From 8a5610a1e46751c333959137694b1806ac6d4ae7 Mon Sep 17 00:00:00 2001 From: Debian Date: Tue, 6 Jan 2026 08:35:24 +0000 Subject: [PATCH] Remove torch.compile from model loaders entirely MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Empty backend string still triggered inductor. Removing compile_args connection from WanVideoModelLoader nodes (131, 132) ensures no torch.compile is applied to the transformer blocks. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 --- workflows/Wan22-I2V-Remix-API.json | 8 -------- 1 file changed, 8 deletions(-) diff --git a/workflows/Wan22-I2V-Remix-API.json b/workflows/Wan22-I2V-Remix-API.json index 481670b..8661377 100644 --- a/workflows/Wan22-I2V-Remix-API.json +++ b/workflows/Wan22-I2V-Remix-API.json @@ -85,10 +85,6 @@ "load_device": "offload_device", "attention_mode": "sageattn", "rms_norm_function": "default", - "compile_args": [ - "127", - 0 - ], "block_swap_args": [ "128", 0 @@ -107,10 +103,6 @@ "load_device": "offload_device", "attention_mode": "sageattn", "rms_norm_function": "default", - "compile_args": [ - "127", - 0 - ], "block_swap_args": [ "128", 0