• Search by Module
• Search by Words
• Search Projects

• Java
• Python
• JavaScript
• TypeScript
• C++
• Scala
• Blog

• allRank-master
  • Makefile
  • LICENSE
  • setup.py
  • setup.cfg
  • README.md
  • tests
    • click_models
      • test_random_click_model.py
      • test_base_cascade_model.py
      • test_masked_click_model.py
      • test_diverse_clicks_model.py
      • test_fixed_click_model.py
      • test_duplicate_click_model.py
      • __init__.py
      • test_alternative_click_models.py
      • test_feature_click_model.py
      • test_apply_click_model.py
    • losses
      • test_lambdaloss.py
      • test_mrr.py
      • test_loss_ordinal.py
      • test_loss_pointwise.py
      • test_approxndcg.py
      • test_binary_listnet.py
      • test_listmle.py
      • test_ndcg.py
      • __init__.py
      • test_listnet.py
      • test_ranknet.py
    • __init__.py
    • test_rank_slates.py
  • scripts
    • run_tests.sh
    • ci.sh
    • local_config.json
    • run_in_docker_click.sh
    • local_config_click_model.json
    • run_in_docker.sh
    • run_example.sh
  • requirements.txt
  • Dockerfile
  • .gitignore
  • allrank
    • utils
      • tensorboard_utils.py
      • ltr_logging.py
      • config_utils.py
      • experiments.py
      • file_utils.py
      • command_executor.py
      • python_utils.py
      • __init__.py
      • args_utils.py
    • training
      • early_stop.py
      • train_utils.py
      • __init__.py
    • config.py
    • click_models
      • click_utils.py
      • duplicate_aware.py
      • __init__.py
      • base.py
      • cascade_models.py
    • models
      • transformer.py
      • metrics.py
      • losses
        • pointwise.py
        • rankNet.py
        • listMLE.py
        • binary_listNet.py
        • bce.py
        • __init__.py
        • lambdaLoss.py
        • ordinal.py
        • approxNDCG.py
        • listNet.py
      • model.py
      • __init__.py
      • positional.py
      • model_utils.py
    • rank_and_click.py
    • data
      • dataset_loading.py
      • generate_dummy_data.py
      • __init__.py
      • dataset_saving.py
    • __init__.py
    • main.py
    • config_template.json
    • inference
      • inference_utils.py
      • __init__.py

import math

import numpy as np
import torch
from pytest import approx
from scipy.special import softmax

from allrank.data.dataset_loading import PADDED_Y_VALUE
from allrank.models.losses import DEFAULT_EPS, listNet


def listNet_wrap(y_pred, y_true, eps=1e-10):
    return listNet(torch.tensor([y_pred]), torch.tensor([y_true]), eps).item()


def test_listnet_simple():
    y_pred = [0.5, 0.2]
    y_true = [1.0, 0.0]

    result = listNet_wrap(y_pred, y_true, eps=0.0)
    expected = - np.sum(softmax(y_true) * np.log(softmax(y_pred)))

    assert not math.isnan(result) and not math.isinf(result)
    assert (result == approx(expected))


def test_listnet_stable_for_very_small_prediction():
    y_pred = [0.5, -1e30]
    y_true = [1.0, 0.0]

    result = listNet_wrap(y_pred, y_true)
    expected = - np.sum(softmax(y_true) * np.log(softmax(y_pred) + DEFAULT_EPS))

    assert not math.isnan(result) and not math.isinf(result)
    assert (result == approx(expected))


def test_listnet_ignores_padded_value():
    y_pred = [0.5, 0.2, 0.5]
    y_true = [1.0, 0.0, PADDED_Y_VALUE]

    result = listNet_wrap(y_pred, y_true)
    expected = - np.sum(softmax(y_true[:2]) * np.log(softmax(y_pred[:2]) + DEFAULT_EPS))

    assert not math.isnan(result) and not math.isinf(result)
    assert (result == approx(expected))