Merge pull request allegro#18 from allegro/dependency_update

Dependency update + two minor fixes

allrank/click_models/click_utils.py

@@ -1,4 +1,4 @@
-from typing import Tuple, List
+from typing import List, Tuple, Union
 
 import numpy as np
 import torch
@@ -7,8 +7,8 @@
 from allrank.data.dataset_loading import PADDED_Y_VALUE
 
 
-def click_on_slates(slates: Tuple[torch.Tensor, torch.Tensor], click_model: ClickModel, include_empty: bool) \
-        -> Tuple[List[torch.Tensor], List[List[int]]]:
+def click_on_slates(slates: Union[Tuple[np.ndarray, np.ndarray], Tuple[torch.Tensor, torch.Tensor]],
+                    click_model: ClickModel, include_empty: bool) -> Tuple[List[Union[np.ndarray, torch.Tensor]], List[List[int]]]:
     """
     This metod runs a click model on a list of slates and returns new slates with `y` taken from clicks
 
@@ -41,7 +41,7 @@ def __init__(self, inner_click_model: ClickModel):
         """
         self.inner_click_model = inner_click_model
 
-    def click(self, documents: Tuple[torch.Tensor, torch.Tensor]):
+    def click(self, documents: Union[Tuple[np.ndarray, np.ndarray], Tuple[torch.Tensor, torch.Tensor]]) -> np.ndarray:
         X, y = documents
         padded_values_mask = y == PADDED_Y_VALUE
         real_X = X[~padded_values_mask]

allrank/config.py

@@ -106,12 +106,12 @@ def _parse_metrics(metrics):
         metrics_dict = defaultdict(list)  # type: Dict[str, list]
         for metric_string in metrics:
             try:
-                name, at = metric_string.split("@")
+                name, at = metric_string.split("_")
                 metrics_dict[name].append(int(at))
             except (ValueError, TypeError):
                 raise MetricConfigError(
                     metric_string,
-                    "Wrong formatting of metric in config. Expected format: <name>@<at> where name is valid metric name and at is and int")
+                    "Wrong formatting of metric in config. Expected format: <name>_<at> where name is valid metric name and at is and int")
         return metrics_dict
 
 

allrank/config_template.json

@@ -46,7 +46,7 @@
     "early_stopping_patience": <int, number of epochs for early stopping>
     "gradient_clipping_norm": <Optional[float]
   },
-  "metrics": <list of metrics for evaluation of the form name@k where name is the name of the metric
+  "metrics": <list of metrics for evaluation of the form name_k where name is the name of the metric
               as defined in models/metrics.py and k is the rank used for evaluation>,
 "loss":
   {

allrank/training/train_utils.py

@@ -50,7 +50,7 @@ def compute_metrics(metrics, model, dl, dev):
         metric_func = getattr(metrics_module, metric_name)
         metric_func_with_ats = partial(metric_func, ats=ats)
         metrics_values = metric_on_epoch(metric_func_with_ats, model, dl, dev)
-        metrics_names = ["{metric_name}@{at}".format(metric_name=metric_name, at=at) for at in ats]
+        metrics_names = ["{metric_name}_{at}".format(metric_name=metric_name, at=at) for at in ats]
         metric_values_dict.update(dict(zip(metrics_names, metrics_values)))
 
     return metric_values_dict

scripts/local_config.json

@@ -45,9 +45,9 @@
     "early_stopping_patience": 100,
     "gradient_clipping_norm": null
   },
-  "val_metric": "ndcg@5",
+  "val_metric": "ndcg_5",
   "metrics": [
-    "ndcg@5"
+    "ndcg_5"
   ],
   "loss": {
     "name": "ordinal",
@@ -57,7 +57,7 @@
   },
   "expected_metrics" : {
     "val": {
-      "ndcg@5": 0.76
+      "ndcg_5": 0.76
     }
   }
 }

scripts/local_config_click_model.json

@@ -45,9 +45,9 @@
     "early_stopping_patience": 100,
     "gradient_clipping_norm": null
   },
-  "val_metric": "ndcg@5",
+  "val_metric": "ndcg_5",
   "metrics": [
-    "ndcg@5"
+    "ndcg_5"
   ],
   "loss": {
     "name": "ordinal",
@@ -57,7 +57,7 @@
   },
   "expected_metrics" : {
     "val": {
-      "ndcg@5": 0.76
+      "ndcg_5": 0.76
     }
   },
   "click_model": {

setup.py

@@ -7,21 +7,21 @@
 README = (HERE / "README.md").read_text()
 
 reqs = [
-    "torch==1.4.0",
-    "torchvision==0.5.0",
-    "scikit-learn==0.22.1",
-    "pandas==0.25.3",
-    "numpy==1.18.1",
-    "scipy==1.4.1",
-    "attrs==19.3.0",
-    "flatten_dict==0.2.0",
-    "tensorboardX==2.0.0",
-    "gcsfs>=0.6.0"
+    "torch==1.6.0",
+    "torchvision==0.7.0",
+    "scikit-learn>=0.23.0",
+    "pandas>=1.0.5",
+    "numpy>=1.18.5",
+    "scipy>=1.4.1",
+    "attrs>=19.3.0",
+    "flatten_dict>=0.3.0",
+    "tensorboardX>=2.1.0",
+    "gcsfs==0.6.2"
 ]
 
 setup(
     name="allRank",
-    version="1.3.0",
+    version="1.3.1",
     description="allRank is a framework for training learning-to-rank neural models",
     long_description=README,
     long_description_content_type="text/markdown",