MCPcopy Create free account

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

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

↓ 1 callersFunctiongather_qmm_nax
mlx/backend/metal/quantized.cpp:576
↓ 1 callersFunctiongather_qmm_rhs
mlx/backend/metal/quantized.cpp:1215
↓ 1 callersFunctiongather_qmm_rhs_nax
mlx/backend/metal/quantized.cpp:1084
↓ 1 callersFunctiongather_qvm
mlx/backend/metal/quantized.cpp:1026
↓ 1 callersFunctiongelu
r"""Applies the Gaussian Error Linear Units function. .. math:: \textrm{GELU}(x) = x * \Phi(x) where :math:`\Phi(x)` is the Gaussian
python/mlx/nn/layers/activations.py:154
↓ 1 callersFunctiongelu_approx
r"""An approximation to Gaussian Error Linear Unit. See :func:`gelu` for the exact computation. This function approximates ``gelu`` with a m
python/mlx/nn/layers/activations.py:169
↓ 1 callersFunctiongelu_fast_approx
r"""A fast approximation to Gaussian Error Linear Unit. See :func:`gelu` for the exact computation. This function approximates ``gelu`` with
python/mlx/nn/layers/activations.py:186
↓ 1 callersFunctiongemv_axbpy
mlx/backend/metal/matmul.cpp:1043
↓ 1 callersFunctiongemv_masked
Batch ndim > 1 */
mlx/backend/metal/kernels/gemv_masked.h:644
↓ 1 callersFunctiongen_hadamard_codelet
mlx/backend/metal/hadamard.cpp:17
↓ 1 callersFunctionget_active_memory
mlx/backend/metal/allocator.cpp:252
↓ 1 callersMethodget_active_memory
mlx/backend/no_gpu/allocator.cpp:30
↓ 1 callersMethodget_active_memory
mlx/backend/metal/allocator.h:26
↓ 1 callersFunctionget_all_optimizers
()
python/tests/test_optimizers.py:25
↓ 1 callersFunctionget_arange_kernel
mlx/backend/metal/jit_kernels.cpp:11
↓ 1 callersFunctionget_binary_kernel
mlx/backend/metal/jit_kernels.cpp:104
↓ 1 callersFunctionget_binary_two_kernel
mlx/backend/metal/jit_kernels.cpp:121
↓ 1 callersMethodget_cache_memory
mlx/backend/metal/allocator.h:36
↓ 1 callersFunctionget_command_buffer
mlx/backend/metal/device.h:100
↓ 1 callersFunctionget_conv_settings
mlx/backend/cuda/conv.cpp:49
↓ 1 callersFunctionget_cpu_name
Get CPU device name (brand string)
mlx/backend/cpu/device_info.cpp:51
↓ 1 callersFunctionget_cublas_handles
mlx/backend/cuda/cublas_utils.cpp:59
↓ 1 callersFunctionget_device_name
()
benchmarks/python/masked_scatter.py:28
↓ 1 callersFunctionget_dynamic_copy_kernel
mlx/backend/metal/jit_kernels.cpp:258
↓ 1 callersFunctionget_fft_kernel
mlx/backend/metal/jit_kernels.cpp:832
↓ 1 callersFunctionget_function_address
mlx/compile.cpp:292
↓ 1 callersFunctionget_gather_qmm_kernel
mlx/backend/metal/jit_kernels.cpp:868
↓ 1 callersFunctionget_gather_qmm_nax_kernel
mlx/backend/metal/jit_kernels.cpp:1069
↓ 1 callersFunctionget_gemv_masked_kernel
mlx/backend/metal/jit_kernels.cpp:701
↓ 1 callersFunctionget_gflop_count
(B, M, N, K)
benchmarks/python/blas/bench_gemm.py:157
↓ 1 callersFunctionget_graph_limits
Can be tuned with MLX_MAX_OPS_PER_BUFFER, MLX_MAX_MB_PER_BUFFER
mlx/backend/cuda/device.cpp:181
↓ 1 callersFunctionget_handle_of_object
python/src/utils.h:57
↓ 1 callersMethodget_kernel_and_dims
mlx/backend/cuda/jit_module.cpp:414
↓ 1 callersFunctionget_kernel_name
mlx/backend/metal/binary.cpp:21
↓ 1 callersFunctionget_libmpi_name
mlx/distributed/mpi/mpi.cpp:23
↓ 1 callersFunctionget_logsumexp_kernel
mlx/backend/metal/jit_kernels.cpp:310
↓ 1 callersMethodget_memory_limit
mlx/backend/no_gpu/allocator.cpp:40
↓ 1 callersMethodget_memory_limit
mlx/backend/metal/allocator.cpp:85
↓ 1 callersMethodget_memory_limit
mlx/backend/cuda/allocator.cpp:360
↓ 1 callersFunctionget_memory_size
mlx/backend/no_gpu/allocator.cpp:14
↓ 1 callersMethodget_mesh_connectivity
mlx/distributed/jaccl/lib/jaccl/jaccl.cpp:145
↓ 1 callersFunctionget_metal_version
mlx/backend/metal/device.cpp:36
↓ 1 callersFunctionget_mpi_libname
()
python/mlx/_distributed_utils/launch.py:396
↓ 1 callersMethodget_peak_memory
mlx/backend/no_gpu/allocator.cpp:33
↓ 1 callersMethodget_peak_memory
mlx/backend/metal/allocator.h:29
↓ 1 callersMethodget_prefer_ring
mlx/distributed/jaccl/lib/jaccl/jaccl.h:37
↓ 1 callersFunctionget_qmm_nax_kernel
mlx/backend/metal/jit_kernels.cpp:1049
↓ 1 callersFunctionget_quant_mode_config
mlx/backend/cuda/quantized/cublas_qqmm.cpp:22
↓ 1 callersFunctionget_quantized_kernel
mlx/backend/metal/jit_kernels.cpp:848
↓ 1 callersFunctionget_reduce_init_kernel
mlx/backend/metal/jit_kernels.cpp:432
↓ 1 callersFunctionget_scan_kernel
mlx/backend/metal/jit_kernels.cpp:329
↓ 1 callersFunctionget_shape
python/src/convert.cpp:345
↓ 1 callersFunctionget_softmax_kernel
mlx/backend/metal/jit_kernels.cpp:290
↓ 1 callersFunctionget_sort_kernel
mlx/backend/metal/jit_kernels.cpp:364
↓ 1 callersFunctionget_steel_attention_kernel
mlx/backend/metal/jit_kernels.cpp:1113
↓ 1 callersFunctionget_steel_attention_nax_kernel
mlx/backend/metal/jit_kernels.cpp:1147
↓ 1 callersFunctionget_steel_conv_3d_kernel
mlx/backend/metal/jit_kernels.cpp:773
↓ 1 callersFunctionget_steel_conv_general_kernel
mlx/backend/metal/jit_kernels.cpp:802
↓ 1 callersFunctionget_steel_conv_kernel
mlx/backend/metal/jit_kernels.cpp:742
↓ 1 callersFunctionget_steel_gemm_fused_kernel
mlx/backend/metal/jit_kernels.cpp:487
↓ 1 callersFunctionget_steel_gemm_fused_nax_kernel
mlx/backend/metal/jit_kernels.cpp:909
↓ 1 callersFunctionget_steel_gemm_gather_nax_kernel
mlx/backend/metal/jit_kernels.cpp:943
↓ 1 callersFunctionget_steel_gemm_masked_kernel
mlx/backend/metal/jit_kernels.cpp:581
↓ 1 callersFunctionget_steel_gemm_segmented_kernel
mlx/backend/metal/jit_kernels.cpp:664
↓ 1 callersFunctionget_steel_gemm_segmented_nax_kernel
mlx/backend/metal/jit_kernels.cpp:1015
↓ 1 callersFunctionget_steel_gemm_splitk_kernel
mlx/backend/metal/jit_kernels.cpp:521
↓ 1 callersFunctionget_steel_gemm_splitk_nax_kernel
mlx/backend/metal/jit_kernels.cpp:981
↓ 1 callersFunctionget_ternary_kernel
mlx/backend/metal/jit_kernels.cpp:137
↓ 1 callersFunctionget_unary_kernel
mlx/backend/metal/jit_kernels.cpp:25
↓ 1 callersFunctiongguf_load_quantized
mlx/io/gguf_quants.cpp:100
↓ 1 callersFunctiongguf_type_to_dtype
mlx/io/gguf.cpp:33
↓ 1 callersFunctionglu
r"""Applies the gated linear unit function. This function splits the ``axis`` dimension of the input into two halves (:math:`a` and :math:`b`
python/mlx/nn/layers/activations.py:207
↓ 1 callersFunctiongrad_fun
tests/compile_tests.cpp:40
↓ 1 callersFunctiongreedy_path
mlx/einsum.cpp:161
↓ 1 callersFunctionhadamard
mlx/backend/cpu/hadamard.cpp:78
↓ 1 callersFunctionhas_primitive
Check if the array has an attached primitive or is a leaf node. */
mlx/array.h:280
↓ 1 callersMethodhas_values
mlx/backend/cuda/quantized/qqmm_impl.h:15
↓ 1 callersFunctionidentity
r"""An initializer that returns an identity matrix. Args: dtype (Dtype, optional): The data type of the array. Default: ``float
python/mlx/nn/init.py:99
↓ 1 callersFunctionifft2
Compute the two-dimensional inverse Fourier Transform. */
mlx/fft.h:103
↓ 1 callersFunctionimag
mlx/backend/cpu/simd/base_simd.h:154
↓ 1 callersFunctionimplicit_gemm_conv_2D_general_gpu
mlx/backend/metal/conv.cpp:324
↓ 1 callersFunctionin_grad_tracing
Return true if we are in a gradient trace (vjp, jvp, etc). */
mlx/transforms_impl.h:80
↓ 1 callersMethodin_pool
mlx/backend/cuda/allocator.cpp:153
↓ 1 callersMethodinit
(self, parameters: dict)
python/mlx/optimizers/optimizers.py:198
↓ 1 callersFunctioninit_array
python/src/array.cpp:100
↓ 1 callersFunctioninit_constants
python/src/constants.cpp:8
↓ 1 callersFunctioninit_cublas_handles_cache
mlx/backend/cuda/cublas_utils.cpp:91
↓ 1 callersFunctioninit_cuda
python/src/cuda.cpp:10
↓ 1 callersFunctioninit_cudnn_conv_cache
mlx/backend/cuda/conv.cpp:255
↓ 1 callersFunctioninit_cudnn_handles_cache
mlx/backend/cuda/cudnn_utils.cpp:89
↓ 1 callersFunctioninit_cudnn_sdpa_cache
mlx/backend/cuda/scaled_dot_product_attention.cpp:310
↓ 1 callersFunctioninit_device
python/src/device.cpp:19
↓ 1 callersFunctioninit_distributed
python/src/distributed.cpp:19
↓ 1 callersFunctioninit_export
python/src/export.cpp:134
↓ 1 callersFunctioninit_fast
python/src/fast.cpp:80
↓ 1 callersFunctioninit_fft
python/src/fft.cpp:39
↓ 1 callersFunctioninit_linalg
python/src/linalg.cpp:18
↓ 1 callersFunctioninit_memory
python/src/memory.cpp:10
↓ 1 callersFunctioninit_metal
python/src/metal.cpp:28
↓ 1 callersFunctioninit_mlx_func
python/src/mlx_func.cpp:111
← previousnext →1,401–1,500 of 4,946, ranked by callers