Update Docker CI (apache#8193)

* add failing onnx tets

* point jenkins at new docker

* support convtranspose opset 11 autopadding

* Don't force output shape for conv transpose tests, add 1D and 3D cases

* disable test until CI update complete

* try updating docker images again

* skip a test until update complete

* next try at docker images

* manage TF memory use in TF1 tests

* support explicit padding for NCHW TF padding test

* Update to tagged tlcpack images

Thanks, Andrew!

Co-authored-by: Andrew Reusch <areusch@gmail.com>

Co-authored-by: Andrew Reusch <areusch@gmail.com>

Jenkinsfile

@@ -45,12 +45,12 @@
 
 // NOTE: these lines are scanned by docker/dev_common.sh. Please update the regex as needed. -->
 ci_lint = "tlcpack/ci-lint:v0.66"
-ci_gpu = "tlcpack/ci-gpu:v0.75"
-ci_cpu = "tlcpack/ci-cpu:v0.74"
+ci_gpu = "tlcpack/ci-gpu:v0.76"
+ci_cpu = "tlcpack/ci-cpu:v0.75"
 ci_wasm = "tlcpack/ci-wasm:v0.71"
 ci_i386 = "tlcpack/ci-i386:v0.73"
-ci_qemu = "tlcpack/ci-qemu:v0.05"
-ci_arm = "tlcpack/ci-arm:v0.05"
+ci_qemu = "tlcpack/ci-qemu:v0.06"
+ci_arm = "tlcpack/ci-arm:v0.06"
 // <--- End of regex-scanned config.
 
 // tvm libraries

python/tvm/relay/frontend/onnx.py

@@ -585,6 +585,70 @@ def _impl_v1(cls, inputs, attr, params):
             out = _op.nn.bias_add(out, inputs[2])
         return out
 
+    @classmethod
+    def _impl_v11(cls, inputs, attr, params):
+        # get number of channels
+        out_type = infer_type(inputs[1])
+        out_shapes = [get_const_tuple(out_type.checked_type.shape)]
+        channels = out_shapes[0][1]
+        attr["channels"] = channels
+        groups = attr.get("group", 1)
+
+        if "kernel_shape" not in attr:
+            attr["kernel_shape"] = out_shapes[0][2:]
+
+        attr["groups"] = groups
+        # infer pads for auto_pad
+        data = inputs[0]
+        input_shape = infer_shape(data)
+        ndim = len(input_shape)
+        if "auto_pad" in attr:
+            attr["auto_pad"] = attr["auto_pad"].decode("utf-8")
+            if attr["auto_pad"] in ("SAME_UPPER", "SAME_LOWER"):
+                # Warning: Convolution does not yet support dynamic shapes,
+                # one will need to run dynamic_to_static on this model after import
+                kernel_shape = attr["kernel_shape"]
+                kndim = len(kernel_shape)
+                dilations = attr.get("dilations", [1] * kndim)
+                output_padding = attr.get("output_padding", [0] * kndim)
+                strides = attr["strides"]
+                total_pad = [0] * kndim
+                for i in range(kndim):
+                    total_pad[i] = (
+                        output_padding[i] + ((kernel_shape[i] - 1) * dilations[i] + 1) - strides[i]
+                    )
+                left = [p // 2 for p in total_pad]
+                right = [total_pad[i] - left[i] for i in range(kndim)]
+                if "LOWER" in attr["auto_pad"]:
+                    pad = left + right
+                else:
+                    pad = right + left
+                attr["pads"] = pad
+            elif attr["auto_pad"] == "VALID":
+                attr["pads"] = tuple([0 for i in range(ndim - 2)])
+            elif attr["auto_pad"] == "NOTSET":
+                pass
+            else:
+                msg = 'Value {} in attribute "auto_pad" of operator Conv is invalid.'
+                raise tvm.error.OpAttributeInvalid(msg.format(attr["auto_pad"]))
+            attr.pop("auto_pad")
+
+        out = AttrCvt(
+            op_name=dimension_picker("conv", "_transpose"),
+            transforms={
+                "kernel_shape": "kernel_size",
+                "dilations": ("dilation", 1),
+                "pads": ("padding", 0),
+                "group": ("groups", 1),
+            },
+            disables=["output_shape"],
+            custom_check=dimension_constraint(),
+        )([data, inputs[1]], attr, params)
+        use_bias = len(inputs) == 3
+        if use_bias:
+            out = _op.nn.bias_add(out, inputs[2])
+        return out
+
 
 class GlobalAveragePool(OnnxOpConverter):
     """Operator converter for GlobalAveragePool"""