From a4d02c2b7044be394607f5cf6dee6f22621c68a7 Mon Sep 17 00:00:00 2001
From: kingbri <8082010+kingbri1@users.noreply.github.com>
Date: Sun, 17 Aug 2025 23:09:27 -0400
Subject: [PATCH] Model: Add log messages for model loading

It's useful to know the split method that the model is being loaded
on.

Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
---
 backends/exllamav3/model.py | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/backends/exllamav3/model.py b/backends/exllamav3/model.py
index 53d2e91..217f5bf 100644
--- a/backends/exllamav3/model.py
+++ b/backends/exllamav3/model.py
@@ -188,6 +188,7 @@ class ExllamaV3Container(BaseModelContainer):
             # Set GPU split options
             # Enable manual GPU split if provided
             if gpu_split:
+                self.gpu_split_auto = False
                 self.gpu_split = gpu_split
 
                 # Causes crash if set with GPU split
@@ -464,6 +465,15 @@ class ExllamaV3Container(BaseModelContainer):
                 if value:
                     yield value
 
+        logger.info("Loading model: " + str(self.model_dir))
+
+        if self.use_tp:
+            logger.info("Loading with tensor parallel")
+        elif self.gpu_split_auto:
+            logger.info("Loading with autosplit")
+        else:
+            logger.info("Loading with a manual GPU split (or a one GPU setup)")
+
         for value in self.model.load_gen(
             tensor_p=self.use_tp,
             tp_backend=self.tp_backend,