Christopher Glaze
commited on
Commit
·
c5d744a
1
Parent(s):
6440c9b
Update device for simcse generator
Browse files- handler.py +7 -9
handler.py
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
|
2 |
-
from typing import Dict,
|
3 |
from pathlib import Path
|
4 |
import json
|
5 |
import joblib
|
@@ -12,11 +12,12 @@ from sklearn.base import TransformerMixin
|
|
12 |
|
13 |
class SimcseGenerator(TransformerMixin):
|
14 |
def __init__(
|
15 |
-
self,
|
16 |
) -> None:
|
17 |
|
18 |
self.model_name = model_name
|
19 |
-
|
|
|
20 |
|
21 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
22 |
model = AutoModel.from_pretrained(model_name).to(self.device)
|
@@ -53,13 +54,10 @@ class SimcseGenerator(TransformerMixin):
|
|
53 |
return embeddings
|
54 |
|
55 |
class EndpointHandler():
|
56 |
-
def __init__(self
|
57 |
-
# Preload all the elements you are going to need at inference.
|
58 |
-
# pseudo:
|
59 |
-
# self.model= load_model(path)
|
60 |
|
61 |
local_path = Path(__file__).parent
|
62 |
-
|
63 |
with open(local_path/'stop_words.json','r') as fp:
|
64 |
self.stop_words = set(json.load(fp))
|
65 |
|
@@ -70,7 +68,7 @@ class EndpointHandler():
|
|
70 |
self.instruction_pipeline = joblib.load(local_path/'instruction_classification_pipeline.joblib')
|
71 |
self.response_pipeline = joblib.load(local_path/'response_quality_pipeline.joblib')
|
72 |
|
73 |
-
self.simcse_generator = SimcseGenerator(
|
74 |
|
75 |
def _get_stop_word_proportion(self, s):
|
76 |
s = s.lower()
|
|
|
1 |
|
2 |
+
from typing import Dict, Union, Optional
|
3 |
from pathlib import Path
|
4 |
import json
|
5 |
import joblib
|
|
|
12 |
|
13 |
class SimcseGenerator(TransformerMixin):
|
14 |
def __init__(
|
15 |
+
self, batch_size: int =16, model_name: str = "princeton-nlp/unsup-simcse-bert-base-uncased"
|
16 |
) -> None:
|
17 |
|
18 |
self.model_name = model_name
|
19 |
+
|
20 |
+
self.device = torch.device('cpu')
|
21 |
|
22 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
23 |
model = AutoModel.from_pretrained(model_name).to(self.device)
|
|
|
54 |
return embeddings
|
55 |
|
56 |
class EndpointHandler():
|
57 |
+
def __init__(self):
|
|
|
|
|
|
|
58 |
|
59 |
local_path = Path(__file__).parent
|
60 |
+
|
61 |
with open(local_path/'stop_words.json','r') as fp:
|
62 |
self.stop_words = set(json.load(fp))
|
63 |
|
|
|
68 |
self.instruction_pipeline = joblib.load(local_path/'instruction_classification_pipeline.joblib')
|
69 |
self.response_pipeline = joblib.load(local_path/'response_quality_pipeline.joblib')
|
70 |
|
71 |
+
self.simcse_generator = SimcseGenerator()
|
72 |
|
73 |
def _get_stop_word_proportion(self, s):
|
74 |
s = s.lower()
|