Skip to content

[WIP/AI-assisted] fix: embedding parameter naming consistency #2335

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

Draft
lxcxjxhx wants to merge 4 commits into
base: main
Choose a base branch
from
+16 −6
Draft

[WIP/AI-assisted] fix: embedding parameter naming consistency #2335

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
fdcaad0
fix: handle deprecated 'embedding' kwarg with warning (#2210)
lxcxjxhx Jul 9, 2026
ad036a8
fix: handle deprecated 'embedding' kwarg with warning (#2210)
lxcxjxhx Jul 9, 2026
9aa447f
fix: rename 'embedding' parameter to 'embeddings' with deprecation wa…
lxcxjxhx Jul 9, 2026
4b67660
fix: rename 'embedding' parameter to 'embeddings' for consistency
lxcxjxhx Jul 9, 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
22 changes: 16 additions & 6 deletions llama_cpp/llama.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -90,7 +90,7 @@ def __init__(
yarn_beta_slow: float = 1.0,
yarn_orig_ctx: int = 0,
logits_all: bool = False,
embedding: bool = False,
embeddings: bool = False,
offload_kqv: bool = True,
flash_attn: bool = False,
op_offload: Optional[bool] = None,
Expand Down Expand Up @@ -199,8 +199,18 @@ def __init__(
A Llama instance.
"""
self.verbose = verbose
self._stack = contextlib.ExitStack()

# Handle deprecated 'embedding' kwarg (singular) as alias for 'embeddings' (plural)
if 'embedding' in kwargs:
warnings.warn(
"The 'embedding' parameter is deprecated. Use 'embeddings' instead. "
"Support for 'embedding' will be removed in a future version.",
DeprecationWarning,
stacklevel=2
)
embeddings = kwargs.pop('embedding')

self._stack = contextlib.ExitStack()
set_verbose(verbose)

if not Llama.__backend_initialized:
Expand Down Expand Up @@ -342,7 +352,7 @@ def __init__(
)
self.context_params.yarn_orig_ctx = yarn_orig_ctx if yarn_orig_ctx != 0 else 0
self._logits_all = logits_all if draft_model is None else True
self.context_params.embeddings = embedding # TODO: Rename to embeddings
self.context_params.embeddings = embeddings
self.context_params.offload_kqv = offload_kqv
self.context_params.flash_attn_type = (
llama_cpp.LLAMA_FLASH_ATTN_TYPE_ENABLED
Expand Down Expand Up @@ -397,7 +407,7 @@ def __init__(
self.context_params.n_batch = self.n_batch
self.context_params.n_ubatch = min(self.n_batch, n_ubatch)

if embedding:
if embeddings:
self.context_params.n_seq_max = min(
self.n_batch,
llama_cpp.llama_max_parallel_sequences(),
Expand Down Expand Up @@ -1088,7 +1098,7 @@ def embed(

if self.context_params.embeddings is False:
raise RuntimeError(
"Llama model must be created with embedding=True to call this method"
"Llama model must be created with embeddings=True to call this method"
)

if self.verbose:
Expand Down Expand Up @@ -2163,7 +2173,7 @@ def __getstate__(self):
yarn_beta_slow=self.context_params.yarn_beta_slow,
yarn_orig_ctx=self.context_params.yarn_orig_ctx,
logits_all=self._logits_all,
embedding=self.context_params.embeddings,
embeddings=self.context_params.embeddings,
offload_kqv=self.context_params.offload_kqv,
flash_attn=(
self.context_params.flash_attn_type
Expand Down