From b038659889e8907d856e6c647575bd78101a59f1 Mon Sep 17 00:00:00 2001
From: Clybius <clybius@gmail.com>
Date: Mon, 16 Jun 2025 16:06:40 -0500
Subject: [PATCH] chore: Add github workflow & Remove comment & Change defaults

---
 .github/workflows/publish.yml | 22 ++++++++++++++++++++++
 chroma_NAG.py                 |  8 ++++----
 2 files changed, 26 insertions(+), 4 deletions(-)
 create mode 100644 .github/workflows/publish.yml

diff --git a/.github/workflows/publish.yml b/.github/workflows/publish.yml
new file mode 100644
index 0000000..a4fd892
--- /dev/null
+++ b/.github/workflows/publish.yml
@@ -0,0 +1,22 @@
+name: Publish to Comfy registry
+on:
+  workflow_dispatch:
+  push:
+    branches:
+      - main
+      - master
+    paths:
+      - "pyproject.toml"
+
+jobs:
+  publish-node:
+    name: Publish Custom Node to registry
+    runs-on: ubuntu-latest
+    steps:
+      - name: Check out code
+        uses: actions/checkout@v4
+      - name: Publish Custom Node
+        uses: Comfy-Org/publish-node-action@main
+        with:
+          ## Add your own personal access token to your Github Repository secrets and reference it here.
+          personal_access_token: ${{ secrets.REGISTRY_ACCESS_TOKEN }}
diff --git a/chroma_NAG.py b/chroma_NAG.py
index d541e7c..4e7b52e 100644
--- a/chroma_NAG.py
+++ b/chroma_NAG.py
@@ -186,8 +186,8 @@ class ChromaNAG:
         return {
             "required": {
                 "model": ("MODEL",),
-                "conditioning": ("CONDITIONING",), # Positive conditioning
-                "nag_scale": ("FLOAT", {"default": 11.0, "min": 0.0, "max": 100.0, "step": 0.01}),
+                "conditioning": ("CONDITIONING",),
+                "nag_scale": ("FLOAT", {"default": 5.0, "min": -100.0, "max": 100.0, "step": 0.01}),
                 "nag_alpha": ("FLOAT", {"default": 0.25, "min": 0.0, "max": 1.0, "step": 0.01}),
                 "nag_tau": ("FLOAT", {"default": 2.5, "min": 0.0, "max": 10.0, "step": 0.01}),
            }
@@ -205,7 +205,7 @@ class ChromaNAG:
         device = mm.get_torch_device()
         #dtype = mm.unet_dtype()
         
-        # The NAG context is derived from the positive prompt's embeddings.
+        # The NAG context is derived from the negative prompt's embeddings.
         # For FLUX/Chroma, the conditioning input is already embedded.
         # Shape: [1, sequence_length, embedding_dim]
         nag_context = conditioning[0][0].clone()
@@ -213,7 +213,7 @@ class ChromaNAG:
         model_clone = model.clone()
         diffusion_model = model_clone.get_model_object("diffusion_model")
         diffusion_model.txt_in.to(device)
-        txt = diffusion_model.txt_in(nag_context.to(device))
+        txt = diffusion_model.txt_in(nag_context.to(device, diffusion_model.txt_in.dtype))
 
         # Chroma models have `double_blocks` where image and text tokens interact.
         # This is the equivalent of a cross-attention stage in other models.