Repository navigation
Conversation
Signed-off-by: Frank Lai <frlai@nvidia.com>
Signed-off-by: Frank Lai <frlai@nvidia.com>
Signed-off-by: Marc Alwan <malwan@nvidia.com>
Signed-off-by: Marc Alwan <malwan@nvidia.com>
Signed-off-by: Marc Alwan <malwan@nvidia.com>
Signed-off-by: Marc Alwan <malwan@nvidia.com>
malwan-nv
force-pushed
the
malwan-nv/leapp-compiled-runtime
branch
from
October 14, 2026 17:26
7325dc2 to
2b281a0
Compare
frlai
reviewed
Oct 16, 2026
replace the hand-maintained in-place allowlist with cached ATen schema discovery preserve eager receiver identity while recording deterministic functional graph operations support metadata-mutating views such as squeeze_, unsqueeze_, and transpose_ cover masked_fill_, index_copy_, cumsum_, and other schema-compatible methods reject nondeterministic or incompatible methods before mutation consolidate overlapping traced-tensor regression coverage while retaining distinct semantics
Signed-off-by: Frank Lai <frlai@nvidia.com>
Collaborator
|
another thing. the runtime needs a reset call. reset needs to clear all the buffers and load in the initial values. |
* docs: clarify whole-pipeline export and optional stage boundaries Signed-off-by: Frank Lai <frlai@nvidia.com> * docs: streamline pipeline pitch and single-model quick start Signed-off-by: Frank Lai <frlai@nvidia.com> * docs: correct capture scope and getting started references Signed-off-by: Frank Lai <frlai@nvidia.com> * docs: focus README on general pipeline export workflow Signed-off-by: Frank Lai <frlai@nvidia.com> * docs: clarify stateful cached validation and sequential replay Signed-off-by: Frank Lai <frlai@nvidia.com> * docs: remove em dashes from README Signed-off-by: Frank Lai <frlai@nvidia.com> * docs: describe capture flexibility as a capability Signed-off-by: Frank Lai <frlai@nvidia.com> * docs: remove redundant supported computation overview Signed-off-by: Frank Lai <frlai@nvidia.com> * docs: restore concise YAML artifact summary Signed-off-by: Frank Lai <frlai@nvidia.com> --------- Signed-off-by: Frank Lai <frlai@nvidia.com>
Signed-off-by: Frank Lai <frlai@nvidia.com>
# Conflicts: # README.md # docs/source/conf.py
Signed-off-by: Marc Alwan <malwan@nvidia.com>
Signed-off-by: Marc Alwan <malwan@nvidia.com>
Signed-off-by: Marc Alwan <malwan@nvidia.com>
Signed-off-by: Marc Alwan <malwan@nvidia.com>
Signed-off-by: Marc Alwan <malwan@nvidia.com>
Signed-off-by: Frank Lai <frlai@nvidia.com>
Signed-off-by: Frank Lai NV <frlai@nvidia.com>
Signed-off-by: Frank Lai NV <frlai@nvidia.com>
* feat: lower NumPy calls through ATen and support reverse slices Signed-off-by: Frank Lai <frlai@nvidia.com> * docs: remove NumPy ATen experiment report Signed-off-by: Frank Lai <frlai@nvidia.com> * test: consolidate NumPy tracing coverage Signed-off-by: Frank Lai <frlai@nvidia.com> * docs: explain NumPy ATen capture and supported limits Signed-off-by: Frank Lai <frlai@nvidia.com> * docs: describe NumPy capture without historical comparisons Signed-off-by: Frank Lai <frlai@nvidia.com> * fix: defer recoverable NumPy capture errors until output validation Signed-off-by: Frank Lai <frlai@nvidia.com> * perf: validate NumPy numerics at exported node outputs Signed-off-by: Frank Lai <frlai@nvidia.com> * test: fold NumPy comparison probe into replay coverage Signed-off-by: Frank Lai <frlai@nvidia.com> * docs: simplify NumPy capture explanation Signed-off-by: Frank Lai <frlai@nvidia.com> * test: use portable cosine dtypes and remove duplicate NumPy checks Signed-off-by: Frank Lai <frlai@nvidia.com> --------- Signed-off-by: Frank Lai <frlai@nvidia.com>
Move the bundle regression target under tests/unit_tests and remove the redundant C++ fixture-loading check. Keep the inspection-only build instructions aligned with the test layout. Signed-off-by: Marc Alwan <malwan@nvidia.com>
Export Session, construction options, result/error types, and backend-independent DLPack ownership. Compile the common binding wrapper against the real core API. Execution returns Unsupported in this contract revision and is supplied by the runtime integration. Signed-off-by: Marc Alwan <malwan@nvidia.com>
malwan-nv
marked this pull request as ready for review
October 21, 2026 22:29
Bring the latest develop fixes and formatting requirements through the PR stack while preserving published commit history. Apply required Ruff formatting without changing Python syntax trees. Signed-off-by: Marc Alwan <malwan@nvidia.com>
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds
packages/leapp-runtime-rustas a public API for loading and inspecting exported LEAPP bundles from Rust, Python, and C++.The Rust loader validates graph connections, artifact paths, optional SHA-256 hashes, and safetensors initial state. Python exposes detached inspection values and structured errors.