[6/N] pass whole config to inner model (#10205)
Signed-off-by: youkaichao <youkaichao@gmail.com>
This commit is contained in:
@@ -24,7 +24,8 @@ from vllm.sequence import IntermediateTensors
|
||||
|
||||
from .interfaces import SupportsPP
|
||||
from .utils import (is_pp_missing_parameter,
|
||||
make_empty_intermediate_tensors_factory, make_layers)
|
||||
make_empty_intermediate_tensors_factory, make_layers,
|
||||
maybe_prefix)
|
||||
|
||||
|
||||
def load_column_parallel_weight(param: torch.nn.Parameter,
|
||||
@@ -299,14 +300,13 @@ class Phi3SmallDecoderLayer(nn.Module):
|
||||
|
||||
class Phi3SmallModel(nn.Module):
|
||||
|
||||
def __init__(
|
||||
self,
|
||||
config: PretrainedConfig,
|
||||
cache_config: Optional[CacheConfig] = None,
|
||||
quant_config: Optional[QuantizationConfig] = None,
|
||||
prefix: str = "",
|
||||
):
|
||||
def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
|
||||
super().__init__()
|
||||
|
||||
config = vllm_config.model_config.hf_config
|
||||
cache_config = vllm_config.cache_config
|
||||
quant_config = vllm_config.quant_config
|
||||
|
||||
self.config = config
|
||||
self.embed_tokens = VocabParallelEmbedding(config.vocab_size,
|
||||
config.hidden_size)
|
||||
@@ -363,18 +363,14 @@ class Phi3SmallModel(nn.Module):
|
||||
class Phi3SmallForCausalLM(nn.Module, SupportsPP):
|
||||
_tied_weights_keys = ["lm_head.weight"]
|
||||
|
||||
def __init__(
|
||||
self,
|
||||
vllm_config: VllmConfig,
|
||||
prefix: str = "",
|
||||
) -> None:
|
||||
def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
|
||||
super().__init__()
|
||||
config = vllm_config.model_config.hf_config
|
||||
cache_config = vllm_config.cache_config
|
||||
quant_config = vllm_config.quant_config
|
||||
self.config = config
|
||||
self.quant_config = quant_config
|
||||
self.model = Phi3SmallModel(config, cache_config, quant_config)
|
||||
self.model = Phi3SmallModel(vllm_config=vllm_config,
|
||||
prefix=maybe_prefix(prefix, "model"))
|
||||
self.vocab_size = config.vocab_size
|
||||
self.mup_width_multiplier = config.mup_width_multiplier
|
||||
self.lm_head = ParallelLMHead(
|
||||
|
||||
Reference in New Issue
Block a user