diff --git a/README.md b/README.md
index 20e9ba52..8f51a72b 100644
--- a/README.md
+++ b/README.md
@@ -17,6 +17,7 @@ https://invoke-ai.github.io/invoke-training/
   - DreamBooth LoRA
   - Textual Inversion
 - Stable Diffusion XL
+  - Full finetuning
   - LoRA
   - DreamBooth LoRA
   - Textual Inversion
diff --git a/docs/tutorials/stable_diffusion/robocats_finetune_sdxl.md b/docs/tutorials/stable_diffusion/robocats_finetune_sdxl.md
index de529262..fa8287be 100644
--- a/docs/tutorials/stable_diffusion/robocats_finetune_sdxl.md
+++ b/docs/tutorials/stable_diffusion/robocats_finetune_sdxl.md
@@ -124,7 +124,7 @@ python src/invoke_training/scripts/_experimental/lora_extraction/extract_lora_fr
   --lora-rank 32
 ```
 
-## 5 - Import into InvokeAI
+## 6 - Import into InvokeAI
 
 If you haven't already, setup [InvokeAI](https://github.com/invoke-ai/InvokeAI) by following its documentation.
 
@@ -132,7 +132,7 @@ Import your finetuned diffusers model or your extracted LoRA from the 'Models' t
 
 Congratulations, you can now use your new robocat model! 🎉
 
-## 6 - Comparison: Finetune vs. LoRA Extraction
+## 7 - Comparison: Finetune vs. LoRA Extraction
 
 As noted earlier, the LoRA extraction process is lossy for a number of reasons. (There are some significant improvements coming to the LoRA extraction process soon to improve the fidelity of the extracted LoRA model.)
 
diff --git a/src/invoke_training/pipelines/stable_diffusion_xl/finetune/train.py b/src/invoke_training/pipelines/stable_diffusion_xl/finetune/train.py
index b6e4e04d..7370f1b2 100644
--- a/src/invoke_training/pipelines/stable_diffusion_xl/finetune/train.py
+++ b/src/invoke_training/pipelines/stable_diffusion_xl/finetune/train.py
@@ -131,6 +131,7 @@ def train(config: SdxlFinetuneConfig, callbacks: list[PipelineCallbacks] | None
 
     logger.info("Loading models.")
     tokenizer_1, tokenizer_2, noise_scheduler, text_encoder_1, text_encoder_2, vae, unet = load_models_sdxl(
+        logger=logger,
         model_name_or_path=config.model,
         hf_variant=config.hf_variant,
         vae_model=config.vae_model,