Feature support matrix

Note

This page is generated from dev/backend_support_data.json, which is produced by dev/backend_support_sweep.py — an automated sweep that invokes every public function in brainunit.math, brainunit.linalg, brainunit.fft, and every public Quantity method across each locally-installed backend and records the outcome. Re-run the sweep and the renderer to refresh this page.

Cell legend

Glyph	Meaning
✓	Verified: the call returned a value of the expected backend kind.
⊘	Skipped: the backend’s array-API surface does not expose the underlying op, or it rejects a keyword brainunit forwards (e.g. JAX-only precision=).
✗	Failed: the call raised an unexpected exception on this backend.
⚠	Works with a caveat (e.g. lazy result on dask, expected BackendError for materialization on dask).
🅙	JAX-only by design — gated by brainunit._jax_guard.require_jax_backend. Raises BackendError on any non-jax backend.
—	Not applicable to backend dispatch (dtype factories, dimension predicates).
?	Not tested in this report or not mapped by the automated sweep. The single unmapped Quantity method (tree_unflatten) is also ? because automated invocation requires a hand-crafted aux/children pair.

Sweep environment. Backends invoked: numpy, jax, cupy, torch, dask, ndonnx. Backends shown but not tested: none.

High-level summary

Per-subpackage rating

Subpackage	numpy	jax	cupy	torch	dask	ndonnx
brainunit.math	Mostly ⚠	Full ✓	Mostly ⚠	Partial ⚠	Partial ⚠	Partial ⚠
brainunit.linalg	Mostly ⚠	Full ✓	Mostly ⚠	Mostly ⚠	Partial ⚠	Limited ✗
brainunit.fft	Full ✓	Full ✓	Full ✓	Full ✓	Full ✓	Limited ✗
Quantity methods	Mostly ⚠	Mostly ⚠	Mostly ⚠	Partial ⚠	Partial ⚠	Partial ⚠
brainunit.lax	JAX-only 🅙	Full ✓	JAX-only 🅙	JAX-only 🅙	JAX-only 🅙	JAX-only 🅙
brainunit.autograd	JAX-only 🅙	Full ✓	JAX-only 🅙	JAX-only 🅙	JAX-only 🅙	JAX-only 🅙
brainunit.sparse	JAX-only 🅙	Full ✓	JAX-only 🅙	JAX-only 🅙	JAX-only 🅙	JAX-only 🅙

Rating thresholds: Full ≥ 95 % pass and zero fail; Mostly ≥ 80 % pass; Partial ≥ 30 % pass; Limited < 30 % pass; JAX-only = gated by require_jax_backend.

Backend-specific notes

• jax — full feature set; default backend. All JAX-only subpackages (brainunit.lax, brainunit.autograd, brainunit.sparse) require this backend.

• numpy — eager CPU computation through array_api_compat.numpy. A handful of reductions (amax, amin, mean, nan* variants) fail when brainunit forwards a where=None kwarg numpy can’t interpret. These are listed with footnotes in the math tables below.

• cupy — measured on a CUDA device through array_api_compat.cupy. CuPy 14.1.1; device NVIDIA GeForce RTX 3060 Laptop GPU; CUDA driver/runtime 13030 / 12090. Cells and summary ratings below are generated from the measured sweep.

• torch — through array_api_compat.torch. The torch array-API surface lacks several ops brainunit dispatches to (cbrt, digamma, some einops reductions, axes= for n-D FFTs) and rejects JAX-flavored kwargs (precision, symmetrize_input, tol). Affected calls are recorded as skip rather than fail.

• dask — lazy arrays. Reductions and most array ops succeed but the result remains lazy until .compute(). Per brainunit._base_quantity, the Python casts float(q) / int(q) / operator.index(q) / np.asarray(q) / hash(q) and the Quantity.tolist method raise BackendError to avoid silent materialization — these cells are ⚠ with the BackendError text in the footnote. Quantity.item on dask raises a different error (the dask Array has no .item() method) so it appears as ⊘ rather than ⚠. Methods like Quantity.float / .double are .astype in disguise and stay lazy on dask, so they pass.

• ndonnx — symbolic graph-building backend. Many array-API ops (fft.*, several linalg.*, complex / specialty math) are not implemented and surface as ⊘ skip rows. Saiunit does not encode unit information into the ONNX graph.

JAX-only subpackages

These subpackages dispatch directly to JAX primitives that have no array-API equivalent. Each entry point is wrapped with brainunit._jax_guard.require_jax_backend, which raises BackendError on any non-jax mantissa.

brainunit.lax — 101 public callable(s); all require jax.

Probe result	numpy	jax	cupy	torch	dask	ndonnx
all functions	🅙	✓	🅙	🅙	🅙	🅙

List of brainunit.lax functions

acos acosh approx_max_k approx_min_k asin asinh atan atan2 atanh batch_matmul bessel_i0e bessel_i1e betainc bitcast_convert_type broadcast broadcast_in_dim broadcast_shapes broadcast_to_rank broadcasted_iota cholesky clamp clz collapse complex conv conv_transpose

convert_element_type cumlogsumexp cummax cummin cumsum digamma div dot_general dynamic_index_in_dim dynamic_slice dynamic_slice_ind_dim dynamic_update_index_in_dim dynamic_update_slice dynamic_update_slice_in_dim eig eigh eq erf erf_inv erfc fft gather ge gt hessenberg

householder_product igamma igamma_grad_a igammac index_in_dim index_take integer_pow iota le lgamma logistic lt lu mul ne neg pad polygamma population_count pow qdwh qr random_gamma_grad reduce reduce_precision

rem rsqrt scatter scatter_add scatter_apply scatter_max scatter_min scatter_mul scatter_sub schur shift_left shift_right_arithmetic shift_right_logical slice slice_in_dim sort sort_key_val sub svd top_k triangular_solve tridiagonal tridiagonal_solve zeros_like_array zeta

brainunit.autograd — 7 public callable(s); all require jax.

Probe result	numpy	jax	cupy	torch	dask	ndonnx
all functions	✓	✓	✗	✗	✗	✗

List of brainunit.autograd functions

grad hessian

jacfwd jacobian

jacrev value_and_grad

vector_grad

brainunit.sparse — 10 public callable(s); all require jax.

Probe result	numpy	jax	cupy	torch	dask	ndonnx
all functions	🅙	✓	🅙	🅙	🅙	✗

List of brainunit.sparse functions

COO CSC CSR

SparseMatrix coo_fromdense coo_todense

csc_fromdense csc_todense

csr_fromdense csr_todense

brainunit.math

Public callables in brainunit.math that go through the multi-backend dispatcher. Grouped by the same unit semantics used by the math API reference.

Array Creation and Conversion

brainunit.math — Array Creation and Conversion

Function	numpy	jax	cupy	torch	dask	ndonnx
full	✓	✓	✓	✓	✓	✓
eye	✓	✓	✓	✓	✓	✓
identity	✓	✓	✓	⊘ [1]	⊘ [2]	⊘ [3]
tri	✓	✓	✓	⊘ [4]	✓	⊘ [5]
empty	✓	✓	✓	✓	✓	✓
ones	✓	✓	✓	✓	✓	✓
zeros	✓	✓	✓	✓	✓	✓
full_like	✓	✓	✓	✓	✓	⊘ [6]
diag	✓	✓	✓	✓	✓	⊘ [7]
tril	✓	✓	✓	✓	✓	✓
triu	✓	✓	✓	✓	✓	✓
empty_like	✓	✓	✓	✓	✓	⊘ [8]
ones_like	✓	✓	✓	✓	✓	⊘ [9]
zeros_like	✓	✓	✓	✓	✓	⊘ [10]
fill_diagonal	✓	✓	✓	⊘ [11]	⊘ [12]	⊘ [13]
array	✓	✓	✓	✓	✓	✓
asarray	✓	✓	✓	✓	✓	✓
arange	✓	✓	✓	✓	✓	✓
linspace	✓	✓	✓	✓	✓	⊘ [14]
logspace	✓	✓	✓	⊘ [15]	⊘ [16]	⊘ [17]
meshgrid	✓	✓	✓	✓	✓	⊘ [18]
vander	✓	✓	✓	✓	⊘ [19]	⊘ [20]
tril_indices	✓	✓	✓	✓	✓	✓
tril_indices_from	✓	✓	✓	⊘ [21]	✓	⊘ [22]
triu_indices	✓	✓	✓	✓	✓	✓
triu_indices_from	✓	✓	✓	⊘ [23]	✓	⊘ [24]
from_numpy	✓	✓	✓	✓	✓	✓
as_numpy	✓	✓	✓	✓	✓	✓
tree_ones_like	✓	✓	✓	✓	✓	⊘ [9]
tree_zeros_like	✓	✓	✓	✓	✓	⊘ [10]

Unit-preserving Operations

brainunit.math — Unit-preserving Operations

Function	numpy	jax	cupy	torch	dask	ndonnx
row_stack	✓	✓	✓	✓	✓	⊘ [25]
concatenate	✓	✓	✓	✓	✓	⊘ [26]
stack	✓	✓	✓	✓	✓	✓
vstack	✓	✓	✓	✓	✓	⊘ [25]
hstack	✓	✓	✓	✓	✓	⊘ [27]
dstack	✓	✓	✓	✓	✓	⊘ [28]
column_stack	✓	✓	✓	✓	⊘ [29]	⊘ [30]
block	✓	✓	⊘ [31]	⊘ [32]	✓	⊘ [33]
append	✓	✓	✓	⊘ [34]	✓	⊘ [35]
split	✓	✓	✓	⊘ [36]	⊘ [37]	⊘ [38]
array_split	✓	✓	✓	⊘ [36]	⊘ [37]	⊘ [38]
dsplit	✓	✓	✓	✓	⊘ [39]	⊘ [40]
hsplit	✓	✓	✓	✓	⊘ [41]	⊘ [42]
vsplit	✓	✓	✓	✓	⊘ [43]	⊘ [44]
atleast_1d	✓	✓	✓	✓	✓	⊘ [45]
atleast_2d	✓	✓	✓	✓	✓	⊘ [46]
atleast_3d	✓	✓	✓	✓	✓	⊘ [47]
broadcast_arrays	✓	✓	✓	✓	✓	✓
broadcast_to	✓	✓	✓	✓	✓	✓
reshape	✓	✓	✓	✓	✓	⊘ [48]
moveaxis	✓	✓	✓	✓	✓	✓
transpose	✓	✓	✓	⊘ [49]	✓	⊘ [50]
swapaxes	✓	✓	✓	✓	✓	⊘ [51]
tile	✓	✓	✓	✓	✓	✓
repeat	✓	✓	✓	✓	✓	✓
flip	✓	✓	✓	✓	✓	✓
fliplr	✓	✓	✓	✓	✓	⊘ [52]
flipud	✓	✓	✓	✓	✓	⊘ [53]
roll	✓	✓	✓	✓	✓	✓
expand_dims	✓	✓	✓	✓	✓	✓
squeeze	✓	✓	✓	⊘ [54]	✓	⊘ [54]
sort	✓	✓	✓	✓	✓	✓
max	✓	✓	✓	✓	✓	✓
min	✓	✓	✓	✓	✓	✓
amax	✓	✓	✓	✓	✓	✓
amin	✓	✓	✓	✓	✓	✓
diagflat	✓	✓	✓	✓	⊘ [55]	⊘ [56]
diagonal	✓	✓	✓	✓	✓	⊘ [57]
choose	✓	✓	⊘ [58]	⊘ [59]	⊘ [60]	⊘ [61]
ravel	✓	✓	✓	✓	⊘ [62]	⊘ [63]
flatten	✓	✓	✓	✓	✓	✓
unflatten	✓	✓	✓	✓	✓	✓
remove_diag	✓	✓	✓	✓	✓	✓
astype	✓	✓	✓	✓	✓	✓
real	✓	✓	✓	✓	✓	✗ [64]
imag	✓	✓	✓	✓	✓	✗ [64]
conj	✓	✓	✓	✓	✓	✗ [64]
conjugate	✓	✓	✓	⊘ [65]	⊘ [66]	✗ [64]
negative	✓	✓	✓	✓	✓	✓
positive	✓	✓	✓	✓	✓	✓
abs	✓	✓	✓	✓	✓	✓
sum	✓	✓	✓	✓	✓	✓
nancumsum	✓	✓	✓	⊘ [67]	✓	⊘ [68]
nansum	✓	✓	✓	✓	✓	⊘ [69]
cumsum	✓	✓	✓	✓	✓	⊘ [70]
ediff1d	✓	✓	✓	⊘ [71]	✓	⊘ [72]
absolute	✓	✓	✓	✓	✓	⊘ [73]
fabs	✓	✓	✓	⊘ [74]	✓	⊘ [75]
median	✓	✓	✓	⊘ [76]	⊘ [77]	⊘ [78]
nanmin	✓	✓	✓	⊘ [79]	✓	⊘ [80]
nanmax	✓	✓	✓	⊘ [81]	✓	⊘ [82]
ptp	✓	✓	✓	⊘ [83]	⊘ [84]	⊘ [85]
average	✓	✓	✓	⊘ [86]	✓	⊘ [87]
mean	✓	✓	✓	✓	✓	✓
std	✓	✓	✓	✓	✓	✓
nanmedian	✓	✓	✓	⊘ [88]	⊘ [89]	⊘ [90]
nanmean	✓	✓	✓	✓	✓	⊘ [91]
nanstd	✓	✓	✓	⊘ [92]	✓	⊘ [93]
diff	✓	✓	✓	✓	✓	✓
rot90	✓	✓	✓	✓	✓	⊘ [94]
intersect1d	✓	✓	✓	⊘ [95]	⊘ [96]	⊘ [97]
nan_to_num	✓	✓	✓	✓	✓	⊘ [98]
percentile	✓	✓	✓	⊘ [99]	✓	⊘ [100]
nanpercentile	✓	✓	⊘ [101]	⊘ [102]	✓	⊘ [103]
quantile	✓	✓	✓	⊘ [104]	✓	⊘ [105]
nanquantile	✓	✓	⊘ [106]	⊘ [107]	✓	⊘ [108]
round	✓	✓	✓	✓	✓	⊘ [109]
around	✓	✓	✓	⊘ [110]	✓	⊘ [111]
rint	✓	✓	✓	⊘ [112]	✓	⊘ [113]
floor	✓	✓	✓	✓	✓	✓
ceil	✓	✓	✓	✓	✓	✓
trunc	✓	✓	✓	✓	✓	✓
fix	✓	✓	✓	✓	✓	✓
modf	✓	✓	✓	⊘ [114]	✓	⊘ [115]
fmod	✓	✓	✓	✓	✓	⊘ [116]
mod	✓	✓	✓	⊘ [117]	✓	⊘ [118]
copysign	✓	✓	✓	✓	✓	✗ [119]
remainder	✓	✓	✓	✓	✓	✓
maximum	✓	✓	✓	✓	✓	✓
minimum	✓	✓	✓	✓	✓	✓
fmax	✓	✓	✓	✓	✓	⊘ [120]
fmin	✓	✓	✓	✓	✓	⊘ [121]
lcm	✓	✓	✓	✓	⊘ [122]	⊘ [123]
gcd	✓	✓	✓	✓	⊘ [124]	⊘ [125]
trace	✓	✓	✓	✓	✓	⊘ [126]
add	✓	✓	✓	✓	✓	✓
subtract	✓	✓	✓	✓	✓	✓
nextafter	✓	✓	✓	✓	✓	✗ [119]
promote_dtypes	✓	✓	✓	✓	✓	✓
interp	✓	✓	✓	⊘ [127]	⊘ [128]	⊘ [129]
clip	✓	✓	✓	✓	✓	✓
histogram	✓	✓	✓	✓	✗ [130]	⊘ [131]
compress	✓	✓	✓	⊘ [132]	✓	⊘ [133]
extract	✓	✓	✓	⊘ [134]	✓	⊘ [135]
take	⊘ [136]	✓	⊘ [136]	✓	⊘ [136]	⊘ [136]
select	✓	✓	✓	✗ [137]	✓	⊘ [138]
where	✓	✓	✓	✓	✓	✓
unique	✓	✓	✓	✓	⊘ [139]	⊘ [140]
gather	✓	✓	✓	✓	✗ [141]	⊘ [142]
ldexp	✓	✓	✓	✓	✓	⊘ [143]

Unit-changing Operations

brainunit.math — Unit-changing Operations

Function	numpy	jax	cupy	torch	dask	ndonnx
reciprocal	✓	✓	✓	✓	✓	✓
prod	✓	✓	✓	✓	✓	✓
product	✓	✓	✓	✓	✓	✓
nancumprod	✓	✓	✓	⊘ [144]	✓	⊘ [145]
nanprod	✓	✓	✓	⊘ [146]	✓	⊘ [147]
cumprod	✓	✓	✓	✓	✓	⊘ [148]
cumproduct	✓	✓	✓	✓	✓	⊘ [148]
var	✓	✓	✓	✓	✓	✓
nanvar	✓	✓	✓	⊘ [149]	✓	⊘ [150]
cbrt	✓	✓	✓	⊘ [151]	✓	⊘ [152]
square	✓	✓	✓	✓	✓	✓
sqrt	✓	✓	✓	✓	✓	✓
multiply	✓	✓	✓	✓	✓	✓
divide	✓	✓	✓	✓	✓	✓
power	✓	✓	✓	⊘ [153]	✓	⊘ [154]
cross	✓	✓	✓	✓	⊘ [155]	⊘ [156]
true_divide	✓	✓	✓	✓	✓	⊘ [157]
floor_divide	✓	✓	✓	✓	✓	✓
float_power	✓	✓	✓	✓	✓	⊘ [158]
divmod	✓	✓	✓	⊘ [159]	✓	⊘ [160]
convolve	✓	✓	✓	⊘ [161]	⊘ [162]	⊘ [163]
trapezoid	?	?	?	?	?	?
dot	✓	✓	✓	✓	✓	⊘ [164]
multi_dot	✓	✓	⊘ [165]	✓	⊘ [166]	⊘ [167]
vdot	✓	✓	✓	✓	✓	⊘ [168]
vecdot	✓	✓	✓	✓	✓	✓
inner	✓	✓	✓	✓	⊘ [169]	⊘ [170]
outer	✓	✓	✓	✓	✓	⊘ [171]
kron	✓	✓	✓	✓	⊘ [172]	⊘ [173]
matmul	✓	✓	✓	✓	✓	✓
tensordot	✓	✓	✓	✓	✓	✓
matrix_power	✓	✓	✓	✓	⊘ [174]	⊘ [175]
correlate	✓	✓	✓	⊘ [176]	⊘ [177]	⊘ [178]
cov	✓	✓	✓	✓	✓	⊘ [179]

Dimensionless-input Operations

brainunit.math — Dimensionless-input Operations

Function	numpy	jax	cupy	torch	dask	ndonnx
exprel	✗ [180]	✓	🅙	🅙	🅙	🅙
exp	✓	✓	✓	✓	✓	✓
exp2	✓	✓	✓	✓	✓	⊘ [181]
expm1	✓	✓	✓	✓	✓	✗ [119]
log	✓	✓	✓	✓	✓	✓
log10	✓	✓	✓	✓	✓	✓
log1p	✓	✓	✓	✓	✓	✗ [119]
log2	✓	✓	✓	✓	✓	✓
arccos	✓	✓	✓	✓	✓	⊘ [182]
arccosh	✓	✓	✓	✓	✓	⊘ [183]
arcsin	✓	✓	✓	✓	✓	⊘ [184]
arcsinh	✓	✓	✓	✓	✓	⊘ [185]
arctan	✓	✓	✓	✓	✓	⊘ [186]
arctanh	✓	✓	✓	✓	✓	⊘ [187]
cos	✓	✓	✓	✓	✓	✓
cosh	✓	✓	✓	✓	✓	✓
sin	✓	✓	✓	✓	✓	✓
sinc	✓	✓	✓	✓	✓	⊘ [188]
sinh	✓	✓	✓	✓	✓	✓
tan	✓	✓	✓	✓	✓	✓
tanh	✓	✓	✓	✓	✓	✓
frexp	✓	✓	✓	✓	✓	⊘ [189]
hypot	✓	✓	✓	✓	✓	✗ [119]
arctan2	✓	✓	✓	✓	✓	⊘ [190]
logaddexp	✓	✓	✓	✓	✓	✓
logaddexp2	✓	✓	✓	✓	✓	⊘ [191]
corrcoef	✓	✓	✓	✓	✓	⊘ [192]
bitwise_not	✓	✓	✓	✓	✓	⊘ [193]
invert	✓	✓	✓	⊘ [194]	✓	⊘ [195]
bitwise_and	✓	✓	✓	✓	✓	✓
bitwise_or	✓	✓	✓	✓	✓	✓
bitwise_xor	✓	✓	✓	✓	✓	✓
left_shift	✓	✓	✓	⊘ [196]	✓	⊘ [197]
right_shift	✓	✓	✓	⊘ [198]	✓	⊘ [199]

Angle and Phase Operations

brainunit.math — Angle and Phase Operations

Function	numpy	jax	cupy	torch	dask	ndonnx
deg2rad	✓	✓	✓	✓	✓	⊘ [200]
rad2deg	✓	✓	✓	✓	✓	⊘ [201]
degrees	✓	✓	✓	⊘ [202]	✓	⊘ [203]
radians	✓	✓	✓	⊘ [204]	✓	⊘ [205]
angle	✓	✓	✓	✓	✓	✗ [64]

Unit-removing Operations

brainunit.math — Unit-removing Operations

Function	numpy	jax	cupy	torch	dask	ndonnx
heaviside	✓	✓	✓	✓	⊘ [206]	⊘ [207]
sign	✓	✓	✓	✓	✓	✓
get_promote_dtypes	✓	✓	✓	✓	✓	✓
iscomplexobj	✓	✓	✓	⊘ [208]	⊘ [209]	⊘ [210]
signbit	✓	✓	✓	✓	✓	✗ [119]
equal	✓	✓	✓	✓	✓	✓
not_equal	✓	✓	✓	✓	✓	✓
greater	✓	✓	✓	✓	✓	✓
greater_equal	✓	✓	✓	✓	✓	✓
less	✓	✓	✓	✓	✓	✓
less_equal	✓	✓	✓	✓	✓	✓
array_equal	✓	✓	✓	⊘ [211]	⊘ [212]	⊘ [213]
isclose	✓	✓	✓	✓	✓	⊘ [214]
allclose	✓	✓	✓	✓	✓	⊘ [215]
all	✓	✓	✓	✓	✓	✓
any	✓	✓	✓	✓	✓	✓
logical_not	✓	✓	✓	✓	✓	✓
logical_and	✓	✓	✓	✓	✓	✓
logical_or	✓	✓	✓	✓	✓	✓
logical_xor	✓	✓	✓	✓	✓	✓
alltrue	✓	✓	✓	✓	✓	✓
sometrue	✓	✓	✓	✓	✓	✓
bincount	✓	✓	✓	✓	✓	⊘ [216]
digitize	✓	✓	✓	⊘ [217]	✓	⊘ [218]
argsort	✓	✓	✓	✓	✓	✓
argmax	✓	✓	✓	✓	✓	✓
argmin	✓	✓	✓	✓	✓	✓
nanargmax	✓	✓	✓	⊘ [219]	✓	⊘ [220]
nanargmin	✓	✓	✓	⊘ [221]	✓	⊘ [222]
argwhere	✓	✓	✓	✓	✓	⊘ [223]
nonzero	✓	✓	✓	✓	✓	✓
flatnonzero	✓	✓	✓	⊘ [224]	✓	⊘ [225]
searchsorted	✓	✓	✓	✓	✓	✓
count_nonzero	✓	✓	✓	✓	✓	✓
diag_indices_from	✓	✓	✓	⊘ [226]	⊘ [227]	⊘ [228]

Activation Functions

brainunit.math — Activation Functions

Function	numpy	jax	cupy	torch	dask	ndonnx
relu	✗ [180]	✓	🅙	🅙	🅙	🅙
relu6	✗ [180]	✓	🅙	🅙	🅙	🅙
sigmoid	✗ [180]	✓	🅙	🅙	🅙	🅙
softplus	✗ [180]	✓	🅙	🅙	🅙	🅙
sparse_plus	✗ [180]	✓	🅙	🅙	🅙	🅙
sparse_sigmoid	✗ [180]	✓	🅙	🅙	🅙	🅙
soft_sign	✗ [180]	✓	🅙	🅙	🅙	🅙
silu	✗ [180]	✓	🅙	🅙	🅙	🅙
swish	✗ [180]	✓	🅙	🅙	🅙	🅙
log_sigmoid	✗ [180]	✓	🅙	🅙	🅙	🅙
leaky_relu	✓	✓	✓	✓	✓	✓
hard_sigmoid	✗ [180]	✓	🅙	🅙	🅙	🅙
hard_silu	✗ [180]	✓	🅙	🅙	🅙	🅙
hard_swish	✗ [180]	✓	🅙	🅙	🅙	🅙
hard_tanh	✗ [180]	✓	🅙	🅙	🅙	🅙
elu	✗ [180]	✓	🅙	🅙	🅙	🅙
celu	✗ [180]	✓	🅙	🅙	🅙	🅙
selu	✗ [180]	✓	🅙	🅙	🅙	🅙
gelu	✗ [180]	✓	🅙	🅙	🅙	🅙
glu	✗ [180]	✓	🅙	🅙	🅙	🅙
squareplus	✗ [180]	✓	🅙	🅙	🅙	🅙
mish	✗ [180]	✓	🅙	🅙	🅙	🅙

Einstein Operations

brainunit.math — Einstein Operations

Function	numpy	jax	cupy	torch	dask	ndonnx
einreduce	✓	✓	✓	✓	✓	✓
einrearrange	✓	✓	✓	✗ [229]	✓	⊘ [50]
einrepeat	✓	✓	✓	✓	✓	✓
einshape	✓	✓	✓	✓	✓	✓
einsum	✗ [180]	✓	✗ [230]	⊘ [231]	✗ [232]	⊘ [233]

Dtypes, Constants, and Utilities

brainunit.math — Dtypes, Constants, and Utilities

Function	numpy	jax	cupy	torch	dask	ndonnx
dtype	✓	✓	✓	✓	✓	✓
finfo	✓	✓	✓	✓	✓	✓
iinfo	✓	✓	✓	✓	✓	✓
issubdtype	✓	✓	✓	✓	✓	✓
result_type	✓	✓	✓	✓	✓	✓
ndim	✓	✓	✓	✓	✓	✓
isreal	✓	✓	✓	✓	✓	⊘ [234]
isscalar	✓	✓	✓	✓	✓	✓
isfinite	✓	✓	✓	✓	✓	✓
isinf	✓	✓	✓	✓	✓	✓
isnan	✓	✓	✓	✓	✓	✓
shape	✓	✓	✓	✓	✓	✓
size	✓	✓	✓	✓	✓	✓
broadcast_shapes	✓	✓	✓	✓	✓	✓
gradient	✓	✓	✓	✓	✓	⊘ [235]
bartlett	✗ [180]	✓	✗ [236]	✗ [237]	✗ [238]	✗ [239]
blackman	✗ [180]	✓	✗ [236]	✗ [237]	✗ [238]	✗ [239]
hamming	✗ [180]	✓	✗ [236]	✗ [237]	✗ [238]	✗ [239]
hanning	✗ [180]	✓	✗ [236]	✗ [237]	✗ [238]	✗ [239]
kaiser	✗ [180]	✓	✗ [236]	✗ [237]	✗ [238]	✗ [239]

Non-dispatched helpers

These names live under brainunit.math for convenience but do not dispatch on backend — they are dtype factories (re-exported from jax.numpy) or pure-Python predicates / introspection helpers over Quantity / Unit objects. Behavior is identical on every backend.

assert_quantity bfloat16 bool_ cdouble check_dims check_units complex128 complex64 complex_ csingle display_in_unit double

fail_for_dimension_mismatch fail_for_unit_mismatch float16 float32 float64 float_ get_dim get_dtype get_magnitude get_mantissa get_or_create_dimension get_unit

inexact int16 int2 int32 int4 int64 int8 int_ is_dimensionless is_float is_int is_quantity

is_unitless maybe_decimal set_exprel_order single uint uint16 uint2 uint32 uint4 uint64 uint8

brainunit.linalg

brainunit.linalg

Function	numpy	jax	cupy	torch	dask	ndonnx
cholesky	✓	✓	✓	✓	✓	✗ [119]
cond	✓	✓	✓	✓	⊘ [240]	⊘ [241]
cross	✓	✓	✓	✓	⊘ [155]	⊘ [156]
det	✓	✓	✓	✓	⊘ [242]	⊘ [243]
diagonal	✓	✓	✓	✓	✓	⊘ [244]
dot	✓	✓	✓	✓	✓	⊘ [164]
eig	✗ [180]	✓	✗ [245]	⊘ [246]	⊘ [247]	✗ [248]
eigh	✗ [180]	✓	⊘ [249]	⊘ [250]	⊘ [251]	⊘ [252]
eigvals	✗ [180]	✓	✗ [245]	⊘ [246]	⊘ [247]	✗ [248]
eigvalsh	✗ [180]	✓	⊘ [249]	⊘ [250]	⊘ [251]	⊘ [252]
inner	✓	✓	✓	✓	⊘ [169]	⊘ [170]
inv	✓	✓	✓	✓	✓	⊘ [253]
kron	✓	✓	✓	✓	⊘ [172]	⊘ [173]
lstsq	✓	✓	✓	✓	⊘ [254]	⊘ [255]
matmul	✓	✓	✓	✓	✓	✓
matrix_norm	✓	✓	✓	✓	✓	⊘ [256]
matrix_power	✓	✓	✓	✓	⊘ [174]	⊘ [175]
matrix_rank	✓	✓	✓	✓	⊘ [257]	⊘ [258]
matrix_transpose	✓	✓	✓	✓	✓	⊘ [259]
multi_dot	✓	✓	⊘ [165]	✓	⊘ [166]	⊘ [167]
norm	✓	✓	✓	✓	✓	⊘ [260]
outer	✓	✓	✓	✓	✓	⊘ [171]
pinv	✓	✓	✓	✓	⊘ [261]	⊘ [262]
qr	✓	✓	✓	✓	✓	⊘ [255]
slogdet	✓	✓	✓	✓	⊘ [263]	⊘ [255]
solve	✓	✓	✓	✓	✓	⊘ [264]
svd	✗ [180]	✓	✗ [245]	⊘ [265]	⊘ [266]	✗ [248]
svdvals	✗ [180]	✓	✗ [245]	⊘ [265]	⊘ [266]	✗ [248]
tensordot	✓	✓	✓	✓	✓	✓
tensorinv	✓	✓	✓	✓	⊘ [267]	⊘ [268]
tensorsolve	✓	✓	✓	✓	⊘ [269]	⊘ [270]
trace	✓	✓	✓	✓	✓	⊘ [271]
vdot	✓	✓	✓	✓	✓	⊘ [168]
vecdot	✓	✓	✓	✓	✓	✓
vector_norm	✓	✓	✓	✓	✓	⊘ [272]

brainunit.fft

Routing varies inside brainunit.fft: _fft_change_unit.py calls brainunit._backend.get_backend() directly (e.g. for fftfreq / rfftfreq), while _fft_keep_unit.py delegates to the math package’s _fun_keep_unit_unary helper and inherits its dispatch.

brainunit.fft

Function	numpy	jax	cupy	torch	dask	ndonnx
fft	✓	✓	✓	✓	✓	⊘ [273]
fft2	✓	✓	✓	✓	✓	⊘ [274]
fftfreq	✓	✓	✓	✓	✓	⊘ [275]
fftn	✓	✓	✓	✓	✓	⊘ [276]
fftshift	✓	✓	✓	✓	✓	⊘ [277]
ifft	✓	✓	✓	✓	✓	⊘ [278]
ifft2	✓	✓	✓	✓	✓	⊘ [279]
ifftn	✓	✓	✓	✓	✓	⊘ [280]
ifftshift	✓	✓	✓	✓	✓	⊘ [281]
irfft	✓	✓	✓	✓	✓	⊘ [282]
irfft2	✓	✓	✓	✓	✓	⊘ [283]
irfftn	✓	✓	✓	✓	✓	⊘ [284]
rfft	✓	✓	✓	✓	✓	⊘ [285]
rfft2	✓	✓	✓	✓	✓	⊘ [286]
rfftfreq	✓	✓	✓	✓	✓	⊘ [275]
rfftn	✓	✓	✓	✓	✓	⊘ [287]

Quantity methods

Methods on brainunit.Quantity itself. .to_<backend>() methods ignore the current backend and convert to the named one — cells show ⊘ when the target backend isn’t installed in the sweep environment.

Materialization is documented above (see Backend-specific notes). Quantity.tolist on dask is the one method that raises BackendError from brainunit’s own guard (⚠). .item reports ⊘ on dask / ndonnx because the underlying array object does not expose .item().

Quantity public methods

Function	numpy	jax	cupy	torch	dask	ndonnx
all	✓	✓	✓	✓	✓	✓
any	✓	✓	✓	✓	✓	✓
argmax	✓	✓	✓	✓	✓	✓
argmin	✓	✓	✓	✓	✓	✓
argsort	✓	✓	✓	✓	✓	✓
astype	✓	✓	✓	✓	✓	✓
clip	✓	✓	✓	✓	✓	✓
clone	✓	✓	✓	⊘ [288]	⊘ [289]	⊘ [290]
conj	✓	✓	✓	✓	✓	✗ [119]
conjugate	✓	✓	✓	✓	✓	✗ [119]
copy	✓	✓	✓	⊘ [288]	⊘ [289]	⊘ [290]
cpu	✓	✓	✗ [291]	✗ [292]	✗ [293]	✗ [294]
cross	✓	✓	✓	⊘ [295]	⊘ [296]	⊘ [297]
cuda	⊘ [298]	⊘ [298]	⊘ [298]	⊘ [298]	⊘ [298]	⊘ [298]
cumprod	⊘ [299]	⊘ [299]	⊘ [299]	⊘ [299]	⊘ [299]	⊘ [299]
cumsum	✓	✓	✓	⊘ [300]	✓	⊘ [70]
diagonal	✓	✓	✓	⊘ [301]	✓	⊘ [302]
dot	✓	✓	✓	✓	✓	⊘ [303]
double	✓	✓	✓	✓	✓	✓
expand_as	✓	✓	✓	✓	✓	✓
expand_dims	✓	✓	✓	✓	✓	✓
factorless	✓	✓	✓	✓	✓	✓
fill	✓	✓	✗ [304]	✗ [292]	✗ [304]	✗ [305]
flatten	✓	✓	✓	✓	✓	✓
float	✓	✓	✓	✓	✓	✓
half	✓	✓	✓	✓	⊘ [306]	✓
has_same_unit	✓	✓	✓	✓	✓	✓
in_unit	✓	✓	✓	✓	✓	✓
item	✓	✓	✓	✓	⊘ [307]	⊘ [307]
max	✓	✓	✓	✓	✓	✓
mean	✓	✓	✓	✓	✓	✓
min	✓	✓	✓	✓	✓	✓
nancumprod	⊘ [308]	⊘ [308]	⊘ [308]	⊘ [308]	⊘ [308]	⊘ [308]
nanprod	✓	✓	✓	⊘ [146]	✓	⊘ [147]
nonzero	✓	✓	✓	✓	✓	✓
outer	✓	✓	✓	✓	✓	⊘ [309]
pow	✓	✓	✓	✓	✓	✓
prod	✓	✓	✓	✓	✓	✓
ptp	✓	✓	✓	⊘ [83]	✓	⊘ [85]
put	✓	✓	✗ [304]	✗ [292]	⊘ [310]	✗ [305]
ravel	✓	✓	✓	✓	✓	⊘ [63]
repeat	✓	✓	✓	✓	✓	✓
repr_in_unit	✓	✓	✓	✓	✓	✓
reshape	✓	✓	✓	✓	✓	✓
resize	✓	✓	✓	✗ [311]	✗ [312]	✗ [313]
round	✓	✓	✓	✓	✓	✓
scatter_add	✓	✓	✓	✓	⊘ [314]	✗ [305]
scatter_div	✓	✓	⊘ [315]	✓	⊘ [316]	✗ [305]
scatter_max	✓	✓	✓	✓	⊘ [317]	✗ [305]
scatter_min	✓	✓	✓	✓	⊘ [318]	✗ [305]
scatter_mul	✓	✓	⊘ [319]	✓	⊘ [320]	✗ [305]
scatter_sub	✓	✓	✓	✓	⊘ [314]	✗ [305]
searchsorted	✓	✓	✓	✓	✓	✓
sort	✓	✓	✗ [304]	✗ [292]	✗ [304]	✗ [294]
split	✓	✓	✓	⊘ [321]	✗ [312]	✗ [322]
squeeze	✓	✓	✓	✗ [323]	✓	✗ [324]
std	✓	✓	✓	✓	✓	✓
sum	✓	✓	✓	✓	✓	✓
swapaxes	✓	✓	✓	✓	✓	⊘ [325]
take	✓	✓	✓	✓	✗ [326]	✓
tile	✓	✓	✓	✗ [327]	✓	✗ [328]
to	✓	✓	✓	✓	✓	✓
to_cupy	✓	✓	✓	✓	✓	⊘ [329]
to_dask	✓	✓	✓	✗ [292]	✓	✗ [294]
to_decimal	✓	✓	✓	✓	✓	✓
to_jax	✓	✓	✗ [245]	✓	✓	✗ [330]
to_ndonnx	✓	⊘ [331]	⊘ [331]	⊘ [332]	⊘ [333]	✓
to_numpy	✓	✓	✓	✓	✓	✓
to_torch	✓	✓	✗ [334]	✓	✗ [335]	✗ [336]
tolist	✓	✓	✓	✓	⚠ [337]	⊘ [338]
trace	✓	✓	✓	⊘ [339]	✓	⊘ [340]
transpose	✓	✓	✓	⊘ [49]	✓	⊘ [341]
tree_flatten	✓	✓	✓	✓	✓	✓
tree_unflatten	?	?	?	?	?	?
unsqueeze	✓	✓	✓	✓	✓	✓
update_mantissa	✓	✓	✗ [304]	✗ [292]	✗ [304]	✗ [294]
var	✓	✓	✓	✓	✓	✓
view	✓	✓	✓	⊘ [342]	✓	⊘ [343]
with_unit	✓	✓	✓	✓	✓	✓

Coverage statistic

Subpackage	Mapped	Non-dispatched	Unmapped	Total
brainunit.math	294	47	1	342
brainunit.linalg	35	0	0	35
brainunit.fft	16	0	0	16
Quantity	78	0	1	79

Mapped = functions the sweep actually invoked. Non-dispatched = type factories / predicates that don’t go through backend dispatch. Unmapped = no call pattern registered (will appear as ? in tables).

How this was produced

dev/backend_support_sweep.py walks every public callable in the subpackages above, picks a calling pattern from an in-script registry, and invokes the function under with brainunit.using_backend(b) for each backend b in the local environment. Outcomes are classified as pass / skip / fail / warn / unmapped / na and written to dev/backend_support_data.json.

dev/backend_support_render.py (this script’s source) reads that JSON and emits the rst file you are currently reading. To refresh:

PYTHONPATH=. python dev/backend_support_sweep.py
PYTHONPATH=. python dev/backend_support_render.py

JAX-only subpackages are probed with one representative function per subpackage rather than enumerated — every entry point in brainunit.lax / .autograd / .sparse is gated identically.

Footnotes

[1]

AttributeError: module ‘array_api_compat.torch’ has no attribute ‘identity’

[2]

AttributeError: module ‘array_api_compat.dask.array’ has no attribute ‘identity’

[3]

AttributeError: module ‘ndonnx’ has no attribute identity

[4]

AttributeError: module ‘array_api_compat.torch’ has no attribute ‘tri’

[5]

AttributeError: module ‘ndonnx’ has no attribute tri

[6]

TypeError: ‘full_like’ got an unexpected keyword argument ‘shape’

[7]

AttributeError: module ‘ndonnx’ has no attribute diag

[8]

TypeError: ‘empty_like’ got an unexpected keyword argument ‘shape’

[9] (1,2)

TypeError: ‘ones_like’ got an unexpected keyword argument ‘shape’

[10] (1,2)

TypeError: ‘zeros_like’ got an unexpected keyword argument ‘shape’

[11]

AttributeError: module ‘array_api_compat.torch’ has no attribute ‘fill_diagonal’

[12]

AttributeError: module ‘array_api_compat.dask.array’ has no attribute ‘fill_diagonal’

[13]

AttributeError: module ‘ndonnx’ has no attribute fill_diagonal

[14]

TypeError: ‘linspace’ got an unexpected keyword argument ‘retstep’

[15]

TypeError: logspace() received an invalid combination of arguments - got (float, float), but expected one of: * (Tensor start, Tensor end, int steps, float base = 10.0, *, Tensor out = None, torch….

[16]

AttributeError: module ‘array_api_compat.dask.array’ has no attribute ‘logspace’

[17]

AttributeError: module ‘ndonnx’ has no attribute logspace

[18]

AttributeError: ‘list’ object has no attribute ‘ndim’

[19]

AttributeError: module ‘array_api_compat.dask.array’ has no attribute ‘vander’

[20]

AttributeError: module ‘ndonnx’ has no attribute vander

[21]

AttributeError: module ‘array_api_compat.torch’ has no attribute ‘tril_indices_from’

[22]

AttributeError: module ‘ndonnx’ has no attribute tril_indices_from

[23]

AttributeError: module ‘array_api_compat.torch’ has no attribute ‘triu_indices_from’

[24]

AttributeError: module ‘ndonnx’ has no attribute triu_indices_from

[25] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘vstack’

[26]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘concatenate’

[27]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘hstack’

[28]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘dstack’

[29]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘column_stack’

[30]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘column_stack’

[31]

AttributeError: brainunit: backend ‘array_api_compat.cupy’ has no operation ‘block’

[32]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘block’

[33]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘block’

[34]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘append’

[35]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘append’

[36] (1,2)

TypeError: ‘split’ got an unexpected keyword argument ‘axis’

[37] (1,2)

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘split’

[38] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘split’

[39]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘dsplit’

[40]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘dsplit’

[41]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘hsplit’

[42]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘hsplit’

[43]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘vsplit’

[44]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘vsplit’

[45]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘atleast_1d’

[46]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘atleast_2d’

[47]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘atleast_3d’

[48]

TypeError: ‘reshape’ got an unexpected keyword argument ‘order’

[49] (1,2)

TypeError: transpose() missing 2 required positional argument: “dim0”, “dim1”

[50] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘transpose’

[51]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘swapaxes’

[52]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fliplr’

[53]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘flipud’

[54] (1,2)

TypeError: squeeze() missing 1 required positional argument: ‘axis’

[55]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘diagflat’

[56]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘diagflat’

[57]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘diagonal’

[58]

AttributeError: ‘list’ object has no attribute ‘shape’

[59]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘choose’

[60]

TypeError: ‘choose’ got an unexpected keyword argument ‘mode’

[61]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘choose’

[62]

TypeError: ‘ravel’ got an unexpected keyword argument ‘order’

[63] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘ravel’

[64] (1,2,3,4,5)

ValueError: ‘complex128’ does not have a corresponding ndonnx data type

[65]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘conjugate’

[66]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘conjugate’

[67]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘nancumsum’

[68]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nancumsum’

[69]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nansum’

[70] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘cumsum’

[71]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘ediff1d’

[72]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘ediff1d’

[73]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘absolute’

[74]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘fabs’

[75]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fabs’

[76]

TypeError: median() missing 1 required positional arguments: “dim”

[77]

TypeError: ‘median’ got an unexpected keyword argument ‘overwrite_input’

[78]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘median’

[79]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘nanmin’

[80]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nanmin’

[81]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘nanmax’

[82]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nanmax’

[83] (1,2)

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘ptp’

[84]

TypeError: ‘ptp’ got an unexpected keyword argument ‘keepdims’

[85] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘ptp’

[86]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘average’

[87]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘average’

[88]

TypeError: nanmedian() missing 1 required positional arguments: “dim”

[89]

TypeError: ‘nanmedian’ got an unexpected keyword argument ‘overwrite_input’

[90]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nanmedian’

[91]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nanmean’

[92]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘nanstd’

[93]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nanstd’

[94]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘rot90’

[95]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘intersect1d’

[96]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘intersect1d’

[97]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘intersect1d’

[98]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nan_to_num’

[99]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘percentile’

[100]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘percentile’

[101]

AttributeError: brainunit: backend ‘array_api_compat.cupy’ has no operation ‘nanpercentile’

[102]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘nanpercentile’

[103]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nanpercentile’

[104]

TypeError: quantile() received an invalid combination of arguments - got (Tensor), but expected one of: * (Tensor input, Tensor q, int dim = None, bool keepdim = False, *, str interpolation = “line…

[105]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘quantile’

[106]

AttributeError: brainunit: backend ‘array_api_compat.cupy’ has no operation ‘nanquantile’

[107]

TypeError: nanquantile() received an invalid combination of arguments - got (Tensor), but expected one of: * (Tensor input, Tensor q, int dim = None, bool keepdim = False, *, str interpolation = “l…

[108]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nanquantile’

[109]

TypeError: ‘round’ got an unexpected keyword argument ‘decimals’

[110]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘around’

[111]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘around’

[112]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘rint’

[113]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘rint’

[114]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘modf’

[115]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘modf’

[116]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fmod’

[117]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘mod’

[118]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘mod’

[119] (1,2,3,4,5,6,7,8,9)

NotImplementedError:

[120]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fmax’

[121]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fmin’

[122]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘lcm’

[123]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘lcm’

[124]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘gcd’

[125]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘gcd’

[126]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘trace’

[127]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘interp’

[128]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘interp’

[129]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘interp’

[130]

expected dask result backend; got dask, numpy

[131]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘histogram’

[132]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘compress’

[133]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘compress’

[134]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘extract’

[135]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘extract’

[136] (1,2,3,4)

TypeError: ‘take’ got an unexpected keyword argument ‘unique_indices’

[137]

TypeError: select(): argument ‘input’ (position 1) must be Tensor, not list

[138]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘select’

[139]

TypeError: ‘unique’ got an unexpected keyword argument ‘equal_nan’

[140]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘unique’

[141]

NotImplementedError: Don’t yet support nd fancy indexing

[142]

AttributeError: ‘Array’ object has no attribute ‘reshape’

[143]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘ldexp’

[144]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘nancumprod’

[145]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nancumprod’

[146] (1,2)

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘nanprod’

[147] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nanprod’

[148] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘cumprod’

[149]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘nanvar’

[150]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nanvar’

[151]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘cbrt’

[152]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘cbrt’

[153]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘power’

[154]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘power’

[155] (1,2)

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘cross’

[156] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘cross’

[157]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘true_divide’

[158]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘float_power’

[159]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘divmod’

[160]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘divmod’

[161]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘convolve’

[162]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘convolve’

[163]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘convolve’

[164] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘dot’

[165] (1,2)

AttributeError: brainunit: backend ‘array_api_compat.cupy’ has no operation ‘linalg.multi_dot’

[166] (1,2)

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘linalg.multi_dot’

[167] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.multi_dot’

[168] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘vdot’

[169] (1,2)

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘inner’

[170] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘inner’

[171] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘outer’

[172] (1,2)

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘kron’

[173] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘kron’

[174] (1,2)

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘linalg.matrix_power’

[175] (1,2)

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.matrix_power’

[176]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘correlate’

[177]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘correlate’

[178]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘correlate’

[179]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘cov’

[180] (1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34)

expected numpy result backend; got jax

[181]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘exp2’

[182]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘arccos’

[183]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘arccosh’

[184]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘arcsin’

[185]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘arcsinh’

[186]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘arctan’

[187]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘arctanh’

[188]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘sinc’

[189]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘frexp’

[190]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘arctan2’

[191]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘logaddexp2’

[192]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘corrcoef’

[193]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘bitwise_not’

[194]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘invert’

[195]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘invert’

[196]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘left_shift’

[197]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘left_shift’

[198]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘right_shift’

[199]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘right_shift’

[200]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘deg2rad’

[201]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘rad2deg’

[202]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘degrees’

[203]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘degrees’

[204]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘radians’

[205]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘radians’

[206]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘heaviside’

[207]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘heaviside’

[208]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘iscomplexobj’

[209]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘iscomplexobj’

[210]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘iscomplexobj’

[211]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘array_equal’

[212]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘array_equal’

[213]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘array_equal’

[214]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘isclose’

[215]

AttributeError: module ‘ndonnx’ has no attribute allclose

[216]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘bincount’

[217]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘digitize’

[218]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘digitize’

[219]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘nanargmax’

[220]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nanargmax’

[221]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘nanargmin’

[222]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘nanargmin’

[223]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘argwhere’

[224]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘flatnonzero’

[225]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘flatnonzero’

[226]

AttributeError: brainunit: backend ‘array_api_compat.torch’ has no operation ‘diag_indices_from’

[227]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘diag_indices_from’

[228]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘diag_indices_from’

[229]

TypeError: transpose(): argument ‘dim0’ (position 2) must be int, not list

[230]

TypeError: Argument ‘[[1. 2.] [3. 4.]]’ of type <class ‘cupy.ndarray’> is not a valid JAX type.

[231]

TypeError: Error interpreting argument to <function _einsum at 0x…> as an abstract array. The problematic value is of type <class ‘torch.Tensor’> and was passed to the function at path operands[0…

[232]

TypeError: Argument ‘dask.array<array, shape=(2, 2), dtype=float64, chunksize=(2, 2), chunktype=numpy.ndarray>’ of type <class ‘dask.array.core.Array’> is not a valid JAX type.

[233]

TypeError: Error interpreting argument to <function _einsum at 0x…> as an abstract array. The problematic value is of type <class ‘ndonnx._array.Array’> and was passed to the function at path ope…

[234]

AttributeError: module ‘ndonnx’ has no attribute isreal

[235]

AttributeError: module ‘ndonnx’ has no attribute gradient

[236] (1,2,3,4,5)

expected cupy result backend; got jax

[237] (1,2,3,4,5)

expected torch result backend; got jax

[238] (1,2,3,4,5)

expected dask result backend; got jax

[239] (1,2,3,4,5)

expected ndonnx result backend; got jax

[240]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘linalg.cond’

[241]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.cond’

[242]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘linalg.det’

[243]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.det’

[244]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.diagonal’

[245] (1,2,3,4,5)

JaxRuntimeError: This operation requires CUDA support from jaxlib or jax cuda plugin.

[246] (1,2)

TypeError: Error interpreting argument to eig as a JAX value. The problematic value is of type <class ‘torch.Tensor’> and was passed to eig at position 0.

[247] (1,2)

TypeError: Error interpreting argument to eig as a JAX value. The problematic value is of type <class ‘dask.array.core.Array’> and was passed to eig at position 0.

[248] (1,2,3,4)

TypeError: Value ‘array(data: [[1.0, 2.0], [3.0, 4.0]], dtype=float64)’ with dtype object is not a valid JAX array type. Only arrays of numeric types are supported by JAX.

[249] (1,2)

TypeError: Error interpreting argument to transpose as a JAX value. The problematic value is of type <class ‘cupy.ndarray’> and was passed to transpose at position 0.

[250] (1,2)

TypeError: Error interpreting argument to transpose as a JAX value. The problematic value is of type <class ‘torch.Tensor’> and was passed to transpose at position 0.

[251] (1,2)

TypeError: Error interpreting argument to transpose as a JAX value. The problematic value is of type <class ‘dask.array.core.Array’> and was passed to transpose at position 0.

[252] (1,2)

TypeError: Error interpreting argument to transpose as a JAX value. The problematic value is of type <class ‘ndonnx._array.Array’> and was passed to transpose at position 0.

[253]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.inv’

[254]

TypeError: lstsq() got an unexpected keyword argument ‘rcond’

[255] (1,2,3)

AttributeError: module ‘ndonnx’ has no attribute linalg

[256]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.matrix_norm’

[257]

TypeError: svd() got an unexpected keyword argument ‘compute_uv’

[258]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.matrix_rank’

[259]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.matrix_transpose’

[260]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.norm’

[261]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘linalg.pinv’

[262]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.pinv’

[263]

AttributeError: module ‘array_api_compat.dask.array.linalg’ has no attribute ‘slogdet’

[264]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.solve’

[265] (1,2)

TypeError: Error interpreting argument to svd as a JAX value. The problematic value is of type <class ‘torch.Tensor’> and was passed to svd at position 0.

[266] (1,2)

TypeError: Error interpreting argument to svd as a JAX value. The problematic value is of type <class ‘dask.array.core.Array’> and was passed to svd at position 0.

[267]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘linalg.tensorinv’

[268]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.tensorinv’

[269]

AttributeError: brainunit: backend ‘array_api_compat.dask.array’ has no operation ‘linalg.tensorsolve’

[270]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.tensorsolve’

[271]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.trace’

[272]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘linalg.vector_norm’

[273]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.fft’

[274]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.fft2’

[275] (1,2)

AttributeError: module ‘ndonnx’ has no attribute fft

[276]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.fftn’

[277]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.fftshift’

[278]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.ifft’

[279]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.ifft2’

[280]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.ifftn’

[281]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.ifftshift’

[282]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.irfft’

[283]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.irfft2’

[284]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.irfftn’

[285]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.rfft’

[286]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.rfft2’

[287]

AttributeError: brainunit: backend ‘ndonnx’ has no operation ‘fft.rfftn’

[288] (1,2)

AttributeError: module ‘array_api_compat.torch’ has no attribute ‘copy’

[289] (1,2)

AttributeError: module ‘array_api_compat.dask.array’ has no attribute ‘copy’

[290] (1,2)

AttributeError: module ‘ndonnx’ has no attribute copy

[291]

InvalidInputException: Argument ‘[1. 2. 3.]’ of type <class ‘cupy.ndarray’> is not a valid JAX type.

[292] (1,2,3,4,5,6)

TypeError: Cannot interpret ‘torch.float64’ as a data type

[293]

InvalidInputException: Argument ‘dask.array<array, shape=(3,), dtype=float64, chunksize=(3,), chunktype=numpy.ndarray>’ of type <class ‘dask.array.core.Array’> is not a valid JAX type.

[294] (1,2,3,4)

TypeError: Cannot interpret ‘Float64’ as a data type

[295]

TypeError: cross() got an unexpected keyword argument ‘axisa’

[296]

AttributeError: module ‘array_api_compat.dask.array’ has no attribute ‘cross’

[297]

AttributeError: module ‘ndonnx’ has no attribute cross

[298] (1,2,3,4,5,6)

RuntimeError: Unknown backend cuda. Available backends are [‘cpu’]

[299] (1,2,3,4,5,6)

TypeError: cumprod is not supported for quantities with units (has unit m), because each element of the result would have a different unit exponent. Use .prod() for a single reduction, or convert t…

[300]

TypeError: cumsum() missing 1 required positional arguments: “dim”

[301]

TypeError: diagonal() got an unexpected keyword argument ‘axis1’

[302]

AttributeError: module ‘ndonnx’ has no attribute diagonal

[303]

TypeError: Error interpreting argument to <function dot at 0x…> as an abstract array. The problematic value is of type <class ‘ndonnx._array.Array’> and was passed to the function at path a. This…

[304] (1,2,3,4,5,6,7)

ValueError: The dtype of the original data is float32, while we got float64.

[305] (1,2,3,4,5,6,7,8)

BackendError: Quantity.at indexed-update is not supported on the ndonnx backend. Call .to_numpy() (or another concrete backend) on the input first.

[306]

AttributeError: module ‘array_api_compat.dask.array’ has no attribute ‘float16’

[307] (1,2)

AttributeError: ‘Array’ object has no attribute ‘item’

[308] (1,2,3,4,5,6)

TypeError: nancumprod is not supported for quantities with units (has unit m), because each element of the result would have a different unit exponent. Use .nanprod() for a single reduction, or con…

[309]

TypeError: Error interpreting argument to <function outer at 0x…> as an abstract array. The problematic value is of type <class ‘ndonnx._array.Array’> and was passed to the function at path a. Th…

[310]

NotImplementedError: Quantity.at on dask backend does not support index type ‘Array’ for op ‘set’. Use a boolean mask of the source shape, a slice, an int, or a 1D int array; or call .to_numpy() to…

[311]

expected torch result backend; got numpy

[312] (1,2)

expected dask result backend; got numpy

[313]

expected ndonnx result backend; got numpy

[314] (1,2)

NotImplementedError: Quantity.at on dask backend does not support index type ‘Array’ for op ‘add’. Use a boolean mask of the source shape, a slice, an int, or a 1D int array; or call .to_numpy() to…

[315]

NotImplementedError: cupy_true_divide.at is not supported yet

[316]

NotImplementedError: Quantity.at on dask backend does not support index type ‘Array’ for op ‘divide’. Use a boolean mask of the source shape, a slice, an int, or a 1D int array; or call .to_numpy()…

[317]

NotImplementedError: Quantity.at on dask backend does not support index type ‘Array’ for op ‘max’. Use a boolean mask of the source shape, a slice, an int, or a 1D int array; or call .to_numpy() to…

[318]

NotImplementedError: Quantity.at on dask backend does not support index type ‘Array’ for op ‘min’. Use a boolean mask of the source shape, a slice, an int, or a 1D int array; or call .to_numpy() to…

[319]

NotImplementedError: cupy_multiply.at is not supported yet

[320]

NotImplementedError: Quantity.at on dask backend does not support index type ‘Array’ for op ‘multiply’. Use a boolean mask of the source shape, a slice, an int, or a 1D int array; or call .to_numpy…

[321]

TypeError: split() got an unexpected keyword argument ‘axis’

[322]

AxisError: axis1: axis 0 is out of bounds for array of dimension 0

[323]

TypeError: ‘NoneType’ object is not iterable

[324]

TypeError: int() argument must be a string, a bytes-like object or a real number, not ‘NoneType’

[325]

AttributeError: module ‘ndonnx’ has no attribute swapaxes

[326]

TypeError: Axis value must be an integer, got None

[327]

TypeError: tile(): argument ‘dims’ (position 2) must be tuple of ints, not int

[328]

TypeError: object of type ‘int’ has no len()

[329]

ValueError: Unsupported dtype object

[330]

TypeError: Value ‘array(data: [1.0, 2.0, 3.0], dtype=float64)’ with dtype object is not a valid JAX array type. Only arrays of numeric types are supported by JAX.

[331] (1,2)

ValueError: unable to infer dtype from [1. 2. 3.]

[332]

ValueError: unable to infer dtype from tensor([1., 2., 3.], dtype=torch.float64)

[333]

ValueError: unable to infer dtype from dask.array<array, shape=(3,), dtype=float64, chunksize=(3,), chunktype=numpy.ndarray>

[334]

RuntimeError: Cannot initialize CUDA without ATen_cuda library. PyTorch splits its backend into two shared libraries: a CPU library and a CUDA library; this error has occurred because you are tryin…

[335]

TypeError: len() of unsized object

[336]

ValueError: ONNX provides no control over the used device

[337]

BackendError (expected): Quantity.tolist() would materialize a dask-backed Quantity. Call q.mantissa.compute() first.

[338]

AttributeError: ‘Array’ object has no attribute ‘tolist’

[339]

TypeError: trace() got an unexpected keyword argument ‘offset’

[340]

AttributeError: module ‘ndonnx’ has no attribute trace

[341]

AttributeError: module ‘ndonnx’ has no attribute transpose

[342]

TypeError: view() received an invalid combination of arguments - got (type), but expected one of: * (torch.dtype dtype) didn’t match because some of the arguments have invalid types: (!type!) * (tu…

[343]

AttributeError: ‘Array’ object has no attribute ‘view’