xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-06-07 11:02:17 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Chore] clean up debug info

Browse Source
This commit is contained in:
i-yuanyukun 2025-12-22 15:25:28 +08:00
parent 2a98ab3c8e
commit 27ae2e761c
2 changed files with 17 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
vllm/distributed/afd_transfer/afd_connector/p2p_connector.py
View File

@ -60,7 +60,7 @@ class P2PAFDConnector(AFDConnectorBase):
self.num_hidden_layers: int = ( self.num_hidden_layers: int = (
self.config.model_config.hf_config.num_hidden_layers self.config.model_config.hf_config.num_hidden_layers
) )
self.recv_attn_output_counter: int = 0 self.recv_attn_output_counter: int = 0
self.recv_ffn_output_counter: int = 0 self.recv_ffn_output_counter: int = 0
self.dp_metadata_list: dict[int, DPMetadata] = {} self.dp_metadata_list: dict[int, DPMetadata] = {}
@ -139,11 +139,9 @@ class P2PAFDConnector(AFDConnectorBase):
for idx in range(num_of_stages): for idx in range(num_of_stages):
if idx == 0: if idx == 0:
tensor_metadata_list[0] = tensor_metadata tensor_metadata_list[0] = tensor_metadata
logger.info(f"build tensor metadata: stage_{idx=}, size={tensor_metadata.size}")
else: else:
new_size = list(tensor_metadata.size) new_size = list(tensor_metadata.size)
new_size[0] = connector_metadata.afd_tokens_lens[idx] new_size[0] = connector_metadata.afd_tokens_lens[idx]
logger.info(f"build tensor metadata: stage_{idx=}, {new_size=}, {connector_metadata.afd_tokens_lens=}")
tensor_metadata_list[idx] = TensorMetadata( tensor_metadata_list[idx] = TensorMetadata(
tensor_metadata.device, tensor_metadata.device,
tensor_metadata.dtype, tensor_metadata.dtype,
@ -167,7 +165,6 @@ class P2PAFDConnector(AFDConnectorBase):
) )
metadata_tuple = (metadata, tensor_metadata) metadata_tuple = (metadata, tensor_metadata)
process_group.send_object(metadata_tuple, dst=dst) process_group.send_object(metadata_tuple, dst=dst)
logger.info(f"_send_metadata called build tensor metadata")
self._tensor_metadata_list = self._build_tensor_metadata_list( self._tensor_metadata_list = self._build_tensor_metadata_list(
tensor_metadata, metadata tensor_metadata, metadata
) )
@ -180,23 +177,32 @@ class P2PAFDConnector(AFDConnectorBase):
(self._current_afd_connector_metadata, tensor_metadata) = ( (self._current_afd_connector_metadata, tensor_metadata) = (
process_group.recv_object(src=src) process_group.recv_object(src=src)
) )
logger.info(f"_recv_metadata called build tensor metadata")
self._tensor_metadata_list = self._build_tensor_metadata_list( self._tensor_metadata_list = self._build_tensor_metadata_list(
tensor_metadata, self._current_afd_connector_metadata tensor_metadata, self._current_afd_connector_metadata
) )
logger.info(f"{self.config.parallel_config.data_parallel_size=}")
if self.config.parallel_config.data_parallel_size > 1: if self.config.parallel_config.data_parallel_size > 1:
logger.info("jcz recv_metadata num_of_stages:{}".format(self._current_afd_connector_metadata.num_of_stages)) logger.info(
"jcz recv_metadata num_of_stages:{}".format(
self._current_afd_connector_metadata.num_of_stages
)
)
for stage_idx in range(self._current_afd_connector_metadata.num_of_stages): for stage_idx in range(self._current_afd_connector_metadata.num_of_stages):
num_tokens_per_ubatch = self._tensor_metadata_list[stage_idx].size[0] num_tokens_per_ubatch = self._tensor_metadata_list[stage_idx].size[0]
logger.info(f"{stage_idx=}, {num_tokens_per_ubatch=}")
self.dp_metadata_list[stage_idx] = DPMetadata.make( self.dp_metadata_list[stage_idx] = DPMetadata.make(
self.config.parallel_config, self.config.parallel_config,
num_tokens_per_ubatch, num_tokens_per_ubatch,
torch.tensor([num_tokens_per_ubatch] * self.config.parallel_config.data_parallel_size, torch.tensor(
device="cpu", dtype=torch.int32), [num_tokens_per_ubatch]
* self.config.parallel_config.data_parallel_size,
device="cpu",
dtype=torch.int32,
),
) )
logger.info("jcz recv_metadata self.dp_metadata_list:{}".format(self.dp_metadata_list)) logger.info(
"jcz recv_metadata self.dp_metadata_list:{}".format(
self.dp_metadata_list
)
)
def _send_hidden_states( def _send_hidden_states(
self, self,
@ -229,7 +235,6 @@ class P2PAFDConnector(AFDConnectorBase):
dtype=tensor_metadata.dtype, dtype=tensor_metadata.dtype,
device=tensor_metadata.device, device=tensor_metadata.device,
) )
# logger.info(f"{__file__}: p2p recv hidden states: {hidden_states.shape=}, {tensor_metadata.size=}")
torch.distributed.recv( torch.distributed.recv(
hidden_states, hidden_states,
src=process_group.ranks[src], src=process_group.ranks[src],
@ -267,7 +272,6 @@ class P2PAFDConnector(AFDConnectorBase):
self.recv_ffn_output_counter self.recv_ffn_output_counter
% self._current_afd_connector_metadata.num_of_stages % self._current_afd_connector_metadata.num_of_stages
) )
logger.info(f"{stage_idx=}")
hidden_states, work_list = self._recv_hidden_states( hidden_states, work_list = self._recv_hidden_states(
src, src,
self.e2a_group, self.e2a_group,

15
vllm/model_executor/models/step3_text.py
View File

@ -303,10 +303,6 @@ class Step3TextDecoderLayer(nn.Module):
else: else:
hidden_states, residual = self.input_layernorm(hidden_states, residual) hidden_states, residual = self.input_layernorm(hidden_states, residual)
# query, key and positions must have the same number of tokens
# /model_executor/layers/rotary_embedding/base.py
# positions.shape=torch.Size([8192]), hidden_states.shape=torch.Size([4096, 3712]）
logger.info(f"{positions.shape=}, {hidden_states.shape=}")
hidden_states = self.self_attn( hidden_states = self.self_attn(
positions=positions, positions=positions,
hidden_states=hidden_states, hidden_states=hidden_states,
@ -342,7 +338,6 @@ class Step3TextDecoderLayer(nn.Module):
hidden_states = share_output + moe_output hidden_states = share_output + moe_output
else: else:
hidden_states = self.mlp(hidden_states) hidden_states = self.mlp(hidden_states)
logger.info(f"{type(hidden_states)=}")
return hidden_states return hidden_states
@ -353,7 +348,6 @@ class Step3TextModel(nn.Module):
config = vllm_config.model_config.hf_config config = vllm_config.model_config.hf_config
cache_config = vllm_config.cache_config cache_config = vllm_config.cache_config
quant_config = vllm_config.quant_config quant_config = vllm_config.quant_config
logger.info(f"{quant_config=}")
afd_config = vllm_config.afd_config afd_config = vllm_config.afd_config
self.vocab_size = config.vocab_size self.vocab_size = config.vocab_size
self.config = config self.config = config
@ -440,7 +434,6 @@ class Step3TextModel(nn.Module):
ubatch_hidden_states[stage_i] = hidden_states ubatch_hidden_states[stage_i] = hidden_states
ubatch_residual[stage_i] = residual ubatch_residual[stage_i] = residual
logger.info(f"create attn metadata:, {afd_metadata.afd_tokens_lens=}")
metadata = AFDConnectorMetadata.create_attention_metadata( metadata = AFDConnectorMetadata.create_attention_metadata(
layer_idx=layer.layer_idx, layer_idx=layer.layer_idx,
stage_idx=stage_i, stage_idx=stage_i,
@ -515,7 +508,6 @@ class Step3TextModel(nn.Module):
hidden_states, hidden_states,
layer_idx, layer_idx,
) -> torch.Tensor | IntermediateTensors: ) -> torch.Tensor | IntermediateTensors:
logger.info(f"{type(self.layers)=}, {type(layer_idx)=}")
hidden_states = self.layers[layer_idx].compute_ffn_output(hidden_states) hidden_states = self.layers[layer_idx].compute_ffn_output(hidden_states)
return hidden_states return hidden_states
@ -582,7 +574,6 @@ class Step3TextForCausalLM(nn.Module, SupportsPP):
return logits return logits
def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]) -> set[str]: def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]) -> set[str]:
logger.info(f"{__file__}: load_weights!")
qkv_params_mapping = [ qkv_params_mapping = [
# (param_name, shard_name, relative_start_idx, relative_end_idx) # (param_name, shard_name, relative_start_idx, relative_end_idx)
( (
@ -615,7 +606,6 @@ class Step3TextForCausalLM(nn.Module, SupportsPP):
(".gate_up_proj", ".up_proj", 1), (".gate_up_proj", ".up_proj", 1),
] ]
params_dict = dict(self.named_parameters()) params_dict = dict(self.named_parameters())
# logger.info(f"{params_dict.keys()=}")
loaded_params: set[str] = set() loaded_params: set[str] = set()
expert_params_mapping = [ expert_params_mapping = [
@ -627,10 +617,6 @@ class Step3TextForCausalLM(nn.Module, SupportsPP):
disable_moe_stacked_params = [data[1] for data in expert_params_mapping] disable_moe_stacked_params = [data[1] for data in expert_params_mapping]
for name, loaded_weight in weights: for name, loaded_weight in weights:
# logger.info(
# f"{self.afd_role=}, {name=}, is_moe: {self.is_moe_weight(name)}, "
# f"is_common: {self.is_common_weight(name)}"
# )
if self.afd_role == "attention" and self.is_moe_weight(name): if self.afd_role == "attention" and self.is_moe_weight(name):
continue continue
@ -695,7 +681,6 @@ class Step3TextForCausalLM(nn.Module, SupportsPP):
start_idx, start_idx,
end_idx, end_idx,
) in qkv_params_mapping: ) in qkv_params_mapping:
# logger.info(f"{weight_name=}, {name=}")
if weight_name not in name: if weight_name not in name:
continue continue
name = name.replace(weight_name, param_name) name = name.replace(weight_name, param_name)