Skip to content
This repository has been archived by the owner on Oct 11, 2024. It is now read-only.

Commit

Permalink
[misc] update tpu int8 to use new vLLM Parameters (vllm-project#7973)
Browse files Browse the repository at this point in the history
  • Loading branch information
dsikka committed Aug 31, 2024
1 parent 6b163e4 commit 935a2dd
Show file tree
Hide file tree
Showing 2 changed files with 13 additions and 11 deletions.
3 changes: 2 additions & 1 deletion vllm/model_executor/layers/linear.py
Original file line number Diff line number Diff line change
Expand Up @@ -23,7 +23,8 @@
WEIGHT_LOADER_V2_SUPPORTED = [
"CompressedTensorsLinearMethod", "AWQMarlinLinearMethod",
"AWQLinearMethod", "GPTQMarlinLinearMethod", "Fp8LinearMethod",
"MarlinLinearMethod", "QQQLinearMethod", "GPTQMarlin24LinearMethod"
"MarlinLinearMethod", "QQQLinearMethod", "GPTQMarlin24LinearMethod",
"TPUInt8LinearMethod"
]


Expand Down
21 changes: 11 additions & 10 deletions vllm/model_executor/layers/quantization/tpu_int8.py
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@
from vllm.model_executor.layers.linear import LinearBase, LinearMethodBase
from vllm.model_executor.layers.quantization.base_config import (
QuantizationConfig)
from vllm.model_executor.utils import set_weight_attrs
from vllm.model_executor.parameter import ModelWeightParameter

ACTIVATION_SCHEMES = ["none"]

Expand Down Expand Up @@ -64,16 +64,16 @@ def create_weights(self, layer: Module, input_size_per_partition: int,
output_partition_sizes: List[int], input_size: int,
output_size: int, params_dtype: torch.dtype,
**extra_weight_attrs):
weight = Parameter(torch.empty(sum(output_partition_sizes),
input_size_per_partition,
dtype=params_dtype),
requires_grad=False)

weight_loader = extra_weight_attrs.get("weight_loader")
weight = ModelWeightParameter(data=torch.empty(
sum(output_partition_sizes),
input_size_per_partition,
dtype=params_dtype),
input_dim=1,
output_dim=0,
weight_loader=weight_loader)
layer.register_parameter("weight", weight)
set_weight_attrs(weight, {
**extra_weight_attrs,
"input_dim": 1,
"output_dim": 0,
})

def _quantize_weight(
self, weight: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
Expand All @@ -92,6 +92,7 @@ def _quantize_weight(
return qweight, qscale

def process_weights_after_loading(self, layer: Module) -> None:
layer.weight = Parameter(layer.weight.data, requires_grad=False)
device = layer.weight.device
qweight, qscale = self._quantize_weight(layer.weight)
qweight = qweight.to(device)
Expand Down

0 comments on commit 935a2dd

Please sign in to comment.