biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Models] Add remaining model PP support (#7168)

Browse Source 
Signed-off-by: Muralidhar Andoorveedu <muralidhar.andoorveedu@centml.ai>
Signed-off-by: Murali Andoorveedu <muralidhar.andoorveedu@centml.ai>
Co-authored-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Murali Andoorveedu
2024-10-03 19:56:58 -07:00
committed by GitHub
parent 303d44790a
commit 0f6d7a9a34
69 changed files with 2585 additions and 1344 deletions
Download Patch File Download Diff File Expand all files Collapse all files

62
vllm/model_executor/models/olmo.py
View File

@@ -21,7 +21,7 @@
# See the License for the specific language governing permissions and
# limitations under the License.
"""Inference-only OLMo model compatible with HuggingFace weights."""
from typing import Iterable, List, Optional, Tuple
from typing import Iterable, List, Optional, Tuple, Union
import torch
from torch import nn
@@ -29,14 +29,13 @@ from transformers import OlmoConfig
from vllm.attention import Attention, AttentionMetadata
from vllm.config import CacheConfig
from vllm.distributed import get_tensor_model_parallel_world_size
from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size
from vllm.model_executor.layers.activation import SiluAndMul
from vllm.model_executor.layers.linear import (MergedColumnParallelLinear,
QKVParallelLinear,
RowParallelLinear)
from vllm.model_executor.layers.logits_processor import LogitsProcessor
from vllm.model_executor.layers.quantization.base_config import (
QuantizationConfig)
from vllm.model_executor.layers.quantization import QuantizationConfig
from vllm.model_executor.layers.rotary_embedding import get_rope
from vllm.model_executor.layers.sampler import Sampler, SamplerOutput
from vllm.model_executor.layers.vocab_parallel_embedding import (
@@ -45,6 +44,10 @@ from vllm.model_executor.model_loader.weight_utils import default_weight_loader
from vllm.model_executor.sampling_metadata import SamplingMetadata
from vllm.sequence import IntermediateTensors
from .interfaces import SupportsPP
from .utils import (is_pp_missing_parameter,
make_empty_intermediate_tensors_factory, make_layers)
class OlmoAttention(nn.Module):
"""
@@ -223,19 +226,24 @@ class OlmoModel(nn.Module):
def __init__(self,
config: OlmoConfig,
cache_config: Optional[CacheConfig] = None,
quant_config: Optional[QuantizationConfig] = None):
quant_config: Optional[QuantizationConfig] = None,
prefix: str = ""):
super().__init__()
self.config = config
self.embed_tokens = VocabParallelEmbedding(config.vocab_size,
config.hidden_size)
self.layers = nn.ModuleList([
OlmoDecoderLayer(config, cache_config, quant_config)
for layer_idx in range(config.num_hidden_layers)
])
self.start_layer, self.end_layer, self.layers = make_layers(
config.num_hidden_layers,
lambda prefix: OlmoDecoderLayer(config, cache_config, quant_config
),
prefix=f"{prefix}.layers")
self.norm = nn.LayerNorm(config.hidden_size,
elementwise_affine=False,
bias=False)
self.make_empty_intermediate_tensors = (
make_empty_intermediate_tensors_factory(["hidden_states"],
config.hidden_size))
def forward(
self,
@@ -243,34 +251,41 @@ class OlmoModel(nn.Module):
positions: torch.Tensor,
kv_caches: List[torch.Tensor],
attn_metadata: AttentionMetadata,
) -> torch.Tensor:
intermediate_tensors: Optional[IntermediateTensors],
) -> Union[torch.Tensor, IntermediateTensors]:
"""
:param input_ids: A tensor of shape `(batch_size, seq_len)`.
"""
# Get embeddings of input.
# shape: (batch_size, seq_len, d_model)
inputs_embeds = self.embed_tokens(input_ids)
if get_pp_group().is_first_rank:
# Get embeddings of input.
# shape: (batch_size, seq_len, d_model)
inputs_embeds = self.embed_tokens(input_ids)
# embed positions
hidden_states = inputs_embeds
# embed positions
hidden_states = inputs_embeds
else:
assert intermediate_tensors is not None
hidden_states = intermediate_tensors["hidden_states"]
# Apply blocks one-by-one.
for layer_idx, decoder_layer in enumerate(self.layers):
for i in range(self.start_layer, self.end_layer):
# shape: (batch_size, seq_len, d_model)
hidden_states = decoder_layer(
hidden_states = self.layers[i](
positions,
hidden_states,
kv_caches[layer_idx],
kv_caches[i - self.start_layer],
attn_metadata,
)
if not get_pp_group().is_last_rank:
return IntermediateTensors({"hidden_states": hidden_states})
# Apply final layer norm.
# shape: (batch_size, seq_len or 1, d_model)
hidden_states = self.norm(hidden_states)
return hidden_states
class OlmoForCausalLM(nn.Module):
class OlmoForCausalLM(nn.Module, SupportsPP):
"""
Extremely barebones HF model wrapper.
"""
@@ -294,6 +309,8 @@ class OlmoForCausalLM(nn.Module):
)
self.logits_processor = LogitsProcessor(config.vocab_size)
self.sampler = Sampler()
self.make_empty_intermediate_tensors = (
self.model.make_empty_intermediate_tensors)
def forward(
self,
@@ -302,12 +319,13 @@ class OlmoForCausalLM(nn.Module):
kv_caches: List[torch.Tensor],
attn_metadata: AttentionMetadata,
intermediate_tensors: Optional[IntermediateTensors] = None,
) -> torch.Tensor:
) -> Union[torch.Tensor, IntermediateTensors]:
hidden_states = self.model(
input_ids=input_ids,
positions=positions,
kv_caches=kv_caches,
attn_metadata=attn_metadata,
intermediate_tensors=intermediate_tensors,
)
return hidden_states
@@ -358,6 +376,8 @@ class OlmoForCausalLM(nn.Module):
# Skip loading extra bias for GPTQ models.
if name.endswith(".bias") and name not in params_dict:
continue
if is_pp_missing_parameter(name, self):
continue
param = params_dict[name]
weight_loader = param.weight_loader
weight_loader(param, loaded_weight, shard_id)
@@ -366,6 +386,8 @@ class OlmoForCausalLM(nn.Module):
# Skip loading extra bias for GPTQ models.
if name.endswith(".bias") and name not in params_dict:
continue
if is_pp_missing_parameter(name, self):
continue
param = params_dict[name]
weight_loader = getattr(param, "weight_loader",
default_weight_loader)