Merge pull request #18 from jiangkuaixue123/afd-step3-merge

[Feature] adapt step3 model with AFD

vllm/distributed/afd_transfer/afd_connector/p2p_connector.py

@@ -52,9 +52,15 @@ class P2PAFDConnector(AFDConnectorBase):
         self._need_recv_metadata: bool = True
         self._tensor_metadata_list: dict[int, TensorMetadata] = {}
         self._current_afd_connector_metadata: AFDConnectorMetadata | None = None
-        self.num_hidden_layers: int = (
+        if getattr(self.config.model_config.hf_config, "text_config", None) is not None:
-            self.config.model_config.hf_config.num_hidden_layers
+            self.num_hidden_layers: int = (
-        )
+                self.config.model_config.hf_config.text_config.num_hidden_layers
+            )
+        else:
+            self.num_hidden_layers: int = (
+                self.config.model_config.hf_config.num_hidden_layers
+            )
+
         self.recv_attn_output_counter: int = 0
         self.recv_ffn_output_counter: int = 0
         self.dp_metadata_list: dict[int, DPMetadata] = {}
@@ -175,16 +181,28 @@ class P2PAFDConnector(AFDConnectorBase):
             tensor_metadata, self._current_afd_connector_metadata
         )
         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):
                 num_tokens_per_ubatch = self._tensor_metadata_list[stage_idx].size[0]
                 self.dp_metadata_list[stage_idx] = DPMetadata.make(
                     self.config.parallel_config,
                     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(
         self,

vllm/model_executor/models/deepseek_v2.py

@@ -1670,7 +1670,7 @@ class DeepseekV2ForCausalLM(
         return hidden_states
 
     def compute_ffn_output(
-        self, current_layer_idx, hidden_states
+        self, hidden_states, current_layer_idx
     ) -> torch.Tensor | IntermediateTensors:
         hidden_states = self.model.compute_ffn_output(hidden_states, current_layer_idx)
         return hidden_states

vllm/model_executor/models/step3_text.py

@@ -11,12 +11,16 @@ from torch import nn
 
 from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
-from vllm.config import CacheConfig, ModelConfig, VllmConfig
+from vllm.config import AFDConfig, CacheConfig, ModelConfig, VllmConfig
 from vllm.distributed import (
     get_pp_group,
     get_tensor_model_parallel_world_size,
     tensor_model_parallel_all_reduce,
 )
+from vllm.distributed.afd_transfer.afd_connector.metadata import (
+    AFDConnectorMetadata,
+)
+from vllm.forward_context import AFDMetadata, get_forward_context
 from vllm.logger import init_logger
 from vllm.model_executor.layers.activation import SiluAndMul
 from vllm.model_executor.layers.fused_moe import FusedMoE
@@ -37,6 +41,7 @@ from vllm.model_executor.layers.vocab_parallel_embedding import (
 from vllm.model_executor.model_loader.weight_utils import default_weight_loader
 from vllm.sequence import IntermediateTensors
 from vllm.transformers_utils.configs.step3_vl import Step3TextConfig
+from vllm.v1.worker.ubatching import dbo_current_ubatch_id, dbo_enabled, dbo_yield
 
 from .interfaces import SupportsPP
 from .utils import (
@@ -228,54 +233,59 @@ class Step3TextDecoderLayer(nn.Module):
         config: Step3TextConfig,
         cache_config: CacheConfig | None = None,
         quant_config: QuantizationConfig | None = None,
+        afd_config: AFDConfig | None = None,
         prefix: str = "",
     ) -> None:
         super().__init__()
         self.hidden_size = config.hidden_size
+        self.afd_role = afd_config.afd_role if afd_config is not None else None
 
-        self.self_attn = Step3TextAttention(
+        if self.afd_role is None or self.afd_role == "attention":
-            hidden_size=self.hidden_size,
+            self.self_attn = Step3TextAttention(
-            num_heads=config.num_attention_heads,
-            num_kv_heads=1,
-            cache_config=cache_config,
-            quant_config=quant_config,
-            norm_eps=config.rms_norm_eps,
-            max_position_embedding=config.max_position_embedding,
-            head_dim=config.head_dim,
-            share_q_dim=config.share_q_dim,
-            rope_parameters=config.rope_parameters,
-            prefix=f"{prefix}.self_attn",
-        )
-
-        layer_idx = int(prefix.split("layers.")[1].split(".")[0])
-        moe_layers_enum = getattr(config, "moe_layers_enum", None)
-        if moe_layers_enum is not None:
-            moe_layers_idx = [int(i) for i in moe_layers_enum.strip().split(",")]
-        else:
-            # Default to 1dense.
-            moe_layers_idx = [i for i in range(1, config.num_hidden_layers)]
-
-        if layer_idx in moe_layers_idx:
-            self.moe = FusedMoEBlock(
-                config=config, quant_config=quant_config, prefix=f"{prefix}.moe"
-            )
-            self.share_expert = Step3TextMLP(
                 hidden_size=self.hidden_size,
-                intermediate_size=config.share_expert_dim,
+                num_heads=config.num_attention_heads,
-                hidden_act="silu",
+                num_kv_heads=1,
+                cache_config=cache_config,
                 quant_config=quant_config,
-                prefix=f"{prefix}.share_expert",
+                norm_eps=config.rms_norm_eps,
+                max_position_embedding=config.max_position_embedding,
+                head_dim=config.head_dim,
+                share_q_dim=config.share_q_dim,
+                rope_parameters=config.rope_parameters,
+                prefix=f"{prefix}.self_attn",
             )
-            self.use_moe = True
+
-        else:
+        self.layer_idx = int(prefix.split("layers.")[1].split(".")[0])
-            self.mlp = Step3TextMLP(
+
-                hidden_size=config.hidden_size,
+        if self.afd_role is None or self.afd_role == "ffn":
-                intermediate_size=config.intermediate_size,
+            moe_layers_enum = getattr(config, "moe_layers_enum", None)
-                hidden_act="silu",
+            if moe_layers_enum is not None:
-                quant_config=quant_config,
+                moe_layers_idx = [int(i) for i in moe_layers_enum.strip().split(",")]
-                prefix=f"{prefix}.mlp",
+            else:
-            )
+                # Default to 1dense.
-            self.use_moe = False
+                moe_layers_idx = [i for i in range(1, config.num_hidden_layers)]
+
+            if self.layer_idx in moe_layers_idx:
+                self.moe = FusedMoEBlock(
+                    config=config, quant_config=quant_config, prefix=f"{prefix}.moe"
+                )
+                self.share_expert = Step3TextMLP(
+                    hidden_size=self.hidden_size,
+                    intermediate_size=config.share_expert_dim,
+                    hidden_act="silu",
+                    quant_config=quant_config,
+                    prefix=f"{prefix}.share_expert",
+                )
+                self.use_moe = True
+            else:
+                self.mlp = Step3TextMLP(
+                    hidden_size=config.hidden_size,
+                    intermediate_size=config.intermediate_size,
+                    hidden_act="silu",
+                    quant_config=quant_config,
+                    prefix=f"{prefix}.mlp",
+                )
+                self.use_moe = False
         self.input_layernorm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
         self.post_attention_layernorm = RMSNorm(
             config.hidden_size, eps=config.rms_norm_eps
@@ -300,6 +310,9 @@ class Step3TextDecoderLayer(nn.Module):
 
         hidden_states, residual = self.post_attention_layernorm(hidden_states, residual)
 
+        if self.afd_role == "attention":
+            return hidden_states, residual
+
         if self.use_moe:
             share_output = self.share_expert(hidden_states)
             moe_output = self.moe(hidden_states)
@@ -309,6 +322,16 @@ class Step3TextDecoderLayer(nn.Module):
 
         return hidden_states, residual
 
+    def compute_ffn_output(self, hidden_states):
+        assert self.afd_role == "ffn"
+        if self.use_moe:
+            share_output = self.share_expert(hidden_states)
+            moe_output = self.moe(hidden_states)
+            hidden_states = share_output + moe_output
+        else:
+            hidden_states = self.mlp(hidden_states)
+        return hidden_states
+
 
 @support_torch_compile
 class Step3TextModel(nn.Module):
@@ -317,6 +340,7 @@ class Step3TextModel(nn.Module):
         config = vllm_config.model_config.hf_config
         cache_config = vllm_config.cache_config
         quant_config = vllm_config.quant_config
+        afd_config = vllm_config.afd_config
         self.vocab_size = config.vocab_size
         self.config = config
 
@@ -336,6 +360,7 @@ class Step3TextModel(nn.Module):
                 config=config,
                 cache_config=cache_config,
                 quant_config=quant_config,
+                afd_config=afd_config,
                 prefix=prefix,
             ),
             prefix=f"{prefix}.layers",
@@ -352,6 +377,81 @@ class Step3TextModel(nn.Module):
     def embed_input_ids(self, input_ids: torch.Tensor) -> torch.Tensor:
         return self.embed_tokens(input_ids)
 
+    def forward_with_afd(
+        self,
+        hidden_states: torch.Tensor,
+        residual: torch.Tensor,
+        positions: torch.Tensor,
+        afd_metadata: AFDMetadata,
+    ) -> tuple[torch.Tensor, torch.Tensor]:
+        forward_conext = get_forward_context()
+        recv_handle = None
+
+        ubatch_hidden_states = []
+        ubatch_residual = []
+
+        start_idx = 0
+        for pos in afd_metadata.positions_list:
+            num_tokens = pos.shape[1] if pos.ndim == 2 else pos.shape[0]
+            end_idx = start_idx + num_tokens
+            ubatch_hidden_states.append(hidden_states[start_idx:end_idx])
+            ubatch_residual.append(
+                residual[start_idx:end_idx] if residual is not None else None
+            )
+            start_idx = end_idx
+
+        for layer in islice(self.layers, self.start_layer, self.end_layer):
+            for stage_i in range(forward_conext.afd_metadata.num_of_stages):
+                afd_connector = afd_metadata.afd_connector
+                forward_conext.attn_metadata = afd_metadata.attn_metadata_list[stage_i]
+                forward_conext.dp_metadata = afd_metadata.dp_metadata_list[stage_i]
+
+                residual = ubatch_residual[stage_i]
+
+                if layer.layer_idx > 0:
+                    hidden_states, recv_metadata = afd_connector.recv_ffn_output()
+                    if recv_metadata.recv_handle_list is not None:
+                        recv_handle = recv_metadata.recv_handle_list
+                else:
+                    hidden_states = ubatch_hidden_states[stage_i]
+
+                if recv_handle is not None:
+                    for work in recv_handle:
+                        work.wait()
+
+                current_positions = afd_metadata.positions_list[stage_i]
+                hidden_states, residual = layer(
+                    current_positions, hidden_states, residual
+                )
+
+                ubatch_hidden_states[stage_i] = hidden_states
+                ubatch_residual[stage_i] = residual
+                metadata = AFDConnectorMetadata.create_attention_metadata(
+                    layer_idx=layer.layer_idx,
+                    stage_idx=stage_i,
+                    seq_len=hidden_states.shape[0],
+                    dtype=hidden_states.dtype,
+                    device=hidden_states.device,
+                    num_of_stages=afd_metadata.num_of_stages,
+                    afd_tokens_lens=afd_metadata.afd_tokens_lens,
+                )
+                afd_connector.send_attn_output(hidden_states, metadata)
+
+        # Recv last layer FFN output.
+        for stage_i in range(afd_metadata.num_of_stages):
+            ubatch_hidden_states[stage_i], recv_metadata = (
+                afd_connector.recv_ffn_output()
+            )
+
+        # Re-assemble the batch
+        hidden_states = torch.cat(ubatch_hidden_states, dim=0)
+        if any(r is not None for r in ubatch_residual):
+            residual = torch.cat(ubatch_residual, dim=0)
+        else:
+            residual = None
+
+        return hidden_states, residual
+
     def forward(
         self,
         input_ids: torch.Tensor,
@@ -370,8 +470,19 @@ class Step3TextModel(nn.Module):
             hidden_states = intermediate_tensors["hidden_states"]
             residual = intermediate_tensors["residual"]
 
-        for layer in islice(self.layers, self.start_layer, self.end_layer):
+        forward_ctx = get_forward_context()
-            hidden_states, residual = layer(positions, hidden_states, residual)
+        afd_metadata = forward_ctx.afd_metadata if forward_ctx is not None else None
+
+        if afd_metadata is not None:
+            hidden_states, residual = self.forward_with_afd(
+                hidden_states,
+                residual,
+                positions,
+                afd_metadata,
+            )
+        else:
+            for layer in islice(self.layers, self.start_layer, self.end_layer):
+                hidden_states, residual = layer(positions, hidden_states, residual)
 
         if not get_pp_group().is_last_rank:
             return IntermediateTensors(
@@ -384,6 +495,14 @@ class Step3TextModel(nn.Module):
         hidden_states, _ = self.norm(hidden_states, residual)
         return hidden_states
 
+    def compute_ffn_output(
+        self,
+        hidden_states,
+        layer_idx,
+    ) -> torch.Tensor | IntermediateTensors:
+        hidden_states = self.layers[layer_idx].compute_ffn_output(hidden_states)
+        return hidden_states
+
 
 class Step3TextForCausalLM(nn.Module, SupportsPP):
     def __init__(
@@ -398,6 +517,11 @@ class Step3TextForCausalLM(nn.Module, SupportsPP):
         self.config = config
         self.vllm_config = vllm_config
 
+        self.afd_config = vllm_config.afd_config
+        self.afd_role = (
+            self.afd_config.afd_role if self.afd_config is not None else None
+        )
+
         self.model = Step3TextModel(vllm_config=vllm_config, prefix=prefix)
 
         if get_pp_group().is_last_rank:
@@ -429,6 +553,14 @@ class Step3TextForCausalLM(nn.Module, SupportsPP):
         )
         return hidden_states
 
+    def compute_ffn_output(
+        self,
+        hidden_states,
+        current_layer_idx,
+    ) -> torch.Tensor | IntermediateTensors:
+        hidden_states = self.model.compute_ffn_output(hidden_states, current_layer_idx)
+        return hidden_states
+
     def compute_logits(self, hidden_states: torch.Tensor) -> torch.Tensor:
         logits = self.logits_processor(self.lm_head, hidden_states)
         return logits
@@ -477,9 +609,13 @@ class Step3TextForCausalLM(nn.Module, SupportsPP):
         disable_moe_stacked_params = [data[1] for data in expert_params_mapping]
 
         for name, loaded_weight in weights:
+            if self.afd_role == "attention" and self.is_moe_weight(name):
+                continue
+
             for param_name, weight_name, shard_id in stacked_params_mapping:
                 if weight_name not in name:
                     continue
+
                 if any(
                     disable_moe_stacked_param in name
                     for disable_moe_stacked_param in disable_moe_stacked_params
@@ -498,6 +634,10 @@ class Step3TextForCausalLM(nn.Module, SupportsPP):
                     param_name, weight_name, shard_id = mapping
                     if weight_name not in name:
                         continue
+
+                    if self.afd_role is not None and self.afd_role == "attention":
+                        continue
+
                     name = name.replace(weight_name, param_name)
                     # Skip layers on other devices.
                     if is_pp_missing_parameter(name, self):
@@ -521,6 +661,12 @@ class Step3TextForCausalLM(nn.Module, SupportsPP):
                     loaded_params.add(name)
                     break
                 else:
+                    if (
+                        self.afd_role == "ffn"
+                        and not self.is_moe_weight(name)
+                        and not self.is_common_weight(name)
+                    ):
+                        continue
                     for (
                         param_name,
                         weight_name,
@@ -552,3 +698,25 @@ class Step3TextForCausalLM(nn.Module, SupportsPP):
                         weight_loader(param, loaded_weight)
                         loaded_params.add(name)
         return loaded_params
+
+    def is_moe_weight(self, name):
+        if (
+            "shared_expert" in name
+            or "experts" in name
+            or "gate" in name
+            or "up" in name
+            or "down" in name
+        ):
+            return True
+        return False
+
+    def is_common_weight(self, name):
+        if (
+            "lm_head" in name
+            or "model.norm.weight" in name
+            or "embed" in name
+            or "input_layernorm" in name
+            or "post_attention_layernorm" in name
+        ):
+            return True
+        return False

vllm/model_executor/models/step3_vl.py

@@ -1126,6 +1126,16 @@ class Step3VLForConditionalGeneration(nn.Module, SupportsMultiModal, SupportsPP)
 
         return hidden_states
 
+    def compute_ffn_output(
+        self,
+        hidden_states,
+        current_layer_idx,
+    ) -> torch.Tensor | IntermediateTensors:
+        hidden_states = self.language_model.compute_ffn_output(
+            hidden_states, current_layer_idx
+        )
+        return hidden_states
+
     def compute_logits(
         self,
         hidden_states: torch.Tensor,

vllm/v1/worker/gpu_ffn_model_runner.py

@@ -130,15 +130,17 @@ class GPUFFNModelRunner(LoRAModelRunnerMixin):
 
         try:
             hidden_states, recv_metadata = self.connector.recv_attn_output()
-            if hasattr(self.connector, 'dp_metadata_list'):
+            if hasattr(self.connector, "dp_metadata_list"):
-                dp_metadata = self.connector.dp_metadata_list.get(recv_metadata.stage_idx, None)
+                dp_metadata = self.connector.dp_metadata_list.get(
+                    recv_metadata.stage_idx, None
+                )
             else:
                 dp_metadata = None
             current_layer_idx = recv_metadata.layer_idx
-            logger.info(
+            # logger.info(
-                f"layer {current_layer_idx} moe recv hidden states type:{type(hidden_states)}, shape:{hidden_states.shape}"
+            #     f"layer {current_layer_idx} moe recv hidden states type:{type(hidden_states)}, shape:{hidden_states.shape}"
-                f" dp_metadata: {dp_metadata}"
+            #     f" dp_metadata: {dp_metadata}"
-            )
+            # )
             num_tokens = hidden_states.shape[0]
             if recv_metadata is not None and recv_metadata.recv_handle_list is not None:
                 for work in recv_metadata.recv_handle_list:
@@ -220,7 +222,7 @@ class GPUFFNModelRunner(LoRAModelRunnerMixin):
                 hidden_states, dim=0
             )
             ffn_output = self.model.compute_ffn_output(
-                current_layer_idx, gathered_hidden_states
+                gathered_hidden_states, current_layer_idx
             )
             # Extract the output corresponding to current rank
             start_idx = hidden_states.shape[0] * get_tensor_model_parallel_rank()
@@ -229,7 +231,7 @@ class GPUFFNModelRunner(LoRAModelRunnerMixin):
         else:
             # Single TP case
             rank_ffn_output = self.model.compute_ffn_output(
-                current_layer_idx, hidden_states
+                hidden_states, current_layer_idx
             )
 
         return rank_ffn_output
@@ -349,7 +351,7 @@ class GPUFFNModelRunner(LoRAModelRunnerMixin):
                 hidden_states, dim=0
             )
             ffn_output = self.model.compute_ffn_output(
-                current_layer_idx, gathered_hidden_states
+                gathered_hidden_states, current_layer_idx
             )
 
             # Extract the output corresponding to current rank
@@ -359,7 +361,7 @@ class GPUFFNModelRunner(LoRAModelRunnerMixin):
         else:
             # Single TP case
             rank_ffn_output = self.model.compute_ffn_output(
-                current_layer_idx, hidden_states
+                hidden_states, current_layer_idx
             )
 
         return rank_ffn_output

vllm/v1/worker/gpu_model_runner.py

@@ -3141,7 +3141,9 @@ class GPUModelRunner(
             # Mark KV scales as calculated after the first forward pass
             self.calculate_kv_scales = False
 
-        afd_metadata = self._build_afd_metadata(ubatch_slices_padded, num_tokens_unpadded)
+        afd_metadata = self._build_afd_metadata(
+            ubatch_slices_padded, num_tokens_unpadded
+        )
 
         self.profiler.step()
         # Run the model.
@@ -3160,8 +3162,9 @@ class GPUModelRunner(
             record_function_or_nullcontext("gpu_model_runner: forward"),
             self.maybe_get_kv_connector_output(scheduler_output) as kv_connector_output,
         ):
-            logger.info(f"input_ids: {input_ids.shape}")
+            if input_ids is not None:
-            if inputs_embeds:
+                logger.info(f"input_ids: {input_ids.shape}")
+            if inputs_embeds is not None:
                 logger.info(f"inputs_embeds: {inputs_embeds.shape}")
             model_output = self._model_forward(
                 input_ids=input_ids,
@@ -4274,7 +4277,9 @@ class GPUModelRunner(
                 if num_tokens_across_dp is not None:
                     num_tokens_across_dp[:] = num_tokens_padded
 
-            afd_metadata = self._build_afd_metadata(ubatch_slices_padded, num_tokens_unpadded)
+            afd_metadata = self._build_afd_metadata(
+                ubatch_slices_padded, num_tokens_unpadded
+            )
 
             with (
                 self.maybe_randomize_inputs(input_ids, inputs_embeds),

vllm/v1/worker/gpu_ubatch_wrapper.py

@@ -405,9 +405,12 @@ class UBatchWrapper:
                 afd_metadata.input_ids_list.append(sliced_input_ids)
                 afd_metadata.positions_list.append(sliced_positions)
                 afd_metadata.inputs_embeds_list.append(sliced_inputs_embeds)
-                afd_metadata.intermediate_tensors_list.append(sliced_intermediate_tensors)
+                afd_metadata.intermediate_tensors_list.append(
+                    sliced_intermediate_tensors
+                )
                 afd_metadata.attn_metadata_list.append(
-                    attn_metadata[i] if attn_metadata is not None else None)
+                    attn_metadata[i] if attn_metadata is not None else None
+                )
                 afd_metadata.dp_metadata_list.append(ubatch_dp_metadata)
 
         return afd_metadata
@@ -481,7 +484,7 @@ class UBatchWrapper:
             # num_tokens, we don't have a non-ubatched one. Without this
             # check, the cudagraph wrapper will try to capture a cudagraph
             # for this shape during a normal run.
-            
+
             if cudagraph_runtime_mode is CUDAGraphMode.FULL:
                 assert batch_descriptor is not None
                 if batch_descriptor.num_tokens in self.cudagraphs: