Memory Exhaustion and Crash Issue when Running ComfyUI on Google Colab with SDXL Workflow

I'm encountering a persistent issue while attempting to run ComfyUI on Google Colab using the suggested SDXL workflow (utilizing the image with the bottle). The problem lies in the memory management during image generation, which consistently leads to the exhaustion of available RAM.

Here's a breakdown of the issue I'm facing:

1. **Initial Generation**: Upon initiating the image generation process, RAM usage steadily climbs and peaks at around 8GB during the first generation.

2. **Second Generation**: With each subsequent generation, the RAM usage continues to increase. After the second generation, RAM usage reaches approximately 19GB, and it remains at this level throughout the generation process.

3. **Third Generation and Crash**: As the image generation progresses to the third iteration, RAM consumption reaches its peak at around 24GB. Unfortunately, this causes the process to crash, necessitating the use of the ^C command to terminate it.

I've tested both the normal RAM and high RAM modes, and the issue persists in both scenarios.

I'm seeking assistance in resolving this memory management problem, as it severely limits my ability to effectively utilize ComfyUI for image generation in this workflow. Any insights or guidance on how to address this issue would be greatly appreciated. Thank you for your attention and assistance in advance.