[Doc]: fix typos in Python comments (#24294)
Signed-off-by: Didier Durand <durand.didier@gmail.com> Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
This commit is contained in:
co-authored by
Wentao Ye
parent
35efa70297
commit
35bf193864
@@ -534,7 +534,7 @@ def invoke_fused_moe_kernel(A: torch.Tensor,
|
||||
EM = sorted_token_ids.size(0)
|
||||
if A.size(0) < config["BLOCK_SIZE_M"]:
|
||||
# optimize for small batch_size.
|
||||
# We assume that top_ids of each token is unique, so
|
||||
# We assume that top_ids of each token is unique,
|
||||
# so num_valid_experts <= batch_size <= BLOCK_SIZE_M,
|
||||
# and we can skip some invalid blocks.
|
||||
EM = min(sorted_token_ids.size(0),
|
||||
|
||||
@@ -710,7 +710,7 @@ def determine_expert_map(
|
||||
|
||||
# Create a tensor of size num_experts filled with -1
|
||||
expert_map = torch.full((global_num_experts, ), -1, dtype=torch.int32)
|
||||
# Create a expert map for the local experts
|
||||
# Create an expert map for the local experts
|
||||
start_idx = ep_rank * base_experts + min(ep_rank, remainder)
|
||||
expert_map[start_idx:start_idx + local_num_experts] = torch.arange(
|
||||
0, local_num_experts, dtype=torch.int32)
|
||||
@@ -806,7 +806,7 @@ class FusedMoE(CustomOp):
|
||||
|
||||
self.global_num_experts = num_experts + num_redundant_experts
|
||||
|
||||
# we padding globally so EP buffer allocation works
|
||||
# we are padding globally so EP buffer allocation works
|
||||
if quant_config and quant_config.get_name() == "mxfp4":
|
||||
from vllm.model_executor.layers.quantization.mxfp4 import ( # noqa: E501
|
||||
should_use_flashinfer_mxfp4)
|
||||
|
||||
@@ -469,7 +469,7 @@ class GPTQMarlinMoEMethod(FusedMoEMethodBase):
|
||||
)
|
||||
layer.register_parameter("w2_scales", w2_scales)
|
||||
set_weight_attrs(w2_scales, extra_weight_attrs)
|
||||
# dont shard the w2 scales when running act order
|
||||
# don't shard the w2 scales when running act order
|
||||
set_weight_attrs(w2_scales,
|
||||
{"load_full_w2": self.quant_config.desc_act})
|
||||
# up_proj scales
|
||||
@@ -493,7 +493,7 @@ class GPTQMarlinMoEMethod(FusedMoEMethodBase):
|
||||
)
|
||||
layer.register_parameter("w2_qzeros", w2_qzeros)
|
||||
set_weight_attrs(w2_qzeros, extra_weight_attrs)
|
||||
# dont shard the w2 scales when running act order
|
||||
# don't shard the w2 scales when running act order
|
||||
set_weight_attrs(w2_qzeros,
|
||||
{"load_full_w2": self.quant_config.desc_act})
|
||||
w13_g_idx = torch.nn.Parameter(
|
||||
|
||||
Reference in New Issue
Block a user