Skip to content

add session-based example with pretrained embeddings #1102

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.

Sign up for GitHub

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

Merged
radekosmulski merged 22 commits into from
Jun 19, 2023
Merged

add session-based example with pretrained embeddings #1102

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
22 commits
Select commit Hold shift + click to select a range
7ea4156
add SIGIR nb
radekosmulski May 19, 2023
c9a2023
update
radekosmulski May 22, 2023
eb21bb4
WIP
radekosmulski Jun 2, 2023
b24c4d6
implement review changes
radekosmulski Jun 6, 2023
37dd36c
update
radekosmulski Jun 6, 2023
ea844c1
fix issue
radekosmulski Jun 6, 2023
726ee1d
update WIP
radekosmulski Jun 9, 2023
abe2f4a
update
radekosmulski Jun 12, 2023
2c27e0a
breaking changes
radekosmulski Jun 13, 2023
4dd89e6
clean up/create reproducer
radekosmulski Jun 13, 2023
07f5af3
fix typo
radekosmulski Jun 13, 2023
00b9797
update reproducer
radekosmulski Jun 13, 2023
c41aa0d
update example
radekosmulski Jun 19, 2023
ba61544
fix typo
radekosmulski Jun 19, 2023
dc451cb
add test
radekosmulski Jun 19, 2023
880c55c
clean up test
radekosmulski Jun 19, 2023
cd2a065
update test
radekosmulski Jun 19, 2023
1b5a6d9
update test
radekosmulski Jun 19, 2023
435a5f2
fix issues
radekosmulski Jun 19, 2023
00c142e
implement review suggestions
radekosmulski Jun 19, 2023
5f4e99b
update test
radekosmulski Jun 19, 2023
c2ee0b2
fix typo
radekosmulski Jun 19, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16,781 changes: 16,781 additions & 0 deletions examples/usecases/transformers-next-item-prediction-with-pretrained-embeddings.ipynb
View file
Open in desktop

Large diffs are not rendered by default.

Empty file added merlin/datasets/ecommerce/sigir/__init__.py
View file
Open in desktop
Empty file.
Empty file added merlin/datasets/ecommerce/sigir/browsing_train/__init__.py
View file
Open in desktop
Empty file.
209 changes: 209 additions & 0 deletions merlin/datasets/ecommerce/sigir/browsing_train/schema.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,209 @@
{
"feature": [
{
"name": "session_id_hash",
"type": "INT",
"intDomain": {
"name": "session_id_hash",
"max": "999",
"isCategorical": true
},
"annotation": {
"tag": [
"item_id",
"item",
"categorical",
"id"
],
"extraMetadata": [
{
"num_buckets": null,
"freq_threshold": 0.0,
"max_size": 1000.0,
"start_index": 0.0,
"cat_path": ".//categories/unique.session_id_hash.parquet",
"embedding_sizes": {
"cardinality": 1000.0,
"dimension": 77.0
},
"_dims": [
[
0.0,
null
]
],
"is_list": false,
"is_ragged": false,
"dtype_item_size": 64.0
}
]
}
},
{
"name": "event_type",
"type": "INT",
"intDomain": {
"name": "event_type",
"max": "2",
"isCategorical": true
},
"annotation": {
"tag": [
"categorical"
],
"extraMetadata": [
{
"num_buckets": null,
"freq_threshold": 0.0,
"max_size": 1000.0,
"start_index": 0.0,
"cat_path": ".//categories/unique.event_type.parquet",
"embedding_sizes": {
"cardinality": 3.0,
"dimension": 16.0
},
"_dims": [
[
0.0,
null
]
],
"is_list": false,
"is_ragged": false,
"dtype_item_size": 64.0
}
]
}
},
{
"name": "product_action",
"type": "INT",
"intDomain": {
"name": "product_action",
"max": "4",
"isCategorical": true
},
"annotation": {
"tag": [
"categorical"
],
"extraMetadata": [
{
"num_buckets": null,
"freq_threshold": 0.0,
"max_size": 1000.0,
"start_index": 0.0,
"cat_path": ".//categories/unique.product_action.parquet",
"embedding_sizes": {
"cardinality": 5.0,
"dimension": 16.0
},
"_dims": [
[
0.0,
null
]
],
"is_list": false,
"is_ragged": false,
"dtype_item_size": 64.0
}
]
}
},
{
"name": "product_sku_hash",
"type": "INT",
"intDomain": {
"name": "product_sku_hash",
"max": "999",
"isCategorical": true
},
"annotation": {
"tag": [
"categorical"
],
"extraMetadata": [
{
"num_buckets": null,
"freq_threshold": 0.0,
"max_size": 1000.0,
"start_index": 0.0,
"cat_path": ".//categories/unique.product_sku_hash.parquet",
"embedding_sizes": {
"cardinality": 1000.0,
"dimension": 77.0
},
"_dims": [
[
0.0,
null
]
],
"is_list": false,
"is_ragged": false,
"dtype_item_size": 64.0
}
]
}
},
{
"name": "hashed_url",
"type": "INT",
"intDomain": {
"name": "hashed_url",
"max": "999",
"isCategorical": true
},
"annotation": {
"tag": [
"categorical"
],
"extraMetadata": [
{
"num_buckets": null,
"freq_threshold": 0.0,
"max_size": 1000.0,
"start_index": 0.0,
"cat_path": ".//categories/unique.hashed_url.parquet",
"embedding_sizes": {
"cardinality": 1000.0,
"dimension": 77.0
},
"_dims": [
[
0.0,
null
]
],
"is_list": false,
"is_ragged": false,
"dtype_item_size": 64.0
}
]
}
},
{
"name": "server_timestamp_epoch_ms",
"type": "FLOAT",
"annotation": {
"tag": [
"continuous"
],
"extraMetadata": [
{
"_dims": [
[
0.0,
null
]
],
"is_list": false,
"is_ragged": false,
"dtype_item_size": 64.0
}
]
}
}
]
}
Empty file added merlin/datasets/ecommerce/sigir/sku_information/__init__.py
View file
Open in desktop
Empty file.
132 changes: 132 additions & 0 deletions merlin/datasets/ecommerce/sigir/sku_information/schema.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,132 @@
{
"feature": [
{
"name": "product_sku_hash",
"type": "INT",
"intDomain": {
"name": "product_sku_hash",
"max": "999",
"isCategorical": true
},
"annotation": {
"tag": [
"id",
"categorical",
"item"
],
"extraMetadata": [
{
"num_buckets": null,
"freq_threshold": 0.0,
"max_size": 1000.0,
"cat_path": ".//categories/unique.product_sku_hash.parquet",
"embedding_sizes": {
"cardinality": 1000.0,
"dimension": 77.0
},
"_dims": [
[
0.0,
null
]
],
"is_list": false,
"is_ragged": false,
"dtype_item_size": 64.0
}
]
}
},
{
"name": "description_vector",
"type": "FLOAT",
"floatDomain": {
"min": -0.44,
"max": 0.603
},
"annotation": {
"tag": [
"item"
],
"extraMetadata": [
{
"_dims": [
[
0.0,
null
],
[
50,
50
]
],
"is_list": true,
"is_ragged": true,
"dtype_item_size": 64.0
}
]
}
},
{
"name": "category_hash",
"type": "INT",
"intDomain": {
"name": "category_hash",
"max": "174",
"isCategorical": true
},
"annotation": {
"tag": [
"item",
"item_id",
"categorical",
"id"
],
"extraMetadata": [
{
"num_buckets": null,
"freq_threshold": 0.0,
"max_size": 1000.0,
"start_index": 0.0,
"cat_path": ".//categories/unique.category_hash.parquet",
"embedding_sizes": {
"cardinality": 175.0,
"dimension": 29.0
},
"_dims": [
[
0.0,
null
]
],
"is_list": false,
"is_ragged": false,
"dtype_item_size": 64.0
}
]
}
},
{
"name": "price_bucket",
"type": "FLOAT",
"annotation": {
"tag": [
"continuous"
],
"extraMetadata": [
{
"_dims": [
[
0.0,
null
]
],
"is_list": false,
"is_ragged": false,
"dtype_item_size": 64.0
}
]
}
}
]
}
2 changes: 2 additions & 0 deletions merlin/datasets/synthetic.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -49,6 +49,8 @@
"booking.com": HERE / "ecommerce/booking/transformed/",
"booking.com-raw": HERE / "ecommerce/booking/raw/",
"transactions": HERE / "ecommerce/transactions",
"sigir-browsing": HERE / "ecommerce/sigir/browsing_train",
"sigir-sku": HERE / "ecommerce/sigir/sku_information",
}


Expand Down
38 changes: 38 additions & 0 deletions .../tf/examples/test_usecase_transformers_next_item_prediction_with_pretrained_embeddings.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,38 @@
import shutil

import pytest
from testbook import testbook

from tests.conftest import REPO_ROOT

pytest.importorskip("transformers")
utils = pytest.importorskip("merlin.systems.triton.utils")

TRITON_SERVER_PATH = shutil.which("tritonserver")


@pytest.mark.skipif(not TRITON_SERVER_PATH, reason="triton server not found")
@testbook(
REPO_ROOT
/ "examples/usecases/transformers-next-item-prediction-with-pretrained-embeddings.ipynb",
timeout=720,
execute=False,
)
@pytest.mark.notebook
def test_next_item_prediction(tb, tmpdir):
tb.inject(
f"""
import os, random
os.environ["OUTPUT_DATA_DIR"] = "{tmpdir}"
os.environ["NUM_EPOCHS"] = "1"
os.environ["NUM_EXAMPLES"] = "1_500"
os.environ["MINIMUM_SESSION_LENGTH"] = "2"
"""
)
tb.execute_cell(list(range(0, 48)))

with utils.run_triton_server(f"{tmpdir}/ensemble", grpc_port=8001):
tb.execute_cell(list(range(48, len(tb.cells))))

predicted_hashed_url_id = tb.ref("predicted_hashed_url_id").item()
assert predicted_hashed_url_id >= 0 and predicted_hashed_url_id <= 1002