From f3d96852cef1ad5737b5500606b7f9f2e6b39fa9 Mon Sep 17 00:00:00 2001
From: Robrecht Cannoodt <rcannood@gmail.com>
Date: Tue, 4 Jun 2024 17:56:28 +0200
Subject: [PATCH] apply suggestions from Elior (See
 Eliorkalfon/single_cell_pb#11)

---
 src/task/methods/transformer_ensemble/script.py | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/src/task/methods/transformer_ensemble/script.py b/src/task/methods/transformer_ensemble/script.py
index 79764953..398e504a 100644
--- a/src/task/methods/transformer_ensemble/script.py
+++ b/src/task/methods/transformer_ensemble/script.py
@@ -55,6 +55,7 @@
         "uncommon": False,
         "sampling_strategy": "random",
         "weight": 0.5,
+        "validation_percentage": 0.1,
     },
     # Note by author - weight_df2: 0.25 (excluding uncommon elements, resulting in 0.559)
     {
@@ -62,11 +63,13 @@
         "uncommon": True,
         "sampling_strategy": "random",
         "weight": 0.25,
+        "validation_percentage": 0.1,
+
     },
     # Note by author - weight_df3: 0.25 (leveraging clustering sampling, achieving 0.575)
     {
         "mean_std": "mean_std",
-        "uncommon": False, # should this be set to False or True?
+        "uncommon": False,
         "sampling_strategy": "k-means",
         "weight": 0.25,
     },
@@ -76,6 +79,7 @@
         "uncommon": False,
         "sampling_strategy": "random",
         "weight": 0.3,
+        "validation_percentage": 0.1,
     }
 ]
 
@@ -113,8 +117,8 @@
             num_epochs=par["num_train_epochs"],
             early_stopping=par["early_stopping"],
             batch_size=par["batch_size"],
-            device=device,
             mean_std=argset["mean_std"],
+            device=device,
         )
     elif argset["sampling_strategy"] == "random":
         label_reducer, scaler, transformer_model = train_non_k_means_strategy(
@@ -125,8 +129,9 @@
             num_epochs=par["num_train_epochs"],
             early_stopping=par["early_stopping"],
             batch_size=par["batch_size"],
-            device=device,
             mean_std=argset["mean_std"],
+            validation_percentage=argset["validation_percentage"],
+            device=device,
         )
     else:
         raise ValueError("Invalid sampling_strategy argument")