Spaces:
Running
on
Zero
Running
on
Zero
Lord-Raven
commited on
Commit
·
f63295c
1
Parent(s):
2b2a5e4
Messing with fastAPI.
Browse files
app.py
CHANGED
@@ -4,6 +4,7 @@ import gradio
|
|
4 |
import json
|
5 |
import onnxruntime
|
6 |
import time
|
|
|
7 |
from transformers import pipeline
|
8 |
from fastapi import FastAPI
|
9 |
from fastapi.middleware.cors import CORSMiddleware
|
@@ -32,7 +33,8 @@ print(f"CUDA device: {torch.cuda.get_device_name(torch.cuda.current_device())}")
|
|
32 |
model_name = "MoritzLaurer/deberta-v3-base-zeroshot-v2.0"
|
33 |
tokenizer_name = "MoritzLaurer/deberta-v3-base-zeroshot-v2.0"
|
34 |
|
35 |
-
|
|
|
36 |
# classifier = pipeline(task="zero-shot-classification", model=model_name, tokenizer=tokenizer_name)
|
37 |
|
38 |
def classify(data_string, request: gradio.Request):
|
@@ -42,19 +44,27 @@ def classify(data_string, request: gradio.Request):
|
|
42 |
data = json.loads(data_string)
|
43 |
|
44 |
# Prevent batch suggestion warning in log.
|
45 |
-
|
|
|
46 |
|
47 |
# if 'task' in data and data['task'] == 'few_shot_classification':
|
48 |
# return few_shot_classification(data)
|
49 |
# else:
|
50 |
start_time = time.time()
|
51 |
-
result =
|
52 |
-
|
|
|
|
|
|
|
|
|
53 |
return json.dumps(result)
|
54 |
|
|
|
|
|
|
|
55 |
@spaces.GPU(duration=3)
|
56 |
-
def
|
57 |
-
return
|
58 |
|
59 |
def create_sequences(data):
|
60 |
return [data['sequence'] + '\n' + data['hypothesis_template'].format(label) for label in data['candidate_labels']]
|
|
|
4 |
import json
|
5 |
import onnxruntime
|
6 |
import time
|
7 |
+
from datetime import datetime
|
8 |
from transformers import pipeline
|
9 |
from fastapi import FastAPI
|
10 |
from fastapi.middleware.cors import CORSMiddleware
|
|
|
33 |
model_name = "MoritzLaurer/deberta-v3-base-zeroshot-v2.0"
|
34 |
tokenizer_name = "MoritzLaurer/deberta-v3-base-zeroshot-v2.0"
|
35 |
|
36 |
+
classifier_cpu = pipeline(task="zero-shot-classification", model=model_name, tokenizer=tokenizer_name)
|
37 |
+
classifier_gpu = pipeline(task="zero-shot-classification", model=model_name, tokenizer=tokenizer_name, device="cuda:0")
|
38 |
# classifier = pipeline(task="zero-shot-classification", model=model_name, tokenizer=tokenizer_name)
|
39 |
|
40 |
def classify(data_string, request: gradio.Request):
|
|
|
44 |
data = json.loads(data_string)
|
45 |
|
46 |
# Prevent batch suggestion warning in log.
|
47 |
+
classifier_cpu.call_count = 0
|
48 |
+
classifier_gpu.call_count = 0
|
49 |
|
50 |
# if 'task' in data and data['task'] == 'few_shot_classification':
|
51 |
# return few_shot_classification(data)
|
52 |
# else:
|
53 |
start_time = time.time()
|
54 |
+
result = {}
|
55 |
+
if (data['cpu'])
|
56 |
+
result = zero_shot_classification_cpu(data)
|
57 |
+
else
|
58 |
+
result = zero_shot_classification_gpu(data)
|
59 |
+
print(f"Classification @ [{datetime.now().strftime('%Y-%m-%d %H:%M:%S')}] took {time.time() - start_time}.")
|
60 |
return json.dumps(result)
|
61 |
|
62 |
+
def zero_shot_classification_cpu(data):
|
63 |
+
return classifier_cpu(data['sequence'], candidate_labels=data['candidate_labels'], hypothesis_template=data['hypothesis_template'], multi_label=data['multi_label'])
|
64 |
+
|
65 |
@spaces.GPU(duration=3)
|
66 |
+
def zero_shot_classification_gpu(data):
|
67 |
+
return classifier_gpu(data['sequence'], candidate_labels=data['candidate_labels'], hypothesis_template=data['hypothesis_template'], multi_label=data['multi_label'])
|
68 |
|
69 |
def create_sequences(data):
|
70 |
return [data['sequence'] + '\n' + data['hypothesis_template'].format(label) for label in data['candidate_labels']]
|