Update README.md

README.md

@@ -59,8 +59,8 @@ The captions associated with the training data leveraged the language and taggin
 
 This model is intended for generating short video clips (typically a few seconds) from descriptive text prompts.
 
-1.  **Select an Epoch Checkpoint:** Experiment with different `wan_1.3B_e{i}.safetensors` files. While later epochs generally offer more refined results for common themes, `wan_1.3B_e10.safetensors` is specifically recommended if you plan to use the motion helper LoRA.
-2.  **(Highly Recommended) Apply the Motion Helper LoRA:** For significantly improved video quality, motion, and style, apply the `NSFW_Wan_1.3b_motion_helper` LoRA. This LoRA was trained using the epoch 10 checkpoint.
+1.  **Select an Epoch Checkpoint:** Experiment with different `wan_1.3B_e{i}.safetensors` files. While later epochs generally offer more refined results for common themes, `wan_1.3B_e10.safetensors` is specifically recommended if you plan to use the motion helper LoRA. From epoch 11 onward, the model began training on a video dataset.
+2.  **(Highly Recommended) Apply the Motion Helper LoRA (only for epoch 10 and lower):** For significantly improved video quality, motion, and style, apply the `NSFW_Wan_1.3b_motion_helper` LoRA. This LoRA was trained using the epoch 10 checkpoint.
     *   Download it from: [NSFW_Wan_1.3b_motion_helper on Hugging Face](https://huggingface.co/NSFW-API/NSFW_Wan_1.3b_motion_helper)
     *   Ensure it is correctly loaded and applied in your generation pipeline along with the main model checkpoint.
 3.  **Craft Your Prompt:** Utilize natural language to describe the desired scene, subjects, actions, and style.