Repository Analysis

jax-ml/jax

Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more

18.9 Moderate AI signal View on GitHub

Analysis Overview

This report presents the forensic synthetic code analysis of jax-ml/jax, a Python project with 36,007 GitHub stars. SynthScan v2.0 examined 669,651 lines of code across 1402 source files, recording 12595 pattern matches distributed across 26 syntactic categories. The overall adjusted score of 18.9 places this repository in the Moderate AI signal band.

The scanner applied 160+ deterministic lexical heuristics, multi-line block detectors, abstract syntax tree depth profilers, and a cross-file Jaccard similarity matrix to construct a statistically normalised synthetic code estimate. All matches are individually weighted by severity coefficient and contextual multiplier before summation, and the resulting headline score is temporally discounted to account for the repository's development history relative to the commercial emergence of large language model coding tooling (November 2022 onward).

18.9
Adjusted Score
18.9
Raw Score
100%
Time Factor
2026-07-14
Last Push
36.0K
Stars
Python
Language
669.7K
Lines of Code
1.4K
Files
12.6K
Pattern Hits
2026-07-14
Scan Date
0.45
HC Hit Rate

What These Metrics Mean

Adjusted Score
Primary synthetic code indicator. Raw score normalised per 1,000 lines of code and multiplied by the temporal discount factor. This is the definitive comparative metric — use it to rank repositories by AI authorship density.
Raw Score
The unmodified sum of all severity-weighted, context-multiplied pattern match scores before temporal discounting. Reflects the absolute signal strength independent of when the repository was last active.
Time Factor
The temporal discount multiplier (0–100%) applied to the raw score. Repositories last updated before ChatGPT's launch (Nov 2022) receive a 5% factor. Full signal is only assigned to repositories active in the post-adoption era (Jan 2024+).
Pattern Hits
Total count of individual pattern matches across all files and categories. A high hit count with a low score may indicate a very large codebase with isolated AI snippets; a low count with a high score indicates dense, concentrated AI signatures.
HC Hit Rate
High+Critical pattern hits per file, averaged across the repository. This orthogonal signal catches repositories where a few files are densely packed with high-severity AI tells — a strong indicator even when the normalised score appears moderate due to codebase size.
Lines of Code / Files
Total lines and files analysed. The scanner examines 94 file extensions. These denominators are used to normalise the score, enabling fair comparison between repositories of vastly different sizes.

Score History

This chart maps the temporal evolution of the adjusted synthetic code score across successive scan runs. An upward trajectory indicates ongoing incorporation of AI-generated code or expanding LLM-assisted scaffolding; a stable or declining trajectory may reflect active human refactoring, code removal, or the adoption of stricter authorship policies. The dashed secondary line (right axis) independently tracks total raw pattern hit count, which can diverge from the normalised score when codebase size changes significantly between scans.

Severity Breakdown

Classifies detected patterns by their diagnostic confidence and structural impact. CRITICAL patterns (coefficient 10) represent definitive synthetic signatures — hallucinated imports, explicit LLM attribution metadata — virtually never produced by human authors. HIGH (5) indicates strong structural tells such as cross-file repetition or cross-linguistic idioms. MEDIUM (2) covers recognisable conversational padding and AI-specific vocabulary. LOW (1) captures subtle indicators like tautological comments and generic boilerplate that require density to carry independent signal.

CRITICAL 3HIGH 628MEDIUM 309LOW 11655

Directory Score Breakdown

This horizontal bar chart decomposes the repository's raw synthetic code score by top-level directory, allowing you to pinpoint precisely which modules or components carry the highest AI authorship density. Directories with disproportionately high scores relative to their size warrant targeted manual review: concentrated AI signatures often trace back to mass-generated configuration layers, auto-ported test suites, LLM-scaffolded boilerplate classes, or entire subsystems authored under heavy copilot assistance. Use this view to prioritise your human code-review effort.

Pattern Findings

The scanner identified 12595 distinct pattern matches across 26 syntactic categories. Each entry below represents a discrete location in the source code where the engine recorded a statistically significant AI authorship indicator. Expand any category row to inspect the individual file paths, line numbers, code snippets, and the lexical context (CODE, COMMENT, or STRING) in which each match was detected.

Reading the findings table: The Severity column indicates the diagnostic confidence level (CRITICAL / HIGH / MEDIUM / LOW). The Context column identifies whether the match occurred inside executable code, an inline comment, or a string literal — comment-context matches receive a ×1.5 weight because LLMs systematically over-annotate. The ⚡ bolt icon marks clustered matches: three or more patterns within a 10-line window, each receiving an additional ×1.5 density multiplier as dense clusters constitute far stronger evidence of synthetic authorship than isolated hits.

Hyper-Verbose Identifiers5799 hits · 4489 pts
SeverityFileLineSnippetContext
LOWci/parse_wheel_metadata.py34def _extract_wheel_package_name(wheel_name: str) -> str | None:CODE
LOWci/parse_wheel_metadata.py52def parse_expected_wheel_versions(wheel_dir: Path) -> dict[str, str]:CODE
LOWtests/random_test.py569 def test_key_construction_with_dtype(self, dtype_spec):CODE
LOWtests/random_test.py573 def test_key_construction_with_both_impl_and_dtype(self):CODE
LOWtests/random_test.py583 def test_wrap_key_data_with_dtype(self, dtype_spec):CODE
LOWtests/random_test.py591 def test_wrap_key_data_with_both_impl_and_dtype(self):CODE
LOWtests/random_test.py1215 def test_key_make_like_other_key_via_dtype(self, prng_name):CODE
LOWtests/random_test.py1223 def test_key_wrap_like_other_key_via_dtype(self, prng_name):CODE
LOWtests/random_test.py1231 def test_key_impl_from_string_error(self):CODE
LOWtests/random_test.py1235 def test_key_impl_from_object_error(self):CODE
LOWtests/random_test.py1242 def test_key_impl_builtin_is_string_name(self, name):CODE
LOWtests/random_test.py199 def test_config_prngs_registered(self):CODE
LOWtests/random_test.py412 def test_threefry_gpu_kernel_lowering(self):CODE
LOWtests/random_test.py445 def test_threefry_split_fold_in_symmetry(self, make_key):CODE
LOWtests/random_test.py458 def test_threefry_split_vmapped_fold_in_symmetry(self, make_key):CODE
LOWtests/random_test.py472 def test_loggamma_nan_corner_case(self):CODE
LOWtests/random_test.py545 def test_default_prng_selection(self, make_key, name, impl):CODE
LOWtests/random_test.py557 def test_key_construction_with_explicit_impl_name(self, make_key, name, impl):CODE
LOWtests/random_test.py609 def test_legacy_prng_key_flag(self):CODE
LOWtests/random_test.py637 def test_seed_no_implicit_transfers(self, make_key):CODE
LOWtests/random_test.py713 def test_construction_upgrade_flag(self):CODE
LOWtests/random_test.py759 def test_key_dtype_attributes(self):CODE
LOWtests/random_test.py814 def test_cpp_dispatch_aot_normal(self):CODE
LOWtests/random_test.py827 def test_cpp_dispatch_aot_split(self):CODE
LOWtests/random_test.py905 def test_eval_shape_keys_in_out(self):CODE
LOWtests/random_test.py963 def test_dynamic_update_slice(self):CODE
LOWtests/random_test.py1053 def test_device_put_replicated(self):CODE
LOWtests/random_test.py1059 def test_make_array_from_callback(self):CODE
LOWtests/random_test.py1071 def test_make_array_from_single_device_arrays(self):CODE
LOWtests/random_test.py1081 def test_key_array_custom_jvp(self):CODE
LOWtests/random_test.py1111 def test_key_array_indexing_nd(self):CODE
LOWtests/random_test.py1136 def test_array_impl_attributes(self):CODE
LOWtests/random_test.py1194 def test_key_make_like_other_key(self, prng_name):CODE
LOWtests/random_test.py1204 def test_key_wrap_like_other_key(self, prng_name):CODE
LOWtests/random_test.py1262 def test_keyarray_custom_vjp_symbolic_zeros(self):CODE
LOWtests/random_test.py1277 def test_keyarray_array_conversion_fails(self):CODE
LOWtests/random_test.py1301def _double_threefry_random_bits(key, bit_width, shape):CODE
LOWtests/random_test.py1565 def test_full_like_with_key_fillvalue(self):CODE
LOWtests/random_test.py1586 def test_full_with_key_fillvalue(self):CODE
LOWtests/extend_test.py84 def test_key_make_with_custom_impl(self):CODE
LOWtests/extend_test.py92 def test_key_wrap_with_custom_impl(self):CODE
LOWtests/extend_test.py102 def test_key_make_with_custom_impl_via_dtype(self):CODE
LOWtests/extend_test.py109 def test_key_wrap_with_custom_impl_via_dtype(self):CODE
LOWtests/extend_test.py118 def test_key_dtype_and_spec_with_custom_impl(self):CODE
LOWtests/extend_test.py130 def test_unknown_platform_error(self):CODE
LOWtests/extend_test.py139 def test_hlo_sharding_roundtrip(self):CODE
LOWtests/scheduling_groups_test.py67 def test_xla_metadata_call_inlineable(self):CODE
LOWtests/scheduling_groups_test.py88 def test_xla_metadata_call_inlineable_remat_in_scan(self):CODE
LOWtests/scheduling_groups_test.py105 def test_xla_metadata_call_deduplication(self):CODE
LOWtests/scheduling_groups_test.py133 def test_xla_metadata_call_deduplication_remat(self):CODE
LOWtests/scheduling_groups_test.py162 def test_xla_metadata_call_deduplication_kwargs(self):CODE
LOWtests/jax_to_ir_test.py65 def test_jax_to_hlo_with_constants(self):CODE
LOWtests/jax_to_ir_test.py85 def test_parse_shape_str_invalid(self):CODE
LOWtests/ragged_collective_test.py134 def test_ragged_all_to_all_grad(self, axis_name, mesh_axes):CODE
LOWtests/ragged_collective_test.py210 def test_ragged_all_to_all_axis_index_groups(self, axis_name, mesh_axes):CODE
LOWtests/ragged_collective_test.py304 def test_ragged_all_to_all_degenerate_groups(self, axis_name, mesh_axes):CODE
LOWtests/ragged_collective_test.py387 def test_ragged_all_to_all_vmap_multi_dim_operand(self):CODE
LOWtests/ragged_collective_test.py483 def test_ragged_all_to_all_vmap(CODE
LOWtests/ragged_collective_test.py584 def test_ragged_all_to_all_vmap_unsupported_axis_index_groups(self):CODE
LOWtests/ragged_collective_test.py642 def test_ragged_all_to_all_errors(self):CODE
5739 more matches not shown…
Unused Imports3518 hits · 2479 pts
SeverityFileLineSnippetContext
LOWtests/random_test.py15CODE
LOWtests/xla_transform_test.py17CODE
LOWtests/batching_test.py15CODE
LOWtests/debug_info_test.py15CODE
LOWtests/mutable_array_test.py15CODE
LOWtests/lax_numpy_test.py15CODE
LOWtests/shard_map_test.py15CODE
LOWtests/export_harnesses_multi_platform_test.py22CODE
LOWtests/shape_poly_test.py15CODE
LOWtests/lax_vmap_test.py15CODE
LOWtests/lax_numpy_indexing_test.py15CODE
LOWtests/typing_test.py21CODE
LOWtests/array_api_test.py20CODE
LOWtests/state_test.py15CODE
LOWtests/export_test.py14CODE
LOWtests/export_test.py38CODE
LOWtests/export_test.py45CODE
LOWtests/export_test.py51CODE
LOWtests/pmap_test.py15CODE
LOWtests/colocated_python_test.py36CODE
LOWtests/lru_cache_test.py15CODE
LOWtests/compilation_cache_test.py15CODE
LOWtests/hijax_test.py15CODE
LOWtests/export_serialization_back_compat_test.py53CODE
LOWtests/api_test.py15CODE
LOWtests/roofline_test.py14CODE
LOWtests/lax_test.py14CODE
LOWtests/mosaic/gpu_torch_test.py29CODE
LOWtests/mosaic/flash_attention_test.py24CODE
LOWtests/mosaic/gpu_test.py53CODE
LOWtests/multiprocess/colocated_python_test.py24CODE
LOWtests/pallas/tpu_all_gather_test.py16CODE
LOWtests/pallas/tpu_splash_attention_mask_test.py15CODE
LOWtests/pallas/tpu_pallas_pipeline_test.py16CODE
LOWtests/pallas/tpu_pallas_random_test.py26CODE
LOWtests/pallas/indexing_test.py15CODE
LOWtests/pallas/tpu_splash_attention_kernel_test.py15CODE
LOWtests/pallas/pallas_test.py14CODE
LOWdocs/parallel.py61CODE
LOWdocs/autodidax.py2156CODE
LOWjaxlib/xla_client.py17CODE
LOWjaxlib/init.py15CODE
LOWjaxlib/mosaic/gpu/wheel/__init__.py17CODE
LOWjaxlib/mosaic/python/tpu.py20CODE
LOWjaxlib/mosaic/python/tpu.py22CODE
LOWjaxlib/mosaic/python/tpu.py24CODE
LOWjaxlib/mosaic/python/mosaic_gpu.py23CODE
LOWjaxlib/mosaic/python/mosaic_gpu.py25CODE
LOWjaxlib/mosaic/python/mosaic_gpu.py26CODE
LOWjaxlib/tools/build_utils.py17CODE
LOWjaxlib/triton/dialect.py19CODE
LOWjaxlib/triton/dialect.py23CODE
LOWjaxlib/triton/dialect.py23CODE
LOWjaxlib/triton/dialect.py30CODE
LOWjaxlib/triton/dialect.py31CODE
LOWjax/sharding.py18CODE
LOWjax/sharding.py19CODE
LOWjax/sharding.py19CODE
LOWjax/sharding.py19CODE
LOWjax/sharding.py19CODE
3458 more matches not shown…
Docstring Block Structure567 hits · 2308 pts
SeverityFileLineSnippetContext
HIGHjax/experimental/multihost_utils.py549A context manager for atomically running code on the set of live devices. THIS API IS UNDER ACTIVE DEVELOPMENT AND ISSTRING
HIGHjax/experimental/mosaic/gpu/profiler.py137Measures the GPU runtime of a function using CUPTI. ``measure`` is a higher-order function that wraps a function ``f`STRING
HIGH…ops/tpu/splash_attention/splash_attention_mask_info.py84Downcast numpy array. If possible, downcast the data-type of the input array to the smallest numpy type (among np.iSTRING
HIGH…ops/tpu/splash_attention/splash_attention_mask_info.py326Similar to `_process_mask` but the mask must be a dynamic array. Since the mask is dynamic, we can't know the exact nSTRING
HIGH…ops/tpu/splash_attention/splash_attention_mask_info.py528Transform a dense mask into a sparse representation. The number of head and Q sequence shards are needed to create a STRING
HIGH…llas/ops/tpu/splash_attention/splash_attention_mask.py97Makes a chunked causal attention mask. Args: shape: The desired shape of the mask (q_seq_len, kv_seq_len). chSTRING
HIGHjax/experimental/sparse/linalg.py43Compute the top-k standard eigenvalues using the LOBPCG routine. LOBPCG [1] stands for Locally Optimal Block PrecondiSTRING
HIGHjax/experimental/sparse/bcoo.py2437Experimental batched COO matrix implemented in JAX Args: (data, indices) : data and indices in batched COO formatSTRING
HIGHjax/_src/callback.py269Calls a pure Python callback. Works under :func:`jit`/:func:`~vmap`/etc. For more explanation, see `External CallbackSTRING
HIGHjax/_src/tree.py26Call all() over the leaves of a tree. Args: tree: the pytree to evaluate is_leaf : an optionally specified fuSTRING
HIGHjax/_src/tree.py55Flattens a pytree. The flattening order (i.e. the order of elements in the output list) is deterministic, corresponSTRING
HIGHjax/_src/tree.py91Gets the leaves of a pytree. Args: tree: the pytree for which to get the leaves is_leaf : an optionally speciSTRING
HIGHjax/_src/tree.py120Maps a multi-input function over pytree args to produce a new pytree. Args: f: function that takes ``1 + len(restSTRING
HIGHjax/_src/tree.py163Call reduce() over the leaves of a tree. Args: function: the reduction function tree: the pytree to reduce ovSTRING
HIGHjax/_src/tree.py203Perform a reduction over a pytree with an associative binary operation. This function exploits the fact that the operSTRING
HIGHjax/_src/tree.py245Gets the treedef for a pytree. Args: tree: the pytree for which to get the leaves is_leaf : an optionally speSTRING
HIGHjax/_src/tree.py273Transform a tree having tree structure (outer, inner) into one having structure (inner, outer). Args: outer_treedSTRING
HIGHjax/_src/tree.py303Reconstructs a pytree from the treedef and the leaves. The inverse of :func:`tree_flatten`. Args: treedef: theSTRING
HIGHjax/_src/tree.py335Flattens a pytree like ``tree_flatten``, but also returns each leaf's key path. Args: tree: a pytree to flatten. STRING
HIGHjax/_src/tree.py366Gets the leaves of a pytree like ``tree_leaves`` and returns each leaf's key path. Args: tree: a pytree. If it coSTRING
HIGHjax/_src/tree.py395Maps a multi-input function over pytree key path and args to produce a new pytree. This is a more powerful alternativSTRING
HIGHjax/_src/tree.py433Broadcasts a tree prefix into the full structure of a given tree. Args: prefix_tree: a pytree that is a tree STRING
HIGHjax/_src/xla_bridge.py317Creates a CPU client with the requested collectives implementation. The implementation of CPU collectives used by theSTRING
HIGHjax/_src/config.py376Set up thread-local state and return a contextmanager for managing it. This function is a convenience wrapper. It defSTRING
HIGHjax/_src/core.py2967Invalidate a given reference and return its final value. For more information about mutable array references, refer tSTRING
HIGHjax/_src/dtypes.py882Returns the type to which a binary operation should cast its arguments. JAX implementation of :func:`numpy.promote_tySTRING
HIGHjax/_src/dtypes.py1130Check if a dtype/value is safe to cast to another dtype/value Args: input_dtype_or_value: a dtype or value (to beSTRING
HIGHjax/_src/api.py215Sets up ``fun`` for just-in-time compilation with XLA. Args: fun: Function to be jitted. ``fun`` should be a pureSTRING
HIGHjax/_src/api.py2304Transfer array shards to specified devices and form Array(s). Args: shards: A sequence of arrays, scalars, or (neSTRING
HIGHjax/_src/api.py2384Transfer array(s) to each specified device and form Array(s). Args: x: an array, scalar, or (nested) standard PytSTRING
HIGHjax/_src/api.py2453Transfer ``x`` to host. If ``x`` is a pytree, then the individual buffers are copied in parallel. Args: x: An STRING
HIGHjax/_src/api.py2602A context manager that adds a user specified name to the JAX name stack. When staging out computations for just-in-tiSTRING
HIGHjax/_src/distributed.py233Initializes the JAX distributed system. Calling :func:`~jax.distributed.initialize` prepares JAX for execution on mSTRING
HIGHjax/_src/sharding_impls.py654Computes the global shape given the per process if possible. The returned shape will have the size of the global tensSTRING
HIGHjax/_src/tree_util.py127Makes a tuple treedef from an iterable of child treedefs. Args: treedefs: iterable of PyTree structures ReturnSTRING
HIGHjax/_src/tree_util.py153Return a list of treedefs for immediate children Args: treedef: a single PyTreeDef Returns: a list of PyTrSTRING
HIGHjax/_src/tree_util.py178Return True if the treedef represents a leaf. Args: treedef: tree to check Returns: True if treedef is a lSTRING
HIGHjax/_src/tree_util.py206Tests whether all elements in the given iterable are all leaves. This function is useful in advanced cases, for exampSTRING
HIGHjax/_src/tree_util.py338Extends the set of types that are considered internal nodes in pytrees. This function is a thin wrapper around ``regiSTRING
HIGHjax/_src/tree_util.py690Flatten the given pytree node by one level. Args: tree: A valid pytree node, either built-in or registered via STRING
HIGHjax/_src/tree_util.py836Helper to pretty-print a tuple of keys. Args: keys: A tuple of ``KeyEntry`` or any class that can be converted toSTRING
HIGHjax/_src/tree_util.py953Extends the set of types that are considered internal nodes in pytrees. This function is similar to ``register_pytreeSTRING
HIGHjax/_src/tree_util.py1007Extends the set of types that are considered internal nodes in pytrees. This differs from ``register_pytree_with_keysSTRING
HIGHjax/_src/tree_util.py1193Registers `cls` as a pytree with no leaves. Instances are treated as static by :func:`jax.jit`, :func:`jax.pmap`, etcSTRING
HIGHjax/_src/custom_partitioning_sharding_rule.py215Parses the LHS or RHS of an Einsum notation like string. Converts each operand or result in the Einsum notation like STRING
HIGHjax/_src/array.py711Returns a ``jax.Array`` via data fetched from ``data_callback``. ``data_callback`` is used to fetch the data for eachSTRING
HIGHjax/_src/array.py846Creates distributed tensor using the data available in process. This function is a common special case of `make_arraySTRING
HIGHjax/_src/array.py1046Returns a ``jax.Array`` from a sequence of ``jax.Array``\s each on a single device. Every device in input ``shardiSTRING
HIGHjax/_src/profiler.py94Registers a subprocess's profiler server to be profiled alongside the current process. When the current process colleSTRING
HIGHjax/_src/mesh_utils.py789Creates a performant device mesh for jax.sharding.Mesh. Args: mesh_shape: shape of logical mesh, ordered by increSTRING
HIGHjax/_src/mesh_utils.py870Creates a device mesh for hybrid (e.g., ICI and DCN) parallelism. Args: mesh_shape: shape of the logical mesh forSTRING
HIGHjax/_src/pmap.py475Infer axis size from the first mapped argument. shard_map already does a check on all arguments, so just look at firsSTRING
HIGHjax/_src/pmap.py526Extract dynamic args and argnums after handling static args. Args: wrapped_f: The wrapped function. static_brSTRING
HIGHjax/_src/pmap.py564Compute flat in_axes tuple from in_axes prefix and args structure. Args: in_axes: The original in_axes specificatSTRING
HIGHjax/_src/pmap.py671Compute effective mesh devices based on context. Args: devices: The mesh devices tuple. backend: The backend STRING
HIGHjax/_src/custom_derivatives.py187Define a custom JVP rule for the function represented by this instance. Args: jvp: a Python callable represenSTRING
HIGHjax/_src/custom_derivatives.py236Convenience wrapper for defining JVPs for each argument separately. This convenience wrapper cannot be used togetheSTRING
HIGHjax/_src/custom_derivatives.py621Define a custom VJP rule for the function represented by this instance. Args: fwd: a Python callable represenSTRING
HIGHjax/_src/interpreters/mlir.py259Translate a Python ``val`` to an IR constant. See https://docs.jax.dev/en/latest/internals/constants.html. Args: STRING
HIGHjax/_src/interpreters/mlir.py291Translate a Python ``val`` to a sequence of IR constants. See https://docs.jax.dev/en/latest/internals/constants.htmlSTRING
507 more matches not shown…
Over-Commented Block1484 hits · 1410 pts
SeverityFileLineSnippetContext
LOWconftest.py1# Copyright 2021 The JAX Authors.COMMENT
LOWconftest.py41# The pytest_collection hook can be used to overwrite the collection logic, butCOMMENT
LOWsetup.py1# Copyright 2018 The JAX Authors.COMMENT
LOWbuild_wheel.py1# Copyright 2025 The JAX Authors.COMMENT
LOWci/run_pytest_tpu.sh1#!/bin/bashCOMMENT
LOWci/run_pytest_rocm.sh1#!/bin/bashCOMMENT
LOWci/run_bazel_test_cuda_non_rbe.sh1#!/bin/bashCOMMENT
LOWci/run_bazel_cuda_targeted_tests.sh1#!/bin/bashCOMMENT
LOWci/run_bazel_cuda_targeted_tests.sh21# JAXCI_BAZEL_TARGETS: newline-separated Bazel targets.COMMENT
LOWci/run_bazel_test_rocm_rbe.sh1#!/bin/bashCOMMENT
LOWci/build_artifacts.sh1#!/bin/bashCOMMENT
LOWci/build_rocm_artifacts.sh1#!/bin/bashCOMMENT
LOWci/run_bazel_test_cpu_rbe.sh1#!/bin/bashCOMMENT
LOWci/upload_rocm_logs.sh1#!/bin/bashCOMMENT
LOWci/run_bazel_test_cuda_rbe.sh1#!/bin/bashCOMMENT
LOWci/run_pytest_cuda.sh1#!/bin/bashCOMMENT
LOWci/build_oneapi_artifacts.sh1#!/bin/bashCOMMENT
LOWci/parse_wheel_metadata.py1# Copyright 2026 The JAX Authors.COMMENT
LOWci/run_bazel_test_tpu.sh1#!/bin/bashCOMMENT
LOWci/run_bazel_test_tpu.sh221 //tests/pallas:tpu_pallas_call_print_test_tpu \COMMENT
LOWci/run_pytest_cpu.sh1#!/bin/bashCOMMENT
LOWci/utilities/install_wheels_locally.sh1#!/bin/bashCOMMENT
LOWci/utilities/report_resultstore_link.py1# Copyright 2026 The JAX Authors.COMMENT
LOWci/utilities/run_docker_container.sh1#!/bin/bashCOMMENT
LOWci/utilities/run_docker_container.sh21# Note: While GitHub action workflows use the same Docker images, they do notCOMMENT
LOWci/utilities/convert_msys_paths_to_win_paths.py1# Copyright 2024 The JAX Authors.COMMENT
LOWci/utilities/collect_bazel_test_xmls.sh1#!/bin/bashCOMMENT
LOWci/utilities/collect_bazel_test_xmls.sh21# bazel-testlogs/tests/cpu_tests/test.xml). Pytest, by contrast, writes aCOMMENT
LOWci/utilities/setup_build_environment.sh1#!/bin/bashCOMMENT
LOWci/utilities/run_auditwheel.sh1#!/bin/bashCOMMENT
LOWci/utilities/generate_invocation_id.py1# Copyright 2026 The JAX Authors.COMMENT
LOWci/utilities/setup_portserver.sh1#!/bin/bashCOMMENT
LOWci/postprocess/process_test_results.sh1#!/bin/bashCOMMENT
LOWci/postprocess/xml2json.py1# Copyright 2026 The JAX Authors.COMMENT
LOWtests/random_test.py1# Copyright 2018 The JAX Authors.COMMENT
LOWtests/random_test.py641COMMENT
LOWtests/extend_test.py1# Copyright 2023 The JAX Authors.COMMENT
LOWtests/svd_test.py1# Copyright 2022 The JAX Authors.COMMENT
LOWtests/scipy_spatial_test.py1# Copyright 2023 The JAX Authors.COMMENT
LOWtests/scheduling_groups_test.py1# Copyright 2025 The JAX Authors.COMMENT
LOWtests/jax_to_ir_test.py1# Copyright 2019 The JAX Authors.COMMENT
LOWtests/ragged_collective_test.py1# Copyright 2025 The JAX Authors.COMMENT
LOWtests/pytorch_interoperability_test.py1# Copyright 2020 The JAX Authors.COMMENT
LOWtests/fft_test.py1# Copyright 2019 The JAX Authors.COMMENT
LOWtests/export_back_compat_test.py1# Copyright 2023 The JAX Authors.COMMENT
LOWtests/export_back_compat_test.py1041 # stablehlo.dynamic_rbg_bit_generator is used temporarily for aCOMMENT
LOWtests/mesh_utils_test.py1# Copyright 2021 The JAX Authors. All Rights Reserved.COMMENT
LOWtests/lax_vmap_op_test.py1# Copyright 2020 The JAX Authors.COMMENT
LOWtests/debug_nans_test.py1# Copyright 2019 The JAX Authors.COMMENT
LOWtests/lobpcg_test.py1# Copyright 2022 The JAX Authors.COMMENT
LOWtests/heap_profiler_test.py1# Copyright 2021 The JAX Authors.COMMENT
LOWtests/custom_root_test.py1# Copyright 2018 The JAX Authors.COMMENT
LOWtests/mosaic_test.py1# Copyright 2023 The JAX Authors.COMMENT
LOWtests/lax_control_flow_test.py1# Copyright 2018 The JAX Authors.COMMENT
LOWtests/lax_control_flow_test.py2221COMMENT
LOWtests/array_interoperability_test.py1# Copyright 2020 The JAX Authors.COMMENT
LOWtests/xla_transform_test.py1# Copyright 2026 The JAX Authors.COMMENT
LOWtests/lax_scipy_test.py1# Copyright 2018 The JAX Authors.COMMENT
LOWtests/errors_test.py1# Copyright 2020 The JAX Authors.COMMENT
LOWtests/distributed_initialize_test.py1# Copyright 2025 The JAX Authors.COMMENT
1424 more matches not shown…
Decorative Section Separators129 hits · 387 pts
SeverityFileLineSnippetContext
MEDIUMci/run_pytest_tpu.sh15# ==============================================================================COMMENT
MEDIUMci/run_pytest_rocm.sh15# ==============================================================================COMMENT
MEDIUMci/run_pytest_rocm.sh52# ==============================================================================COMMENT
MEDIUMci/run_pytest_rocm.sh54# ==============================================================================COMMENT
MEDIUMci/run_pytest_rocm.sh61# ==============================================================================COMMENT
MEDIUMci/run_pytest_rocm.sh64# ==============================================================================COMMENT
MEDIUMci/run_pytest_rocm.sh117# ==============================================================================COMMENT
MEDIUMci/run_pytest_rocm.sh119# ==============================================================================COMMENT
MEDIUMci/run_bazel_test_cuda_non_rbe.sh15# ==============================================================================COMMENT
MEDIUMci/run_bazel_cuda_targeted_tests.sh15# ==============================================================================COMMENT
MEDIUMci/run_bazel_test_rocm_rbe.sh15# ==============================================================================COMMENT
MEDIUMci/build_artifacts.sh15# ==============================================================================COMMENT
MEDIUMci/build_rocm_artifacts.sh15# ==============================================================================COMMENT
MEDIUMci/run_bazel_test_cpu_rbe.sh15# ==============================================================================COMMENT
MEDIUMci/upload_rocm_logs.sh15# ==============================================================================COMMENT
MEDIUMci/run_bazel_test_cuda_rbe.sh15# ==============================================================================COMMENT
MEDIUMci/run_pytest_cuda.sh15# ==============================================================================COMMENT
MEDIUMci/run_pytest_cuda.sh45# ==============================================================================COMMENT
MEDIUMci/run_pytest_cuda.sh47# ==============================================================================COMMENT
MEDIUMci/run_pytest_cuda.sh54# ==============================================================================COMMENT
MEDIUMci/run_pytest_cuda.sh57# ==============================================================================COMMENT
MEDIUMci/run_pytest_cuda.sh107# ==============================================================================COMMENT
MEDIUMci/run_pytest_cuda.sh109# ==============================================================================COMMENT
MEDIUMci/build_oneapi_artifacts.sh15# ==============================================================================COMMENT
MEDIUMci/run_bazel_test_tpu.sh15# ==============================================================================COMMENT
MEDIUMci/run_pytest_cpu.sh15# ==============================================================================COMMENT
MEDIUMci/utilities/install_wheels_locally.sh15# ==============================================================================COMMENT
MEDIUMci/utilities/report_resultstore_link.py14# ==============================================================================COMMENT
MEDIUMci/utilities/run_docker_container.sh15# ==============================================================================COMMENT
MEDIUMci/utilities/convert_msys_paths_to_win_paths.py14# ==============================================================================COMMENT
MEDIUMci/utilities/collect_bazel_test_xmls.sh15# ==============================================================================COMMENT
MEDIUMci/utilities/setup_build_environment.sh15# ==============================================================================COMMENT
MEDIUMci/utilities/run_auditwheel.sh15# ==============================================================================COMMENT
MEDIUMci/utilities/generate_invocation_id.py14# ==============================================================================COMMENT
MEDIUMci/utilities/setup_portserver.sh15# ==============================================================================COMMENT
MEDIUMci/postprocess/process_test_results.sh15# ==============================================================================COMMENT
MEDIUMtests/mesh_utils_test.py14# ==============================================================================COMMENT
MEDIUMtests/hijax_test.py2530 # ------------COMMENT
MEDIUMtests/hijax_test.py2532 # ------------COMMENT
MEDIUMtests/hijax_test.py2603 #------------COMMENT
MEDIUMtests/hijax_test.py2605 #------------COMMENT
MEDIUMtests/mosaic/gpu_constraints_test.py14# ==============================================================================COMMENT
MEDIUMtests/mosaic/gpu_torch_distributed_test.py14# ==============================================================================COMMENT
MEDIUMtests/mosaic/gpu_dialect_test.py14# ==============================================================================COMMENT
MEDIUMtests/mosaic/gpu_torch_test.py14# ==============================================================================COMMENT
MEDIUMtests/mosaic/profiler_cupti_test.py14# ==============================================================================COMMENT
MEDIUMtests/mosaic/flash_attention_test.py14# ==============================================================================COMMENT
MEDIUMtests/mosaic/gpu_multidevice_test.py14# ==============================================================================COMMENT
MEDIUMtests/mosaic/gpu_layout_inference_test.py14# ==============================================================================COMMENT
MEDIUMtests/mosaic/matmul_test.py14# ==============================================================================COMMENT
MEDIUMtests/mosaic/gpu_test.py14# ==============================================================================COMMENT
MEDIUMtests/mosaic/gpu_distributed_test.py14# ==============================================================================COMMENT
MEDIUMtests/pallas/mgpu_ragged_dot_test.py14# ==============================================================================COMMENT
MEDIUMtests/pallas/mgpu_torch_test.py14# ==============================================================================COMMENT
MEDIUMtests/pallas/mgpu_examples_test.py14# ==============================================================================COMMENT
MEDIUMtests/pallas/mgpu_attention_test.py14# ==============================================================================COMMENT
MEDIUMtests/pallas/mgpu_collective_matmul_test.py14# ==============================================================================COMMENT
MEDIUMtests/pallas/mgpu_matmul_test.py14# ==============================================================================COMMENT
MEDIUMdocs/notebooks/cute_dsl_jax/cute_dsl_jax_kernels.py284 # ── Vector Add ────────────────────────────────────────────────────COMMENT
MEDIUMdocs/notebooks/cute_dsl_jax/cute_dsl_jax_kernels.py297 # ── SAXPY ─────────────────────────────────────────────────────────COMMENT
69 more matches not shown…
Deep Nesting452 hits · 355 pts
SeverityFileLineSnippetContext
LOWconftest.py55CODE
LOWci/postprocess/xml2json.py70CODE
LOWtests/random_test.py690CODE
LOWtests/export_back_compat_test.py557CODE
LOWtests/lax_control_flow_test.py2376CODE
LOWtests/xla_transform_test.py326CODE
LOWtests/errors_test.py44CODE
LOWtests/ffi_test.py45CODE
LOWtests/lax_numpy_einsum_test.py295CODE
LOWtests/debug_info_test.py119CODE
LOWtests/api_util_test.py28CODE
LOWtests/dtypes_test.py494CODE
LOWtests/fused_attention_stablehlo_test.py121CODE
LOWtests/lax_numpy_test.py2588CODE
LOWtests/lax_numpy_test.py6768CODE
LOWtests/lax_numpy_reducers_test.py164CODE
LOWtests/linalg_test.py708CODE
LOWtests/linalg_test.py1786CODE
LOWtests/linalg_test.py1888CODE
LOWtests/linalg_test.py1789CODE
LOWtests/linalg_test.py1896CODE
LOWtests/profiler_test.py54CODE
LOWtests/profiler_test.py133CODE
LOWtests/lax_numpy_indexing_test.py1491CODE
LOWtests/export_test.py2395CODE
LOWtests/control_deps_test.py50CODE
LOWtests/pgle_test.py135CODE
LOWtests/pgle_test.py221CODE
LOWtests/pgle_test.py367CODE
LOWtests/fused_attention_stablehlo_multigpu_test.py122CODE
LOWtests/sparse_bcoo_bcsr_test.py57CODE
LOWtests/sparse_bcoo_bcsr_test.py101CODE
LOWtests/device_test.py24CODE
LOWtests/device_test.py46CODE
LOWtests/compilation_cache_test.py577CODE
LOWtests/transfer_guard_test.py190CODE
LOWtests/transfer_guard_test.py201CODE
LOWtests/api_test.py5001CODE
LOWtests/api_test.py5322CODE
LOWtests/api_test.py5330CODE
LOWtests/documentation_coverage_test.py141CODE
LOWtests/sparse_test.py1006CODE
LOWtests/lax_test.py118CODE
LOWtests/lax_test.py1613CODE
LOWtests/lax_test.py4588CODE
LOWtests/mosaic/gpu_test.py682CODE
LOWtests/mosaic/gpu_test.py1169CODE
LOWtests/mosaic/gpu_test.py5464CODE
LOWtests/mosaic/gpu_test.py6168CODE
LOWtests/mosaic/gpu_test.py7390CODE
LOWtests/multiprocess/thread_guard_test.py42CODE
LOWtests/multiprocess/pgle_test.py44CODE
LOWtests/pallas/tpu_pallas_interpret_test.py313CODE
LOWtests/pallas/tpu_pallas_interpret_test.py409CODE
LOWtests/pallas/triton_pallas_test.py150CODE
LOW…allas/tpu_fusible_matmul_with_stateful_fusions_test.py84CODE
LOW…ts/pallas/gpu_pallas_interpret_cluster_barrier_test.py375CODE
LOW…ts/pallas/gpu_pallas_interpret_cluster_barrier_test.py380CODE
LOWtests/pallas/einshape_test.py60CODE
LOWtests/pallas/gpu_pallas_interpret_test.py639CODE
392 more matches not shown…
Self-Referential Comments73 hits · 202 pts
SeverityFileLineSnippetContext
MEDIUMci/utilities/run_docker_container.sh50 # Create a temporary file to pass any user defined JAXCI_ / JAX_ / JAXLIB_COMMENT
MEDIUMci/utilities/setup_build_environment.sh105# Create the output directory if it doesn't exist.COMMENT
MEDIUMtests/ragged_collective_test.py772 # Define a mesh with PP + EPCOMMENT
MEDIUMtests/lax_scipy_test.py546 # This function is not defined for negative values, this makes sure they are nanCOMMENT
MEDIUMtests/custom_api_test.py2129 # Create the custom functionCOMMENT
MEDIUMtests/garbage_collection_guard_test.py54 # Create a reference cycle of two jax.Arrays.COMMENT
MEDIUMtests/garbage_collection_guard_test.py67 # Create a reference cycle of two jax.Arrays.COMMENT
MEDIUMtests/layout_test.py361 # Create a custom layout instead of using `arr.layout` to test the API.COMMENT
MEDIUMtests/layout_test.py384 # Create a custom layout instead of using `arr.layout` to test the API.COMMENT
MEDIUMtests/layout_test.py665 # Create a custom layout instead of using `arr.layout` to test the API.COMMENT
MEDIUMtests/layout_test.py697 # Create a custom layout instead of using `arr.layout` to test the API.COMMENT
MEDIUMtests/export_test.py2486 # Create a custom layout instead of using `arr.layout` to test the API.COMMENT
MEDIUMtests/sparse_bcoo_bcsr_test.py1153 # Create a matrix with duplicate indicesCOMMENT
MEDIUMtests/aot_test.py337 # Create a compile-only topology, but DON'T switch to CPU so the realCOMMENT
MEDIUMtests/array_test.py667 # Create a few arraysCOMMENT
MEDIUMtests/hijax_test.py64# Define a typeCOMMENT
MEDIUMtests/api_test.py2043 # Creating an array from per-device JAX arrays calls internalCOMMENT
MEDIUMtests/mosaic/gpu_layout_inference_test.py1527 # Create a var to use in the constraint system.COMMENT
MEDIUMtests/multiprocess/array_test.py613 # Create an array that is non-addressable in processes besides `pid`.COMMENT
MEDIUMtests/multiprocess/array_test.py635 # Create a PRNG key array that is non-addressable in processes besidesCOMMENT
MEDIUMtests/multiprocess/array_test.py669 # Create a sharding that is non-addressable in processes besides `pid`.COMMENT
MEDIUMtests/multiprocess/array_test.py696 # Create a sharding that is non-addressable in processes besides `pid`.COMMENT
MEDIUMtests/multiprocess/array_test.py723 # Create a single device sharding for a device local to process `pid`.COMMENT
MEDIUMtests/pallas/tpu_pallas_interpret_distributed_test.py63 # Create an input array that shards the last dimension acrossCOMMENT
MEDIUMtests/pallas/tpu_pallas_interpret_distributed_test.py149 # Create an input array that shards the first dimension acrossCOMMENT
MEDIUMtests/pallas/mosaic_gpu_test.py7525 # Create an index-invariant output.COMMENT
MEDIUMtests/pallas/pallas_test.py2539 # Create a validity mask for OOB values.COMMENT
MEDIUMdocs/sphinxext/jax_list_config_options.py38 # Create a field list itemCOMMENT
MEDIUMdocs/sphinxext/jax_list_config_options.py41 # Create the field name (label)COMMENT
MEDIUMdocs/sphinxext/jax_list_config_options.py46 # Create the field body (content)COMMENT
MEDIUMdocs/sphinxext/jax_list_config_options.py81 # Create a section for this optionCOMMENT
MEDIUMdocs/sphinxext/jax_list_config_options.py86 # Create a title with the option name (important for TOC)COMMENT
MEDIUMdocs/sphinxext/jax_list_config_options.py92 # Create a field list for side-by-side displayCOMMENT
MEDIUMjaxlib/tools/build_wheel.py226 # This file is required by PEP-561. It marks jaxlib as package containingCOMMENT
MEDIUM.github/workflows/bazel_cuda_b200_mosaic_presubmit.yml87 # Create an array with the file patterns andCOMMENT
MEDIUMjax/version.py15# This file is included as part of both jax and jaxlib. It is alsoCOMMENT
MEDIUM…ops/tpu/splash_attention/splash_attention_mask_info.py587 # Create a collection of the unique head masks in the input multi-head mask.COMMENT
MEDIUM…llas/ops/tpu/splash_attention/splash_attention_mask.py374 # Define the mask function for chunk attentionCOMMENT
MEDIUMjax/experimental/pallas/ops/tpu/megablox/gmm.py180 # Create the group ids for each grid index based on the tile counts for eachCOMMENT
MEDIUMjax/experimental/pallas/ops/tpu/megablox/gmm.py230 # Create the m-dimension tile ids for each grid index based on the visitCOMMENT
MEDIUMjax/experimental/pallas/ops/tpu/megablox/gmm.py383 # Create the metadata we need for computation.COMMENT
MEDIUMjax/experimental/pallas/ops/tpu/megablox/gmm.py632 # Create the metadata we need for computation.COMMENT
MEDIUMjax/experimental/jax2tf/call_tf.py220 # Define the fwd and bwd custom_vjp functionsCOMMENT
MEDIUMjax/experimental/jax2tf/tests/jax2tf_test.py349 x = tf.Variable(4.0, dtype=tf.float32) # Create a Tensorflow variable initialized to 4.0CODE
MEDIUMjax/experimental/jax2tf/examples/mnist_lib.py225 # Create the model and save itCOMMENT
MEDIUMjax/_src/test_util.py2071 # The following function methods operate on mpmath number instances.COMMENT
MEDIUMjax/_src/custom_batching.py178# Define a class, instead of making a function closing over `rule`, soCOMMENT
MEDIUMjax/_src/ad_checkpoint.py411# This function is similar to api_util.argnums_partial, except the errorCOMMENT
MEDIUMjax/_src/ad_checkpoint.py953 >>> # Define a function where we explicitly name an intermediate valueSTRING
MEDIUMjax/_src/ffi.py710 # This method is kept to support the behavior that was previously exposedCOMMENT
MEDIUMjax/_src/interpreters/partial_eval.py253 # Create the input tracers for the staged-out (unknown-value) call.COMMENT
MEDIUMjax/_src/interpreters/partial_eval.py2433 # This function is conceptually the same thing as just calling eval_jaxpr,COMMENT
MEDIUMjax/_src/interpreters/mlir.py1391 # Create a keepalives list that will be mutated during the lowering.COMMENT
MEDIUMjax/_src/lax/utils.py15# This module contains utility functions split out of jax._src.lax.lax toCOMMENT
MEDIUMjax/_src/lax/control_flow/loops.py1359 # Create the staged eqn.COMMENT
MEDIUMjax/_src/pallas/mosaic/lowering.py3887 # Create a scalar constant.COMMENT
MEDIUMjax/_src/pallas/mosaic/lowering.py3890 # Create a vector constant.COMMENT
MEDIUMjax/_src/pallas/mosaic_gpu/lowering.py619 # The below code emission relies on the assumption that the first scratchCOMMENT
MEDIUMjax/_src/lib/__init__.py15# This module is largely a wrapper around `jaxlib` that performs versionCOMMENT
MEDIUMjax/_src/export/serialization_generated.py109 """This method is deprecated. Please switch to GetRootAs."""STRING
13 more matches not shown…
Cross-File Repetition32 hits · 160 pts
SeverityFileLineSnippetContext
HIGHdocs/new_docs/401/cute-dsl.md0per-thread kernel: each thread computes relu of one element.STRING
HIGHdocs/notebooks/cute_dsl_jax.md0per-thread kernel: each thread computes relu of one element.STRING
HIGHdocs/notebooks/cute_dsl_jax/cute_dsl_jax_kernels.py0per-thread kernel: each thread computes relu of one element.STRING
HIGHdocs/new_docs/401/cute-dsl.md0per-thread: out[i] = max(0, x[i] + bias[i % width]).STRING
HIGHdocs/notebooks/cute_dsl_jax.md0per-thread: out[i] = max(0, x[i] + bias[i % width]).STRING
HIGHdocs/notebooks/cute_dsl_jax/cute_dsl_jax_kernels.py0per-thread: out[i] = max(0, x[i] + bias[i % width]).STRING
HIGHdocs/new_docs/401/cute-dsl.md0tiled gemm: each thread accumulates output elements.STRING
HIGHdocs/notebooks/cute_dsl_jax.md0tiled gemm: each thread accumulates output elements.STRING
HIGHdocs/notebooks/cute_dsl_jax/cute_dsl_jax_kernels.py0tiled gemm: each thread accumulates output elements.STRING
HIGHdocs/new_docs/401/cute-dsl.md0per-thread kernel: 2-d element-wise add using flat indexing.STRING
HIGHdocs/notebooks/cute_dsl_jax.md0per-thread kernel: 2-d element-wise add using flat indexing.STRING
HIGHdocs/notebooks/cute_dsl_jax/cute_dsl_jax_kernels.py0per-thread kernel: 2-d element-wise add using flat indexing.STRING
HIGHjaxlib/setup.py0this class makes 'bdist_wheel' include an abi tag on the wheel.STRING
HIGHjax_plugins/cuda/plugin_setup.py0this class makes 'bdist_wheel' include an abi tag on the wheel.STRING
HIGHjax_plugins/rocm/plugin_setup.py0this class makes 'bdist_wheel' include an abi tag on the wheel.STRING
HIGHjax_plugins/oneapi/plugin_setup.py0this class makes 'bdist_wheel' include an abi tag on the wheel.STRING
HIGHjaxlib/tools/build_gpu_plugin_wheel.py0[metadata] license_files = license.txt [bdist_wheel] plat_name={tag}STRING
HIGHjaxlib/tools/build_mosaic_wheel.py0[metadata] license_files = license.txt [bdist_wheel] plat_name={tag}STRING
HIGHjaxlib/tools/build_wheel.py0[metadata] license_files = license.txt [bdist_wheel] plat_name={tag}STRING
HIGHjaxlib/tools/build_gpu_kernels_wheel.py0[metadata] license_files = license.txt [bdist_wheel] plat_name={tag}STRING
HIGH…mental/jax2tf/tests/flax_models/transformer_nlp_seq.py0global hyperparameters used to minimize obnoxious kwarg plumbing.STRING
HIGH…erimental/jax2tf/tests/flax_models/transformer_lm1b.py0global hyperparameters used to minimize obnoxious kwarg plumbing.STRING
HIGH…perimental/jax2tf/tests/flax_models/transformer_wmt.py0global hyperparameters used to minimize obnoxious kwarg plumbing.STRING
HIGH…mental/jax2tf/tests/flax_models/transformer_nlp_seq.py0applies addpositionembs module. by default this layer uses a fixed sinusoidal embedding table. if a learned position embSTRING
HIGH…erimental/jax2tf/tests/flax_models/transformer_lm1b.py0applies addpositionembs module. by default this layer uses a fixed sinusoidal embedding table. if a learned position embSTRING
HIGH…perimental/jax2tf/tests/flax_models/transformer_wmt.py0applies addpositionembs module. by default this layer uses a fixed sinusoidal embedding table. if a learned position embSTRING
HIGH…mental/jax2tf/tests/flax_models/transformer_nlp_seq.py0transformer mlp / feed-forward block. args: config: transformerconfig dataclass containing hyperparameters. out_dim: optSTRING
HIGH…perimental/jax2tf/tests/flax_models/transformer_wmt.py0transformer mlp / feed-forward block. args: config: transformerconfig dataclass containing hyperparameters. out_dim: optSTRING
HIGH…erimental/jax2tf/tests/flax_models/transformer_lm1b.py0transformer mlp / feed-forward block. args: config: transformerconfig dataclass containing hyperparameters. out_dim: optSTRING
HIGHjax/_src/pallas/mosaic/interpret/shared_memory.py0synchronizes the vector clocks for the cores on the given device.STRING
HIGH…/_src/pallas/mosaic/interpret/interpret_pallas_call.py0synchronizes the vector clocks for the cores on the given device.STRING
HIGHjax/_src/pallas/mosaic_gpu/interpret/shared_memory.py0synchronizes the vector clocks for the cores on the given device.STRING
AI Structural Patterns161 hits · 132 pts
SeverityFileLineSnippetContext
LOWtests/lax_control_flow_test.py2874CODE
LOWtests/lax_control_flow_test.py2879CODE
LOWtests/xla_transform_test.py277CODE
LOWtests/scaled_matmul_stablehlo_test.py233CODE
LOWtests/mutable_array_test.py429CODE
LOWtests/custom_api_test.py3220CODE
LOWtests/custom_api_test.py1187CODE
LOWtests/custom_api_test.py2843CODE
LOWtests/dtypes_test.py797CODE
LOWtests/fused_attention_stablehlo_test.py76CODE
LOWtests/fused_attention_stablehlo_test.py88CODE
LOWtests/shard_map_test.py503CODE
LOWtests/shard_map_test.py536CODE
LOWtests/shard_map_test.py573CODE
LOWtests/shard_map_test.py612CODE
LOWtests/shard_map_test.py1069CODE
LOWtests/lax_numpy_reducers_test.py278CODE
LOWtests/lax_numpy_reducers_test.py511CODE
LOWtests/shape_poly_test.py1278CODE
LOWtests/nn_test.py64CODE
LOWtests/linalg_test.py973CODE
LOWtests/state_test.py1606CODE
LOWtests/image_test.py74CODE
LOWtests/image_test.py103CODE
LOWtests/memories_test.py1320CODE
LOWtests/export_test.py760CODE
LOWtests/export_test.py763CODE
LOWtests/pmap_test.py1439CODE
LOWtests/pjit_test.py6288CODE
LOWtests/pjit_test.py7898CODE
LOWtests/pjit_test.py8565CODE
LOWtests/pjit_test.py8788CODE
LOWtests/pjit_test.py8977CODE
LOWtests/pjit_test.py9083CODE
LOWtests/pjit_test.py11390CODE
LOWtests/colocated_python_test.py583CODE
LOWtests/scipy_ndimage_test.py56CODE
LOWtests/fused_attention_stablehlo_multigpu_test.py77CODE
LOWtests/fused_attention_stablehlo_multigpu_test.py89CODE
LOWtests/checkify_test.py1205CODE
LOWtests/pallas/tpu_pallas_test.py5689CODE
LOWtests/pallas/tpu_pallas_test.py5731CODE
LOWtests/pallas/gpu_paged_attention_test.py77CODE
LOWtests/pallas/tpu_pallas_async_test.py86CODE
LOWtests/pallas/tpu_pallas_async_test.py139CODE
LOWtests/pallas/tpu_pallas_async_test.py195CODE
LOWtests/pallas/tpu_pallas_async_test.py636CODE
LOWtests/pallas/tpu_pallas_async_test.py758CODE
LOWtests/pallas/tpu_pallas_pipeline_test.py655CODE
LOWtests/pallas/mgpu_collective_matmul_test.py108CODE
LOWtests/pallas/tpu_ragged_paged_attention_test.py36CODE
LOWtests/pallas/mosaic_gpu_test.py9088CODE
LOW…sts/pallas/tpu_splash_attention_kernel_sharded_test.py263CODE
LOWjaxlib/xla_client.py66CODE
LOWjax/experimental/mosaic/gpu/launch_context.py1191CODE
LOWjax/experimental/mosaic/gpu/examples/matmul.py105CODE
LOWjax/experimental/mosaic/gpu/examples/matmul.py321CODE
LOWjax/experimental/pallas/ops/gpu/attention.py220CODE
LOWjax/experimental/pallas/ops/gpu/decode_attention.py247CODE
LOWjax/experimental/pallas/ops/gpu/decode_attention.py304CODE
101 more matches not shown…
Cross-Language Confusion24 hits · 130 pts
SeverityFileLineSnippetContext
HIGHtests/stack_test.py34 stack = stack.push(jnp.int32(7))CODE
HIGHtests/stack_test.py36 stack = stack.push(jnp.int32(8))CODE
HIGHtests/stack_test.py41 stack = stack.push(jnp.int32(9))CODE
HIGHtests/export_test.py1121 stack.push(self.assertRaisesRegex(Exception, re.escape(expect_error)))CODE
HIGHtests/export_test.py938 stack.push(self.assertRaisesRegex(Exception, expect_error))CODE
HIGHtests/export_test.py1044 stack.push(self.assertRaisesRegex(ValueError, expect_error_outer_exp))CODE
HIGHtests/export_test.py1058 stack.push(self.assertRaisesRegex(Exception, expect_error_run))CODE
HIGHjax/experimental/jax2tf/tests/shape_poly_test.py440 stack.push(self.assertRaisesRegex(Exception, re.escape(expect_error)))CODE
HIGHjax/experimental/colocated_python/func.py363 func_backend.SINGLETON_RESULT_STORE.push(uid, result_leaves)CODE
HIGHjax/_src/jaxpr_util.py648 while ((match = anchorRegex.exec(line)) !== null) {{CODE
HIGHjax/_src/jaxpr_util.py693 const isCurrent = currentMatchIdx !== -1 && lineAbsoluteIdx === matchingLines[currentMatchIdx];CODE
HIGHjax/_src/jaxpr_util.py726 let selectedElement = null;CODE
HIGHjax/_src/jaxpr_util.py734 if (lineIdx !== undefined) {{CODE
HIGHjax/_src/jaxpr_util.py759 while (currentIdx !== null && currentIdx !== undefined) {{CODE
HIGHjax/_src/jaxpr_util.py759 while (currentIdx !== null && currentIdx !== undefined) {{CODE
HIGHjax/_src/jaxpr_util.py764 renderedFrames.push({{file: file, func: func, line: frame.line, col: frame.col}});CODE
HIGHjax/_src/jaxpr_util.py825 matchingLines.push(idx);CODE
HIGHjax/_src/jaxpr_util.py829 matchingLines.push(idx);CODE
HIGHjax/_src/jaxpr_util.py958 if (offsetRight > 100 && offsetRight < document.body.clientWidth - 100) {{CODE
HIGHjax/_src/lax/lax.py9050 b <= a, then the result is undefined, and different implementations maySTRING
HIGHjax/_src/numpy/linalg.py529 discontinuous at det=0 so standard autodiff is undefined; we use sign_dot=0STRING
HIGHjax/_src/tpu/linalg/eigh.py326 agenda = agenda.push(_Subproblem(offset=jnp.array(0, np.int32), size=n))CODE
HIGHjax/_src/tpu/linalg/eigh.py417 agenda.push(_Subproblem(offset, rank)),CODE
HIGHjax/_src/tpu/linalg/eigh.py436 agenda.push(_Subproblem(offset + rank, (b - rank))),CODE
AI Slop Vocabulary48 hits · 116 pts
SeverityFileLineSnippetContext
MEDIUMci/utilities/run_docker_container.sh22# run this script as they leverage built-in containerization features to runCOMMENT
MEDIUMtests/export_back_compat_test.py264 # Compute the inputs to simplify the harnessCOMMENT
MEDIUMtests/lax_numpy_operators_test.py294 # probably add a custom test harness for unwrap that tests the periodCOMMENT
LOWtests/custom_api_test.py366 # useful either: instead of using nondiff_argnums here, a user can just passCOMMENT
MEDIUMtests/lax_numpy_test.py3387 # permissive canonicalization logic in the test harness.COMMENT
MEDIUMtests/export_harnesses_multi_platform_test.py60 # If you want to run this test for only one harness, add parameterCOMMENT
MEDIUMtests/shape_poly_test.py3670 # Pick the dtype with the most harnesses in this group. Some harnessCOMMENT
MEDIUMtests/shape_poly_test.py3742 # If you want to run this test for only one harness that includes "foo"COMMENT
MEDIUMtests/shape_poly_test.py3799 # Update this here rather than in harness object because vmap_random_gamma is derivedCOMMENT
LOWtests/shape_poly_test.py1861 # It is not sufficient to just use the shape of an input; it is still unusedCOMMENT
MEDIUMtests/linalg_test.py927 # This expresses identity function, which makes us robust to, e.g., theCOMMENT
MEDIUMtests/lax_autodiff_test.py159 # TODO(mattjj): make some-equal checks more robust, enable second-orderCOMMENT
MEDIUMtests/sparsify_test.py341 # Note: more comprehensive tests in sparse_test.py:test_bcoo_squeezeCOMMENT
MEDIUMtests/sparsify_test.py354 # Note: more comprehensive tests in sparse_test.py:test_bcoo_revCOMMENT
MEDIUMtests/sparsify_test.py419 # Note: more comprehensive tests in sparse_test.py:test_bcoo_conv_general_dilatedCOMMENT
LOWtests/api_test.py454 # We can't just use `lambda x: x` because JAX simplifies this away to anCOMMENT
LOWtests/mosaic/gpu_layout_inference_test.py91 # TMEM reference, so we can just return a trivial mapping.COMMENT
LOWdocs/array_refs.py115# `jax.ref.swap`, but usually you'd just use NumPy-style array indexing syntax:COMMENT
LOWdocs/hijax_custom_derivatives.py1058# that a primitive with a `jvp` rule can just set `lin = linearize_from_jvp`COMMENT
LOWdocs/autodidax.py1475# call `bind`, in the primitive wrapper we can just use `make_jaxpr` to formCOMMENT
MEDIUM…ops/tpu/splash_attention/splash_attention_mask_info.py892 # maintain the SPMD paradigm.COMMENT
MEDIUM…erimental/jax2tf/tests/jax_primitives_coverage_test.py50 # If you want to run this test for only one harness, add parameterCOMMENT
MEDIUM…erimental/jax2tf/tests/jax_primitives_coverage_test.py90 # f"{[u.description for u in jax_unimpl]} in harness: {harness.fullname}"))COMMENT
MEDIUMjax/experimental/jax2tf/tests/shape_poly_test.py223 # Makes and tests a harness. See PolyHarness documentation.COMMENT
LOWjax/experimental/jax2tf/tests/shape_poly_test.py587 # It is not sufficient to just use the shape of an input; it is still unusedCOMMENT
MEDIUMjax/experimental/jax2tf/tests/model_harness.py77 """Partially apply harness in order to create variables lazily.STRING
MEDIUMjax/experimental/jax2tf/tests/tf_test_util.py221 # Run JAX. Should not fail, we assume that the harness has been filteredCOMMENT
MEDIUMjax/experimental/jax2tf/tests/primitives_test.py91 # If you want to run this test for only one harness, add parameterCOMMENT
LOWjax/_src/compiler.py489# simply return the non-PGLE profiled module from the persistent cache if itCOMMENT
LOWjax/_src/pjit.py1049 # checks and just return the pjit_in_shardings directly. `shard_args` willCOMMENT
LOWjax/_src/checkify.py350 # Default non-HOP case: just call primitive and don't update error.COMMENT
LOWjax/_src/interpreters/batching.py635 # if there's only agreeing batch dims and scalars, just call the primitiveCOMMENT
LOWjax/_src/lax/ann.py367# 1. ApproxTopK is internally a variadic reduce, so we can simply callCOMMENT
MEDIUMjax/_src/lax/control_flow/loops.py2074# lowering. Fundamentally, we'd like to rewrite a while loop that looks likeCOMMENT
MEDIUMjax/_src/numpy/linalg.py1445 # TODO: add custom jvp rule for more robust lstsq differentiationCOMMENT
LOWjax/_src/pallas/pallas_call.py674 # dimensions. For now, we just use 0.COMMENT
LOWjax/_src/pallas/mosaic/lowering.py1714 # `jax.empty_ref`), but lowering expects them to exist---so we just returnCOMMENT
MEDIUM…/_src/pallas/mosaic/interpret/interpret_pallas_call.py1595 # TODO(nrink): It would be more robust if the buffer id, i.e. `src`,COMMENT
LOW…/_src/pallas/mosaic/interpret/interpret_pallas_call.py645 # callback to `get`. Should we just pass the shape to `get`?COMMENT
LOWjax/_src/pallas/mosaic_gpu/interpret/gpu_callbacks.py583 # callback to `get`. Should we just pass the shape to `get`?COMMENT
MEDIUMjax/_src/internal_test_util/test_harnesses.py148 # Descriptive name of the harness, used as a testcase_name. Unique in a group.COMMENT
MEDIUMjax/_src/internal_test_util/test_harnesses.py157 # partially implemented in JAX for this harness.COMMENT
MEDIUMjax/_src/internal_test_util/test_harnesses.py386 # Change the testcase name to include the harness name.COMMENT
MEDIUMjax/_src/internal_test_util/test_harnesses.py2003# This first harness runs the tests for all dtypes using default values forCOMMENT
MEDIUMjax/_src/internal_test_util/test_harnesses.py2589# This first harness runs the tests for all dtypes using default values forCOMMENT
MEDIUMjax/_src/internal_test_util/test_harnesses.py2894# This first harness runs the tests for all dtypes and precisions usingCOMMENT
MEDIUMjax/_src/internal_test_util/test_harnesses.py3095 # This first harness runs the tests for all dtypes and precisions usingCOMMENT
MEDIUMjax/_src/internal_test_util/test_harnesses.py351 self.enabled = enabled # Does it apply to the current harness?CODE
Dead Code56 hits · 106 pts
SeverityFileLineSnippetContext
MEDIUMtests/custom_api_test.py374CODE
MEDIUMtests/custom_api_test.py376CODE
MEDIUMtests/custom_api_test.py379CODE
MEDIUMtests/custom_api_test.py382CODE
MEDIUMtests/custom_api_test.py385CODE
MEDIUMtests/custom_api_test.py386CODE
MEDIUMtests/custom_api_test.py387CODE
MEDIUMtests/custom_api_test.py1843CODE
MEDIUMtests/custom_api_test.py1855CODE
MEDIUMtests/custom_api_test.py1858CODE
MEDIUMtests/custom_api_test.py1859CODE
MEDIUMtests/custom_api_test.py1860CODE
MEDIUMtests/custom_api_test.py1862CODE
MEDIUMtests/custom_api_test.py1863CODE
MEDIUMtests/custom_api_test.py1864CODE
MEDIUMtests/lax_numpy_reducers_test.py907CODE
MEDIUMtests/lax_numpy_reducers_test.py908CODE
MEDIUMtests/lax_numpy_reducers_test.py909CODE
MEDIUMtests/lax_numpy_reducers_test.py919CODE
MEDIUMtests/lax_numpy_reducers_test.py920CODE
MEDIUMtests/lax_numpy_reducers_test.py921CODE
MEDIUMtests/pmap_test.py2077CODE
MEDIUMtests/pmap_test.py2079CODE
MEDIUMtests/pmap_test.py2085CODE
MEDIUMtests/pmap_test.py2095CODE
MEDIUMtests/pmap_test.py2099CODE
MEDIUMtests/pmap_test.py2114CODE
MEDIUMtests/pmap_test.py2120CODE
MEDIUMtests/pmap_test.py2121CODE
MEDIUMtests/pmap_test.py2122CODE
MEDIUMtests/pmap_test.py2123CODE
MEDIUMtests/pmap_test.py2134CODE
MEDIUMtests/pmap_test.py2140CODE
MEDIUMtests/pmap_test.py2141CODE
MEDIUMtests/pmap_test.py2142CODE
MEDIUMtests/pmap_test.py2143CODE
MEDIUMtests/pmap_test.py2149CODE
MEDIUMtests/pmap_test.py2155CODE
MEDIUMtests/pmap_test.py2305CODE
MEDIUMtests/pmap_test.py2307CODE
MEDIUMtests/pmap_test.py2310CODE
MEDIUMtests/pmap_test.py2311CODE
MEDIUMtests/pmap_test.py2313CODE
MEDIUMtests/pmap_test.py2314CODE
MEDIUMtests/pmap_test.py2315CODE
MEDIUMtests/pmap_test.py2316CODE
MEDIUMtests/pmap_test.py2328CODE
MEDIUMtests/lax_test.py1775CODE
MEDIUMtests/lax_test.py1807CODE
MEDIUMtests/mosaic/gpu_test.py969CODE
MEDIUMtests/mosaic/gpu_test.py971CODE
MEDIUMtests/mosaic/gpu_test.py972CODE
MEDIUMtests/mosaic/gpu_test.py973CODE
MEDIUMtests/mosaic/gpu_test.py980CODE
MEDIUMjax/_src/lax/control_flow/loops.py2995CODE
MEDIUMjax/_src/pallas/core.py1705CODE
Excessive Try-Catch Wrapping96 hits · 106 pts
SeverityFileLineSnippetContext
MEDIUMci/utilities/convert_msys_paths_to_win_paths.py40 print("Error: cygpath not found. Make sure it's in your PATH.")CODE
MEDIUMci/utilities/convert_msys_paths_to_win_paths.py43 print(f"Error converting path: {e}")CODE
LOWci/postprocess/xml2json.py87 except Exception as e:CODE
LOWtests/array_interoperability_test.py180 except Exception as e:CODE
LOWtests/debug_info_test.py111 except Exception as e:CODE
LOWtests/shape_poly_test.py2040 except Exception:CODE
LOWtests/profiler_test.py63 except Exception as e:CODE
LOWtests/transfer_guard_test.py117 except Exception as e:CODE
LOWtests/transfer_guard_test.py136 except Exception as e:CODE
LOWtests/api_test.py5083 except Exception as e:CODE
LOWtests/pallas/gpu_pallas_distributed_test.py650 except Exception:CODE
LOWtests/pallas/gpu_pallas_distributed_test.py701 except Exception:CODE
LOWtests/pallas/gpu_pallas_distributed_test.py1318 except Exception:CODE
LOWtests/pallas/ops_test.py726 except Exception as e:CODE
LOWtests/pallas/ops_test.py854 except Exception as e:CODE
LOWtests/pallas/ops_test.py2716 except Exception as e:CODE
LOWdocs/array_refs.py82except Exception as e:CODE
LOWdocs/array_refs.py222except Exception as e:CODE
LOWdocs/array_refs.py231except Exception as e:CODE
LOWdocs/array_refs.py240except Exception as e:CODE
LOWdocs/array_refs.py249except Exception as e:CODE
LOWdocs/array_refs.py289except Exception as e:CODE
LOWdocs/array_refs.py413except Exception as e:CODE
LOWdocs/parallel.md425except Exception as e:CODE
LOWdocs/parallel.md464except Exception as e:CODE
LOWdocs/array_refs.md87except Exception as e:CODE
LOWdocs/array_refs.md222except Exception as e:CODE
LOWdocs/array_refs.md231except Exception as e:CODE
LOWdocs/array_refs.md240except Exception as e:CODE
LOWdocs/array_refs.md249except Exception as e:CODE
LOWdocs/array_refs.md288except Exception as e:CODE
LOWdocs/array_refs.md405except Exception as e:CODE
LOWdocs/parallel.py379except Exception as e:CODE
LOWdocs/parallel.py417except Exception as e:CODE
LOWdocs/sphinxext/source_include.py72 except Exception as e:CODE
MEDIUMdocs/sphinxext/source_include.py40def get_tagged_block(filepath, tag, lines_spec=None):CODE
LOW…atic/fault_tolerance/data_parallelism_with_recovery.py171 except Exception as e:CODE
LOWdocs/_static/fault_tolerance/live_devices.py56 except Exception as e:CODE
LOWdocs/_static/fault_tolerance/data_parallelism.py118 except Exception as e:CODE
LOWdocs/new_docs/301/refs.md144except Exception as e:CODE
LOWdocs/new_docs/201/shard-map.md439except Exception as e:CODE
LOWdocs/new_docs/201/shard-map.md513except Exception as e:CODE
LOWdocs/new_docs/201/sharding.md403except Exception as e:CODE
LOWdocs/new_docs/201/sharding.md442except Exception as e:CODE
MEDIUMdocs/pallas/tpu/sparse.md148print("Error |result - lax.dynamic_slice| =", diff)CODE
LOWjax/__init__.py28except Exception as exc:CODE
LOWjax/experimental/multihost_utils.py679 except Exception as e:CODE
LOWjax/experimental/array_serialization/serialization.py238 except Exception as e:CODE
MEDIUMjax/experimental/array_serialization/serialization.py197def _thread_func(self):CODE
LOWjax/experimental/jax2tf/call_tf.py295 except Exception as e:CODE
LOWjax/experimental/jax2tf/call_tf.py592 except Exception as e:CODE
LOW…erimental/jax2tf/tests/jax_primitives_coverage_test.py72 except Exception as e:CODE
LOWjax/experimental/jax2tf/tests/models_test_main.py206 except Exception as e:CODE
LOWjax/experimental/jax2tf/tests/tf_test_util.py246 except Exception as e:CODE
LOWjax/experimental/jax2tf/tests/primitives_test.py135 except Exception as e:CODE
LOWjax/experimental/jax2tf/examples/mnist_lib.py76 except Exception as e:CODE
LOWjax/experimental/colocated_python/obj_backend.py115 except Exception as exc:CODE
LOWjax/experimental/sparse/bcsr.py892 except Exception:CODE
LOWjax/_src/environment_info.py30 except Exception:CODE
MEDIUMjax/_src/environment_info.py27def try_nvidia_smi() -> str | None:CODE
36 more matches not shown…
Modern Structural Boilerplate63 hits · 64 pts
SeverityFileLineSnippetContext
LOWdocs/sphinxext/jax_list_config_options.py22logger = logging.getLogger(__name__)CODE
LOWjaxlib/xla_client.py61logger = logging.getLogger(__name__)CODE
LOWjax/export.py14__all__ = ["DisabledSafetyCheck", "Exported", "export", "deserialize",CODE
LOWjax/debug.py14__all__ = ["callback", "print", "log", "DebugEffect", "OrderedDebugEffect",CODE
LOWjax/ref.py15__all__ = [CODE
LOWjax/interpreters/xla.py15__all__ = ["apply_primitive", "canonicalize_dtype_handlers", "Backend"]CODE
LOWjax/experimental/mosaic/gpu/utils.py42logger = logging.getLogger(__name__)CODE
LOWjax/experimental/mosaic/gpu/layout_inference.py46logger = logging.getLogger(__name__)CODE
LOWjax/experimental/roofline/roofline.py41logger = logging.getLogger(__name__)CODE
LOW…perimental/array_serialization/pytree_serialization.py44logger = logging.getLogger(__name__)CODE
LOW…perimental/array_serialization/pytree_serialization.py61__all__ = ["save", "load", "load_pytreedef",CODE
LOW…x/experimental/array_serialization/tensorstore_impl.py53logger = logging.getLogger(__name__)CODE
LOW…ntal/array_serialization/pytree_serialization_utils.py33logger = logging.getLogger(__name__)CODE
LOW…ntal/array_serialization/pytree_serialization_utils.py45__all__ = ["serialize_pytreedef", "deserialize_pytreedef",CODE
LOWjax/experimental/array_serialization/serialization.py74logger = logging.getLogger(__name__)CODE
LOWjax/experimental/jax2tf/jax2tf.py64logger = logging.getLogger(__name__)CODE
LOWjax/experimental/jax2tf/call_tf.py55logger = logging.getLogger(__name__)CODE
LOWjax/experimental/colocated_python/obj.py50 def update_devices(self, uid: int, device_set: set[jax.Device]) -> None:CODE
LOWjax/_src/callback.py46logger = logging.getLogger(__name__)CODE
LOWjax/_src/lru_cache.py32logger = logging.getLogger(__name__)CODE
LOWjax/_src/xla_bridge.py48logger = logging.getLogger(__name__)CODE
LOWjax/_src/test_loader.py41logger = logging.getLogger(__name__)CODE
LOWjax/_src/config.py37logger = logging.getLogger(__name__)CODE
LOWjax/_src/logging_config.py62def update_logging_level_global(logging_level: str | None) -> None:CODE
LOWjax/_src/compiler.py69logger = logging.getLogger(__name__)CODE
LOWjax/_src/cache_key.py34logger = logging.getLogger(__name__)CODE
LOWjax/_src/hijax.py183 def update_from_loval(self, state: QDD, val: HiVal, /, *lo_vals: LoVal) -> None:CODE
LOWjax/_src/hijax.py186 def update_from_loval2(self, state, val, lo_vals_ft, /) -> None:CODE
LOWjax/_src/hijax.py269 def update_from_loval(self, box_state: BoxTypeState, box, *lo_vals) -> None: # pyrefly: ignore[bad-override]CODE
LOWjax/_src/hijax.py1207 def update_from_loval2(self, qdd, log: Log, lo_ft) -> None:CODE
LOWjax/_src/util.py38logger = logging.getLogger(__name__)CODE
LOWjax/_src/dispatch.py74logger = logging.getLogger(__name__)CODE
LOWjax/_src/debugging.py56logger = logging.getLogger(__name__)CODE
LOWjax/_src/compilation_cache.py51logger = logging.getLogger(__name__)CODE
LOWjax/_src/compilation_cache.py193def set_cache_dir(path) -> None:CODE
LOWjax/_src/error_check.py129def set_error_if(pred: Array, /, msg: str) -> None:CODE
LOWjax/_src/distributed.py28logger = logging.getLogger(__name__)CODE
LOWjax/_src/public_test_util.py30__all__ = ['check_grads', 'check_jvp', 'check_vjp']CODE
LOWjax/_src/ad_checkpoint.py63logger = logging.getLogger(__name__)CODE
LOWjax/_src/jaxpr_util.py44logger = logging.getLogger(__name__)CODE
LOWjax/_src/hardware_utils.py90logger = logging.getLogger(__name__)CODE
LOWjax/_src/profiler.py46logger = logging.getLogger(__name__)CODE
LOWjax/_src/profiler.py139def set_metadata(key: str, value: str) -> None:CODE
LOWjax/_src/path.py20__all__ = ["Path"]CODE
LOWjax/_src/path.py22logger = logging.getLogger(__name__)CODE
LOWjax/_src/interpreters/pxla.py73logger = logging.getLogger(__name__)CODE
LOWjax/_src/interpreters/partial_eval.py63logger = logging.getLogger(__name__)CODE
LOWjax/_src/lax/fft.py34__all__ = [CODE
LOWjax/_src/clusters/k8s_cluster.py30logger = logging.getLogger(__name__)CODE
LOWjax/_src/clusters/cloud_tpu_cluster.py25logger = logging.getLogger(__name__)CODE
LOWjax/_src/clusters/cluster.py22logger = logging.getLogger(__name__)CODE
LOWjax/_src/state/primitives.py1221 if not isinstance(aval, AbstractLinVal): raise TypeError(aval)CODE
LOWjax/_src/state/types.py598 def _setitem(self, tracer, idx, value) -> None:CODE
LOWjax/_src/numpy/array_constructors.py37logger = logging.getLogger(__name__)CODE
LOWjax/_src/numpy/index_tricks.py37__all__ = ["c_", "index_exp", "mgrid", "ogrid", "r_", "s_"]CODE
LOWjax/_src/numpy/array_methods.py22__all__ = ['register_jax_array_methods']CODE
LOWjax/_src/pallas/mosaic/interpret/shared_memory.py34logger = logging.getLogger(__name__)CODE
LOWjax/_src/pallas/triton/lowering.py1349def _set_attr(v: ir.Value, name: str, attr: ir.Attribute) -> None:CODE
LOWjax/_src/pallas/mosaic_gpu/interpret/shared_memory.py33logger = logging.getLogger(__name__)CODE
LOWjax/_src/export/_export.py61logger = logging.getLogger(__name__)CODE
3 more matches not shown…
Redundant / Tautological Comments26 hits · 38 pts
SeverityFileLineSnippetContext
LOWtests/array_interoperability_test.py209 # Check if the source device is preservedCOMMENT
LOWtests/nn_test.py120 # Check if float8_e8m0fnu is availableCOMMENT
LOWtests/nn_test.py328 # Check if cudnn backend is called (only on CUDA).COMMENT
LOWtests/pgle_test.py277 # Check if FDO profile file of the biggest module is not emptyCOMMENT
LOWtests/pgle_test.py285 # Check if FDO profile file in dump directory is not emptyCOMMENT
LOWtests/multiprocess/array_test.py297 # Check if we can specify that local input actually contains full-spanCOMMENT
LOW…allas/tpu_fusible_matmul_with_stateful_fusions_test.py145 # Check if either input fusion reads from a Ref that is also written by theCOMMENT
LOWtests/pallas/tpu_pallas_call_print_test.py202 # Check if the numbers in the output match the values generated by `arange`.COMMENT
LOW.github/workflows/upload_metadata.yml67 # Check if zip file is empty before unzippingCOMMENT
LOW.github/workflows/verify-squash.yml32 # Check if the skip label is presentCOMMENT
LOW.github/workflows/verify-squash.yml55 # Check if squashedCOMMENT
LOW.github/workflows/build_artifacts.yml169 # Set shell to cmd to avoid path errors when using gcloud commands on WindowsCOMMENT
LOWjax/experimental/mosaic/gpu/fragmented_array.py5033 # Check if input maps exactly to the end (prevents trailing dims).COMMENT
LOWjax/_src/test_util.py1307 # Check if strict hermeticity is already satisfiedCOMMENT
LOWjax/_src/checkify.py867 # Check if the first cond application will error.COMMENT
LOWjax/_src/pretty_printer.py48 # Check if we're in IPython or ColabCOMMENT
LOWjax/_src/lax/control_flow/loops.py2251 # Check if the same Ref is written to in both cond and body.COMMENT
LOWjax/_src/state/primitives.py764 # Check if start is static (which it can be)COMMENT
LOWjax/_src/state/primitives.py777 # Check if we are indexing with a scalar or not. If we are indexingCOMMENT
LOWjax/_src/numpy/setops.py177 # Set mask to zero at locations corresponding to unique() padding.COMMENT
LOWjax/_src/numpy/array_constructors.py252 # Check if object supports any of the data exchange protocolsCOMMENT
LOWjax/_src/pallas/core.py1250 elif grid and isinstance(grid[0], tuple): # Check if we have a named gridCODE
LOWjax/_src/pallas/einshape.py232 # Check if all RHS dims are knownCOMMENT
LOWjax/_src/scipy/signal.py681 # Check if we can broadcast the outer axes togetherCOMMENT
LOWjax/_src/scipy/signal.py730 # Check if x and y are the same length, zero-pad if necessaryCOMMENT
LOWjax/_src/cudnn/fused_attention_stablehlo.py1265 # Check if all required keys are presentCOMMENT
Fake / Example Data21 hits · 30 pts
SeverityFileLineSnippetContext
LOWtests/export_back_compat_test.py96 dummy_data = self.load_testdata(bctu.dummy_data_dict)CODE
LOWtests/export_back_compat_test.py98 dummy_data, platform=self.default_jax_backend())CODE
LOWtests/export_back_compat_test.py103 dummy_data = self.load_testdata(bctu.dummy_data_dict)CODE
LOWtests/export_back_compat_test.py105 dummy_data,CODE
LOWtests/export_back_compat_test.py113 dummy_data = self.load_testdata(bctu.dummy_data_dict)CODE
LOWtests/export_back_compat_test.py115 dummy_data,CODE
LOWtests/shard_map_test.py634 dummy_data = jax.lax.precv(CODE
LOWtests/shard_map_test.py642 dummy_data,CODE
LOWtests/shard_map_test.py646 return dummy_dataCODE
LOWdocs/gpu_performance_tips.md400 dummy_data = jnp.zeros(CODE
LOWdocs/gpu_performance_tips.md403 dummy_data = jax.device_put(CODE
LOWdocs/gpu_performance_tips.md404 dummy_data,CODE
LOWdocs/gpu_performance_tips.md411 carry = weights, input_buffer, output_buffer, dummy_data, dummy_dataCODE
LOWdocs/gpu_performance_tips.md590 weights, input_buffer, dummy_data, meshCODE
LOWdocs/gpu_performance_tips.md599 jax.ShapeDtypeStruct.like(dummy_data),CODE
LOWdocs/gpu_performance_tips.md609 dummy_data,CODE
LOWdocs/gpu_performance_tips.md610 dummy_data,CODE
LOWdocs/gpu_performance_tips.md677 dummy_data = jax.device_put(CODE
LOWdocs/gpu_performance_tips.md692 )(weights, input_buffer, dummy_data)CODE
LOW…perimental/array_serialization/pytree_serialization.py237 dummy_key_path = os.path.join(kvstore_path, "dummy_key")CODE
LOW…src/internal_test_util/export_back_compat_test_util.py128# The dummy_data is used for getting started for adding a new test and forCOMMENT
Hallucination Indicators3 hits · 30 pts
SeverityFileLineSnippetContext
CRITICALtests/mosaic/profiler_cupti_test.py96 jax._src.lib.mosaic_gpu._mosaic_gpu_ext._cupti_init()CODE
CRITICALtests/filecheck/jax_mlir_ext.filecheck.py114 return jax._src.lib._jax.Traceback.get_traceback()CODE
CRITICALjax/_src/pallas/mosaic_gpu/lowering.py4034 for k, v in ctx.launch_ctx.module.operation.attributes.items():CODE
Slop Phrases13 hits · 26 pts
SeverityFileLineSnippetContext
MEDIUMdocs/array_refs.py609# wouldn't be for `vmap` or `shard_map`, as discussed above). Once all theCOMMENT
MEDIUMdocs/parallel.py396# want the compiler to choose them automatically, you can use the `@auto_axes`COMMENT
MEDIUMdocs/ffi.py161# // this handler in a header, you can use the `XLA_FFI_DECLARE_HANDLER_SYMBOL`COMMENT
MEDIUMdocs/ffi.py209# If you're familiar with the legacy "custom call" API, it's worth noting that you can also use {func}`~jax.ffi.registerCOMMENT
MEDIUMdocs/ffi.py209# If you're familiar with the legacy "custom call" API, it's worth noting that you can also use {func}`~jax.ffi.registerCOMMENT
MEDIUMjax/experimental/jax2tf/examples/saved_model_lib.py102 # names, you can use `tree.map_structure_with_path` from the `dm-tree` packageCOMMENT
LOWjax/_src/core.py2899 # but we make sure to reset it to Device because the Ref owns the memory spaceCOMMENT
MEDIUMjax/_src/ad_checkpoint.py289 Here is a simple example:STRING
MEDIUMjax/_src/errors.py201 indices, such as with :code:`.at[...].set(...)`. Here is a simple example::STRING
MEDIUMjax/_src/errors.py666 Here is a simple example of code that would lead to such an error::STRING
MEDIUMjax/_src/lax/windowed_reductions.py154 Here is a simple example of a windowed product over pairs in a 1-dimensional array:STRING
MEDIUMjax/_src/lax/slicing.py371 For example, here is how you can extract values at particular indices usingSTRING
LOWjax/_src/pallas/core.py801 # If you hit this, make sure you take transforms into account and use eitherCOMMENT
Synthetic Comment Markers3 hits · 20 pts
SeverityFileLineSnippetContext
HIGHdocs/contributing.md26## Can I contribute AI generated code?COMMENT
HIGHdocs/contributing.md35You are responsible for any code you contribute to JAX, regardless of whether it was written manually or generated by AICODE
HIGHjaxlib/py_executable.h287 // Python objects to keep alive as requested by user.COMMENT
AI Response Leakage2 hits · 15 pts
SeverityFileLineSnippetContext
HIGHdocs/ffi.py133# // In this example, we treat all leading dimensions as batch dimensions, so thisCOMMENT
HIGHdocs/ffi.py176# In this implementation, we treat all leading dimensions of the buffer as batch dimensions, and perform RMS normalizatiCOMMENT
Structural Annotation Overuse8 hits · 12 pts
SeverityFileLineSnippetContext
LOWdocs/distributed_data_loading.md230# Step 1: setup the Dataset for pure data parallelism (do once)COMMENT
LOWdocs/distributed_data_loading.md239# Step 2: create a jax.Array of per-replica batches from the per-process batchCOMMENT
LOWdocs/distributed_data_loading.md325# Step 1: Set up the Dataset with a different data shard per-process (do once)COMMENT
LOWdocs/distributed_data_loading.md335# Step 2: Create a jax.Array of per-replica batches from the per-process batchCOMMENT
LOWdocs/new_docs/501/data-loading.md221# Step 1: setup the Dataset for pure data parallelism (do once)COMMENT
LOWdocs/new_docs/501/data-loading.md230# Step 2: create a jax.Array of per-replica batches from the per-process batchCOMMENT
LOWdocs/new_docs/501/data-loading.md316# Step 1: Set up the Dataset with a different data shard per-process (do once)COMMENT
LOWdocs/new_docs/501/data-loading.md326# Step 2: Create a jax.Array of per-replica batches from the per-process batchCOMMENT
Overly Generic Function Names10 hits · 10 pts
SeverityFileLineSnippetContext
LOWtests/core_test.py584 def my_function():CODE
LOWtests/util_test.py80 def my_function():CODE
LOWtests/aot_test.py122 def my_function(x):CODE
LOWtests/aot_test.py143 def my_function(x):CODE
LOWtests/source_info_test.py279 def my_function(x):CODE
LOWtests/api_test.py93 def my_function():CODE
LOWtests/api_test.py107 def my_function():CODE
LOWtests/api_test.py1622 def my_function(x, flag):CODE
LOWjax/typing.py47 def my_function(x: ArrayLike) -> Array:STRING
LOWjax/experimental/jax2tf/tests/jax2tf_test.py306 def test_function(self):CODE
Example Usage Blocks3 hits · 5 pts
SeverityFileLineSnippetContext
LOWci/utilities/run_docker_container.sh24# Usage:COMMENT
LOWjaxlib/mosaic/dialect/tpu/util.h110// Example usage:COMMENT
LOWjax/_src/deprecations.py22# Example usage:COMMENT
Modern AI Meta-Vocabulary1 hit · 3 pts
SeverityFileLineSnippetContext
MEDIUM.github/workflows/wheel_tests_continuous.yml5# It orchestrates the following:COMMENT
TODO Padding2 hits · 3 pts
SeverityFileLineSnippetContext
LOWjax/_src/lax/slicing.py822 # TODO: implement this via its own primitive so we can define appropriate autodiff rules.COMMENT
LOWjax/_src/lax/linalg.py2332 # TODO: implement full matrices case, documented here: https://people.maths.ox.ac.uk/gilesm/files/NA-08-01.pdfCOMMENT
Verbosity Indicators1 hit · 2 pts
SeverityFileLineSnippetContext
LOWjaxlib/python_ref_manager.h85 // The purpose of this function is to amortize lock acquisition costs overCOMMENT