torch.OutOfMemoryError in KSampler Node During Sampling with WanTEModel on GTX 1050 Ti

Encountered a torch.OutOfMemoryError while executing the KSampler node (ID: 3) in ComfyUI v0.3.38 with PyTorch 2.7.0+cu128 on a GTX 1050 Ti (4GB VRAM). The error occurs during the sampling process with the WanTEModel, specifically in the patch_embedding step. Logs indicate multiple OOM warnings, including failed VAE encoding attempts, with fallback to tiled VAE encoding. VRAM usage shows 3.48GB free out of 4GB total, suggesting potential memory management issues or model size incompatibility. The issue persists across multiple prompt executions. Please advise on optimizations or fixes for low-VRAM setups. Workflow file is too large to include but can be uploaded manually.