relai-ai/mmcv-standard · Datasets at Hugging Face

Question stringlengths 45 251	Response stringlengths 1 445
Which MMCV operator specifically handles image region pooling that can accommodate rotational variations?	MMCVRoIAlignRotated
What is the shape of the attention map output from MMCVCAWeight, given an input query and key matrix each of shape (N, C', H, W)?	(N, H+W-1, H, W)
What operator is defined in Torch and included for conversion to ONNX?	grid_sampler
What shape does the attention map output by MMCVCAWeight take in the context of Criss-Cross Attention?	(N, H+W-1, H, W)
What shape does the output tensor have when using the MMCVCAWeight operator?	(N, H+W-1, H, W)
What is the order of the sampled border features' arrangement in the pooled features' output tensor for BorderDet?	(top, left, bottom, right)
What PyTorch-based project provides tools for image inpainting, matting, restoration, colorization, and generation within the OpenMMLab framework?	MMagic
What parameter determines the number of input samples to take for each output sample in RoI align pooling for rotated proposals when using the mmcv version mentioned?	sampling_ratio
Which method in the mmcv.ops module is utilized for sampling points that are spaced farthest apart?	furthest_point_sample()
What is the function in mmcv used to pad an image so that each edge becomes a multiple of a specific value?	mmcv.impad_to_multiple
Which optimizer hook from mmcv version 1.7.1 has been moved and is now represented by both AmpOptimWrapper and ApexOptimWrapper in mmengine?	Fp16OptimizerHook
In the context of MMCVRoIAlign, what is the purpose of the `sampling_ratio` parameter when its value is set to `0`?	Take samples densely for current models.
What must be ensured when using a custom data transformation class like `MyFlip` in the configuration file?	The file where the `MyFlip` class is located must be importable at runtime.
What class in mmcv's data transformation framework enables mapping fields within the data dictionary, allowing image processing transforms to handle alternative image field names?	KeyMapper
Which function in mmcv.cnn is specifically designed to bundle convolution, normalization, and activation layers, as per the MMCV version 2.2.0?	ConvModule
For building MMCV on Windows with CUDA support, what CUDA version is specified as a requirement?	CUDA 10.2
For NonMaxSuppression, what are the two possible configurations for describing the box data format?	0 - [y1, x1, y2, x2], 1 - [x_center, y_center, width, height]
Which utility in mmcv version 1.7.1 checks the availability of multiple processing units including IPUs, MLUs, MPS, and NPUs?	mmcv.utils.device_type
What shape does the output tensor take when the `MMCVCAMap` operator processes an input value matrix of shape (N, C, H, W)?	(N, C, H, W)
What is the shape of the pooled features output by the MMCVBorderAlign operator?	[N, C, H*W, 4]
What data type is required for the input tensor in the cumulative maximum operation?	tensor(float32)
Which MMCV operator deals specifically with rotating region of interest alignment in version 2.0?	MMCVRoIAlignRotated
How does the `TransformBroadcaster` ensure the same random transformations are applied to multiple fields simultaneously?	share_random_params=True
What types of systems are supported for the implementation of common CUDA operations mentioned in MMCV?	Linux, Windows, macOS
What is the shape of the `indices` output tensor for the MMCV function `cummin` given an input tensor with non-empty elements?	The same shape as `input`.
Which mmcv function related to checkpoint loading is also present in mmengine according to version 1.7.1?	load_checkpoint
What is the shape of the attention map output by the MMCVCAWeight operator in the mmcv framework?	(N, H+W-1, H, W)
What parameter is explicitly stated to support only stride=1 in the MMCVMaskedConv2d function?	stride
What is the shape of the attention map produced by the MMCVCAWeight operator in terms of N, H, and W?	(N, H+W-1, H, W)
What specific function in mmengine, according to its source code, acts as a counterpart to the mmcv.runner.dist_utils._init_dist_slurm in v1.7.1 of mmcv?	mmengine.dist.utils._init_dist_slurm
What should you set the `TORCH_CUDA_ARCH_LIST` environment variable to, based on the compute capability of an NVIDIA GeForce GTX 1660 SUPER GPU?	7.5
What is the solution if you encounter an "unsupported Microsoft Visual Studio version" error while building mmcv-full with CUDA 9.2 on Windows?	Use Microsoft Visual Studio 2017.
What is required to avoid installing `opencv-python` when setting up MMCV in a minimum container environment?	`opencv-python-headless`
What is the role of the `BaseTransform` class in designing new data transformation classes in MMCV?	The `BaseTransform` class serves as a unified base class for all data transformation classes, requiring them to inherit it and implement the `transform` method.
Which function in the MMCV 1.7.1 version has been moved to both the AmpOptimWrapper and ApexOptimWrapper in MMEngine?	Fp16OptimizerHook
Which module in MMCV v1.x that was responsible for parallel operations has been removed in the transition to MMCV v2.x?	mmcv.parallel
What environment variable should be set to indicate the absence of PyTorch-related modules while building an mmcv lite version?	MMCV_WITH_OPS=0
What is the specific constraint on the stride parameter for MMCVMaskedConv2d according to the mmcv library?	Only support stride=1 in mmcv.
Which function in the MMCV version 1.7.1 identifies if an object is a string?	mmcv.utils.misc.is_str
In the MMCVBorderAlign operator, how are the pooled features' positions sampled along the borders of the bounding boxes?	Uniformly sampled `pool_size`+1 positions on each line, including the start and end points.
What must you ensure about PyTorch before installing any version of MMCV to avoid potential installation issues?	PyTorch must be successfully installed.
What are the two methods mentioned for resolving conflicts in your local branch with the upstream's latest main branch, and what is the advantage of using one of these methods if you are skilled at it?	Rebase, merge; advantage of rebase: keeps commit logs tidy.
What is the shape of the `offset` tensor input required for MMCVModulatedDeformConv2d given in the reference?	(N, deformable_group* 2* kH* kW, outH, outW)
What function in MMCV version 2.2.0 calculates the generalized intersection-over-union between point sets and polygons?	convex_giou
What modules and classes were removed during the upgrade from MMCV v1.x to MMCV v2.x?	Modules and classes removed: `mmcv.fileio`, `mmcv.runner`, `mmcv.parallel`, `mmcv.engine`, `mmcv.device`, and all classes and most functions in `mmcv.utils`.
What function in the MMCV 2.2.0 API is used to calculate the generalized intersection-over-union between point sets and polygons?	convex_giou
What are the possible values for the `padding_mode` parameter in the `grid_sampler` function, and what do they signify?	0: `zeros`, 1: `border`, 2: `reflection`
Which module in MMEngine provides functionality similar to `mmcv.utils.timer.TimerError` in MMCV version 1.7.1?	mmengine.utils.timer.TimerError
What is the constraint on the stride for the masked 2D convolution function in the MMCV library?	Stride=1
Which module from mmcv.cnn in version 2.0.0 is described as having the ability to perform receptive field searching?	Conv2dRFSearchOp
Which module in mmcv.cnn version 2.0.0 is responsible for implementing depthwise separable convolutions?	DepthwiseSeparableConvModule
What is the constraint on the tensor type for the inputs in MMCVBorderAlign?	tensor(float32)
What must be modified in the local files of PyTorch to resolve the "error: member 'torch::jit::ProfileOptionalOp::Kind' may not be initialized" when building mmcv-full on Windows with PyTorch version 1.7.0?	Delete `static constexpr Symbol Kind = ::c10::prim::profile;` and `static constexpr Symbol Kind = ::c10::prim::profile_optional;` at `torch/include\torch/csrc/jit/ir/ir.h`. Replace `explicit operator type&() { return (this->value); }` with `explicit operator type&() { return ((type*)this->value); }` at `torch/include/pybind11/cast.h`. Replace all `CONSTEXPR_EXCEPT_WIN_CUDA` with `const` at `torch/include/torch/csrc/jit/api/module.h`.
What is the equivalent method in mmengine for `mmcv.utils.path.symlink` in version 1.7.1 of MMCV?	mmengine.utils.path.symlink
Which function in mmcv version 1.7.1 has been moved to both mmengine.optimizer.amp_optimizer_wrapper.AmpOptimWrapper and mmengine.optimizer.apex_optimizer_wrapper.ApexOptimWrapper?	mmcv.runner.fp16_utils.cast_tensor_type
Which operator in MMCV explicitly involves manipulating an input mask for modulated deformable convolution, and what is the required shape of this input mask tensor?	MMCVModulatedDeformConv2d; shape (N, deformable_group * kH * kW, outH, outW)
What is the expected shape of the input offset tensor for the MMCVModulatedDeformConv2d operation given a deformable group size of 2?	(N, deformable_group2kH*kW, outH, outW)
In mmcv version 1.7.1, which hook has been moved to mmengine's AmpOptimWrapper and ApexOptimWrapper?	Fp16OptimizerHook
What requirement must be fulfilled before building MMCV on a macOS with an Apple silicon chip to avoid encountering an installation issue?	Install PyTorch 1.13+
What minimum version of the CUDA compiler driver is required when building MMCV on Linux with GPU support?	9.2
What data type is required for the input tensor in the function that outputs cumulative minimum elements along a specified dimension when using MMCV?	tensor(float32)
What is required to ensure `cl.exe` is available in the prompt when building MMCV on Windows?	Add `C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\VC\Tools\MSVC\[version]\bin\Hostx86\x64` to `PATH`.
In the current version of MMCV, what shape and dtype do the outputs of the function that computes the cumulative minimum elements along a dimension have, compared to the input tensor?	Same shape and dtype as the input tensor.
Which mmcv function allows the image size to fit the rotated image when rotating by a certain angle?	imrotate with auto_bound=True
What environment variable must be set to a directory path before building MMCV with existing mlu-ops library according to the specified process for Cambricon MLU devices?	MMCV_MLU_OPS_PATH
Which model in the Model Zoo is specified for Key Information Extraction?	SDMGR
What is the dtype required for the input tensor to output the cumulative minimum elements in the specified dimension using the function described for MMCV operators?	float32
Under what condition is the width or height of a box calculated as (x2 - x1 + 1) when filtering out boxes with high IoU overlap in the NonMaxSuppression operation?	offset = 1
What module was removed in the upgrade from MMCV v1.x to v2.x that caused issues with importing ‘is_list_of’?	`mmcv.utils` module
What is the MMEngine equivalent for `mmcv.runner.dist_utils.master_only` in version 1.7.1?	mmengine.dist.utils.master_only
What is the only supported stride value for MMCVMaskedConv2d in mmcv?	1
What is the data type constraint for the input feature of MMCVPSAMask?	tensor(float32)
What is the shape of the output features from the MMCVBorderAlign operation for given input features with shape [N, 4C, H, W] and corresponding boxes?	[N, C, H*W, 4]
What aspect of the MMCVBorderAlign operation determines the number of sampled positions along the border lines of bounding boxes?	pool_size
What is the procedure to extend the functionality of mmcv.cnn with a custom layer using the registry mechanism?	1. Write and register your custom module using the `MODELS` registry. 2. Import your custom module where needed. 3. Use the appropriate build function (e.g., `build_upsample_layer`) with your custom type.
What is the only stride value supported by the masked 2D convolution in mmcv?	1
What is the required minimum version of CUDA compiler driver (nvcc) to build MMCV on Linux?	9.2
What is the shape of the output tensor when using the MMCVPSAMask operator on an input feature with a batch size of N, height H, and width W?	(N, H * W, H, W)
What might cause the error "RuntimeError: Expected to have finished reduction in the prior iteration before starting a new one" when using DDP mode?	Running different branches in your code in DDP mode.
What is the output shape of the RoI pooled feature map given `num_rois`, `C`, `output_height`, and `output_width`?	(num_rois, C, output_height, output_width)
Which operator in mmcv is involved in region of interest pooling and comes with type constraints according to version v2.0.0?	MMCVDeformRoIPool
What type of output tensor shape does the `grid_sampler` operation produce and what is the type constraint for the input tensor?	Output tensor shape: (N, C, outH, outW) Type constraint for input tensor: tensor(float32, Linear)
What argument must be set to true in `mmcv.imresize` when you need the scaling factor along with the resized image?	return_scale
What happens if a data transformation class marked with `avoid_cache_randomness` is used with `TransformBroadcaster` when `share_random_params` is set to True in mmcv version as per the latest document?	An exception is thrown.
What determines whether the extrema are considered as referring to the center points or corner points of the input's corner pixels in the context of grid sampling?	`align_corners`
What configuration change is recommended to solve the "RuntimeError: Expected to have finished reduction in the prior iteration before starting a new one" error when running code in DDP mode in mmcv version mentioned in the document?	find_unused_parameters = True
What is the output shape of the `MMCVModulatedDeformConv2d` operation when applied to an input feature of shape (N, C, inH, inW)?	(N, output_channel, outH, outW)
Which function in MMCV is used to convert a BGR image to an HLS image, and which module is it part of?	bgr2hls, mmcv.image
What MMCV function is specifically used to resize images while maintaining the aspect ratio, and which module is it part of?	`imrescale` in `mmcv.image`
What is the shape of the pooled features output by the MMCVBorderAlign operation in the mmcv library's current version?	[N, C, H\*W, 4]
Which formatter is used for enforcing a consistent style in Markdown files according to the PR guidelines?	mdformat
What is the constraint on the stride parameter for MMCVMaskedConv2d in mmcv?	stride=1
In the context of MMCVModulatedDeformConv2d, which parameter determines how the input feature is divided into groups, and what condition must be satisfied for this division?	`groups`; `input_channel` should be divisible by `groups`.
Which module in mmengine.strategy provides support for distributed training using multiple devices?	DDPStrategy
In the context of MMCVRoIAlign and MMCVRoIAlignRotated, what mode is used for pooling in each bin?	avg or max
When using the NonMaxSuppression method with a max_output_boxes_per_class set to 0, how does the number of output boxes compare to the number of input boxes?	The number of output boxes is equal to the number of input boxes.
What command verifies successful installation of mmcv after building it from source?	python .dev_scripts/check_installation.py
What is the purpose of mypy in the MMCV development process?	Static type checking.
What tool is recommended to fix common misspellings in text files for Python development according to the document?	codespell
Which formatter is used for Python docstrings in the specified coding guidelines?	docformatter

Which MMCV operator specifically handles image region pooling that can accommodate rotational variations?

MMCVRoIAlignRotated

What is the shape of the attention map output from MMCVCAWeight, given an input query and key matrix each of shape (N, C', H, W)?

(N, H+W-1, H, W)

What operator is defined in Torch and included for conversion to ONNX?

grid_sampler

What shape does the attention map output by MMCVCAWeight take in the context of Criss-Cross Attention?

(N, H+W-1, H, W)

What shape does the output tensor have when using the MMCVCAWeight operator?

(N, H+W-1, H, W)

What is the order of the sampled border features' arrangement in the pooled features' output tensor for BorderDet?

(top, left, bottom, right)

What PyTorch-based project provides tools for image inpainting, matting, restoration, colorization, and generation within the OpenMMLab framework?

MMagic

What parameter determines the number of input samples to take for each output sample in RoI align pooling for rotated proposals when using the mmcv version mentioned?

sampling_ratio

Which method in the mmcv.ops module is utilized for sampling points that are spaced farthest apart?

furthest_point_sample()

What is the function in mmcv used to pad an image so that each edge becomes a multiple of a specific value?

mmcv.impad_to_multiple

Which optimizer hook from mmcv version 1.7.1 has been moved and is now represented by both AmpOptimWrapper and ApexOptimWrapper in mmengine?

Fp16OptimizerHook

In the context of MMCVRoIAlign, what is the purpose of the `sampling_ratio` parameter when its value is set to `0`?

Take samples densely for current models.

What must be ensured when using a custom data transformation class like `MyFlip` in the configuration file?

The file where the `MyFlip` class is located must be importable at runtime.

What class in mmcv's data transformation framework enables mapping fields within the data dictionary, allowing image processing transforms to handle alternative image field names?

KeyMapper

Which function in mmcv.cnn is specifically designed to bundle convolution, normalization, and activation layers, as per the MMCV version 2.2.0?

ConvModule

For building MMCV on Windows with CUDA support, what CUDA version is specified as a requirement?

CUDA 10.2

For NonMaxSuppression, what are the two possible configurations for describing the box data format?

0 - [y1, x1, y2, x2], 1 - [x_center, y_center, width, height]

Which utility in mmcv version 1.7.1 checks the availability of multiple processing units including IPUs, MLUs, MPS, and NPUs?

mmcv.utils.device_type

What shape does the output tensor take when the `MMCVCAMap` operator processes an input value matrix of shape (N, C, H, W)?

(N, C, H, W)

What is the shape of the pooled features output by the MMCVBorderAlign operator?

[N, C, H*W, 4]

What data type is required for the input tensor in the cumulative maximum operation?

tensor(float32)

Which MMCV operator deals specifically with rotating region of interest alignment in version 2.0?

MMCVRoIAlignRotated

How does the `TransformBroadcaster` ensure the same random transformations are applied to multiple fields simultaneously?

share_random_params=True

What types of systems are supported for the implementation of common CUDA operations mentioned in MMCV?

Linux, Windows, macOS

What is the shape of the `indices` output tensor for the MMCV function `cummin` given an input tensor with non-empty elements?

The same shape as `input`.

Which mmcv function related to checkpoint loading is also present in mmengine according to version 1.7.1?

load_checkpoint

What is the shape of the attention map output by the MMCVCAWeight operator in the mmcv framework?

(N, H+W-1, H, W)

What parameter is explicitly stated to support only stride=1 in the MMCVMaskedConv2d function?

stride

What is the shape of the attention map produced by the MMCVCAWeight operator in terms of N, H, and W?

(N, H+W-1, H, W)

What specific function in mmengine, according to its source code, acts as a counterpart to the mmcv.runner.dist_utils._init_dist_slurm in v1.7.1 of mmcv?

mmengine.dist.utils._init_dist_slurm

What should you set the `TORCH_CUDA_ARCH_LIST` environment variable to, based on the compute capability of an NVIDIA GeForce GTX 1660 SUPER GPU?

7.5

What is the solution if you encounter an "unsupported Microsoft Visual Studio version" error while building mmcv-full with CUDA 9.2 on Windows?

Use Microsoft Visual Studio 2017.

What is required to avoid installing `opencv-python` when setting up MMCV in a minimum container environment?

`opencv-python-headless`

What is the role of the `BaseTransform` class in designing new data transformation classes in MMCV?

The `BaseTransform` class serves as a unified base class for all data transformation classes, requiring them to inherit it and implement the `transform` method.

Which function in the MMCV 1.7.1 version has been moved to both the AmpOptimWrapper and ApexOptimWrapper in MMEngine?

Fp16OptimizerHook

Which module in MMCV v1.x that was responsible for parallel operations has been removed in the transition to MMCV v2.x?

mmcv.parallel

What environment variable should be set to indicate the absence of PyTorch-related modules while building an mmcv lite version?

MMCV_WITH_OPS=0

What is the specific constraint on the stride parameter for MMCVMaskedConv2d according to the mmcv library?

Only support stride=1 in mmcv.

Which function in the MMCV version 1.7.1 identifies if an object is a string?

mmcv.utils.misc.is_str

In the MMCVBorderAlign operator, how are the pooled features' positions sampled along the borders of the bounding boxes?

Uniformly sampled `pool_size`+1 positions on each line, including the start and end points.

What must you ensure about PyTorch before installing any version of MMCV to avoid potential installation issues?

PyTorch must be successfully installed.

What are the two methods mentioned for resolving conflicts in your local branch with the upstream's latest main branch, and what is the advantage of using one of these methods if you are skilled at it?

Rebase, merge; advantage of rebase: keeps commit logs tidy.

What is the shape of the `offset` tensor input required for MMCVModulatedDeformConv2d given in the reference?

(N, deformable_group* 2* kH* kW, outH, outW)

What function in MMCV version 2.2.0 calculates the generalized intersection-over-union between point sets and polygons?

convex_giou

What modules and classes were removed during the upgrade from MMCV v1.x to MMCV v2.x?

Modules and classes removed: `mmcv.fileio`, `mmcv.runner`, `mmcv.parallel`, `mmcv.engine`, `mmcv.device`, and all classes and most functions in `mmcv.utils`.

What function in the MMCV 2.2.0 API is used to calculate the generalized intersection-over-union between point sets and polygons?

convex_giou

What are the possible values for the `padding_mode` parameter in the `grid_sampler` function, and what do they signify?

0: `zeros`, 1: `border`, 2: `reflection`

Which module in MMEngine provides functionality similar to `mmcv.utils.timer.TimerError` in MMCV version 1.7.1?

mmengine.utils.timer.TimerError

What is the constraint on the stride for the masked 2D convolution function in the MMCV library?

Stride=1

Which module from mmcv.cnn in version 2.0.0 is described as having the ability to perform receptive field searching?

Conv2dRFSearchOp

Which module in mmcv.cnn version 2.0.0 is responsible for implementing depthwise separable convolutions?

DepthwiseSeparableConvModule

What is the constraint on the tensor type for the inputs in MMCVBorderAlign?

tensor(float32)

What must be modified in the local files of PyTorch to resolve the "error: member 'torch::jit::ProfileOptionalOp::Kind' may not be initialized" when building mmcv-full on Windows with PyTorch version 1.7.0?

Delete `static constexpr Symbol Kind = ::c10::prim::profile;` and `static constexpr Symbol Kind = ::c10::prim::profile_optional;` at `torch/include\torch/csrc/jit/ir/ir.h`. Replace `explicit operator type&() { return *(this->value); }` with `explicit operator type&() { return *((type*)this->value); }` at `torch/include/pybind11/cast.h`. Replace all `CONSTEXPR_EXCEPT_WIN_CUDA` with `const` at `torch/include/torch/csrc/jit/api/module.h`.

What is the equivalent method in mmengine for `mmcv.utils.path.symlink` in version 1.7.1 of MMCV?

mmengine.utils.path.symlink

Which function in mmcv version 1.7.1 has been moved to both mmengine.optimizer.amp_optimizer_wrapper.AmpOptimWrapper and mmengine.optimizer.apex_optimizer_wrapper.ApexOptimWrapper?

mmcv.runner.fp16_utils.cast_tensor_type

Which operator in MMCV explicitly involves manipulating an input mask for modulated deformable convolution, and what is the required shape of this input mask tensor?

MMCVModulatedDeformConv2d; shape (N, deformable_group * kH * kW, outH, outW)

What is the expected shape of the input offset tensor for the MMCVModulatedDeformConv2d operation given a deformable group size of 2?

(N, deformable_group*2*kH*kW, outH, outW)

In mmcv version 1.7.1, which hook has been moved to mmengine's AmpOptimWrapper and ApexOptimWrapper?

Fp16OptimizerHook

What requirement must be fulfilled before building MMCV on a macOS with an Apple silicon chip to avoid encountering an installation issue?

Install PyTorch 1.13+

What minimum version of the CUDA compiler driver is required when building MMCV on Linux with GPU support?

9.2

What data type is required for the input tensor in the function that outputs cumulative minimum elements along a specified dimension when using MMCV?

tensor(float32)

What is required to ensure `cl.exe` is available in the prompt when building MMCV on Windows?

Add `C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\VC\Tools\MSVC\[version]\bin\Hostx86\x64` to `PATH`.

In the current version of MMCV, what shape and dtype do the outputs of the function that computes the cumulative minimum elements along a dimension have, compared to the input tensor?

Same shape and dtype as the input tensor.

Which mmcv function allows the image size to fit the rotated image when rotating by a certain angle?

imrotate with auto_bound=True

What environment variable must be set to a directory path before building MMCV with existing mlu-ops library according to the specified process for Cambricon MLU devices?

MMCV_MLU_OPS_PATH

Which model in the Model Zoo is specified for Key Information Extraction?

SDMGR

What is the dtype required for the input tensor to output the cumulative minimum elements in the specified dimension using the function described for MMCV operators?

float32

Under what condition is the width or height of a box calculated as (x2 - x1 + 1) when filtering out boxes with high IoU overlap in the NonMaxSuppression operation?

offset = 1

What module was removed in the upgrade from MMCV v1.x to v2.x that caused issues with importing ‘is_list_of’?

`mmcv.utils` module

What is the MMEngine equivalent for `mmcv.runner.dist_utils.master_only` in version 1.7.1?

mmengine.dist.utils.master_only

What is the only supported stride value for MMCVMaskedConv2d in mmcv?

1

What is the data type constraint for the input feature of MMCVPSAMask?

tensor(float32)

What is the shape of the output features from the MMCVBorderAlign operation for given input features with shape [N, 4C, H, W] and corresponding boxes?

[N, C, H*W, 4]

What aspect of the MMCVBorderAlign operation determines the number of sampled positions along the border lines of bounding boxes?

pool_size

What is the procedure to extend the functionality of mmcv.cnn with a custom layer using the registry mechanism?

1. Write and register your custom module using the `MODELS` registry. 2. Import your custom module where needed. 3. Use the appropriate build function (e.g., `build_upsample_layer`) with your custom type.

What is the only stride value supported by the masked 2D convolution in mmcv?

1

What is the required minimum version of CUDA compiler driver (nvcc) to build MMCV on Linux?

9.2

What is the shape of the output tensor when using the MMCVPSAMask operator on an input feature with a batch size of N, height H, and width W?

(N, H * W, H, W)

What might cause the error "RuntimeError: Expected to have finished reduction in the prior iteration before starting a new one" when using DDP mode?

Running different branches in your code in DDP mode.

What is the output shape of the RoI pooled feature map given `num_rois`, `C`, `output_height`, and `output_width`?

(num_rois, C, output_height, output_width)

Which operator in mmcv is involved in region of interest pooling and comes with type constraints according to version v2.0.0?

MMCVDeformRoIPool

What type of output tensor shape does the `grid_sampler` operation produce and what is the type constraint for the input tensor?

Output tensor shape: (N, C, outH, outW) Type constraint for input tensor: tensor(float32, Linear)

What argument must be set to true in `mmcv.imresize` when you need the scaling factor along with the resized image?

return_scale

What happens if a data transformation class marked with `avoid_cache_randomness` is used with `TransformBroadcaster` when `share_random_params` is set to True in mmcv version as per the latest document?

An exception is thrown.

What determines whether the extrema are considered as referring to the center points or corner points of the input's corner pixels in the context of grid sampling?

`align_corners`

What configuration change is recommended to solve the "RuntimeError: Expected to have finished reduction in the prior iteration before starting a new one" error when running code in DDP mode in mmcv version mentioned in the document?

find_unused_parameters = True

What is the output shape of the `MMCVModulatedDeformConv2d` operation when applied to an input feature of shape (N, C, inH, inW)?

(N, output_channel, outH, outW)

Which function in MMCV is used to convert a BGR image to an HLS image, and which module is it part of?

bgr2hls, mmcv.image

What MMCV function is specifically used to resize images while maintaining the aspect ratio, and which module is it part of?

`imrescale` in `mmcv.image`

What is the shape of the pooled features output by the MMCVBorderAlign operation in the mmcv library's current version?

[N, C, H\*W, 4]

Which formatter is used for enforcing a consistent style in Markdown files according to the PR guidelines?

mdformat

What is the constraint on the stride parameter for MMCVMaskedConv2d in mmcv?

stride=1

In the context of MMCVModulatedDeformConv2d, which parameter determines how the input feature is divided into groups, and what condition must be satisfied for this division?

`groups`; `input_channel` should be divisible by `groups`.

Which module in mmengine.strategy provides support for distributed training using multiple devices?

DDPStrategy

In the context of MMCVRoIAlign and MMCVRoIAlignRotated, what mode is used for pooling in each bin?

avg or max

When using the NonMaxSuppression method with a max_output_boxes_per_class set to 0, how does the number of output boxes compare to the number of input boxes?

The number of output boxes is equal to the number of input boxes.

What command verifies successful installation of mmcv after building it from source?

python .dev_scripts/check_installation.py

What is the purpose of mypy in the MMCV development process?

Static type checking.

What tool is recommended to fix common misspellings in text files for Python development according to the document?

codespell

Which formatter is used for Python docstrings in the specified coding guidelines?

docformatter