-
Notifications
You must be signed in to change notification settings - Fork 26
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merge branch 'submission-v4.1' into 'master' #949
base: master
Are you sure you want to change the base?
Conversation
co-author: Aswin B <quic_aswib@quicinc.com> co-author: Utkarsh Mishra <quic_utkarshm@quicinc.com>
Final Submission for code for Qualcomm
* Ignore QTI private SDK and libs * Update path for QTI SDK * Download QTI libs for SD task * Fix QTI BUILD file * Update integration test * Set WITH_STABLEDIFFUSION=1 for CI build * Fix QTI pbtxt files * Disable SD task for Apple devices * Update model path and checksum for QTI * Remove tiny groundtruth_path for SD task * Update path for TFLite models * Update groundtruth_path for SD task
* hacks to enable stable diffusion on Pixel backend * buildifer and clang-format * remove unnecessary line * adapt to submission-v4.1
* Add icon for stable_diffusion task * Add description for stable_diffusion task * Sort the order of task based on BenchmarkId.allIds * Fix ios-build-test.yml * Fix ios-build-test.yml
* feat: pass task-specific config to backend (#922) * Add TaskConfig.CustomConfig and pass them to backend * Add CustomConfig for main.cc * Use seed and num_steps from CustomConfig for TFLite backend * Replace std::cout with LOG(INFO) * Format files * feat: add ConvertOutputs() API (#927) * Add ConvertOutputs() API * Add ConvertOutputs() for mobile_back_tflite * Set minimum macos version * Set minimum macos version to 13.1 * Update _kIphoneOnGitHubAction * feat: timestamp-embedding-parser (WIP) * disabled bitcode to be able compile with new XCode * chore: formatting * refactor: use custom setting in Core ML backend to detect NCHW input. (#924) * Add GetConfigValue() * Add custom setting data-format for Core ML * Use GetConfigValue() to get stable_diffusion_seed and stable_diffusion_num_steps * fix: resolve crash due to permission denied on Android Play Store version (#930) * Set android:extractNativeLibs="true" * Set android.bundle.enableUncompressedNativeLibs=false * chore: increase Android minSdkVersion from 21 to 30 (#859) Increase minSdkVersion to 30 * feat: finalized SD pipeline to use embedding from the binary file. * refactor: updated embedding_utils to parse pkl file * chore: linting * fix: fixed lint issue in neuron * chore: BUILD cleanup * chore: cleanup * chore: ignore .fvm * chore: updated model paths and checksums for stable diffusion benchmark: tflite_settings_android.pbtxt * chore: reverse timesteps and embeddings to support descending order of the timesteps and embeddings * chore: fixed formatting * chore: added links to the sd models and timestep embeddings file * chore: add the proper name for the embedding_timesteps file * chore: added missed declaration for backend_convert_outputs * chore: clang formatting * chore: added missed files * chore: fixed build file for the pixel backend * chore: bazel formatting * fix: added missed interface implementation for pixel * chore: clang formatting --------- Co-authored-by: Anh <anh.app.dev@gmail.com>
# Conflicts: # flutter/assets/tasks.pbtxt # flutter/cpp/datasets/coco_gen.cc # flutter/cpp/proto/mlperf_task.proto # flutter/integration_test/expected_accuracy.dart # flutter/integration_test/expected_throughput.dart # mobile_back_apple/dev-utils/Makefile # mobile_back_qti/.gitignore # mobile_back_qti/cpp/backend_qti/settings/qti_settings_default_cpu.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_default_dsp.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_default_gpu.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_gpufp16.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_sd7cxg3.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_sd7g1.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_sd7pg2.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_sd8cxg3.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_sd8g1.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_sd8g2.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_sd8g3.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_sd8pg1.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_sdm778.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_sdm888.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_sm4450.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_sm7550.pbtxt # mobile_back_qti/cpp/backend_qti/settings/qti_settings_sm8635.pbtxt # mobile_back_tflite/cpp/backend_tflite/neuron/backend_settings/tflite_settings_mtk_mt6989.pbtxt # mobile_back_tflite/cpp/backend_tflite/stable_diffusion_invoker.cc
MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅ |
@mohitmundhragithub It looks like the QTI backend blocks the fallback to other backends.
log-Samsung Galaxy M32.txt For reference, here is the order of check:
|
@anhappdev let's create 5.0 submission branch from the 4.1 branch if we cannot do it from the master branch. The submission date is next Fri. @mohitmundhragithub and @Mostelk. For submission, we don't need an integrated one working. |
Should we create the 5.0 branch from this PR / branch? Since it contains all the commits from both |
Quality Gate passedIssues Measures |
sounds good to me. |
Hi @anhappdev, Can you please upload the ts-embedding pkl file as well, preferably in mobile_open? |
Ok. I uploaded it here: https://github.com/mlcommons/mobile_open/releases/tag/v4.1-sd-tflite |
the pkl file you uploaded is different from the one that is generated in this comment. |
Ok. I replaced the file with the new one. Please check again. |
The branch
anh/merge-v4.1-into-master
is used to resolve merge conflicts when mergingsubmission-v4.1
intomaster
.