mirror of
https://github.com/modelscope/DiffSynth-Studio.git
synced 2026-03-22 16:50:47 +00:00
[feature]:Add adaptation of all models to zero3
This commit is contained in:
@@ -348,13 +348,12 @@ class Flux2Unit_Qwen3PromptEmbedder(PipelineUnit):
|
||||
attention_mask = torch.cat(all_attention_masks, dim=0).to(device)
|
||||
|
||||
# Forward pass through the model
|
||||
with torch.no_grad():
|
||||
output = text_encoder(
|
||||
input_ids=input_ids,
|
||||
attention_mask=attention_mask,
|
||||
output_hidden_states=True,
|
||||
use_cache=False,
|
||||
)
|
||||
output = text_encoder(
|
||||
input_ids=input_ids,
|
||||
attention_mask=attention_mask,
|
||||
output_hidden_states=True,
|
||||
use_cache=False,
|
||||
)
|
||||
|
||||
# Only use outputs from intermediate layers and stack them
|
||||
out = torch.stack([output.hidden_states[k] for k in self.hidden_states_layers], dim=1)
|
||||
|
||||
@@ -1334,15 +1334,13 @@ def model_fn_wan_video(
|
||||
x, x_vap = torch.utils.checkpoint.checkpoint(
|
||||
create_custom_forward_vap(block, vap),
|
||||
x, context, t_mod, freqs, x_vap, context_vap, t_mod_vap, freqs_vap, block_id,
|
||||
use_reentrant=False,
|
||||
determinism_check="none"
|
||||
use_reentrant=False
|
||||
)
|
||||
elif use_gradient_checkpointing:
|
||||
x, x_vap = torch.utils.checkpoint.checkpoint(
|
||||
create_custom_forward_vap(block, vap),
|
||||
x, context, t_mod, freqs, x_vap, context_vap, t_mod_vap, freqs_vap, block_id,
|
||||
use_reentrant=False,
|
||||
determinism_check="none"
|
||||
use_reentrant=False
|
||||
)
|
||||
else:
|
||||
x, x_vap = vap(block, x, context, t_mod, freqs, x_vap, context_vap, t_mod_vap, freqs_vap, block_id)
|
||||
|
||||
Reference in New Issue
Block a user