Merge branch 'dev' of https://github.com/KevinMusgrave/pytorch-metric…

…-learning into dev
KevinMusgrave · Aug 12, 2022 · a5819d1 · a5819d1
2 parents cb89426 + de6bfa4
commit a5819d1
Show file tree

Hide file tree

Showing 3 changed files with 12 additions and 2 deletions.
diff --git a/conda_build/pytorch-metric-learning/meta.yaml b/conda_build/pytorch-metric-learning/meta.yaml
@@ -1,13 +1,13 @@
 {% set name = "pytorch-metric-learning" %}
-{% set version = "1.5.0" %}
+{% set version = "1.5.2" %}
 
 package:
   name: "{{ name|lower }}"
   version: "{{ version }}"
 
 source:
   url: "https://pypi.io/packages/source/{{ name[0] }}/{{ name }}/{{ name }}-{{ version }}.tar.gz"
-  sha256: c5189be2bd587bb44b0eea286f62623288138c7e791a5b415a39a0f0730d820b
+  sha256: ccf227dfc9858ece1f4dac96e1c5224627fde6efce80008de347415d5836287e
 
 build:
   number: 0

diff --git a/docs/faq.md b/docs/faq.md
@@ -0,0 +1,9 @@
+# Frequently Asked Questions
+
+## Large batch sizes and the `INT_MAX` error
+
+A large batch size results in a huge number of pairs/triplets. When the `loss_and_miner_utils` code processes a huge number of tuples, it can cause a PyTorch error:
+
+`RuntimeError: nonzero is not supported for tensors with more than INT_MAX element`
+
+To fix this error, lower your batch size.
diff --git a/mkdocs.yml b/mkdocs.yml
@@ -18,6 +18,7 @@ nav:
     - How to extend this library:
       - Custom losses: extend/losses.md
       - Custom miners: extend/miners.md
+    - Frequently Asked Questions: faq.md
 theme:
   name: 'material'
   palette: