|
from typing import List |
|
from transformers import CLIPTokenizer, CLIPModel |
|
|
|
|
|
class EndpointHandler(): |
|
def __init__(self, path=""): |
|
|
|
hf_model_path = "openai/clip-vit-large-patch14" |
|
self.model = CLIPModel.from_pretrained(hf_model_path) |
|
self.tokenizer = CLIPTokenizer.from_pretrained(hf_model_path) |
|
|
|
def __call__(self, inputs: str) -> List[float]: |
|
""" |
|
data args: |
|
inputs (:obj: `str` | `PIL.Image` | `np.array`) |
|
kwargs |
|
Return: |
|
A :obj:`list` | `dict`: will be serialized and returned |
|
""" |
|
|
|
|
|
|
|
token_inputs = self.tokenizer([inputs], padding=True, return_tensors="pt") |
|
query_embed = self.model.get_text_features(**token_inputs) |
|
np_query_embed = query_embed.detach().cpu().numpy()[0].tolist() |
|
return np_query_embed |
|
|
|
|
|
|
|
|
|
|
|
|