Summary:
There is a module called `2to3` which you can target for future specifically to remove these, the directory of `caffe2` has the most redundant imports:
```2to3 -f future -w caffe2```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/45033
Reviewed By: seemethere
Differential Revision: D23808648
Pulled By: bugra
fbshipit-source-id: 38971900f0fe43ab44a9168e57f2307580d36a38
Summary:
## TLDR
Support using NaN default value for missing dense features in RawInputProcessor for DPER2. In preparation for subsequent support for null flag features in compute meta. For train_eval this is already supported in DPER3 and we do not plan to support this in DPER2 train eval.
Differential Revision: D22439142
fbshipit-source-id: 99ae9755bd41a5d5f43bf5a9a2819d64f3883005
Summary:
Original commit changeset: 46c59d849fa8
The original commit is breaking DPER3 release pipeline with the following failures:
https://www.internalfb.com/intern/chronos/jobinstance?jobinstanceid=9007207344413239&smc=chronos_gp_admin_client&offset=0
```
Child workflow f 202599639 failed with error: c10::Error: [enforce fail at operator.cc:76] blob != nullptr. op Save: Encountered a non-existing input blob: feature_preproc/feature_sparse_to_dense/default_float_value
```
https://www.internalfb.com/intern/chronos/jobinstance?jobinstanceid=9007207344855973&smc=chronos_gp_admin_client&offset=0
```
Child workflow f 202629391 failed with error: c10::Error: [enforce fail at operator.cc:76] blob != nullptr. op Save: Encountered a non-existing input blob: tum_preproc/inductive/feature_sparse_to_dense/default_float_value
```
Related UBN tasks: T69529846, T68986110
Test Plan: Build a DPER3 package on top of this commit, and check that DPER3 release test `model_deliverability_test` is passing.
Differential Revision: D22396317
fbshipit-source-id: 92d5b30cc146c005d6159a8d5bfe8973e2c546dd
Summary:
## TLDR
Support using NaN default value for missing dense features in RawInputProcessor for DPER2. In preparation for subsequent support for null flag features in compute meta. For train_eval this is already supported in DPER3 and we do not plan to support this in DPER2 train eval.
## Overview
Intern project plan to support adding dense flags for missing feature values instead of replacing with zero.
## Project plan :
https://docs.google.com/document/d/1OsPUTjpJycwxWLCue3Tnb1mx0uDC_2KKWvC1Rwpo2NI/edit?usp=sharing
## Code paths:
See https://fb.quip.com/eFXUA0tbDmNw for the call stack for all affected code paths.
Test Plan:
## fblearner flow test
1. `flow-cli clone f197867430 --run-as-secure-group ads_personalization_systems --force-build` to build a ephemeral package and start a fblearner flow run (may fail)
2. Clone the new run and change the secure_group to `XXXX` and entitlement to `default` in the UI
3. Adds explicit_null_min_coverage flag
4. Optionally reduce `max_examples` since we only test pass/fail instead of quality.
5. Submit the run to test the change
Example:
f198538878
## compare output coverages to daiquery runs
1. Randomly select null flag features from compute meta workflow output
2. Look up the feature id in feature metadata using feature name
3. Check against a daiquery sample of coverage to see if the coverage falls within guidelines.
https://www.internalfb.com/intern/daiquery/workspace/275342740223489/192619942076136/
## Sampled features:
GFF_C66_ADS_USER_SUM_84_PAGE_TYPE_RATIO_EVENT_LIKE_IMPRESSION: 15694257
- original feature compute meta coverage: 0.999992
- daiquery feature coverage (10k rows): 0.69588
- null flag compute meta coverage: 0.293409
GFF_R1303_ADS_USER_SUM_7_PAGE_TYPE_COUNTER_CONVERSION: 16051183
- original feature compute meta coverage: 0.949868
- daiquery feature coverage: 0.82241
- null flag compute meta coverage: 0.151687
## Unit tests:
`buck test fblearner/flow/projects/dper/tests/workflows:ads_test`
https://www.internalfb.com/intern/testinfra/testconsole/testrun/6192449504303863/
Differential Revision: D22026450
fbshipit-source-id: 46c59d849fa89253f14dc2b035c4c677cd6e3a4c
Summary:
## TLDR
Support using NaN default value for missing dense features in RawInputProcessor for *DPER2*. In preparation for subsequent support for null flag features in *compute meta*. For train_eval this is already supported in DPER3 and we do not plan to support this in DPER2 train eval.
## Overview
Intern project plan to support adding dense flags for missing feature values instead of replacing with zero.
Project plan :
https://docs.google.com/document/d/1OsPUTjpJycwxWLCue3Tnb1mx0uDC_2KKWvC1Rwpo2NI/edit?usp=sharing
## Code paths:
See https://fb.quip.com/eFXUA0tbDmNw for the call stack for all affected code paths.
Test Plan:
# A. DPER3 blob value inspection
## 1. Build local bento kernel in fbcode folder
`buck build mode/dev-nosan //bento/kernels:bento_kernel_ads_ranking`
## 2. Use kernel `ads_ranking (local)` to print dense feature blob values
n280239
## 2.1 Try `default_dense_value = "0.0"` (default)
```
preproc_6/feature_preproc_6/dper_feature_processor_7/raw_input_proc_7/float_feature_sparse_to_dense_7/float_features [[0. ]
[0. ]
[0. ]
[0. ]
[0. ]
[0. ]
[0. ]
[1. ]
[1.7857143]
[1.7777778]
[1. ]
[0. ]
[0.5625 ]
[0. ]
[0. ]
[0.8 ]
[0. ]
[1. ]
[0.56 ]
[0. ]]
```
## 2.2 Try `default_dense_value = "123"`
```
preproc_2/feature_preproc_2/dper_feature_processor_3/raw_input_proc_3/float_feature_sparse_to_dense_3/float_features [[123. ]
[123. ]
[123. ]
[123. ]
[123. ]
[123. ]
[123. ]
[ 1. ]
[ 1.7857143]
[ 1.7777778]
[ 1. ]
[123. ]
[ 0.5625 ]
[123. ]
[123. ]
[ 0.8 ]
[123. ]
[ 1. ]
[ 0.56 ]
[123. ]]
```
## 2.3 Try `default_dense_value = float("nan")`
```
RuntimeError: [enforce fail at enforce_finite_op.h:40] std::isfinite(input_data[i]). Index 0 is not finite (e.g., NaN, Inf): -nan (Error from operator:
input: "unary_4/logistic_regression_loss_4/average_loss_4/average_loss" name: "" type: "EnforceFinite" device_option { random_seed: 54 })
```
which is expected due to nan input.
# B. Unit test
`buck test fblearner/flow/projects/dper/tests/preprocs:raw_feature_extractor_test`
https://www.internalfb.com/intern/testinfra/testconsole/testrun/5348024586274923/
{F241336814}
Differential Revision: D21961595
fbshipit-source-id: 3dcb153b3c7f42f391584f5e7f52f3d9c76de31f
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/23983
While testing I realized that model layers can extract different types of features from the same column. For example, MultifeedFeaturesTransform uses float and ID list features from the "features" column.
get_accessed_features returns a map from column to AccessedFeatures, and AccessedFeatures only has the feature IDs for one feature type. This is incompatible with have multiple types of features per column, one type ends up overwriting another in the map.
To fix this, I've modified get_accessed_features to return a map from column to a list of AccessedFeatures objects.
Reviewed By: itomatik
Differential Revision: D16693845
fbshipit-source-id: 2099aac8dc3920dd61de6b6ad5cf343c864803bc
Summary:
We need a way to figure get a complete list fo features that are used in training a model. One way to do this is to make it possible to get the list of features used in each Model Layer. Then once the model is complete we can go through the layers and aggregate the features.
I've introduced a function to expose that information here, get_accessed_features, and implemented it in the FeatureSparseToDense layer to start with.
I've tried to include the minimum amount of information to make this useful, while making it easy to integrate into the variety of model layers. This is, for example, why AccessedFeatures does not contain feature_names which is not always present in a model layer. I debated whether or not to include feature_type, but I think that's useful enough, and easy enough to figure out in a model layer, that it's worth including.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/23036
Test Plan:
Added a unit test to verify the behavior of get_accessed_features in FeatureSparseToDense.
aml_dper2-fblearner-flow-integration-tests failed due to a known issue D16355865
aml_dper3-fblearner-flow-integration-tests failed due to a known issue T47197113
I verified no tests in the integration tests failed to issues other than those known ones.
DPER2 canaries: https://fburl.com/fblearner/1217voga
Reviewed By: volkhin
Differential Revision: D16365380
Pulled By: kevinwilfong
fbshipit-source-id: 2dbb4d832628180336533f29f7d917cbad171950
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/16191
logdevice related modifications for generic feature type
we directly convert the generic feature structures to json strings, which corresponds to the column input in offline and dper
Reviewed By: itomatik
Differential Revision: D13551909
fbshipit-source-id: 807830c50bee569de202530bc3700374757793a2
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/10197
Support generic feature in DPER2
For now since we only have one generic type 1, we are directly adding the parsed feature record to embedding feature.
For new feature types with specific structure, there should also be corresponding coding changes expected.
Reviewed By: itomatik
Differential Revision: D8788177
fbshipit-source-id: 9aaa6f35ece382acb4072ec5e57061bb0727f184
Summary:
To achive this, I modified the blob name scheme defined in a layer.
Before it was scope/fc_w and scope/fc_w_auto_0 (if there is another fc
within the same scope).
Now I change it to scope/fc/w and scope/fc_auto_0/w.
That is, we rely on the uniqueness of the scoped layer name to define
names for blobs.
I also overwrote the create_param method in LayerModelHelper to let it
use the resolved name for blobs given the sharingparameter context.
There are some details such as making the initializer more structured
that I need to finalize.
Reviewed By: kennyhorror
Differential Revision: D5435132
fbshipit-source-id: a0525f5ea0977e255dd5ea765b38913f5951d455
Summary:
The SparseToDense layer is essentially calling the SparseToDenseMask op.
This makes it impossible to call the functional layer with the true SparseToDense op.
This diff is to rename the layer.
Please let me know if I missed anything or you have a better name suggestion.
Differential Revision: D5169353
fbshipit-source-id: 724d3c6dba81448a6db054f044176ffc7f708bdb
