You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
* Add a second monitor stage to report progress for the back-half of the pipeline
* Increase threads to match number of stages
* Change triton port to 8000 avoiding a warning about the grpc port
* Add EdgeConnectors for derived classes of MultiMessage to MultiMessage
* Remove filter stage from example #83 & #197fixes#186
Authors:
- David Gardner (https://github.com/dagardner-nv)
Approvers:
- Michael Demoret (https://github.com/mdemoret-nv)
URL: #193
* `FilterDetectionStage` now optionally performs a copy of matching rows (now the default behavior), this resolves the performance issues discovered in #83
* Adds a `copy_ranges` method to `MultiMessage` which parallels the `get_slice` method except that it takes a full copy and is able to select multiple ranges from the table.
* `MultiMessage` now inherits from `DerivedMultiMessage` which implements the curiously recurring template pattern. This allows the public versions of `get_slice` and `copy_ranges` to return a `std::shared_ptr<T>` which removes the need of implementing `get_slice` in multiple derived classes.
* Adds new `TensorMemory` class which serves as the base class for `InferenceMemory` & `ResponseMemory`.
* For `MultiInference` and `MultiResponce` this also copies the input and output tensors respectively.
* Adds `copy_rows` method to `TensorObject` and `Tensor` classes.
* Drive-by fix for `MultiMessage.get_meta_list` which was broken by our recent cudf upgrade, and never had a C++ impl (still doesn't but I added it to the binding).
Fixes#83Fixes#197Fixes#217
Authors:
- David Gardner (https://github.com/dagardner-nv)
- Michael Demoret (https://github.com/mdemoret-nv)
Approvers:
- Michael Demoret (https://github.com/mdemoret-nv)
URL: #231
Report incorrect documentation
#193 Removed the filter stage from this example as a work-around for #83.
Once that is resolved we should add the filter stage back in.
Location of incorrect documentation
examples/nlp_si_detection/README.md
The text was updated successfully, but these errors were encountered: