[Misc] Update config loading for Qwen2-VL and remove Granite (#8837)

This commit is contained in:
Roger Wang
2024-09-26 07:45:30 -07:00
committed by GitHub
parent 7193774b1f
commit 4bb98f2190
7 changed files with 144 additions and 224 deletions
+1 -1
View File
@@ -25,6 +25,7 @@ from typing import Any, Dict, Iterable, List, Optional, Tuple, Union
import torch
from torch import nn
from transformers import GraniteConfig
from vllm.attention import Attention, AttentionMetadata
from vllm.config import CacheConfig, LoRAConfig
@@ -48,7 +49,6 @@ from vllm.model_executor.model_loader.weight_utils import (
default_weight_loader, kv_cache_scales_loader, maybe_remap_kv_scale_name)
from vllm.model_executor.sampling_metadata import SamplingMetadata
from vllm.sequence import IntermediateTensors
from vllm.transformers_utils.configs.granite import GraniteConfig
from vllm.utils import is_hip
from .interfaces import SupportsLoRA
+2 -3
View File
@@ -31,12 +31,9 @@ import torch.nn as nn
import torch.nn.functional as F
from einops import rearrange, repeat
from PIL import Image
from transformers import Qwen2VLConfig
from transformers.image_utils import (get_image_size,
infer_channel_dimension_format,
to_numpy_array)
from transformers.models.qwen2_vl.configuration_qwen2_vl import (
Qwen2VLVisionConfig)
from transformers.models.qwen2_vl.image_processing_qwen2_vl import (
make_batched_images, make_batched_videos, smart_resize)
@@ -66,6 +63,8 @@ from vllm.multimodal.base import MultiModalData
from vllm.multimodal.image import cached_get_image_processor
from vllm.platforms import current_platform
from vllm.sequence import IntermediateTensors, SequenceData
from vllm.transformers_utils.configs.qwen2vl import (Qwen2VLConfig,
Qwen2VLVisionConfig)
from vllm.transformers_utils.processor import get_processor
from vllm.utils import is_cpu