clarify some language in the help content

trainer_alltalk/finetune_content.py

@@ -604,19 +604,17 @@ class FinetuneContent:
         * Start with 4-8 workers
 
     #### Shared Memory Option:
-    The shared memory option affects how your system handles memory overflow:
+    Controls GPU memory allocation during training:
 
-    - **Disable Shared Memory Use:**
+    **Disable Shared Memory Use (Enabled/Checked):**
-        * Limits GPU to 95% VRAM usage
+    * Reserves 5% of GPU VRAM, limiting maximum usage to 95%
-        * Prevents system RAM spillover
+    * Can help prevent some out-of-memory crashes
-        * Recommended for high-VRAM GPUs
+    * May be useful on systems experiencing memory issues
-        * Clearer error messages
 
-    - **Keep Enabled for:**
+    * **Disable Shared Memory Use (Default/Unchecked):**
-        * Windows with low VRAM GPUs
+    * Allows PyTorch to use full GPU VRAM
-        * Systems with plenty of RAM
+    * Maximum training performance
-        * When encountering OOM errors
+    * May use more VRAM
-        * Better fallback options
     """
 
     STEP2_BATCH_SIZE = """