Vivek Panyam
Vivek Panyam
Neuropod's release cadence has been a bit odd in that we've released a bunch of RCs since the initial public release, but haven't put out a new "stable" release (although...
## Context The user provided `node_name_mapping` is not used when running TF SavedModels. It's only used when running TF frozen graphs. TF SavedModels have `SignaureDef`s. These are quite similar to...
### Summary: Neuropod did not initially use C++14 so we needed our own `make_unique` implementation. We upgraded past C++14 a while ago so this is no longer necessary. ### Test...
### Summary: Add initial support for running infer using `asyncio`. This implementation currently blocks the event loop so it shouldn't be used just yet ### Test Plan: Added a test...
### Summary: Refactor code under `source/neuropod/multiprocess` to adhere to the stricter set of warnings introduced in #437 ### Test Plan: CI
### Summary: Allow users to configure intra-op and inter-op thread pool sizes for the underlying frameworks. **Note: This API is experimental and may change in the future.** ### Test Plan:
This PR implements Torch specific seed and determinism options. Closes #325
This PR ensures that we're running on the correct device even if something else calls `cudaSetDevice` before running inference. This fixes a class of issues where another piece of code...
This check currently ensures that `neuropod_path` is not a subdirectory of `python_root`. However, it misses cases where they are the same directory https://github.com/uber/neuropods/blob/9c4b58b84429b4390e8dfa658963d65312b6b52c/source/neuropods/python/backends/python/packager.py#L152-L153
Currently, we only have saved model tests for TF 2.x ([here](https://github.com/uber/neuropod/blob/38091314e8544abf0b50403344a9b1b1f3fa4a6f/source/python/neuropod/tests/test_tensorflow_v2_packaging.py)). The Python API for creating a savedmodel is different in TF 1.x vs 2.x so we should add a...