse additional exception strings
FrozenGene commented on issue #4785: [FFI][Windows] Parse additional exception strings URL: https://github.com/apache/incubator-tvm/pull/4785#issuecomment-580071366 @soiferj if you have no other comment, please approve this change according to this doc: https://docs.tvm.ai/contribute/code_review.html#approve-and-request-changes-explicitly 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] shoubhik commented on issue #4764: [CI] ci-gpu update blockers
shoubhik commented on issue #4764: [CI] ci-gpu update blockers URL: https://github.com/apache/incubator-tvm/issues/4764#issuecomment-580066405 @tqchen, in that case, can we create a new docker instance specifically for testing the qnn parser? 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] FrozenGene commented on a change in pull request #4790: Fast exponent
FrozenGene commented on a change in pull request #4790: Fast exponent URL: https://github.com/apache/incubator-tvm/pull/4790#discussion_r372741774 ## File path: topi/include/topi/elemwise.h ## @@ -360,5 +359,71 @@ inline Tensor full_like(const Tensor& x, }, name, tag); } + /* + * \brief Fast exponential function implementation from Eigen + * https://github.com/eigenteam/eigen-git-mirror/blob/master/Eigen/src/Core/arch/Default/GenericPacketMathFunctions.h#L183 Review comment: What is the license of libeigen? If you understand and write the code by yourself, could remove it. 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] vinx13 commented on a change in pull request #4775: conv3d_ndhwc schedule
vinx13 commented on a change in pull request #4775: conv3d_ndhwc schedule URL: https://github.com/apache/incubator-tvm/pull/4775#discussion_r372740925 ## File path: topi/python/topi/x86/conv3d.py ## @@ -0,0 +1,112 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. +# pylint: disable=invalid-name, unused-variable, too-many-locals +# pylint: disable=unused-argument, redefined-builtin, no-else-return +"""Conv3D operators""" +import tvm +from tvm import autotvm +from .. import generic, tag +from ..nn.conv3d import conv3d, conv3d_ndhwc, conv3d_ncdhw +from ..generic.nn import schedule_conv3d_ndhwc + +@autotvm.register_topi_compute(conv3d, 'cpu', ['direct']) +def conv3d_x86(cfg, input, filter, strides, padding, dilation, layout='NCDHW', out_dtype=None): +if layout == 'NCDHW': +return conv3d_ncdhw(input, filter, strides, padding, dilation, out_dtype) +elif layout == 'NDHWC': +return conv3d_ndhwc(input, filter, strides, padding, dilation, out_dtype) + +@autotvm.register_topi_schedule(schedule_conv3d_ndhwc, 'cpu', ['direct']) +def schedule_conv3d_ndhwc_x86(cfg, outs): +"""TOPI schedule callback for conv2d Review comment: @alexgl-github this comment hasn't been resolved 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] FrozenGene commented on a change in pull request #4788: [FRONTEND][TFLITE]Gather, StridedSlice op support added
FrozenGene commented on a change in pull request #4788:
[FRONTEND][TFLITE]Gather, StridedSlice op support added
URL: https://github.com/apache/incubator-tvm/pull/4788#discussion_r372740391
##
File path: tests/python/frontend/tflite/test_forward.py
##
@@ -244,6 +244,57 @@ def test_forward_slice():
_test_slice(np.arange(8, dtype=np.int32).reshape((2, 4)), begin=[0,
1], size=[-1, -1])
_test_slice(np.arange(5, dtype=np.int32).reshape((5, )), begin=[4],
size=[-1])
+###
+# Gather
+# --
+
+def _test_gather(dshape, indices, axis, dtype):
+""" One iteration of Gather """
+data = np.random.uniform(1, 10, size=dshape).astype(dtype)
+indices = np.asarray(indices).astype('int32')
+
+with tf.Graph().as_default():
+in_data = array_ops.placeholder(shape=data.shape, dtype=data.dtype)
+out = array_ops.gather(in_data, indices, axis=axis)
+compare_tflite_with_tvm(data, 'Placeholder:0', [in_data], [out])
+
+def test_forward_gather():
Review comment:
What happened when the input data is quantized?
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:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] jroesch merged pull request #4783: Make sure to visit the arguments of inlined functions
jroesch merged pull request #4783: Make sure to visit the arguments of inlined functions URL: https://github.com/apache/incubator-tvm/pull/4783 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: us...@infra.apache.org With regards, Apache Git Services
[incubator-tvm] branch master updated (1b8522e -> 6798ba8)
This is an automated email from the ASF dual-hosted git repository. jroesch pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git. from 1b8522e [AUTOTVM] Fix a bug in generating the search space (#4779) add 6798ba8 Make sure to visit the arguments of inlined functions (#4783) No new revisions were added by this update. Summary of changes: src/relay/backend/vm/inline_primitives.cc | 14 -- 1 file changed, 12 insertions(+), 2 deletions(-)
[GitHub] [incubator-tvm] icemelon9 commented on issue #4764: [CI] ci-gpu update blockers
icemelon9 commented on issue #4764: [CI] ci-gpu update blockers URL: https://github.com/apache/incubator-tvm/issues/4764#issuecomment-580047528 @shoubhik @tqchen I found out that mxnet-mkl RNN layer has some bug. If we use mxnet, the rnn layer test can pass. But if we install mxnet-mkl, even for 1.5.0, the mxnet result is wrong. 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] zxy844288792 commented on a change in pull request #4787: [Relay] Conv2D padding representation
zxy844288792 commented on a change in pull request #4787: [Relay] Conv2D padding representation URL: https://github.com/apache/incubator-tvm/pull/4787#discussion_r372718146 ## File path: python/tvm/relay/op/nn/nn.py ## @@ -202,6 +203,8 @@ def conv2d(data, dilation = (dilation, dilation) if isinstance(padding, int): padding = (padding, padding) Review comment: thanks, removed 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] Laurawly commented on a change in pull request #4644: [WIP] Relay op strategy
Laurawly commented on a change in pull request #4644: [WIP] Relay op strategy
URL: https://github.com/apache/incubator-tvm/pull/4644#discussion_r372695578
##
File path: python/tvm/autotvm/task/task.py
##
@@ -116,43 +149,134 @@ def __repr__(self):
self.name, self.args, self.kwargs, self.workload
)
-TASK_TABLE = {
-}
+TASK_TABLE2 = {}
Review comment:
Can we remove 2 in task_table as well?
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:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] Laurawly commented on a change in pull request #4644: [WIP] Relay op strategy
Laurawly commented on a change in pull request #4644: [WIP] Relay op strategy
URL: https://github.com/apache/incubator-tvm/pull/4644#discussion_r372700311
##
File path: python/tvm/relay/quantize/_annotate.py
##
@@ -53,11 +53,11 @@ def simulated_quantize_compute(attrs, inputs, out_type,
target):
return [rdata]
-_reg.register_schedule("relay.op.annotation.simulated_quantize",
- _reg.schedule_injective)
+# _reg.register_schedule("relay.op.annotation.simulated_quantize",
Review comment:
Remove
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:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] Laurawly commented on a change in pull request #4644: [WIP] Relay op strategy
Laurawly commented on a change in pull request #4644: [WIP] Relay op strategy URL: https://github.com/apache/incubator-tvm/pull/4644#discussion_r372704546 ## File path: topi/python/topi/intel_graphics/conv2d.py ## @@ -443,8 +354,32 @@ def _schedule_cl_spatialpack_NCHWc(cfg, s, op): s[output].compute_inline() conv = s.outputs[0] SCHEDULE_OUTPUT = False +else: # conv2d_NCHWc_unpack +conv = op.input_tensors[0] +temp = s[conv].op.input_tensors[0] +kernel = s[conv].op.input_tensors[1] +temp_W = s.cache_read(temp, "warp", [conv]) +conv_L = s.cache_write(conv, "local") +SCHEDULE_OUTPUT = True kernel_L = s.cache_read(kernel, "local", [conv_L]) +if temp.name == "pad_temp": +data = temp.op.input_tensors[0] +# TODO(@Laurawly): Do we need to schedule pad op here? Review comment: Yeah we need to schedule temp. 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] zxy844288792 commented on a change in pull request #4787: [Relay] Conv2D padding representation
zxy844288792 commented on a change in pull request #4787: [Relay] Conv2D padding representation URL: https://github.com/apache/incubator-tvm/pull/4787#discussion_r372702055 ## File path: python/tvm/relay/op/nn/nn.py ## @@ -202,6 +203,8 @@ def conv2d(data, dilation = (dilation, dilation) if isinstance(padding, int): padding = (padding, padding) +# convert 2-way padding to 4-way padding +padding = get_pad_tuple(padding) Review comment: in topi/nn/util, we have get_pad_tuple1d and get_pad_tuple3d, so maybe I will rename it as get_pad_tuple2d for consistency 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] comaniac commented on a change in pull request #4644: [WIP] Relay op strategy
comaniac commented on a change in pull request #4644: [WIP] Relay op strategy
URL: https://github.com/apache/incubator-tvm/pull/4644#discussion_r372636908
##
File path: python/tvm/relay/backend/compile_engine.py
##
@@ -63,6 +83,316 @@ def _get_cache_key(source_func, target):
return source_func
+def get_shape(shape):
+"""Convert the shape to correct dtype and vars."""
+ret = []
+for dim in shape:
+if isinstance(dim, tvm.expr.IntImm):
+val = int(dim)
+assert val <= np.iinfo(np.int32).max
+ret.append(tvm.expr.IntImm("int32", val))
+elif isinstance(dim, tvm.expr.Any):
+ret.append(tvm.var("any_dim", "int32"))
+else:
+ret.append(dim)
+return ret
+
+
+def get_valid_implements(op, attrs, inputs, out_type, target):
+"""Get all valid implementations from the op strategy.
+
+Note that this function doesn't support op that has symbolic input shapes.
+
+Parameters
+--
+op : relay.op.Op
+Relay operator.
+
+attrs : object
+The op attribute.
+
+inputs : list of tvm.Tensor
+Input tensors to the op.
+
+out_type : relay.Type
+The output type.
+
+target : tvm.Target
+The target to compile the op.
+
+Returns
+---
+ret : list of relay.op.OpImplement
+The list of op implementations.
+"""
+fstrategy = op.get_attr("FTVMStrategy")
+assert fstrategy is not None, "%s doesn't have FTVMStrategy registered" %
op.name
+with target:
+strategy = fstrategy(attrs, inputs, out_type, target)
+ret = []
+for spec in strategy.specializations:
+if spec.condition:
+flag = True
+for clause in spec.condition.clauses:
+clause = tvm.ir_pass.Simplify(clause)
+if isinstance(clause, tvm.expr.IntImm) and clause.value:
Review comment:
Could you add comments to this statement, or use a better name for `flag`?
I don't quite understand the logic here.
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:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] comaniac commented on a change in pull request #4644: [WIP] Relay op strategy
comaniac commented on a change in pull request #4644: [WIP] Relay op strategy
URL: https://github.com/apache/incubator-tvm/pull/4644#discussion_r372683444
##
File path: python/tvm/relay/backend/compile_engine.py
##
@@ -63,6 +83,316 @@ def _get_cache_key(source_func, target):
return source_func
+def get_shape(shape):
+"""Convert the shape to correct dtype and vars."""
+ret = []
+for dim in shape:
+if isinstance(dim, tvm.expr.IntImm):
+val = int(dim)
+assert val <= np.iinfo(np.int32).max
+ret.append(tvm.expr.IntImm("int32", val))
+elif isinstance(dim, tvm.expr.Any):
+ret.append(tvm.var("any_dim", "int32"))
+else:
+ret.append(dim)
+return ret
+
+
+def get_valid_implements(op, attrs, inputs, out_type, target):
+"""Get all valid implementations from the op strategy.
+
+Note that this function doesn't support op that has symbolic input shapes.
+
+Parameters
+--
+op : relay.op.Op
+Relay operator.
+
+attrs : object
+The op attribute.
+
+inputs : list of tvm.Tensor
+Input tensors to the op.
+
+out_type : relay.Type
+The output type.
+
+target : tvm.Target
+The target to compile the op.
+
+Returns
+---
+ret : list of relay.op.OpImplement
+The list of op implementations.
+"""
+fstrategy = op.get_attr("FTVMStrategy")
+assert fstrategy is not None, "%s doesn't have FTVMStrategy registered" %
op.name
+with target:
+strategy = fstrategy(attrs, inputs, out_type, target)
+ret = []
+for spec in strategy.specializations:
+if spec.condition:
+flag = True
+for clause in spec.condition.clauses:
+clause = tvm.ir_pass.Simplify(clause)
+if isinstance(clause, tvm.expr.IntImm) and clause.value:
+continue
+flag = False
+break
+if flag:
+for impl in spec.implements:
+ret.append(impl)
+else:
+for impl in spec.implements:
+ret.append(impl)
+return ret
+
+
+def select_implement(op, attrs, inputs, out_type, target, use_autotvm=True):
+"""Select the best implement from the op strategy.
+
+If use_autotvm is True, it'll first try to find the best implementation
+based on AutoTVM profile results. If no AutoTVM profile result is found,
+it'll choose the implementation with highest plevel.
+
+If use_autotvm is False, it'll directly choose the implementation with
+highest plevel.
+
+Note that this function doesn't support op that has symbolic input shapes.
+
+Parameters
+--
+op : relay.op.Op
+Relay operator.
+
+attrs : object
+The op attribute.
+
+inputs : list[tvm.Tensor]
+Input tensors to the op.
+
+out_type : relay.Type
+The output type.
+
+target : tvm.Target
+The target to compile the op.
+
+use_autotvm : bool
+Whether query AutoTVM to pick the best.
+
+Returns
+---
+ret : tuple(relay.op.OpImplement, list[tvm.Tensor])
+The best op implementation and the corresponding output tensors.
+"""
+all_impls = get_valid_implements(op, attrs, inputs, out_type, target)
+
+best_plevel_impl = None
+for impl in all_impls:
+if best_plevel_impl is None or int(impl.plevel) >
int(best_plevel_impl.plevel):
Review comment:
Why we need to cast `plevel` to int here?
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:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] comaniac commented on a change in pull request #4644: [WIP] Relay op strategy
comaniac commented on a change in pull request #4644: [WIP] Relay op strategy
URL: https://github.com/apache/incubator-tvm/pull/4644#discussion_r372640446
##
File path: python/tvm/build_module.py
##
@@ -425,6 +425,8 @@ def lower(sch,
stmt = ir_pass.InstrumentBoundCheckers(stmt)
if simple_mode:
return stmt
+# print('='*80)
+# print(stmt)
Review comment:
Remove
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:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] comaniac commented on a change in pull request #4644: [WIP] Relay op strategy
comaniac commented on a change in pull request #4644: [WIP] Relay op strategy URL: https://github.com/apache/incubator-tvm/pull/4644#discussion_r372693044 ## File path: python/tvm/autotvm/task/dispatcher.py ## @@ -481,8 +412,12 @@ def _query_inside(self, target, workload): """ if self._counter < len(self._records): cfg = self._records[self._counter][0].config +wkl = self._records[self._counter][0].task.workload +if workload is not None: +assert wkl == workload self._counter += 1 -self.update(target, workload, cfg) +self.update(target, wkl, cfg) +cfg.workload = wkl Review comment: Where is `cfg.workload` initialized? I didn't find a definition in `ConfigSpace`. Also what's the purpose to have a workload field in a config space? 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] huajsj opened a new pull request #4791: [TOPI] upsample operator 'NCHWinic' format support.
huajsj opened a new pull request #4791: [TOPI] upsample operator 'NCHWinic' format support. URL: https://github.com/apache/incubator-tvm/pull/4791 some hardware accelerator ask packed format data like NCHWinic to fit the hardware resource, here add 'upsample' 'NCHWinic' format support to help such requirement. 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexwong edited a comment on issue #4497: [WIP] [Relay] Add a PyTorch to Relay Parser
alexwong edited a comment on issue #4497: [WIP] [Relay] Add a PyTorch to Relay Parser URL: https://github.com/apache/incubator-tvm/pull/4497#issuecomment-579987071 I cleaned up the code based off the simpler feedback and will focus on getting the CI to pass (w/ refactored test code based off @jwfromm's comment). 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] mbarrett97 commented on issue #4543: [FRONTEND][TFLITE] Add support for TFLite_Detection_PostProcess
mbarrett97 commented on issue #4543: [FRONTEND][TFLITE] Add support for TFLite_Detection_PostProcess URL: https://github.com/apache/incubator-tvm/pull/4543#issuecomment-579987271 @FrozenGene are there any other changes you want? 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexwong commented on issue #4497: [WIP] [Relay] Add a PyTorch to Relay Parser
alexwong commented on issue #4497: [WIP] [Relay] Add a PyTorch to Relay Parser URL: https://github.com/apache/incubator-tvm/pull/4497#issuecomment-579987071 I cleaned up all of the simpler fixes and will focus on getting the CI to pass (w/ refactored test code based off @jwfromm's comment). 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] jmorrill commented on issue #4785: [FFI][Windows] Parse additional exception strings
jmorrill commented on issue #4785: [FFI][Windows] Parse additional exception strings URL: https://github.com/apache/incubator-tvm/pull/4785#issuecomment-579897309 Clean fix was all you. I just did the clerical work. Thanks man! 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] comaniac commented on a change in pull request #4787: [Relay] Conv2D padding representation
comaniac commented on a change in pull request #4787: [Relay] Conv2D padding representation URL: https://github.com/apache/incubator-tvm/pull/4787#discussion_r372523989 ## File path: python/tvm/relay/op/nn/nn.py ## @@ -202,6 +203,8 @@ def conv2d(data, dilation = (dilation, dilation) if isinstance(padding, int): padding = (padding, padding) +# convert 2-way padding to 4-way padding +padding = get_pad_tuple(padding) Review comment: Since we also have conv1d and conv3d, maybe it's better to rename this function to `get_2d_pad_tuple`? 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] comaniac commented on a change in pull request #4787: [Relay] Conv2D padding representation
comaniac commented on a change in pull request #4787: [Relay] Conv2D padding representation URL: https://github.com/apache/incubator-tvm/pull/4787#discussion_r372528150 ## File path: topi/python/topi/nn/conv2d.py ## @@ -62,6 +62,8 @@ def conv2d(input, filter, strides, padding, dilation, layout='NCHW', out_dtype=N output : tvm.Tensor 4-D with shape [batch, out_channel, out_height, out_width] """ +#only accepts 4-way padding +assert len(padding) == 4, "only accepts 4-way padding" Review comment: Change the function docstring accordingly. 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] comaniac commented on a change in pull request #4787: [Relay] Conv2D padding representation
comaniac commented on a change in pull request #4787: [Relay] Conv2D padding representation URL: https://github.com/apache/incubator-tvm/pull/4787#discussion_r372525767 ## File path: python/tvm/relay/op/nn/util.py ## @@ -0,0 +1,53 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. +# pylint: disable=invalid-name, unused-variable +"""NN operator common utilities""" +from __future__ import absolute_import + +def get_pad_tuple(padding): +"""Common code to get the pad option +Parameters +-- +padding : int or str +Padding size, or ['VALID', 'SAME'] Review comment: The type should be `Union[int, Tuple[int, ...]]`. Also please change the description accordingly. 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] comaniac commented on a change in pull request #4787: [Relay] Conv2D padding representation
comaniac commented on a change in pull request #4787: [Relay] Conv2D padding representation URL: https://github.com/apache/incubator-tvm/pull/4787#discussion_r372526415 ## File path: python/tvm/relay/op/nn/nn.py ## @@ -202,6 +203,8 @@ def conv2d(data, dilation = (dilation, dilation) if isinstance(padding, int): padding = (padding, padding) Review comment: Consider removing this logic since `get_pad_tuple` can also accept `int`. 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexgl-github opened a new pull request #4790: Fast exponent
alexgl-github opened a new pull request #4790: Fast exponent URL: https://github.com/apache/incubator-tvm/pull/4790 Thanks for contributing to TVM! Please refer to guideline https://docs.tvm.ai/contribute/ for useful information and tips. After the pull request is submitted, please request code reviews from [Reviewers](https://github.com/apache/incubator-tvm/blob/master/CONTRIBUTORS.md#reviewers) by @ them in the pull request thread. 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexgl-github commented on issue #4775: conv3d_ndhwc schedule
alexgl-github commented on issue #4775: conv3d_ndhwc schedule URL: https://github.com/apache/incubator-tvm/pull/4775#issuecomment-579823237 > @alexgl-github Thanks for the contribution. > > Almost every PR needs to have a test case in Apache TVM project. In this case, it can be a test case that uses this schedule and compares accuracy with some golden reference. There's already topi conv3d_ndhwc test using this schedule: https://github.com/apache/incubator-tvm/blob/1b8522e475e9a889b80f069a83928cafa3502a74/topi/tests/python/test_topi_conv3d_ndhwc.py#L60 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] FrozenGene commented on a change in pull request #4695: [Relay][Frontend][TFlite] Add parser support for relational ops
FrozenGene commented on a change in pull request #4695: [Relay][Frontend][TFlite] Add parser support for relational ops URL: https://github.com/apache/incubator-tvm/pull/4695#discussion_r372449323 ## File path: python/tvm/relay/frontend/tflite.py ## @@ -705,47 +710,87 @@ def convert_div(self, op): # Check if the input tensor is quantized, call QNN op if self.is_quantized(op): raise tvm.error.OpNotImplemented( -'TFlite quantized div operator is not supported yet.') +'TFlite quantized DIV operator is not supported yet.') return self._convert_elemwise(_op.divide, op) def convert_pow(self, op): +"""Convert TFLite POW""" # Check if the input tensor is quantized, call QNN op if self.is_quantized(op): raise tvm.error.OpNotImplemented( -'TFlite quantized pow operator is not supported yet.') +'TFlite quantized POW operator is not supported yet.') return self._convert_elemwise(_op.power, op) +def convert_squared_difference(self, op): Review comment: This part shouldn’t appear in your pr if you rebase master correctly. Seems that you don’t handle it correctly and move this implementation manually. Please rebase your pr to master correctly referring this: https://docs.tvm.ai/contribute/git_howto.html 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] FrozenGene commented on a change in pull request #4695: [Relay][Frontend][TFlite] Add parser support for relational ops
FrozenGene commented on a change in pull request #4695: [Relay][Frontend][TFlite] Add parser support for relational ops URL: https://github.com/apache/incubator-tvm/pull/4695#discussion_r372450103 ## File path: tests/python/frontend/tflite/test_forward.py ## @@ -843,6 +843,13 @@ def _test_pow(data): """ One iteration of power """ return _test_elemwise(math_ops.pow, data) ### +# Squared_difference +# -- + +def _test_squared_difference(data): Review comment: Same problem as previous comment. Please correct it. 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] u99127 commented on issue #4696: [Relay][Frontend][TFlite] Add support for quantized LOGISTIC
u99127 commented on issue #4696: [Relay][Frontend][TFlite] Add support for quantized LOGISTIC URL: https://github.com/apache/incubator-tvm/pull/4696#issuecomment-579765000 ping for an approving review ? @FrozenGene @anijain2305. 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] u99127 commented on a change in pull request #4788: [FRONTEND][TFLITE]Gather, StridedSlice op support added
u99127 commented on a change in pull request #4788: [FRONTEND][TFLITE]Gather, StridedSlice op support added URL: https://github.com/apache/incubator-tvm/pull/4788#discussion_r372389879 ## File path: python/tvm/relay/frontend/tflite.py ## @@ -747,6 +749,158 @@ def convert_squared_difference(self, op): out = _op.power(difference, relay.const(2, exp_type)) return out +def convert_gather(self, op): +"""Method to Convert TFLite Gather operator""" +# Check if the input tensor is quantized, call QNN op +if self.is_quantized(op): +raise tvm.error.OpNotImplemented( +'TFlite quantized gather operator is not supported yet.') Review comment: Since this is a tensor manipulation operation rather than something that requires any specific qnn dialect, I would prefer we handle the qnn ops here. 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] u99127 commented on a change in pull request #4788: [FRONTEND][TFLITE]Gather, StridedSlice op support added
u99127 commented on a change in pull request #4788: [FRONTEND][TFLITE]Gather,
StridedSlice op support added
URL: https://github.com/apache/incubator-tvm/pull/4788#discussion_r372390838
##
File path: python/tvm/relay/frontend/tflite.py
##
@@ -747,6 +749,158 @@ def convert_squared_difference(self, op):
out = _op.power(difference, relay.const(2, exp_type))
return out
+def convert_gather(self, op):
+"""Method to Convert TFLite Gather operator"""
+# Check if the input tensor is quantized, call QNN op
+if self.is_quantized(op):
+raise tvm.error.OpNotImplemented(
+'TFlite quantized gather operator is not supported yet.')
+input_tensors = self.get_input_tensors(op)
+
+try:
+from tflite.BuiltinOptions import BuiltinOptions
+from tflite.GatherOptions import GatherOptions
+from tflite.TensorType import TensorType
+except ImportError:
+raise ImportError("The tflite package must be installed")
+
+assert op.BuiltinOptionsType() == BuiltinOptions.GatherOptions
+op_options = op.BuiltinOptions()
+gather_options = GatherOptions()
+gather_options.Init(op_options.Bytes, op_options.Pos)
+axis = gather_options.Axis()
+
+data = self.get_expr(input_tensors[0].tensor_idx)
+
+indices = input_tensors[1]
+indices_type = indices.tensor.Type()
+
+assert indices_type in (TensorType.INT32, TensorType.INT64)
+indices_type_str = self.get_tensor_type_str(indices_type)
+indices = self.exp_tab.new_const(self.get_tensor_value(indices),
+ dtype=indices_type_str)
+out = _op.take(data, indices, axis=axis)
+return out
+
+def convert_strided_slice(self, op):
+"""Method to Convert TFLite Strided Slice operator"""
+# Check if the input tensor is quantized, call QNN op
+if self.is_quantized(op):
+raise tvm.error.OpNotImplemented(
+'TFlite quantized strided slice operator is not supported
yet.')
Review comment:
Same comment as above - this is a tensor manipulation op rather than
anything that intrinsically requires quantization knowledge.
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:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] inadob commented on issue #4789: [Frontend][TFLite] Dynamically calculate input_stats of any fake_quant range
inadob commented on issue #4789: [Frontend][TFLite] Dynamically calculate input_stats of any fake_quant range URL: https://github.com/apache/incubator-tvm/pull/4789#issuecomment-579739378 @kevinthesun @anijain2305 can you please take a look 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] inadob opened a new pull request #4789: [Frontend][TFLite] Dynamically calculate input_stats of any fake_quant range
inadob opened a new pull request #4789: [Frontend][TFLite] Dynamically calculate input_stats of any fake_quant range URL: https://github.com/apache/incubator-tvm/pull/4789 * pass the input range to the convertor and caclulate (mean, scale) there * change the range of the second tensor in elemwise operations so that we test inputs with different quant params * change the possible output range for elemwise ops wrt the updated ranges * update the comments for (m, s) calculations * add input range dict to tflite reduce_mean operation 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] siju-samuel opened a new pull request #4788: [FRONTEND][TFLITE]Gather, StridedSlice op support added
siju-samuel opened a new pull request #4788: [FRONTEND][TFLITE]Gather, StridedSlice op support added URL: https://github.com/apache/incubator-tvm/pull/4788 Thanks for contributing to TVM! Please refer to guideline https://docs.tvm.ai/contribute/ for useful information and tips. After the pull request is submitted, please request code reviews from [Reviewers](https://github.com/apache/incubator-tvm/blob/master/CONTRIBUTORS.md#reviewers) by @ them in the pull request thread. @FrozenGene @kevinthesun Could you please help to review this patch. TIA 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: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] zxy844288792 opened a new pull request #4787: [Relay] Conv2D padding representation
zxy844288792 opened a new pull request #4787: [Relay] Conv2D padding representation URL: https://github.com/apache/incubator-tvm/pull/4787 As discussed here: https://discuss.tvm.ai/t/rfc-conv2d-padding-representation/5394. We agree we should enforce topi.nn.conv2d to accept 4-way padding. We also should have relay.nn.conv2d legalizes the padding to 4-way. get_pad_tuple is from topi util.py. I deleted some unuseful code and reuse it for relay.op.nn.conv2d. Thanks for contributing to TVM! Please refer to guideline https://docs.tvm.ai/contribute/ for useful information and tips. After the pull request is submitted, please request code reviews from [Reviewers](https://github.com/apache/incubator-tvm/blob/master/CONTRIBUTORS.md#reviewers) by @ them in the pull request thread. 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: us...@infra.apache.org With regards, Apache Git Services
