manish
commited on
Commit
•
a6c74eb
1
Parent(s):
7222796
add custom handler and gitignore
Browse files- handler.py +26 -0
handler.py
ADDED
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from typing import Dict, List, Any
|
2 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
3 |
+
|
4 |
+
class EndpointHandler():
|
5 |
+
def __init__(self, path=""):
|
6 |
+
# init
|
7 |
+
self.tokenizer = AutoTokenizer.from_pretrained("verseAI/vai-GPT-NeoXT-Chat-Base-20B")
|
8 |
+
self.model = AutoModelForCausalLM.from_pretrained("verseAI/vai-GPT-NeoXT-Chat-Base-20B", device_map="auto", load_in_8bit=True)
|
9 |
+
|
10 |
+
def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
|
11 |
+
"""
|
12 |
+
data args:
|
13 |
+
inputs (:obj: `str`)
|
14 |
+
date (:obj: `str`)
|
15 |
+
Return:
|
16 |
+
A :obj:`list` | `dict`: will be serialized and returned
|
17 |
+
|
18 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
19 |
+
"""
|
20 |
+
# infer
|
21 |
+
inputs = self.tokenizer("<human>: Hello!\n<bot>:", return_tensors='pt').to(self.model.device)
|
22 |
+
outputs = self.model.generate(**inputs, max_new_tokens=10, do_sample=True, temperature=0.8)
|
23 |
+
output_str = self.tokenizer.decode(outputs[0])
|
24 |
+
print(output_str)
|
25 |
+
return output_str
|
26 |
+
|