[Bugfix][CI] fix typos (#34934)
Signed-off-by: 1195343015 <1195343015@qq.com> Signed-off-by: Jiayi Yan <66017932+1195343015@users.noreply.github.com> Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
co-authored by
Harry Mellor
parent
8c760b6ab6
commit
6a895197fa
@@ -44,7 +44,7 @@ class DefaultEplbPolicy(AbstractEplbPolicy):
|
||||
rank_in_pack = np.zeros_like(pack_index, dtype=np.int64)
|
||||
return pack_index, rank_in_pack
|
||||
|
||||
# Sort and get indices in decending order
|
||||
# Sort and get indices in descending order
|
||||
indices = np.argsort(-weight, axis=-1)
|
||||
|
||||
pack_index = np.full((num_layers, num_groups), -1, dtype=np.int64)
|
||||
|
||||
@@ -129,7 +129,7 @@ class KVConnectorRole(enum.Enum):
|
||||
class KVConnectorHandshakeMetadata(ABC): # noqa: B024
|
||||
"""
|
||||
Metadata used for out of band connector handshake between
|
||||
P/D workers. This needs to serializeable.
|
||||
P/D workers. This needs to serializable.
|
||||
"""
|
||||
|
||||
pass
|
||||
|
||||
@@ -398,7 +398,7 @@ class ReqMeta:
|
||||
)
|
||||
|
||||
|
||||
def need_gpu_interm_buffer(lmcache_config: LMCacheEngineConfig):
|
||||
def need_gpu_interim_buffer(lmcache_config: LMCacheEngineConfig):
|
||||
return not lmcache_config.enable_pd
|
||||
|
||||
|
||||
@@ -497,7 +497,7 @@ def _init_lmcache_engine(
|
||||
use_mla,
|
||||
)
|
||||
|
||||
use_gpu = need_gpu_interm_buffer(lmcache_config)
|
||||
use_gpu = need_gpu_interim_buffer(lmcache_config)
|
||||
vllm_gpu_connector: (
|
||||
VLLMBufferLayerwiseGPUConnector
|
||||
| VLLMPagedMemGPUConnectorV2
|
||||
|
||||
@@ -481,7 +481,7 @@ class MooncakeConnectorWorker:
|
||||
)
|
||||
|
||||
self._remote_agents: dict[EngineId, dict[int, dict[int, str]]] = {}
|
||||
self._pending_bootstrap_querys: dict[str, asyncio.Event] = {}
|
||||
self._pending_bootstrap_queries: dict[str, asyncio.Event] = {}
|
||||
self.side_channel_port: int = 0 # we will bind it in register_kv_caches()
|
||||
self.engine_id: EngineId = engine_id
|
||||
self.tp_rank = get_tensor_model_parallel_rank()
|
||||
@@ -1077,7 +1077,7 @@ class MooncakeConnectorWorker:
|
||||
response = self._xfer_resp_decoder.decode(ret_msg)
|
||||
if response.status == MooncakeXferResponseStatus.ERROR:
|
||||
logger.error(
|
||||
"Error happens during tranfering kvcache for %s: %s",
|
||||
"Error happens during transferring kvcache for %s: %s",
|
||||
req_ids,
|
||||
response.err_msg,
|
||||
)
|
||||
@@ -1140,8 +1140,8 @@ class MooncakeConnectorWorker:
|
||||
)
|
||||
|
||||
# Always notify others regardless of connection success or failure.
|
||||
self._pending_bootstrap_querys[remote_bootstrap_addr].set()
|
||||
del self._pending_bootstrap_querys[remote_bootstrap_addr]
|
||||
self._pending_bootstrap_queries[remote_bootstrap_addr].set()
|
||||
del self._pending_bootstrap_queries[remote_bootstrap_addr]
|
||||
|
||||
def receive_kv(
|
||||
self,
|
||||
@@ -1171,11 +1171,11 @@ class MooncakeConnectorWorker:
|
||||
pull_metas: dict[ReqId, PullReqMeta],
|
||||
):
|
||||
remote_bootstrap_addr = next(iter(pull_metas.values())).remote_bootstrap_addr
|
||||
if remote_bootstrap_addr not in self._pending_bootstrap_querys:
|
||||
self._pending_bootstrap_querys[remote_bootstrap_addr] = asyncio.Event()
|
||||
if remote_bootstrap_addr not in self._pending_bootstrap_queries:
|
||||
self._pending_bootstrap_queries[remote_bootstrap_addr] = asyncio.Event()
|
||||
await self._connect_to_prefiller_bootstrap(remote_bootstrap_addr)
|
||||
else:
|
||||
await self._pending_bootstrap_querys[remote_bootstrap_addr].wait()
|
||||
await self._pending_bootstrap_queries[remote_bootstrap_addr].wait()
|
||||
|
||||
if remote_engine_id not in self._remote_agents:
|
||||
logger.error(
|
||||
|
||||
@@ -720,7 +720,7 @@ class OffloadPromMetrics(KVConnectorPromMetrics):
|
||||
per_engine_labelvalues: dict[int, list[object]],
|
||||
):
|
||||
super().__init__(vllm_config, metric_types, labelnames, per_engine_labelvalues)
|
||||
# (engine_idx, transfer_tupe) -> (metric with bounded labels)
|
||||
# (engine_idx, transfer_type) -> (metric with bounded labels)
|
||||
self.histogram_transfer_size: dict[tuple[int, str], PromMetricT] = {}
|
||||
self.counter_kv_bytes: dict[tuple[int, str], PromMetricT] = {}
|
||||
self.counter_kv_transfer_time: dict[tuple[int, str], PromMetricT] = {}
|
||||
|
||||
Reference in New Issue
Block a user