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
↓ 12 callersFunctioncheck_transpose
mlx/backend/metal/matmul.cpp:25
↓ 12 callersMethodfft
mlx/3rdparty/pocketfft.h:2369
↓ 12 callersFunctionfft_impl
mlx/fft.cpp:31
↓ 12 callersFunctionfill_gpu
mlx/backend/metal/copy.cpp:182
↓ 12 callersFunctionfloor
mlx/ops.cpp:2968
↓ 12 callersFunctionget_temp_file
tests/load_tests.cpp:14
↓ 12 callersFunctionis_power_of_2
mlx/utils.h:133
↓ 12 callersFunctionisneginf
mlx/ops.cpp:1867
↓ 12 callersFunctionisposinf
mlx/ops.cpp:1860
↓ 12 callersFunctionnot_equal
mlx/ops.cpp:1766
↓ 12 callersFunctionparse_norm
python/src/fft.cpp:22
↓ 12 callersMethodresize
mlx/threadpool.h:82
↓ 12 callersFunctionroll
mlx/ops.cpp:6317
↓ 12 callersMethodset_value
mlx/event.h:38
↓ 12 callersFunctionslice_update_add
mlx/ops.cpp:950
↓ 12 callersMethodstore
mlx/backend/metal/kernels/steel/gemm/nax.h:681
↓ 11 callersFunctionceil
mlx/ops.cpp:2976
↓ 11 callersFunctionconjugate
mlx/ops.cpp:6190
↓ 11 callersFunctioncopy_gpu
mlx/backend/gpu/copy.cpp:11
↓ 11 callersFunctioncopy_gpu
mlx/backend/metal/copy.cpp:13
↓ 11 callersFunctioncumprod
mlx/ops.cpp:3904
↓ 11 callersMethodencode
mlx/backend/metal/reduce.cpp:58
↓ 11 callersFunctionexport_function
mlx/export.cpp:918
↓ 11 callersMethodgather_qmm
(x, w, s, b, lhs, rhs, trans, sort)
python/tests/test_quantized.py:1069
↓ 11 callersFunctionget_reduce_kernel
mlx/backend/metal/jit_kernels.cpp:452
↓ 11 callersFunctionget_var
mlx/utils.cpp:289
↓ 11 callersFunctionget_work_per_thread
mlx/backend/metal/utils.h:72
↓ 11 callersMethodiinfo
mlx/utils.cpp:340
↓ 11 callersFunctionimport_function
mlx/export.cpp:1034
↓ 11 callersMethodload_weights
Update the model's weights from a ``.npz``, a ``.safetensors`` file, or a list. Args: file_or_weights (str or list(tuple
python/mlx/nn/layers/base.py:123
↓ 11 callersFunctionlogical_not
mlx/ops.cpp:2795
↓ 11 callersFunctionprepare_sdpa_input
mlx/backend/cuda/scaled_dot_product_attention.cpp:15
↓ 11 callersMethodprod
(x)
python/tests/test_autograd.py:970
↓ 11 callersFunctionrandint
mlx/random.h:157
↓ 11 callersMethodread
mlx/io/load.cpp:347
↓ 11 callersFunctionright_shift
mlx/ops.cpp:6257
↓ 11 callersFunctionround
mlx/ops.h:1046
↓ 11 callersFunctionsigmoid
mlx/ops.cpp:3147
↓ 11 callersMethodupdate
Replace the parameters of this Module with the provided ones in the dict of dicts and lists. Commonly used by the optimizer to change
python/mlx/nn/layers/base.py:304
↓ 11 callersFunctionview
mlx/ops.cpp:6289
↓ 10 callersMethod__init__
( self, learning_rate: Union[float, Callable[[mx.array], mx.array]], momentum: float =
python/mlx/optimizers/optimizers.py:248
↓ 10 callersFunctionappend_kv_array
mlx/io/gguf.cpp:261
↓ 10 callersMethodapply
mlx/backend/metal/kernels/steel/gemm/mma.h:426
↓ 10 callersFunctionargmax
mlx/ops.h:765
↓ 10 callersFunctionargmin
mlx/ops.h:755
↓ 10 callersFunctionbinary_op_cpu
mlx/backend/cpu/binary.h:295
↓ 10 callersFunctiondiag
mlx/ops.cpp:5990
↓ 10 callersFunctionenable_tf32
mlx/utils.h:171
↓ 10 callersFunctionensure_row_contiguous_matrix
mlx/backend/metal/quantized.cpp:64
↓ 10 callersFunctionfftshift
mlx/fft.cpp:243
↓ 10 callersFunctiongemm_loop_finalize
mlx/backend/metal/kernels/quantized_utils.h:78
↓ 10 callersFunctiongen_fun
(fun)
benchmarks/python/compile_bench.py:17
↓ 10 callersFunctioninit
mlx/backend/cuda/eval.cpp:15
↓ 10 callersMethodlabel
mlx/io/load.h:169
↓ 10 callersFunctionlaplace
mlx/random.h:248
↓ 10 callersFunctionleft_shift
mlx/ops.cpp:6246
↓ 10 callersMethodload_safe
mlx/backend/metal/kernels/quantized.h:641
↓ 10 callersMethodload_safe
mlx/backend/metal/kernels/fp_quantized.h:218
↓ 10 callersMethodload_unsafe
mlx/backend/metal/kernels/quantized.h:628
↓ 10 callersMethodload_unsafe
mlx/backend/metal/kernels/fp_quantized.h:206
↓ 10 callersMethodlogsumexp
(x, axes=None)
python/tests/test_ops.py:734
↓ 10 callersMethodmake_current
mlx/backend/cuda/device.cpp:59
↓ 10 callersFunctionmin
mlx/ops.cpp:2359
↓ 10 callersMethodnext
python/src/random.cpp:36
↓ 10 callersMethodnext
Iteration helper */
mlx/backend/metal/kernels/steel/conv/loaders/loader_channel_l.h:438
↓ 10 callersMethodregister_to_protection_domain
mlx/distributed/jaccl/lib/jaccl/rdma.cpp:92
↓ 10 callersMethodsave
mlx/export.cpp:461
↓ 10 callersFunctionset_binary_op_output_data
mlx/backend/common/binary.h:37
↓ 10 callersFunctionset_compile_mode
mlx/compile.cpp:1252
↓ 10 callersFunctionshard_linear
Create a new linear layer that has its parameters sharded and also performs distributed communication either in the forward or backward pass.
python/mlx/nn/layers/distributed.py:158
↓ 10 callersFunctionsvd
mlx/linalg.h:68
↓ 10 callersFunctiontile_matmad_nax
mlx/backend/metal/kernels/steel/gemm/nax.h:825
↓ 10 callersFunctiontree_unflatten_from_structure
python/src/trees.cpp:299
↓ 10 callersFunctiontruncated_normal
mlx/random.cpp:324
↓ 10 callersFunctionuse_cuda_graphs
mlx/backend/cuda/device.cpp:17
↓ 9 callersFunctionDEPRECATE
python/src/metal.cpp:20
↓ 9 callersMethod__init__
(self, pooling_function, kernel_size, stride, padding, padding_value)
python/mlx/nn/layers/pooling.py:85
↓ 9 callersFunction_value_or_list
(x, n, msg)
python/mlx/nn/layers/pooling.py:11
↓ 9 callersMethodadd_kernel_node_raw
mlx/backend/cuda/device.cpp:243
↓ 9 callersFunctionadd_strides_and_shapes
mlx/backend/metal/quantized.cpp:128
↓ 9 callersMethodall_gather
mlx/distributed/mpi/mpi.cpp:434
↓ 9 callersMethodargs
mlx/backend/cuda/jit_module.h:29
↓ 9 callersFunctionas_strided
mlx/ops.cpp:270
↓ 9 callersFunctionbinary_op_gpu
mlx/backend/metal/binary.cpp:165
↓ 9 callersFunctioncheck_cpu_stream
mlx/linalg.cpp:13
↓ 9 callersMethodcompute_capability_major
mlx/backend/cuda/device.h:176
↓ 9 callersFunctioncosh
mlx/ops.cpp:3054
↓ 9 callersFunctiondispatch_all_types
mlx/distributed/jaccl/lib/jaccl/types.h:216
↓ 9 callersFunctionexpm1
mlx/ops.cpp:2994
↓ 9 callersFunctionfill_vector
python/src/convert.cpp:262
↓ 9 callersFunctionfloor_divide
mlx/ops.cpp:2899
↓ 9 callersMethodfreeze
Freeze the Module's parameters or some of them. Freezing a parameter means not computing gradients for it. This function is idempoten
python/mlx/nn/layers/base.py:464
↓ 9 callersFunctionget_binary_op_type
mlx/backend/common/binary.h:19
↓ 9 callersFunctionget_cache_memory
mlx/backend/cuda/allocator.cpp:436
↓ 9 callersMethodhadamard_transform
(x)
python/tests/test_ops.py:3043
↓ 9 callersFunctionis_str_or_path
python/src/load.cpp:26
↓ 9 callersFunctionload_gguf
mlx/io/gguf.cpp:241
↓ 9 callersMethodload_safe
mlx/backend/metal/kernels/steel/attn/nax.h:743
↓ 9 callersFunctionmasked_scatter
mlx/ops.cpp:3709
↓ 9 callersFunctionmlx_primitives_sdpa
(q, k, v, scale, mask=None)
python/tests/test_fast_sdpa.py:99
← previousnext →301–400 of 4,946, ranked by callers