[mypy] Misc. typing improvements (#7417)
This commit is contained in:
@@ -165,7 +165,7 @@ def get_internvl_num_patches(image_size: int, patch_size: int,
|
||||
|
||||
|
||||
def get_max_internvl_image_tokens(ctx: InputContext):
|
||||
hf_config = ctx.get_hf_config(PretrainedConfig)
|
||||
hf_config = ctx.get_hf_config()
|
||||
vision_config = hf_config.vision_config
|
||||
|
||||
use_thumbnail = hf_config.use_thumbnail
|
||||
@@ -187,7 +187,7 @@ def input_processor_for_internvl(ctx: InputContext, llm_inputs: LLMInputs):
|
||||
return llm_inputs
|
||||
|
||||
model_config = ctx.model_config
|
||||
hf_config = ctx.get_hf_config(PretrainedConfig)
|
||||
hf_config = ctx.get_hf_config()
|
||||
vision_config = hf_config.vision_config
|
||||
|
||||
image_size = vision_config.image_size
|
||||
@@ -260,7 +260,7 @@ def dummy_data_for_internvl(ctx: InputContext, seq_len: int):
|
||||
|
||||
image_feature_size = get_max_internvl_image_tokens(ctx)
|
||||
model_config = ctx.model_config
|
||||
hf_config = ctx.get_hf_config(PretrainedConfig)
|
||||
hf_config = ctx.get_hf_config()
|
||||
vision_config = hf_config.vision_config
|
||||
tokenizer = cached_get_tokenizer(model_config.tokenizer,
|
||||
trust_remote_code=True)
|
||||
|
||||
@@ -34,7 +34,7 @@ import torch.types
|
||||
from PIL import Image
|
||||
from torch import nn
|
||||
from torch.nn.init import trunc_normal_
|
||||
from transformers.configuration_utils import PretrainedConfig
|
||||
from transformers import PretrainedConfig
|
||||
|
||||
from vllm.attention import AttentionMetadata
|
||||
from vllm.config import CacheConfig, MultiModalConfig
|
||||
@@ -404,7 +404,7 @@ def get_version_by_config(config: PretrainedConfig) -> Tuple[int, ...]:
|
||||
|
||||
|
||||
def get_max_minicpmv_image_tokens(ctx: InputContext):
|
||||
hf_config = ctx.get_hf_config(PretrainedConfig)
|
||||
hf_config = ctx.get_hf_config()
|
||||
return getattr(hf_config, "query_num", 64)
|
||||
|
||||
|
||||
@@ -420,7 +420,7 @@ def dummy_image_for_minicpmv(hf_config: PretrainedConfig):
|
||||
|
||||
|
||||
def dummy_data_for_minicpmv(ctx: InputContext, seq_len: int):
|
||||
hf_config = ctx.get_hf_config(PretrainedConfig)
|
||||
hf_config = ctx.get_hf_config()
|
||||
|
||||
seq_data = dummy_seq_data_for_minicpmv(seq_len)
|
||||
mm_data = dummy_image_for_minicpmv(hf_config)
|
||||
|
||||
@@ -341,7 +341,7 @@ def get_phi3v_image_feature_size(
|
||||
def get_max_phi3v_image_tokens(ctx: InputContext):
|
||||
|
||||
return get_phi3v_image_feature_size(
|
||||
ctx.get_hf_config(PretrainedConfig),
|
||||
ctx.get_hf_config(),
|
||||
input_height=MAX_IMAGE_FEATURE_SIZE_HEIGHT,
|
||||
input_width=MAX_IMAGE_FEATURE_SIZE_WIDTH,
|
||||
)
|
||||
@@ -391,7 +391,7 @@ def input_processor_for_phi3v(ctx: InputContext, llm_inputs: LLMInputs):
|
||||
return llm_inputs
|
||||
|
||||
model_config = ctx.model_config
|
||||
hf_config = ctx.get_hf_config(PretrainedConfig)
|
||||
hf_config = ctx.get_hf_config()
|
||||
|
||||
image_data = multi_modal_data["image"]
|
||||
if isinstance(image_data, Image.Image):
|
||||
|
||||
Reference in New Issue
Block a user