Skip to content

Fix performance issue & bug on mps (Apple Silicon) #25

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
sunnyden wants to merge 3 commits into
base: main
Choose a base branch
from
+7 −3
Open

Fix performance issue & bug on mps (Apple Silicon) #25

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
9343af4
Fix inference issue for mps
sunnyden Jan 25, 2026
da8a71a
Fix unsupported operation
sunnyden Jan 25, 2026
e6cd411
only use fallback for mps
sunnyden Jan 26, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion moshi/moshi/models/lm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -976,7 +976,7 @@ def load_voice_prompt(self, voice_prompt: str):

def load_voice_prompt_embeddings(self, path: str):
self.voice_prompt = path
state = torch.load(path)
state = torch.load(path, map_location=torch.device(self.lm_model.device))

self.voice_prompt_audio = None
self.voice_prompt_embeddings = state["embeddings"].to(self.lm_model.device)
Expand Down
8 changes: 6 additions & 2 deletions moshi/moshi/modules/transformer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -265,8 +265,12 @@ def complete(self, k: torch.Tensor, v: torch.Tensor) -> KVCacheResult:
B, H, T, D = k.shape
indexes = torch.arange(T, device=self.end_offset.device, dtype=self.end_offset.dtype) + self.end_offset
indexes = indexes % self.capacity
self.cache[0].index_copy_(2, indexes, k)
self.cache[1].index_copy_(2, indexes, v)
if k.device.type == "mps":
self.cache[0][:, :, indexes, :] = k
self.cache[1][:, :, indexes, :] = v
else:
self.cache[0].index_copy_(2, indexes, k)
self.cache[1].index_copy_(2, indexes, v)
self.end_offset.add_(T)

keys = self.cache[0]
Expand Down