DRAFT: Trying to improve pycbc_inspiral performance on GPU by spxiwh · Pull Request #5249 · gwastro/pycbc

spxiwh · 2026-01-09T13:28:38Z

@xangma and I have been playing about on and off in the last year with trying to make pycbc_inspiral perform more efficiently on the GPU (largely also using this as a test of how powerful AI-assisted tools can be for stuff like this). We've moved to a batch model (where we process O(100) templates at the same time) and have a number of additional cupy codes and underlying cupy-cuda kernels to enable this in here. The code agrees within the sort of precision we might expect for GPU vs CPU with the standard example/inspiral test code.

HOWEVER, while this is better than what we've had before, we're nowhere near the "ideal" speedups that GPUs should be able to achieve. (I reckon we're seeing about a speedup between 10 and 100x relative to running on a single CPU, but there's all sorts of caveats in those numbers and I don't want to present this as a "proper" benchmark). However, all the code is here if someone else wants to try this out! Just run examples/inspiral/run_cupy.sh. We pulled out most of the main processing loop into a inspiralutils.py to make it easier to poke about with this. (Some bits are not yet there, e.g. we only have TaylorF2 waveforms working, and not all of the less-used chi-squared tests are batched, but this could be added if this would work in principle).

SO @ahnitz @josh-willis @titodalcanton if this is of interest or useful, please feel free to have a look over. Currently the chi-squared seems to be the slowest part of it, but I don't think I know enough about what the GPU actually does under the hood to properly have a sense of where to focus further optimization.

... I seem to have this mixed up with some LISA code as well. I won't clean that up at this point unless we think this gets to the point where it's worth merging.

* add forward zeros after fft * fix generate data lisa ew * remove tlen from PSD function * add doc-string for data generation * update LISA EW model with changes * handle zero noise flag * move PSD functions to pycbc.psd and update to match utils * update model to use reworked PSD functions * remove unused generator object * comment about HorizonDistance * refactor pre-merger waveform to use a common function * rename early_warning_wform.py to pre_merger_waveform.py * update pre-merger model to use update waveform functions * fix broken import * rename lisa_ew.py to lisa_pre_merger.py * rename lisa_ew model to lisa_pre_merger * remove unused imports from lisa_pre_merger.py * remove phase marginalization * update __init__.py for pre-merger renaming * add some basic logging statements * fix return value for apply_pre_merger_kernel * do not copy output for TD to FD in apply_pre_merger_kernel * clarify doc-strings for generate_data_lisa_pre_merger * fix duration and cutoff time in pre-merger model * remove `kernel_length` from `generate_waveform_lisa_pre_merger` * correct doc-string for `generate_pre_merger_psds` * fix doc-strings for PSDFirKernel * raise NotImplementedError when calling `set_phase` * split data generation and pre-processing into two functions * fix missing dict keys in pre-process data * update inference code to read hdf files directly * fix coalescence time

Co-authored-by: Gareth S Cabourn Davies <gareth.cabourndavies@ligo.org>

* fix calcluation of cutoff_deltat Set both forward zeros and cutoff time based on dt * add inline comments explaining times

Changes to make threshold work and CPU work

Avoid memory copies in threshold

Copilot

Pull request overview

This PR attempts to improve GPU performance for pycbc_inspiral by implementing batched template processing using CuPy. The approach processes ~100 templates simultaneously on GPU with custom CUDA kernels for template generation, matched filtering, and chi-squared computation.

Key changes:

Batched template generation via spa_tmplt_batch with CUDA kernels
GPU-optimized matched filtering and chi-squared computation
New inspiral_utils.py with batched GPU workflow

Reviewed changes

Copilot reviewed 21 out of 24 changed files in this pull request and generated 33 comments.

Show a summary per file

File	Description
bin/pycbc_inspiral	Modified to support GPU batch processing; contains hardcoded path and incomplete implementation
pycbc/filter/inspiral_utils.py	New GPU-batched template processing engine
pycbc/waveform/spa_tmplt*.py	Batched TaylorF2 template generation with CUDA kernels
pycbc/vetoes/chisq_cupy.py	Batched chi-squared computation with optimized CUDA kernels
pycbc/filter/matchedfilter.py	Extended for batch operations on GPU
pycbc/waveform/pre_merger_waveform.py	Unrelated LISA code mixed into PR
pycbc/psd/lisa_pre_merger.py	Unrelated LISA code mixed into PR
pycbc/inference/models/lisa_pre_merger.py	Unrelated LISA code mixed into PR
requirements.txt	Commented out markupsafe constraint
pycbc/scheme.py	CUPYScheme initialization updates
pycbc/types/array*.py	CuPy array type handling
pycbc/events/threshold_cupy.py	Batched thresholding with CUDA kernels
examples/inspiral/run_cupy.sh	Example script for GPU execution

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-01-09T13:37:58Z

pycbc/inference/models/brute_marg.py

+    # EW CODE SWAP
+    # def _loglikelihood(self):


Comments like "EW CODE SWAP" suggest this is development/experimental code that shouldn't be merged in this state. These commented-out alternatives should either be removed or properly documented if they're needed for future work.

Suggested change

# EW CODE SWAP

# def _loglikelihood(self):

# Implement the log-likelihood via `_loglr`, consistent with the

# underlying model interface used by this class.

Copilot · 2026-01-09T13:37:59Z

pycbc/filter/matchedfilter.py

 import logging
 from math import sqrt
 import numpy
+import cupy


Importing cupy globally without checking if it's available could cause import errors on systems without cupy installed. The import should be conditional or wrapped in a try/except block that provides a helpful error message.

Suggested change

import cupy

try:

import cupy

except ImportError as e:

raise ImportError(

"The 'cupy' package is required for GPU-accelerated matched filtering "

"in 'pycbc.filter.matchedfilter'. Please install cupy (see "

"https://docs.cupy.dev/ for installation instructions) or configure "

"PyCBC to use a CPU-only backend."

) from e

Copilot · 2026-01-09T13:37:59Z

pycbc/events/threshold_cupy.py

+    raise NotImplementedError("Needs writing properly")
+    # Not sure this function is accessed easily. However, right now, it has not
+    # been properly written. A starting point is here though!


This comment says "Needs writing properly" and raises NotImplementedError, indicating incomplete implementation. The function should either be completed or removed before merging.

Suggested change

raise NotImplementedError("Needs writing properly")

# Not sure this function is accessed easily. However, right now, it has not

# been properly written. A starting point is here though!

Copilot · 2026-01-09T13:37:59Z

pycbc/filter/matchedfilter.py

        snrv, idx = self.threshold_and_clusterers[segnum].threshold_and_cluster(self.snr_threshold / norm, window)

        if len(idx) == 0:
+            print("NO TRIGGERS")


This debug print statement should be removed or replaced with proper logging before merging.

Suggested change

print("NO TRIGGERS")

logger.info("NO TRIGGERS")

Copilot · 2026-01-09T13:38:00Z

pycbc/inference/models/lisa_pre_merger.py

+# Copyright (C) 2018  Collin Capano
+# This program is free software; you can redistribute it and/or modify it
+# under the terms of the GNU General Public License as published by the
+# Free Software Foundation; either version 3 of the License, or (at your
+# option) any later version.
+#
+# This program is distributed in the hope that it will be useful, but
+# WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General
+# Public License for more details.
+#
+# You should have received a copy of the GNU General Public License along
+# with this program; if not, write to the Free Software Foundation, Inc.,
+# 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
+
+"""
+This modules provides models that have analytic solutions for the
+log likelihood.
+"""
+
+import copy
+import logging
+
+import pycbc.types
+
+from .base import BaseModel
+
+import pycbc.psd
+
+from pycbc.waveform.pre_merger_waveform import (
+    pre_process_data_lisa_pre_merger,
+    generate_waveform_lisa_pre_merger,
+)
+from pycbc.psd.lisa_pre_merger import generate_pre_merger_psds
+from pycbc.waveform.waveform import parse_mode_array
+from pycbc.waveform.utils import apply_fseries_time_shift
+from .tools import marginalize_likelihood
+
+
+class LISAPreMergerModel(BaseModel):
+    r"""Model for pre-merger inference in LISA.
+
+    Parameters
+    ----------
+    variable_params : (tuple of) string(s)
+        A tuple of parameter names that will be varied.
+    static_params: Dict[str: Any]
+        Dictionary of static parameters used for waveform generation.
+    psd_file : str
+        Path to the PSD file. Uses the same PSD file for LISA_A and LISA_E
+        channels.
+    **kwargs :
+        All other keyword arguments are passed to ``BaseModel``.
+    """
+    name = "lisa_pre_merger"
+
+    def __init__(
+        self,
+        variable_params,
+        static_params=None,
+        psd_file=None,
+        **kwargs
+    ):
+        # Pop relevant values from kwargs
+        cutoff_time = int(kwargs.pop('cutoff_time'))
+        kernel_length = int(kwargs.pop('kernel_length'))
+        window_length = int(kwargs.pop('window_length'))
+        extra_forward_zeroes = int(kwargs.pop('extra_forward_zeroes'))
+        tlen = int(kwargs.pop('tlen'))
+        sample_rate = float(kwargs.pop('sample_rate'))
+        data_file = kwargs.pop('data_file')
+
+        # set up base likelihood parameters
+        super().__init__(variable_params, **kwargs)
+
+        self.static_params = parse_mode_array(static_params)
+
+        if psd_file is None:
+            raise ValueError("Must specify a PSD file!")
+
+        # Zero phase PSDs for whitening
+        # Only store the frequency-domain PSDs
+        logging.info("Generating pre-merger PSDs")
+        self.whitening_psds = {}
+        self.whitening_psds['LISA_A'] = generate_pre_merger_psds(
+            psd_file,
+            sample_rate=sample_rate,
+            duration=tlen,
+            kernel_length=kernel_length,
+        )["FD"]
+        self.whitening_psds['LISA_E'] = generate_pre_merger_psds(
+            psd_file,
+            sample_rate=sample_rate,
+            duration=tlen,
+            kernel_length=kernel_length,
+        )["FD"]
+
+        # Store data for doing likelihoods.
+        self.kernel_length = kernel_length
+        self.window_length = window_length
+        self.sample_rate = sample_rate
+        self.cutoff_time = cutoff_time
+        self.extra_forward_zeroes = extra_forward_zeroes
+        self.tlen = tlen
+
+        # Load the data from the file
+        data = {}
+        for channel in ["LISA_A", "LISA_E"]:
+            data[channel] = pycbc.types.timeseries.load_timeseries(
+                data_file,
+                group=f"/{channel}",
+            )
+
+        # Pre-process the pre-merger data
+        # Returns time-domain data
+        # Uses UIDs: 4235, 4236
+        logging.info("Pre-processing pre-merger data")
+        pre_merger_data = pre_process_data_lisa_pre_merger(
+            data,
+            sample_rate=sample_rate,
+            psds_for_whitening=self.whitening_psds,
+            window_length=0, 
+            cutoff_time=self.cutoff_time,
+            forward_zeroes=self.kernel_length,
+        )
+
+        self.lisa_a_strain = pre_merger_data["LISA_A"]
+        self.lisa_e_strain = pre_merger_data["LISA_E"]
+
+        # Frequency-domain data for computing log-likelihood
+        self.lisa_a_strain_fd = pycbc.strain.strain.execute_cached_fft(
+            self.lisa_a_strain,
+            copy_output=True,
+            uid=3223965
+        )
+        self.lisa_e_strain_fd = pycbc.strain.strain.execute_cached_fft(
+            self.lisa_e_strain,
+            copy_output=True,
+            uid=3223967
+        )
+        # Data epoch
+        self._epoch = self.lisa_a_strain_fd._epoch
+
+    def get_waveforms(self, params):
+        """Generate the waveforms given the parameters.
+
+        Note: `params` should already include the static parameters.
+        """
+        dt = params["tc"] - self._epoch
+        # Time between the end of the data and the time of coalescence
+        dt_end = params.get("cutoff_deltat", self.tlen - dt)
+        # Actual time between the end of the data and the cutoff time
+        # since cutoff time is specified relative to the merger
+        cutoff_time = self.cutoff_time - dt_end
+        # Additional zeros at the beginning of the data, these:
+        # - manually specified zeroes
+        # - kernel length zeros
+        # - zeros that will be wrapped around when the data is shifted
+        forward_zeroes = (
+            self.extra_forward_zeroes
+            + self.kernel_length
+            + int(dt_end * self.sample_rate)
+        )
+        # Generate the pre-merger waveform
+        # These waveforms are whitened
+        # Uses UIDs: 1235(0), 1236(0)
+        ws = generate_waveform_lisa_pre_merger(
+            params,
+            psds_for_whitening=self.whitening_psds,
+            window_length=self.window_length,
+            sample_rate=self.sample_rate,
+            cutoff_time=cutoff_time,
+            forward_zeroes=forward_zeroes,
+        )
+
+        wf = {}
+        # Adjust epoch to match data and shift merger to the
+        # correct time.
+        # Can safely set copy=False since ws won't be used again.
+        for channel in ws.keys():
+            wf[channel] = apply_fseries_time_shift(
+                ws[channel], float(dt), copy=False,
+            )
+            wf[channel]._epoch = self._epoch
+        return wf
+
+    def _loglikelihood(self):
+        """Compute the pre-merger log-likelihood."""
+        cparams = copy.deepcopy(self.static_params)
+        cparams.update(self.current_params)
+
+        # Generate the waveforms
+        wforms = self.get_waveforms(cparams)
+
+        # Compute <h|d> for each channel
+        snr_A = pycbc.filter.overlap_cplx(
+            wforms["LISA_A"],
+            self.lisa_a_strain_fd,
+            normalized=False,
+        )
+        snr_E = pycbc.filter.overlap_cplx(
+            wforms["LISA_E"],
+            self.lisa_e_strain_fd,
+            normalized=False,
+        )
+        # Compute <h|h> for each channel
+        a_norm = pycbc.filter.sigmasq(wforms["LISA_A"])
+        e_norm = pycbc.filter.sigmasq(wforms["LISA_E"])
+
+        hs = snr_A + snr_E
+        hh = (a_norm + e_norm)
+
+        return marginalize_likelihood(complex(hs), float(hh), phase=False)


This file contains LISA-specific inference model code that is unrelated to the GPU performance optimization goals of this PR. It should be removed.

Copilot · 2026-01-09T13:38:06Z

pycbc/filter/matchedfilter.py

-            # setup up the ifft we will do
-            self.ifft = IFFT(self.corr_mem, self.snr_mem)
+        # Detect if we're using CUPY scheme
+        import pycbc.scheme


This import of module pycbc.scheme is redundant, as it was previously imported on line 37.

Suggested change

import pycbc.scheme

Copilot · 2026-01-09T13:38:06Z

pycbc/fft/cupyfft.py

+    else:
+        raise ValueError(_INV_FFT_MSG.format("IFFT", itype, otype))
+
+class BatchFFT(_BaseFFT):


This class does not call _BaseFFT.init during initialization. (BatchFFT.init may be missing a call to a base class init)

Copilot · 2026-01-09T13:38:06Z

pycbc/fft/cupyfft.py

+
+class BatchIFFT(_BaseIFFT):
+    """Class for performing batched IFFTs via the cupy interface"""
+    def __init__(self, invecs, outvecs, batch_size):


This class does not call _BaseIFFT.init during initialization. (BatchIFFT.init may be missing a call to a base class init)

Suggested change

def __init__(self, invecs, outvecs, batch_size):

def __init__(self, invecs, outvecs, batch_size):

# Initialize base _BaseIFFT with a representative pair and batch size

super(BatchIFFT, self).__init__(invecs[0], outvecs[0], nbatch=batch_size, size=None)

Copilot · 2026-01-09T13:38:07Z

pycbc/waveform/bank.py

+                is_cupy = hasattr(test_arr._data, 'device')
+                use_batched_gen = all_spatmplt and no_compression and is_cupy
+            except ImportError:
+                pass


'except' clause does nothing but pass and there is no explanatory comment.

Suggested change

pass

# Optional GPU/CuPy support is not available; fall back to non-batched generation.

logging.debug("CuPy/CUPYScheme not available; using non-batched waveform generation.")

Copilot · 2026-01-09T13:38:07Z

pycbc/psd/lisa_pre_merger.py

+        kernel, latency, sample_rate = self.psd_to_linear_phase_whitening_fir_kernel(psd)
+        kernel, phase = self.linear_phase_fir_kernel_to_minimum_phase_whitening_fir_kernel(kernel, sample_rate)
+
+        # get merger model for SNR = 1.
+        f_psd = psd.f0 + np.arange(len(psd.data.data)) * psd.deltaF
+        horizon_distance = HorizonDistance(f_low, f_psd[-1], psd.deltaF, m1, m2)
+        f_model, model= horizon_distance(psd, 1.)[1]
+
+        # find the range of frequency bins covered by the merger
+        # model
+        kmin, kmax = f_psd.searchsorted(f_model[0]), f_psd.searchsorted(f_model[-1]) + 1
+
+        # compute SNR=1 model's (d SNR^2 / df) spectral density
+        unit_snr2_density = np.zeros_like(phase)
+        unit_snr2_density[kmin:kmax] = model / psd.data.data[kmin:kmax]
+
+        # integrate across each frequency bin, converting to
+        # snr^2/bin.  NOTE:  this step is here for the record, but
+        # is commented out because it has no effect on the result
+        # given the renormalization that occurs next.
+        #unit_snr2_density *= psd.deltaF
+
+        # take 16th root, then normalize so max=1.  why?  I don't
+        # know, just feels good, on the whole.
+        unit_snr2_density = unit_snr2_density**(1./16)
+        unit_snr2_density /= unit_snr2_density.max()
+
+        # record phase vector and SNR^2 density vector
+        self.target_phase = phase
+        self.target_phase_mask = unit_snr2_density
+


This statement is unreachable.

Suggested change

kernel, latency, sample_rate = self.psd_to_linear_phase_whitening_fir_kernel(psd)

kernel, phase = self.linear_phase_fir_kernel_to_minimum_phase_whitening_fir_kernel(kernel, sample_rate)

# get merger model for SNR = 1.

f_psd = psd.f0 + np.arange(len(psd.data.data)) * psd.deltaF

horizon_distance = HorizonDistance(f_low, f_psd[-1], psd.deltaF, m1, m2)

f_model, model= horizon_distance(psd, 1.)[1]

# find the range of frequency bins covered by the merger

# model

kmin, kmax = f_psd.searchsorted(f_model[0]), f_psd.searchsorted(f_model[-1]) + 1

# compute SNR=1 model's (d SNR^2 / df) spectral density

unit_snr2_density = np.zeros_like(phase)

unit_snr2_density[kmin:kmax] = model / psd.data.data[kmin:kmax]

# integrate across each frequency bin, converting to

# snr^2/bin. NOTE: this step is here for the record, but

# is commented out because it has no effect on the result

# given the renormalization that occurs next.

#unit_snr2_density *= psd.deltaF

# take 16th root, then normalize so max=1. why? I don't

# know, just feels good, on the whole.

unit_snr2_density = unit_snr2_density**(1./16)

unit_snr2_density /= unit_snr2_density.max()

# record phase vector and SNR^2 density vector

self.target_phase = phase

self.target_phase_mask = unit_snr2_density

spxiwh and others added 30 commits December 5, 2024 09:47

Temp playing

270956e

Some more hacks to do LISA EW PE

bfc9ac8

Missing new files!!

ce796ab

Some changes

9b45879

Fixing early warning PE

0058ee1

One more fix

0610738

Some pokes in brute_marg

42e365e

phase marg and modes for LISA EW likelihood

ed4c514

add option to specify PSD path

32fb2f7

update for BBHx plugin and to enable the use of extra_forward_zeroes

a09661c

clean up early warning code

7340fe4

fix psd duration

a19fa3d

add PSD lower frequency cutoff

a68d702

manually cut psd if low freq cutoff is specified

4dbd198

remove mode array and phenomd config

0ec6cc3

handle case where psd low freq cutoff is not specified

d8e492c

remove incorrect PSD lower frequency cutoff

4f94ccb

Fix time shift in inference (#3)

e6f23ee

changes to kernel length

c763c57

avoid extra FFTs

053dcb0

update comments

daece7f

add cutoff_deltat parameter

cd1e2b4

Update doc-string

f268a36

Co-authored-by: Gareth S Cabourn Davies <gareth.cabourndavies@ligo.org>

make cutoff_deltat optional

eaaa0c2

fix calcluation of cutoff_deltat (#6)

1414a23

* fix calcluation of cutoff_deltat Set both forward zeros and cutoff time based on dt * add inline comments explaining times

Some tweaks to premerger likelihood to allow cupy

b6d568e

Force these to floats/complex

0f44fbc

batching partially working

30dc864

corr and ifft results populated and matching

81ed763

xangma and others added 15 commits December 9, 2024 15:27

batched threshold cupy is wrong, but it does run ...

c10adc1

CHanges to make threshold work and CPU work

045a65d

Merge pull request #1 from spxiwh/gpu_dev

0b6b8c8

Changes to make threshold work and CPU work

fixes so results now match cpu results (further verification needed)

3464a92

Filter the -1s out with cupy kernel

f91cb5a

type changes in chisq

2494d6f

Avoid memory copies in threshold

db15eb6

Merge pull request #2 from spxiwh/gpu_dev

bcb2703

Avoid memory copies in threshold

Latest modifications with copilot help

dfc144d

Here we break pycbc_inspiral to do a focused GPU implementation

6b4a2e0

Continuing to work on inspiral_utils.py

83e6508

Fixing the waveform generation to avoid moving stuff around

293dbbd

Checkpoint at faster output + sigmasq ... Chisq next!

f5fe3b5

Some profiling code

8af5398

Add test script

8991665

Copilot AI review requested due to automatic review settings January 9, 2026 13:28

Copilot started reviewing on behalf of spxiwh January 9, 2026 13:29 View session

Copilot AI reviewed Jan 9, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Comments

DRAFT: Trying to improve pycbc_inspiral performance on GPU#5249

DRAFT: Trying to improve pycbc_inspiral performance on GPU#5249
spxiwh wants to merge 45 commits intogwastro:masterfrom
spxiwh:gpu_dev2

spxiwh commented Jan 9, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Jan 9, 2026

Uh oh!

Copilot AI Jan 9, 2026

Uh oh!

Copilot AI Jan 9, 2026

Uh oh!

Copilot AI Jan 9, 2026

Uh oh!

Copilot AI Jan 9, 2026

Uh oh!

Copilot AI Jan 9, 2026

Uh oh!

Copilot AI Jan 9, 2026

Uh oh!

Copilot AI Jan 9, 2026

Uh oh!

Copilot AI Jan 9, 2026

Uh oh!

Copilot AI Jan 9, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

-import cupy
+try:
+    import cupy
+except ImportError as e:
+    raise ImportError(
+        "The 'cupy' package is required for GPU-accelerated matched filtering "
+        "in 'pycbc.filter.matchedfilter'. Please install cupy (see "
+        "https://docs.cupy.dev/ for installation instructions) or configure "
+        "PyCBC to use a CPU-only backend."
+    ) from e

	raise NotImplementedError("Needs writing properly")
	# Not sure this function is accessed easily. However, right now, it has not
	# been properly written. A starting point is here though!

	pass
	# Optional GPU/CuPy support is not available; fall back to non-batched generation.
	logging.debug("CuPy/CUPYScheme not available; using non-batched waveform generation.")

Comments

Conversation

spxiwh commented Jan 9, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants