From 2c783052ca470dca053f40096e6fd4cfebbb519b Mon Sep 17 00:00:00 2001
From: Prajwal Bende <34344073+MrPrajwalB@users.noreply.github.com>
Date: Thu, 10 Oct 2024 01:38:21 +0530
Subject: [PATCH 1/2] Update functional.py

Fixed Binary focal loss when reduced threshold is not None. Focal term should be continuous and equal to 1 at pt equal to reduced threshold
---
 pytorch_toolbelt/losses/functional.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pytorch_toolbelt/losses/functional.py b/pytorch_toolbelt/losses/functional.py
index 1ef62f055..34ccdd55e 100644
--- a/pytorch_toolbelt/losses/functional.py
+++ b/pytorch_toolbelt/losses/functional.py
@@ -70,7 +70,7 @@ def focal_loss_with_logits(
     if reduced_threshold is None:
         focal_term = (1.0 - pt).pow(gamma)
     else:
-        focal_term = ((1.0 - pt) / reduced_threshold).pow(gamma)
+        focal_term = ((1.0 - pt) / (1 - reduced_threshold)).pow(gamma) #the focal term continuity breaks when reduced_threshold < 0.5. At pt == reduced_threshold, focal term should be 1 from both sides.
         focal_term = torch.masked_fill(focal_term, pt < reduced_threshold, 1)
 
     loss = focal_term * ce_loss

From 654e1fd0153922733374d3c86159944522472787 Mon Sep 17 00:00:00 2001
From: Prajwal Bende <34344073+MrPrajwalB@users.noreply.github.com>
Date: Thu, 10 Oct 2024 01:41:36 +0530
Subject: [PATCH 2/2] Update functional.py

---
 pytorch_toolbelt/losses/functional.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pytorch_toolbelt/losses/functional.py b/pytorch_toolbelt/losses/functional.py
index 34ccdd55e..daae4a052 100644
--- a/pytorch_toolbelt/losses/functional.py
+++ b/pytorch_toolbelt/losses/functional.py
@@ -70,7 +70,7 @@ def focal_loss_with_logits(
     if reduced_threshold is None:
         focal_term = (1.0 - pt).pow(gamma)
     else:
-        focal_term = ((1.0 - pt) / (1 - reduced_threshold)).pow(gamma) #the focal term continuity breaks when reduced_threshold < 0.5. At pt == reduced_threshold, focal term should be 1 from both sides.
+        focal_term = ((1.0 - pt) / (1 - reduced_threshold)).pow(gamma) #the focal term continuity breaks when reduced_threshold not equal to 0.5. At pt equal to reduced_threshold, the value of piecewise function of focal term should be 1 from both sides .
         focal_term = torch.masked_fill(focal_term, pt < reduced_threshold, 1)
 
     loss = focal_term * ce_loss