Upload sdxllorapps.py

Scripts/sdxllorapps.py

@@ -67,6 +67,7 @@ def Deps(force_reinstall):
         os.environ['TORCH_HOME'] = '/notebooks/cache/torch'
         os.environ['PYTHONWARNINGS'] = 'ignore'
         call("sed -i 's@text = _formatwarnmsg(msg)@text =\"\"@g' /usr/lib/python3.9/warnings.py", shell=True)
+        #call('pip install --root-user-action=ignore --disable-pip-version-check -qq gradio==3.39.0', shell=True, stdout=open('/dev/null', 'w'))
         if not os.path.exists('/notebooks/diffusers'):
             call('ln -s /diffusers /notebooks', shell=True)
         call("rm -r /deps", shell=True)
@@ -614,7 +615,8 @@ def caption(CAPTIONS_DIR, INSTANCE_DIR):
 
 
 
-def dbtrainxl(Resume_Training, UNet_Training_Epochs, UNet_Learning_Rate, dim, Offset_Noise, Resolution, MODEL_NAME, SESSION_DIR, INSTANCE_DIR, CAPTIONS_DIR, External_Captions,  INSTANCE_NAME, Session_Name, OUTPUT_DIR, ofstnselvl, Save_VRAM):
+def dbtrainxl(Unet_Training_Epochs, Text_Encoder_Training_Epochs, Unet_Learning_Rate, Text_Encoder_Learning_Rate, dim, Offset_Noise, Resolution, MODEL_NAME, SESSION_DIR, INSTANCE_DIR, CAPTIONS_DIR, External_Captions,  INSTANCE_NAME, Session_Name, OUTPUT_DIR, ofstnselvl, Save_VRAM):
+
 
     if os.path.exists(INSTANCE_DIR+"/.ipynb_checkpoints"):
       call('rm -r '+INSTANCE_DIR+'/.ipynb_checkpoints', shell=True)
@@ -622,10 +624,6 @@ def dbtrainxl(Resume_Training, UNet_Training_Epochs, UNet_Learning_Rate, dim, Of
       call('rm -r '+CAPTIONS_DIR+'/.ipynb_checkpoints', shell=True)
 
 
-    while not Resume_Training and not os.path.exists(MODEL_NAME+'/unet/diffusion_pytorch_model.safetensors'):
-        print('No model found, use the "Model Download" cell to download a model.')
-        time.sleep(5) 
-
     Seed=random.randint(1, 999999)
     
     ofstnse=""
@@ -642,22 +640,37 @@ def dbtrainxl(Resume_Training, UNet_Training_Epochs, UNet_Learning_Rate, dim, Of
 
     precision="fp16"
 
-    resume=""
-    if Resume_Training and os.path.exists(SESSION_DIR+'/'+Session_Name+'.safetensors'):
-        resume="--resume"
-        
-        print('Resuming Training...')
-    elif Resume_Training and not os.path.exists(SESSION_DIR+'/'+Session_Name+'.safetensors'):
-        while MODEL_NAME=="":
-            print('No model found, use the "Model Download" cell to download a model.')
-            time.sleep(5)
-        print('Previous model not found, training a new model...')
-
+   
+   
+    def train_only_text(SESSION_DIR, MODEL_NAME, INSTANCE_DIR, OUTPUT_DIR, Seed, Resolution, ofstnse, extrnlcptn, precision, Training_Epochs):
+        print('Training the Text Encoder...')
+        call('accelerate launch /notebooks/diffusers/examples/dreambooth/train_dreambooth_sdxl_TI.py \
+        '+ofstnse+' \
+        '+extrnlcptn+' \
+        --dim='+str(dim)+' \
+        --ofstnselvl='+str(ofstnselvl)+' \
+        --image_captions_filename \
+        --Session_dir='+SESSION_DIR+' \
+        --pretrained_model_name_or_path='+MODEL_NAME+' \
+        --instance_data_dir='+INSTANCE_DIR+' \
+        --output_dir='+OUTPUT_DIR+' \
+        --captions_dir='+CAPTIONS_DIR+' \
+        --seed='+str(Seed)+' \
+        --resolution='+str(Resolution)+' \
+        --mixed_precision='+str(precision)+' \
+        --train_batch_size=1 \
+        --gradient_accumulation_steps=1 '+GC+ ' \
+        --use_8bit_adam \
+        --learning_rate='+str(Text_Encoder_Learning_Rate)+' \
+        --lr_scheduler="cosine" \
+        --lr_warmup_steps=0 \
+        --num_train_epochs='+str(Training_Epochs), shell=True)
+   
    
 
     def train_only_unet(SESSION_DIR, MODEL_NAME, INSTANCE_DIR, OUTPUT_DIR, Seed, Resolution, ofstnse, extrnlcptn, precision, Training_Epochs):
+        print('Training the UNet...')
         call('accelerate launch /notebooks/diffusers/examples/dreambooth/train_dreambooth_sdxl_lora.py \
-        '+resume+' \
         '+ofstnse+' \
         '+extrnlcptn+' \
         --dim='+str(dim)+' \
@@ -674,18 +687,22 @@ def dbtrainxl(Resume_Training, UNet_Training_Epochs, UNet_Learning_Rate, dim, Of
         --train_batch_size=1 \
         --gradient_accumulation_steps=1 '+GC+ ' \
         --use_8bit_adam \
-        --learning_rate='+str(UNet_Learning_Rate)+' \
+        --learning_rate='+str(Unet_Learning_Rate)+' \
         --lr_scheduler="cosine" \
         --lr_warmup_steps=0 \
         --num_train_epochs='+str(Training_Epochs), shell=True)
 
       
 
-    if UNet_Training_Epochs!=0:
-      train_only_unet(SESSION_DIR, MODEL_NAME, INSTANCE_DIR, OUTPUT_DIR, Seed, Resolution, ofstnse, extrnlcptn, precision, Training_Epochs=UNet_Training_Epochs)
+    if Unet_Training_Epochs!=0:
+        if Text_Encoder_Training_Epochs!=0:
+            train_only_text(SESSION_DIR, MODEL_NAME, INSTANCE_DIR, OUTPUT_DIR, Seed, Resolution, ofstnse, extrnlcptn, precision, Training_Epochs=Text_Encoder_Training_Epochs)
+            clear_output()
+        train_only_unet(SESSION_DIR, MODEL_NAME, INSTANCE_DIR, OUTPUT_DIR, Seed, Resolution, ofstnse, extrnlcptn, precision, Training_Epochs=Unet_Training_Epochs)
     else  :
       print('Nothing to do')
 
+
     if os.path.exists(SESSION_DIR+'/'+Session_Name+'.safetensors'):
         clear_output()
         print("DONE, the LoRa model is in the session's folder")