MCPcopy Create free account

hub / github.com/ml-explore/mlx / functions

Functions4,946 in github.com/ml-explore/mlx

↓ 2,145 callersFunctionarray
mlx/array.h:84
↓ 1,866 callersMethodarray
mlx/array.h:533
↓ 1,040 callersMethodsize
mlx/dtype.h:60
↓ 642 callersFunctionarray_equal
mlx/ops.h:494
↓ 613 callersMethodpush_back
mlx/small_vector.h:332
↓ 477 callersMethodsize
mlx/backend/cuda/allocator.cpp:293
↓ 367 callersMethoditem
mlx/array.h:566
↓ 290 callersMethodend
mlx/small_vector.h:247
↓ 273 callersMethodsize
mlx/backend/metal/allocator.cpp:192
↓ 269 callersFunctionreshape
mlx/ops.cpp:392
↓ 266 callersMethodbegin
mlx/small_vector.h:240
↓ 235 callersFunctionastype
mlx/ops.cpp:258
↓ 228 callersFunctionmin
mlx/backend/cpu/simd/neon_fp16_simd.h:189
↓ 228 callersFunctionones
mlx/ops.h:96
↓ 225 callersMethodset_input_array
mlx/backend/metal/device.cpp:300
↓ 225 callersFunctionto_stream
mlx/utils.cpp:15
↓ 222 callersFunctionissubdtype
mlx/dtype.cpp:180
↓ 212 callersMethodappend
mlx/backend/cuda/jit_module.h:33
↓ 191 callersFunctionarange
mlx/ops.cpp:150
↓ 188 callersFunctionmultiply
mlx/ops.cpp:2865
↓ 187 callersMethodset_data
mlx/array.cpp:168
↓ 179 callersFunctionmalloc
mlx/allocator.h:54
↓ 175 callersFunctionzeros
mlx/ops.h:89
↓ 172 callersFunctionslice
mlx/ops.h:181
↓ 169 callersMethodinsert
mlx/export.cpp:558
↓ 167 callersFunctioneval
mlx/transforms.h:22
↓ 158 callersMethodend
mlx/backend/cuda/lru_cache.h:63
↓ 158 callersMethodset_input_array
mlx/backend/cuda/device.cpp:224
↓ 151 callersMethodmax
mlx/types/limits.h:33
↓ 146 callersFunctionfull
mlx/ops.h:67
↓ 145 callersMethodeval
mlx/array.cpp:154
↓ 139 callersMethodbegin
mlx/backend/cuda/lru_cache.h:57
↓ 139 callersMethodvmap
mlx/fast.cpp:45
↓ 136 callersMethodeval
Set the model to evaluation mode. See :func:`train`.
python/mlx/nn/layers/base.py:590
↓ 131 callersFunctionsum
mlx/ops.h:591
↓ 131 callersFunctionvjp
mlx/transforms.cpp:347
↓ 128 callersFunctionallclose
mlx/ops.cpp:1925
↓ 126 callersFunctionto_array
python/src/utils.cpp:8
↓ 119 callersFunctiontranspose
mlx/ops.h:346
↓ 108 callersFunctionabs
mlx/ops.cpp:2770
↓ 106 callersMethodget_kernel
mlx/backend/metal/device.cpp:768
↓ 106 callersFunctionget_type_string
mlx/backend/common/compiled.cpp:47
↓ 105 callersFunctionadd
mlx/ops.cpp:2836
↓ 105 callersMethodget
mlx/distributed/utils.h:15
↓ 105 callersMethodset_output_array
mlx/backend/cuda/device.cpp:233
↓ 103 callersFunctionget_template_definition
mlx/backend/metal/kernels.h:378
↓ 102 callersMethodset_output_array
mlx/backend/metal/device.cpp:314
↓ 99 callersFunctiontype_to_name
mlx/backend/metal/utils.cpp:8
↓ 98 callersMethodemplace_back
mlx/small_vector.h:323
↓ 98 callersFunctionvmap
mlx/transforms.cpp:900
↓ 97 callersFunctionbroadcast_to
mlx/ops.cpp:1601
↓ 96 callersMethodfind
mlx/export.cpp:655
↓ 94 callersFunctionexpand_dims
mlx/ops.cpp:612
↓ 94 callersMethodsplit
mlx/distributed/distributed.cpp:137
↓ 93 callersFunctionmax
Reductions
mlx/backend/cpu/simd/neon_fp16_simd.h:182
↓ 93 callersMethodvalue
mlx/event.h:34
↓ 90 callersFunctionPM
mlx/3rdparty/pocketfft.h:273
↓ 90 callersMethodcopy_shared_buffer
mlx/array.cpp:191
↓ 90 callersFunctionuniform
mlx/random.h:73
↓ 88 callersFunctionmatmul
mlx/ops.cpp:3192
↓ 85 callersFunctionconcatenate
mlx/backend/metal/utils.h:54
↓ 84 callersMethodempty
mlx/compile.cpp:378
↓ 83 callersFunctionfma
mlx/backend/cpu/simd/neon_fp16_simd.h:171
↓ 83 callersFunctionnorm
mlx/linalg.h:32
↓ 81 callersMethodvjp
mlx/fast.cpp:13
↓ 78 callersMethodadd
python/src/array.cpp:43
↓ 72 callersMethodseed
python/src/random.cpp:32
↓ 71 callersMethodsynchronize
mlx/backend/cuda/device.cpp:538
↓ 70 callersMethoddata
mlx/3rdparty/pocketfft.h:226
↓ 70 callersFunctionexp
mlx/ops.cpp:2988
↓ 70 callersFunctionjvp
mlx/transforms.cpp:543
↓ 70 callersMethodmean
(x)
python/tests/test_compile.py:701
↓ 69 callersFunctionf
(x)
python/mlx/nn/layers/distributed.py:20
↓ 69 callersMethodfind
mlx/backend/cuda/lru_cache.h:75
↓ 69 callersFunctionmax
mlx/ops.h:709
↓ 68 callersMethodinsert
mlx/backend/metal/resident.cpp:28
↓ 66 callersMethodassertEqualArray
( self, mx_res: mx.array, expected: mx.array, atol=1e-2, rtol=1e-2,
python/tests/mlx_tests.py:100
↓ 65 callersFunctionall
mlx/ops.h:526
↓ 65 callersMethodreserve
mlx/small_vector.h:431
↓ 65 callersFunctionzeros_like
mlx/ops.cpp:334
↓ 64 callersMethodrank
mlx/distributed/distributed.cpp:129
↓ 61 callersFunctiontime_fn
(fn, *args, **kwargs)
benchmarks/python/time_utils.py:8
↓ 59 callersFunctionisnan
mlx/backend/cpu/simd/base_simd.h:158
↓ 58 callersFunctioncompile
mlx/compile.h:26
↓ 58 callersMethodempty
mlx/backend/cuda/lru_cache.h:48
↓ 58 callersFunctionsqueeze
mlx/ops.cpp:557
↓ 57 callersMethoddispatch_threads
mlx/backend/metal/device.cpp:364
↓ 57 callersFunctiondivide
mlx/ops.cpp:2881
↓ 56 callersMethodcheck_mx_np
(self, op_mx, op_np, a_np, atol=1e-5, rtol=1e-6, **kwargs)
python/tests/test_fft.py:19
↓ 56 callersFunctionflatten
mlx/ops.cpp:457
↓ 55 callersMethodmma
(BM, BK) X (BK, BN) multiply accumulate function */
mlx/backend/metal/kernels/steel/gemm/mma.h:508
↓ 54 callersFunctionsubtract
mlx/ops.cpp:2849
↓ 52 callersMethodadd_temporary
mlx/backend/cuda/device.h:103
↓ 51 callersMethodis_available
mlx/array.cpp:131
↓ 51 callersMethodquantize
(w, transpose=True, group_size=None, bits=None, mode="affine")
python/tests/test_quantized.py:757
↓ 50 callersFunctionsqrt
mlx/ops.cpp:3792
↓ 49 callersFunctionelem_to_loc
mlx/backend/common/utils.h:16
↓ 48 callersMethodinit
Initialize the optimizer's state This function can be used to initialize optimizers which have state (like momentum in :class:`SGD`).
python/mlx/optimizers/optimizers.py:31
↓ 48 callersMethodparameters
Recursively return all the :class:`mlx.core.array` members of this Module as a dict of dicts and lists.
python/mlx/nn/layers/base.py:280
↓ 48 callersFunctiontake
mlx/ops.cpp:3385
next →1–100 of 4,946, ranked by callers