Skip to content

Remove unneed allocation #60

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
roj234 wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Remove unneed allocation #60

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
9a18aa8
Fix unused allocation
roj234 Feb 15, 2026
5340e70
Add runtime checks for logits_all
roj234 Feb 15, 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
21 changes: 15 additions & 6 deletions llama_cpp/llama.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -526,8 +526,8 @@ def free_lora_adapter():
self.n_tokens = 0
self.input_ids: npt.NDArray[np.intc] = np.ndarray((n_ctx,), dtype=np.intc)
self.scores: npt.NDArray[np.single] = np.ndarray(
(n_ctx if logits_all == True else n_batch, self._n_vocab), dtype=np.single
)
(n_ctx, self._n_vocab), dtype=np.single
) if self._logits_all else None

self._mirostat_mu = ctypes.c_float(
2.0 * 5.0
Expand Down Expand Up @@ -638,6 +638,10 @@ def _input_ids(self) -> npt.NDArray[np.intc]:

@property
def _scores(self) -> npt.NDArray[np.single]:
if not self._logits_all:
raise RuntimeError(
"Llama model must be created with logits_all=True to call this method"
)
return self.scores[: self.n_tokens, :]

@property
Expand All @@ -646,6 +650,10 @@ def eval_tokens(self) -> Deque[int]:

@property
def eval_logits(self) -> Deque[List[float]]:
if not self._logits_all:
raise RuntimeError(
"Llama model must be created with logits_all=True to call this method"
)
return deque(
self.scores[: self.n_tokens, :].tolist(),
maxlen=self._n_ctx if self._logits_all else 1,
Expand Down Expand Up @@ -2434,10 +2442,11 @@ def save_state(self) -> LlamaState:
)

def load_state(self, state: LlamaState) -> None:
# Only filling in up to `n_tokens` and then zero-ing out the rest
self.scores[: state.n_tokens, :] = state.scores.copy()
rest = self.scores[state.n_tokens :, :]
rest[rest > 0] = 0.0
if self._logits_all:
# Only filling in up to `n_tokens` and then zero-ing out the rest
self.scores[: state.n_tokens, :] = state.scores.copy()
rest = self.scores[state.n_tokens :, :]
rest[rest > 0] = 0.0
self.input_ids = state.input_ids.copy()
self.n_tokens = state.n_tokens
self._seed = state.seed
Expand Down
Loading