feat: raise TextGenerate max_length cap from 2048 to 16384

comfy/ldm/modules/attention.py

@@ -741,12 +741,12 @@ optimized_attention = attention_basic
 if model_management.sage_attention_enabled():
     logging.info("Using sage attention")
     optimized_attention = attention_sage
-elif model_management.flash_attention_enabled():
-    logging.info("Using Flash Attention")
-    optimized_attention = attention_flash
 elif model_management.xformers_enabled():
     logging.info("Using xformers attention")
     optimized_attention = attention_xformers
+elif model_management.flash_attention_enabled():
+    logging.info("Using Flash Attention")
+    optimized_attention = attention_flash
 elif model_management.pytorch_attention_enabled():
     logging.info("Using pytorch attention")
     optimized_attention = attention_pytorch

comfy_extras/nodes_textgen.py

@@ -35,7 +35,7 @@ class TextGenerate(io.ComfyNode):
                 io.Image.Input("image", optional=True),
                 io.Image.Input("video", optional=True, tooltip="Video frames as image batch. Assumed to be 24 FPS; subsampled to 1 FPS internally."),
                 io.Audio.Input("audio", optional=True),
-                io.Int.Input("max_length", default=256, min=1, max=2048),
+                io.Int.Input("max_length", default=256, min=1, max=16384),
                 io.DynamicCombo.Input("sampling_mode", options=sampling_options, display_name="Sampling Mode"),
                 io.Boolean.Input("thinking", optional=True, default=False, tooltip="Operate in thinking mode if the model supports it."),
                 io.Boolean.Input("use_default_template", optional=True, default=True, tooltip="Use the built in system prompt/template if the model has one.", advanced=True),

requirements.txt

@@ -1,4 +1,4 @@
-comfyui-frontend-package==1.44.19
+comfyui-frontend-package==1.43.18
 comfyui-workflow-templates==0.9.82
 comfyui-embedded-docs==0.5.0
 torch