Skip to content

Add quantized mixtral support #2673

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Jan 31, 2024
Merged

Add quantized mixtral support #2673

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
d9dde90
Add quantized mixtral support
WoosukKwon Jan 30, 2024
d39519c
Check quant_config
WoosukKwon Jan 30, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 9 additions & 4 deletions vllm/model_executor/model_loader.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,6 @@

import torch
import torch.nn as nn
from transformers import PretrainedConfig

from vllm.config import ModelConfig, LoRAConfig
from vllm.model_executor.models import ModelRegistry
Expand All @@ -21,8 +20,14 @@ def _set_default_torch_dtype(dtype: torch.dtype):
torch.set_default_dtype(old_dtype)


def _get_model_architecture(config: PretrainedConfig) -> Type[nn.Module]:
architectures = getattr(config, "architectures", [])
def _get_model_architecture(model_config: ModelConfig) -> Type[nn.Module]:
architectures = getattr(model_config.hf_config, "architectures", [])
# Special handling for quantized Mixtral.
# FIXME(woosuk): This is a temporary hack.
if (model_config.quantization is not None
and "MixtralForCausalLM" in architectures):
architectures = ["QuantMixtralForCausalLM"]

for arch in architectures:
model_cls = ModelRegistry.load_model_cls(arch)
if model_cls is not None:
Expand All @@ -34,7 +39,7 @@ def _get_model_architecture(config: PretrainedConfig) -> Type[nn.Module]:

def get_model(model_config: ModelConfig,
lora_config: Optional[LoRAConfig] = None) -> nn.Module:
model_class = _get_model_architecture(model_config.hf_config)
model_class = _get_model_architecture(model_config)

# Get the (maybe quantized) linear method.
linear_method = None
Expand Down
1 change: 1 addition & 0 deletions vllm/model_executor/models/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,7 @@
"LLaMAForCausalLM": ("llama", "LlamaForCausalLM"),
"MistralForCausalLM": ("mistral", "MistralForCausalLM"),
"MixtralForCausalLM": ("mixtral", "MixtralForCausalLM"),
"QuantMixtralForCausalLM": ("mixtral_quant", "MixtralForCausalLM"),
# transformers's mpt class has lower case
"MptForCausalLM": ("mpt", "MPTForCausalLM"),
"MPTForCausalLM": ("mpt", "MPTForCausalLM"),
Expand Down
Loading