MCPcopy Create free account

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

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

⨍ Functions 4,946◇ Types & classes 770↳ Endpoints 5
↓ 4 callersFunctionprint_constant
mlx/backend/common/compiled.cpp:9
↓ 4 callersFunctionprod
mlx/backend/cpu/simd/neon_fp16_simd.h:203
↓ 4 callersFunctionqr
mlx/linalg.cpp:226
↓ 4 callersFunctionradix5
mlx/backend/metal/kernels/fft/radix.h:69
↓ 4 callersFunctionreal
mlx/ops.cpp:6402
↓ 4 callersFunctionreshape_gpu
mlx/backend/metal/copy.cpp:216
↓ 4 callersMethodrun
(self)
setup.py:153
↓ 4 callersFunctionrun_or_raise
(*args, **kwargs)
benchmarks/python/comparative/compare.py:14
↓ 4 callersFunctionsave_hostfile
(args, hostfile)
python/mlx/_distributed_utils/config.py:72
↓ 4 callersFunctionsave_safetensors
mlx/io/safetensors.cpp:217
↓ 4 callersMethodscatter
(a)
python/tests/test_vmap.py:446
↓ 4 callersFunctionscatter_unsort
(x, inv_order, shape=None)
benchmarks/python/gather_mm_bench.py:21
↓ 4 callersFunctionseed
mlx/random.cpp:26
↓ 4 callersMethodsend
mlx/distributed/utils.cpp:137
↓ 4 callersFunctionset_cache_limit
mlx/backend/cuda/allocator.cpp:439
↓ 4 callersMethodset_cache_limit
mlx/backend/cuda/allocator.cpp:374
↓ 4 callersFunctionset_memory_limit
mlx/backend/cuda/allocator.cpp:430
↓ 4 callersMethodset_wired_limit
mlx/backend/metal/allocator.cpp:89
↓ 4 callersFunctionsimd_shuffle_down
mlx/backend/metal/kernels/utils.h:357
↓ 4 callersFunctionskip_compile
mlx/compile.cpp:1093
↓ 4 callersFunctionslice_gpu
mlx/backend/gpu/slicing.cpp:9
↓ 4 callersFunctionsolve_triangular
mlx/linalg.cpp:698
↓ 4 callersFunctionsqueeze_leading_singletons
python/src/indexing.cpp:501
↓ 4 callersMethodstate
mlx/primitives.h:1092
↓ 4 callersFunctionstatus
mlx/array.h:410
↓ 4 callersFunctionsupported_radices
mlx/backend/metal/fft.cpp:30
↓ 4 callersFunctionsupports_managed_memory
mlx/backend/cuda/allocator.cpp:54
↓ 4 callersFunctionsynchronize
mlx/scheduler.cpp:8
↓ 4 callersFunctiontanh
mlx/ops.cpp:3060
↓ 4 callersFunctiontile_matmad
mlx/backend/metal/kernels/steel/gemm/mma.h:402
↓ 4 callersFunctiontime
examples/cpp/timer.h:16
↓ 4 callersFunctiontime_fn
(fn, *args)
benchmarks/numpy/time_utils.py:6
↓ 4 callersFunctionto_complex
mlx/backend/cpu/eig.cpp:16
↓ 4 callersFunctiontree_visit
python/src/trees.cpp:94
↓ 4 callersFunctiontype_name_str
python/src/transforms.cpp:36
↓ 4 callersFunctionvalidate_eig
mlx/linalg.cpp:502
↓ 4 callersFunctionvalidate_global_scale
mlx/ops.cpp:4449
↓ 4 callersFunctionversion
mlx/version.cpp:7
↓ 4 callersFunctionversion_lower_equal
Return if |device|'s version is not newer than |major|.|minor| version.
mlx/backend/cuda/jit_module.cpp:212
↓ 4 callersFunctionwrite_raw_safetensors
Helper to write a raw safetensors file from a JSON header and data buffer
tests/load_tests.cpp:45
↓ 3 callersFunctionDevice
mlx/device.h:22
↓ 3 callersFunctionGroup
mlx/distributed/distributed.h:28
↓ 3 callersMethod__init__
( self, input_dims: int, output_dims: int, bias: bool = True, group: O
python/mlx/nn/layers/distributed.py:209
↓ 3 callersMethod__init__
(self)
python/tests/test_nn.py:59
↓ 3 callersFunction_all_to_sharded
Simple predicate to shard fully connected layers such that a common representation becomes a sharded representation.
python/mlx/nn/layers/distributed.py:84
↓ 3 callersFunction_defaults_for_mode
(mode, group_size, bits)
python/mlx/nn/layers/quantized.py:11
↓ 3 callersMethod_hadamard
(self, N)
python/tests/test_ops.py:2929
↓ 3 callersFunction_launch_with_io
(command_class, arguments, verbose)
python/mlx/_distributed_utils/launch.py:170
↓ 3 callersFunction_sharded_to_all
Simple predicate to shard fully connected layers such that a sharded representation becomes a common representation.
python/mlx/nn/layers/distributed.py:96
↓ 3 callersFunctionadd_ips
(hosts, verbose=False)
python/mlx/_distributed_utils/config.py:47
↓ 3 callersMethodallocate_protection_domain
mlx/distributed/jaccl/lib/jaccl/rdma.cpp:141
↓ 3 callersFunctionarctan
mlx/ops.cpp:3033
↓ 3 callersFunctionaxpby
* Scale and sum two vectors element-wise * z = alpha * x + beta * y * * Follow numpy style broadcasting between x and y * Inputs are upcasted
examples/extensions/axpby/axpby.cpp:44
↓ 3 callersFunctionbinary_float_op_cpu
mlx/backend/cpu/binary.h:425
↓ 3 callersFunctionblock_sort
mlx/backend/metal/kernels/sort.h:319
↓ 3 callersMethodbuild
mlx/backend/cuda/cudnn_utils.cpp:105
↓ 3 callersFunctioncan_auto_setup
(hosts, sshinfo, auto_setup=False)
python/mlx/_distributed_utils/config.py:111
↓ 3 callersFunctioncategorical_impl
mlx/random.cpp:382
↓ 3 callersFunctioncheck_error
mlx/backend/metal/eval.cpp:21
↓ 3 callersFunctioncheck_rdma
(hosts, verbose=False, strict=True)
python/mlx/_distributed_utils/config.py:82
↓ 3 callersFunctioncheck_valid_ring
(hosts, rings, strict=True)
python/mlx/_distributed_utils/config.py:356
↓ 3 callersFunctioncheckpoint
Transform the passed callable to one that performs gradient checkpointing with respect to the trainable parameters of the module (and the call
python/mlx/nn/utils.py:41
↓ 3 callersMethodclose
mlx/export.cpp:694
↓ 3 callersMethodcommit
mlx/backend/cuda/device.cpp:463
↓ 3 callersFunctioncompiled_allocate_outputs
mlx/backend/common/compiled.cpp:113
↓ 3 callersFunctioncompiled_collapse_contiguous_dims
mlx/backend/common/compiled.cpp:173
↓ 3 callersMethodconcurrent_context
mlx/backend/cuda/device.h:45
↓ 3 callersFunctionconv1d
1D convolution with a filter */
mlx/ops.cpp:4091
↓ 3 callersFunctionconv_transpose_general
Helper function for transposed convolutions
mlx/ops.cpp:4158
↓ 3 callersFunctioncopy_inplace_dispatch
mlx/backend/cpu/copy.cpp:245
↓ 3 callersFunctioncreate_array
python/src/convert.cpp:460
↓ 3 callersMethodcreate_completion_queue
mlx/distributed/jaccl/lib/jaccl/rdma.cpp:148
↓ 3 callersFunctioncreate_connections
mlx/distributed/jaccl/lib/jaccl/rdma.cpp:266
↓ 3 callersMethodcreate_queue_pair
mlx/distributed/jaccl/lib/jaccl/rdma.cpp:155
↓ 3 callersFunctiondepth_first_traversal
mlx/graph_utils.cpp:37
↓ 3 callersFunctiondequantize
mlx/ops.cpp:5171
↓ 3 callersFunctiondo_attention
(f, q, k, v, scale, mask=None, transpose=False)
benchmarks/python/sdpa_bench.py:108
↓ 3 callersFunctiondtype
Get the arrays data type. */
mlx/array.h:143
↓ 3 callersFunctiondtype_to_jaccl_dtype
* Map MLX Dtype to JACCL Dtype for dispatch. */
mlx/distributed/jaccl/jaccl.cpp:20
↓ 3 callersFunctioneigvalsh
mlx/linalg.cpp:521
↓ 3 callersFunctionensure_row_contiguous
mlx/backend/cuda/matmul.cpp:54
↓ 3 callersMethodeval
mlx/backend/common/common.cpp:10
↓ 3 callersFunctionexec
mlx/3rdparty/pocketfft.h:2430
↓ 3 callersFunctionexpand
(mask, rows, cols)
benchmarks/python/block_masked_mm_bench.py:43
↓ 3 callersFunctionexplicit_gemm_conv_ND_cpu
mlx/backend/cpu/conv.cpp:999
↓ 3 callersFunctionexplicit_gemm_conv_ND_gpu
mlx/backend/metal/conv.cpp:34
↓ 3 callersFunctionexplicit_gemm_conv_group_ND_gpu
mlx/backend/metal/conv.cpp:105
↓ 3 callersFunctionextract_quantized_matmul_dims
mlx/ops.cpp:117
↓ 3 callersFunctionextract_rings
(connectivity)
python/mlx/_distributed_utils/config.py:308
↓ 3 callersFunctionfinalize
mlx/backend/cuda/eval.cpp:73
↓ 3 callersFunctionflatten_in_eval
mlx/backend/gpu/copy.cpp:43
↓ 3 callersFunctionforward
(x)
examples/export/eval_mlp.py:39
↓ 3 callersFunctiongather_sort
(x, indices)
benchmarks/python/gather_mm_bench.py:13
↓ 3 callersFunctiongemm_and_bias
mlx/backend/cuda/matmul.cpp:67
↓ 3 callersFunctionget_bytes_per_pack
mlx/backend/common/quantized.h:9
↓ 3 callersMethodget_coordinator
mlx/distributed/jaccl/lib/jaccl/jaccl.h:33
↓ 3 callersFunctionget_cudnn_handle
mlx/backend/cuda/cudnn_utils.cpp:80
↓ 3 callersFunctionget_float_word
mlx/backend/metal/kernels/cexpf.h:32
↓ 3 callersFunctionget_mb_sort_kernel
mlx/backend/metal/jit_kernels.cpp:402
↓ 3 callersMethodget_rank
mlx/distributed/jaccl/lib/jaccl/jaccl.h:25
← previousnext →701–800 of 4,946, ranked by callers