From 3fa1a899aea3ff2700a20a8bc2da17202d3065e5 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 13 May 2025 12:07:59 -0700
Subject: [PATCH] UI: Fix gpu-layers being ignored (closes #6973)

---
 modules/loaders.py         | 2 +-
 modules/models_settings.py | 2 +-
 modules/ui.py              | 2 +-
 modules/ui_model_menu.py   | 2 +-
 4 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/modules/loaders.py b/modules/loaders.py
index 4b76549b..583b65c2 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -5,7 +5,7 @@ import gradio as gr
 
 loaders_and_params = OrderedDict({
     'llama.cpp': [
-        'n_gpu_layers',
+        'gpu_layers',
         'threads',
         'threads_batch',
         'batch_size',
diff --git a/modules/models_settings.py b/modules/models_settings.py
index ae589bb3..4418e3fb 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -67,7 +67,7 @@ def get_model_metadata(model):
             elif k.endswith('rope.scaling.factor'):
                 model_settings['compress_pos_emb'] = metadata[k]
             elif k.endswith('block_count'):
-                model_settings['n_gpu_layers'] = metadata[k] + 1
+                model_settings['gpu_layers'] = metadata[k] + 1
 
         if 'tokenizer.chat_template' in metadata:
             template = metadata['tokenizer.chat_template']
diff --git a/modules/ui.py b/modules/ui.py
index b3d4bccf..eeb6ce92 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -105,7 +105,7 @@ def list_model_elements():
         'filter_by_loader',
         'loader',
         'cpu_memory',
-        'n_gpu_layers',
+        'gpu_layers',
         'threads',
         'threads_batch',
         'batch_size',
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 1e27255b..b63a127c 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -50,7 +50,7 @@ def create_ui():
                     gr.Markdown("## Main options")
                     with gr.Row():
                         with gr.Column():
-                            shared.gradio['n_gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=256, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
+                            shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=256, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')