137 lines
6.1 KiB
Python
137 lines
6.1 KiB
Python
import torch
|
|
import torch.nn as nn
|
|
from .utils import log
|
|
|
|
#based on ComfyUI's and MinusZoneAI's fp8_linear optimization
|
|
def fp8_linear_forward(cls, base_dtype, input):
|
|
weight_dtype = cls.weight.dtype
|
|
if weight_dtype in [torch.float8_e4m3fn, torch.float8_e5m2]:
|
|
if len(input.shape) == 3:
|
|
input_shape = input.shape
|
|
|
|
scale_weight = getattr(cls, 'scale_weight', None)
|
|
if scale_weight is None:
|
|
scale_weight = torch.ones((), device=input.device, dtype=torch.float32)
|
|
else:
|
|
scale_weight = scale_weight.to(input.device)
|
|
|
|
scale_input = torch.ones((), device=input.device, dtype=torch.float32)
|
|
input = torch.clamp(input, min=-448, max=448, out=input)
|
|
inn = input.reshape(-1, input_shape[2]).to(torch.float8_e4m3fn).contiguous() #always e4m3fn because e5m2 * e5m2 is not supported
|
|
|
|
bias = cls.bias.to(base_dtype) if cls.bias is not None else None
|
|
|
|
o = torch._scaled_mm(inn, cls.weight.t(), out_dtype=base_dtype, bias=bias, scale_a=scale_input, scale_b=scale_weight)
|
|
|
|
return o.reshape((-1, input_shape[1], cls.weight.shape[0]))
|
|
else:
|
|
return cls.original_forward(input.to(base_dtype))
|
|
else:
|
|
return cls.original_forward(input)
|
|
|
|
|
|
@torch.compiler.disable()
|
|
def apply_lora(weight, lora, step=None):
|
|
for lora_diff, lora_strength in zip(lora[0], lora[1]):
|
|
if isinstance(lora_strength, list):
|
|
lora_strength = lora_strength[step]
|
|
if lora_strength == 0.0:
|
|
continue
|
|
elif lora_strength == 0.0:
|
|
continue
|
|
patch_diff = torch.mm(
|
|
lora_diff[0].flatten(start_dim=1).to(weight.device),
|
|
lora_diff[1].flatten(start_dim=1).to(weight.device)
|
|
).reshape(weight.shape)
|
|
alpha = lora_diff[2] / lora_diff[1].shape[0] if lora_diff[2] is not None else 1.0
|
|
scale = lora_strength * alpha
|
|
weight = weight.add(patch_diff, alpha=scale)
|
|
return weight
|
|
|
|
|
|
def linear_with_lora_and_scale_forward(cls, input):
|
|
# Handles both scaled and unscaled, with or without LoRA
|
|
has_scale = hasattr(cls, "scale_weight")
|
|
weight = cls.weight.to(input.dtype)
|
|
bias = cls.bias.to(input.dtype) if cls.bias is not None else None
|
|
|
|
if has_scale:
|
|
scale_weight = cls.scale_weight.to(input.device)
|
|
if weight.numel() < input.numel():
|
|
weight = weight * scale_weight
|
|
else:
|
|
input = input * scale_weight
|
|
|
|
lora = getattr(cls, "lora", None)
|
|
if lora is not None:
|
|
weight = apply_lora(weight, lora, cls.step).to(input.dtype)
|
|
|
|
return torch.nn.functional.linear(input, weight, bias)
|
|
|
|
def convert_fp8_linear(module, base_dtype, params_to_keep={}, scale_weight_keys=None):
|
|
log.info("FP8 matmul enabled")
|
|
for name, submodule in module.named_modules():
|
|
if not any(keyword in name for keyword in params_to_keep):
|
|
if isinstance(submodule, nn.Linear):
|
|
if scale_weight_keys is not None:
|
|
scale_key = f"{name}.scale_weight"
|
|
if scale_key in scale_weight_keys:
|
|
print("Setting scale_weight for", name)
|
|
setattr(submodule, "scale_weight", scale_weight_keys[scale_key])
|
|
original_forward = submodule.forward
|
|
setattr(submodule, "original_forward", original_forward)
|
|
setattr(submodule, "forward", lambda input, m=submodule: fp8_linear_forward(m, base_dtype, input))
|
|
|
|
def convert_linear_with_lora_and_scale(module, scale_weight_keys=None, patches=None, params_to_keep={}):
|
|
log.info("Patching Linear layers...")
|
|
for name, submodule in module.named_modules():
|
|
if not any(keyword in name for keyword in params_to_keep):
|
|
# Set scale_weight if present
|
|
if scale_weight_keys is not None:
|
|
scale_key = f"{name}.scale_weight"
|
|
if scale_key in scale_weight_keys:
|
|
setattr(submodule, "scale_weight", scale_weight_keys[scale_key])
|
|
|
|
# Set LoRA if present
|
|
if hasattr(submodule, "lora"):
|
|
#print(f"removing old LoRA in {name}" )
|
|
delattr(submodule, "lora")
|
|
if patches is not None:
|
|
patch_key = f"diffusion_model.{name}.weight"
|
|
patch = patches.get(patch_key, [])
|
|
if len(patch) != 0:
|
|
lora_diffs = []
|
|
for p in patch:
|
|
lora_obj = p[1]
|
|
if "head" in name:
|
|
continue # For now skip LoRA for head layers
|
|
elif hasattr(lora_obj, "weights"):
|
|
lora_diffs.append(lora_obj.weights)
|
|
elif isinstance(lora_obj, tuple) and lora_obj[0] == "diff":
|
|
lora_diffs.append(lora_obj[1])
|
|
else:
|
|
continue
|
|
lora_strengths = [p[0] for p in patch]
|
|
lora = (lora_diffs, lora_strengths)
|
|
setattr(submodule, "lora", lora)
|
|
#print(f"Added LoRA to {name} with {len(lora_diffs)} diffs and strengths {lora_strengths}")
|
|
|
|
# Set forward if Linear and has either scale or lora
|
|
if isinstance(submodule, nn.Linear):
|
|
has_scale = hasattr(submodule, "scale_weight")
|
|
has_lora = hasattr(submodule, "lora")
|
|
if not hasattr(submodule, "original_forward"):
|
|
setattr(submodule, "original_forward", submodule.forward)
|
|
if has_scale or has_lora:
|
|
setattr(submodule, "forward", lambda input, m=submodule: linear_with_lora_and_scale_forward(m, input))
|
|
setattr(submodule, "step", 0) # Initialize step for LoRA scheduling
|
|
|
|
def remove_lora_from_module(module):
|
|
unloaded = False
|
|
for name, submodule in module.named_modules():
|
|
if hasattr(submodule, "lora"):
|
|
if not unloaded:
|
|
log.info("Unloading all LoRAs")
|
|
unloaded = True
|
|
delattr(submodule, "lora")
|