[Doc]: fix typos in various files (#24726)

Signed-off-by: Didier Durand <durand.didier@gmail.com>
This commit is contained in:
Didier Durand
2025-09-12 06:43:12 -07:00
committed by GitHub
parent 0377802c20
commit bcb06d7baf
11 changed files with 11 additions and 11 deletions
+1 -1
View File
@@ -901,7 +901,7 @@ def _get_query_key_seq_metadata(
attn_metadata.encoder_seq_start_loc,
attn_metadata.max_encoder_seq_len)
elif attn_type == AttentionType.ENCODER:
# For encoder attention both the query and the key are same i.e the
# For encoder attention both the query and the key are same i.e. the
# encoder sequence.
return (attn_metadata.encoder_seq_start_loc,
attn_metadata.max_encoder_seq_len,
+1 -1
View File
@@ -551,7 +551,7 @@ class RandomDataset(BenchmarkDataset):
[6880, 6881] -> ['Ġcalls', 'here'] ->
[1650, 939, 486] -> ['Ġcall', 'sh', 'ere']
To avoid uncontrolled change of the prompt length,
the encoded sequence is truncated before being decode again.
the encoded sequence is truncated before being decoded again.
"""
# Build the inner sequence by sampling sequentially from the vocab
inner_seq = ((offset + index + np.arange(input_len))
+1 -1
View File
@@ -242,7 +242,7 @@ def get_logits_processors(processors: Optional[LogitsProcessors],
elif processors:
raise ValueError(
"The `logits_processors` argument is not supported by this "
"server. See --logits-processor-pattern engine argugment "
"server. See --logits-processor-pattern engine argument "
"for more information.")
return None
@@ -324,7 +324,7 @@ class MambaMixer2(MambaBase, CustomOp):
# - the weight already has a "weight_loader" attribute
# which set_weight_attrs will raise if we do not
# delete before trying to override it
# - ditto for the otther two weights below
# - ditto for the other two weights below
delattr(self.conv1d.bias, "weight_loader")
set_weight_attrs(
self.conv1d.bias,
+1 -1
View File
@@ -1117,7 +1117,7 @@ class MiniCPMVBaseModel(nn.Module, SupportsMultiModal, SupportsPP):
def _process_multimodal_inputs(self, modalities: dict):
# The result multimodal_embeddings is tuple of tensors, with each
# tensor correspoending to a multimodal data item (image or video).
# tensor corresponding to a multimodal data item (image or video).
multimodal_embeddings: tuple[torch.Tensor, ...] = ()
# NOTE: It is important to iterate over the keys in this dictionary
+1 -1
View File
@@ -2659,7 +2659,7 @@ class GPUModelRunner(LoRAModelRunnerMixin, KVConnectorModelRunnerMixin):
num_tokens += num_pad
# If cudagraph_mode.decode_mode() == FULL and
# cudagraph_mode.seperate_routine(). This means that we are using
# cudagraph_mode.separate_routine(). This means that we are using
# different graphs and/or modes for mixed prefill-decode batches vs.
# uniform decode batches. A uniform decode batch means that all
# requests have identical query length, except a potential virtual
+1 -1
View File
@@ -392,7 +392,7 @@ class InputBatch:
# NOTE: the following is unsafe
# self.token_ids_cpu[i1, ...], self.token_ids_cpu[i2, ...], =\
# self.token_ids_cpu[i2, ...], self.token_ids_cpu[i1, ...]
# instead, we need to temporiarily copy the data for one of the indices
# instead, we need to temporarily copy the data for one of the indices
# TODO(lucas): optimize this by only copying valid indices
tmp = self.token_ids_cpu[i1, ...].copy()
self.token_ids_cpu[i1, ...] = self.token_ids_cpu[i2, ...]