Spaces:
Sleeping
Sleeping
Commit
·
947409c
1
Parent(s):
2792e5a
chore: removed unused pipeline
Browse files- prediction.py +0 -24
prediction.py
CHANGED
@@ -8,7 +8,6 @@ from xml.parsers.expat import model
|
|
8 |
if os.environ.get("SPACES_ZERO_GPU") is not None:
|
9 |
import spaces
|
10 |
else:
|
11 |
-
|
12 |
class spaces:
|
13 |
@staticmethod
|
14 |
def GPU(func):
|
@@ -67,11 +66,6 @@ class ModelPrediction:
|
|
67 |
"{db_schema}\n"
|
68 |
)
|
69 |
|
70 |
-
def _reset_pipeline(self, model_name):
|
71 |
-
if self._model_name != model_name:
|
72 |
-
self._model_name = model_name
|
73 |
-
self._pipeline = None
|
74 |
-
|
75 |
@staticmethod
|
76 |
def _extract_answer_from_pred(pred: str) -> str:
|
77 |
# extract with regex everything is between <answer> and </answer>
|
@@ -120,24 +114,6 @@ class ModelPrediction:
|
|
120 |
"cost": response._hidden_params["response_cost"],
|
121 |
}
|
122 |
|
123 |
-
@spaces.GPU(duration=20)
|
124 |
-
def predict_with_hf(self, prompt, model_name): # -> dict[str, Any | float]:
|
125 |
-
start_time = time.time()
|
126 |
-
outputs = pipeline(
|
127 |
-
[{"role": "user", "content": prompt}],
|
128 |
-
max_new_tokens=256,
|
129 |
-
)
|
130 |
-
end_time = time.time()
|
131 |
-
elapsed_time = end_time - start_time
|
132 |
-
# inference endpoint costs HF per Hour 3.6$/h -> 0.001 $ per second
|
133 |
-
cost_per_second=0.001
|
134 |
-
response = outputs[0]["generated_text"][-1]['content']
|
135 |
-
# print(response)
|
136 |
-
return {
|
137 |
-
"response": response,
|
138 |
-
"cost": elapsed_time * cost_per_second
|
139 |
-
}
|
140 |
-
|
141 |
def _init_model_prediction(self, model_name):
|
142 |
predict_fun = self.predict_with_api
|
143 |
if "gpt-3.5" in model_name:
|
|
|
8 |
if os.environ.get("SPACES_ZERO_GPU") is not None:
|
9 |
import spaces
|
10 |
else:
|
|
|
11 |
class spaces:
|
12 |
@staticmethod
|
13 |
def GPU(func):
|
|
|
66 |
"{db_schema}\n"
|
67 |
)
|
68 |
|
|
|
|
|
|
|
|
|
|
|
69 |
@staticmethod
|
70 |
def _extract_answer_from_pred(pred: str) -> str:
|
71 |
# extract with regex everything is between <answer> and </answer>
|
|
|
114 |
"cost": response._hidden_params["response_cost"],
|
115 |
}
|
116 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
117 |
def _init_model_prediction(self, model_name):
|
118 |
predict_fun = self.predict_with_api
|
119 |
if "gpt-3.5" in model_name:
|