Automatic CFG - Attention modifiers tester¶
Documentation¶
- Class name:
Automatic CFG - Attention modifiers tester - Category:
model_patches/Automatic_CFG/experimental_attention_modifiers - Output node:
False
This node is designed to test and apply various attention modifier configurations to computational graph models, specifically focusing on enhancing or altering the attention mechanisms within these models. It aims to experiment with different attention modification strategies to optimize model performance or behavior.
Input types¶
Required¶
seed- Specifies the seed for random number generation, ensuring reproducibility of the attention modifier application.
- Comfy dtype:
INT - Python dtype:
int
sigma_start- Defines the starting value of sigma for the attention modifier, indicating the beginning of the range within which the modifier is active.
- Comfy dtype:
FLOAT - Python dtype:
float
sigma_end- Defines the ending value of sigma for the attention modifier, indicating the end of the range within which the modifier is active.
- Comfy dtype:
FLOAT - Python dtype:
float
self_attn_mod_eval- Contains the evaluation code for self-attention modification, allowing for dynamic adjustment of attention mechanisms based on the model's requirements.
- Comfy dtype:
STRING - Python dtype:
str
unet_block_id_input- Identifies the input block IDs where the attention modifier will be applied, enabling targeted modifications.
- Comfy dtype:
STRING - Python dtype:
str
unet_block_id_middle- Identifies the middle block IDs where the attention modifier will be applied, enabling targeted modifications.
- Comfy dtype:
STRING - Python dtype:
str
unet_block_id_output- Identifies the output block IDs where the attention modifier will be applied, enabling targeted modifications.
- Comfy dtype:
STRING - Python dtype:
str
unet_attn- Specifies the type of attention (self, cross, or both) to be modified, allowing for precise control over the modification process.
- Comfy dtype:
COMBO[STRING] - Python dtype:
list
Optional¶
join_parameters- Allows for the combination of attention modifier parameters from another source, enhancing the flexibility of attention modification.
- Comfy dtype:
ATTNMOD - Python dtype:
dict
Output types¶
Attention modifier- Comfy dtype:
ATTNMOD - Outputs the modified attention parameters, reflecting the changes made to the model's attention mechanisms.
- Python dtype:
tuple
- Comfy dtype:
Parameters as string- Comfy dtype:
STRING - Provides a string representation of the attention modifier parameters, facilitating easy review and documentation of the modifications.
- Python dtype:
str
- Comfy dtype:
Usage tips¶
- Infra type:
CPU - Common nodes: unknown
Source code¶
class attentionModifierBruteforceParametersNode:
@classmethod
def INPUT_TYPES(s):
return {"required": {
"seed": ("INT", {"default": 0, "min": 0, "max": 0xffffffffffffffff}),
"sigma_start": ("FLOAT", {"default": 1000, "min": 0.0, "max": 10000.0, "step": 0.1, "round": 0.01}),
"sigma_end": ("FLOAT", {"default": 0, "min": 0.0, "max": 10000.0, "step": 0.1, "round": 0.01}),
"self_attn_mod_eval": ("STRING", {"multiline": True , "default": ""}),
"unet_block_id_input": ("STRING", {"multiline": False, "default": "4,5,7,8"}),
"unet_block_id_middle": ("STRING", {"multiline": False, "default": "0"}),
"unet_block_id_output": ("STRING", {"multiline": False, "default": "0,1,2,3,4,5"}),
"unet_attn": (["attn1","attn2","both"],),
},
"optional":{
"join_parameters": ("ATTNMOD", {"forceInput": True}),
}}
RETURN_TYPES = ("ATTNMOD","STRING",)
RETURN_NAMES = ("Attention modifier", "Parameters as string")
FUNCTION = "exec"
CATEGORY = "model_patches/Automatic_CFG/experimental_attention_modifiers"
def create_sequence_parameters(self, input_str, middle_str, output_str):
input_values = input_str.split(",") if input_str else []
middle_values = middle_str.split(",") if middle_str else []
output_values = output_str.split(",") if output_str else []
result = []
result.extend([{"unet_block_id_input": val, "unet_block_id_middle": "", "unet_block_id_output": ""} for val in input_values])
result.extend([{"unet_block_id_input": "", "unet_block_id_middle": val, "unet_block_id_output": ""} for val in middle_values])
result.extend([{"unet_block_id_input": "", "unet_block_id_middle": "", "unet_block_id_output": val} for val in output_values])
return result
def exec(self, seed, join_parameters=None, **kwargs):
sequence_parameters = self.create_sequence_parameters(kwargs['unet_block_id_input'],kwargs['unet_block_id_middle'],kwargs['unet_block_id_output'])
lenseq = len(sequence_parameters)
current_index = seed % lenseq
current_sequence = sequence_parameters[current_index]
kwargs["unet_block_id_input"] = current_sequence["unet_block_id_input"]
kwargs["unet_block_id_middle"] = current_sequence["unet_block_id_middle"]
kwargs["unet_block_id_output"] = current_sequence["unet_block_id_output"]
if current_sequence["unet_block_id_input"] != "":
current_block_string = f"unet_block_id_input: {current_sequence['unet_block_id_input']}"
elif current_sequence["unet_block_id_middle"] != "":
current_block_string = f"unet_block_id_middle: {current_sequence['unet_block_id_middle']}"
elif current_sequence["unet_block_id_output"] != "":
current_block_string = f"unet_block_id_output: {current_sequence['unet_block_id_output']}"
info_string = f"Progress: {current_index+1}/{lenseq}\n{kwargs['self_attn_mod_eval']}\n{kwargs['unet_attn']} {current_block_string}"
if kwargs['unet_attn'] == "both":
copy_kwargs = kwargs.copy()
kwargs['unet_attn'] = "attn1"
copy_kwargs['unet_attn'] = "attn2"
out_modifiers = [kwargs, copy_kwargs]
else:
out_modifiers = [kwargs]
return (out_modifiers if join_parameters is None else join_parameters + out_modifiers, info_string, )