alphard/stable-diffusion.cpp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: support bf16 lora weights (#82)

Browse Source
This commit is contained in:
Erik Scholz 2023-11-20 15:34:17 +01:00 committed by GitHub
parent ae1d5dcebb
commit c874063408
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 11 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
models/convert.py
View File

@ -101,7 +101,7 @@ QK8_0 = 32
def quantize_q8_0(x): def quantize_q8_0(x):
assert x.shape[-1] % QK8_0 == 0 and x.shape[-1] > QK8_0 assert x.shape[-1] % QK8_0 == 0 and x.shape[-1] > QK8_0
x = x.reshape(-1, QK8_0) x = x.reshape(-1, QK8_0)
amax = np.max(np.abs(x), axis=-1, keepdims=True) amax = np.max(np.abs(x), axis=-1, keepdims=True)
d = amax / ((1 << 7) - 1) d = amax / ((1 << 7) - 1)
qs = (x / d).round().clip(min=-128, max=127).astype(np.int8) qs = (x / d).round().clip(min=-128, max=127).astype(np.int8)
d = d.astype(np.float16).view(np.int8) d = d.astype(np.float16).view(np.int8)
@ -178,7 +178,7 @@ def preprocess(state_dict):
print("no alphas_cumprod in file, generate new one") print("no alphas_cumprod in file, generate new one")
alphas_cumprod = get_alpha_comprod() alphas_cumprod = get_alpha_comprod()
state_dict["alphas_cumprod"] = alphas_cumprod state_dict["alphas_cumprod"] = alphas_cumprod
new_state_dict = {} new_state_dict = {}
for name, w in state_dict.items(): for name, w in state_dict.items():
# ignore unused tensors # ignore unused tensors
@ -192,7 +192,7 @@ def preprocess(state_dict):
if skip: if skip:
continue continue
# # convert BF16 to FP16 # convert BF16 to FP16
if w.dtype == torch.bfloat16: if w.dtype == torch.bfloat16:
w = w.to(torch.float16) w = w.to(torch.float16)
@ -251,7 +251,7 @@ def preprocess(state_dict):
new_state_dict[new_name] = w new_state_dict[new_name] = w
print(f"preprocess {name} => {new_name}") print(f"preprocess {name} => {new_name}")
continue continue
# convert unet transformer linear to conv2d 1x1 # convert unet transformer linear to conv2d 1x1
if name.startswith("model.diffusion_model.") and (name.endswith("proj_in.weight") or name.endswith("proj_out.weight")): if name.startswith("model.diffusion_model.") and (name.endswith("proj_in.weight") or name.endswith("proj_out.weight")):
if len(w.shape) == 2: if len(w.shape) == 2:
@ -342,6 +342,11 @@ def preprocess_lora(state_dict):
for name, w in state_dict.items(): for name, w in state_dict.items():
if not isinstance(w, torch.Tensor): if not isinstance(w, torch.Tensor):
continue continue
# convert BF16 to FP16
if w.dtype == torch.bfloat16:
w = w.to(torch.float16)
name_without_network_parts, network_part = name.split(".", 1) name_without_network_parts, network_part = name.split(".", 1)
new_name_without_network_parts = convert_diffusers_name_to_compvis(name_without_network_parts) new_name_without_network_parts = convert_diffusers_name_to_compvis(name_without_network_parts)
if new_name_without_network_parts == None: if new_name_without_network_parts == None:
@ -421,6 +426,7 @@ def convert(model_path, out_type = None, out_file=None, lora=False):
continue continue
if name in unused_tensors: if name in unused_tensors:
continue continue
data = state_dict[name].numpy() data = state_dict[name].numpy()
n_dims = len(data.shape) n_dims = len(data.shape)
@ -452,7 +458,7 @@ def convert(model_path, out_type = None, out_file=None, lora=False):
else: else:
data = data.astype(np.float32) data = data.astype(np.float32)
ttype = "f32" ttype = "f32"
print("Processing tensor: {} with shape {}, {} -> {}".format(name, data.shape, old_type, ttype)) print("Processing tensor: {} with shape {}, {} -> {}".format(name, data.shape, old_type, ttype))
# header # header