adding model and dataset flags to harness#18
adding model and dataset flags to harness#18stamcenter wants to merge 7 commits intofhe-benchmarking:mainfrom
Conversation
…r multiple models
stamcenter
changed the title
andreea-alexandru
left a comment
andreea-alexandru
left a comment
There was a problem hiding this comment.
Looks good to me, I just left a few minor comments.
harness/generate_input.py
Outdated
| labels_file=LABELS_PATH, | ||
| num_samples=num_samples, | ||
| seed=seed) | ||
| if dataset_name == "mnist": |
There was a problem hiding this comment.
Turn this into a switch statement in preparation for multiple models?
| exec_dir = params.rootdir/ ("submission_remote" if remote_be else "submission") | ||
| exec_dir = params.rootdir/ ("submission_remote" if remote_be else "submissions") | ||
|
|
||
| # check whether the exec_dir contains a subdirector equals to the model name. |
| @@ -28,20 +28,26 @@ def main(): | |||
|
|
|||
There was a problem hiding this comment.
Update the README to reflect the new python3 harness/run_submission.py -h.
harness/utils.py
Outdated
| "total_latency_ms": round(sum(_timestamps.values()), 4), | ||
| "per_stage": _timestampsStr, | ||
| "bandwidth": _bandwidth, | ||
| "mnist_model_quality" : _model_quality, |
There was a problem hiding this comment.
Just "model_quality"?
| #include <chrono> | ||
| #include <fstream> | ||
| #include <iomanip> | ||
| #include <nlohmann/json.hpp> |
There was a problem hiding this comment.
It doesn't seem like something relevant changed in build_task.sh, but now I get a compilation error here
fatal error: nlohmann/json.hpp: No such file or directory
24 | #include <nlohmann/json.hpp>
There was a problem hiding this comment.
It is happening because now the submission is in submission/mlp instead of submission. The relative paths in CMake need to be changed.
include_directories(${CMAKE_CURRENT_SOURCE_DIR}/../../third_party)
2 participants
This pull request extends the harness interface by introducing two new parameters:
--model
--dataset
These parameters allow the harness to dynamically select the model and dataset used during submission execution. It also moved the fault mlp model provided in the submission folder into the submissions folder for testing.