Spaces:
Sleeping
Sleeping
lewiswu1209
commited on
Commit
·
b56b764
1
Parent(s):
c1e6869
add fill_blanks.py
Browse files- bot/skills/fill_blanks.py +46 -0
bot/skills/fill_blanks.py
ADDED
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
import os
|
3 |
+
import re
|
4 |
+
import json
|
5 |
+
import requests
|
6 |
+
|
7 |
+
class Skill:
|
8 |
+
def __init__(self:object) -> None:
|
9 |
+
pass
|
10 |
+
|
11 |
+
def set_tokenizer(self:object, tokenizer:object):
|
12 |
+
self.tokenizer = tokenizer
|
13 |
+
|
14 |
+
def predict(self, question:str, seed=1234, out_seq_length=256, min_gen_length=1, sampling_strategy="BaseStrategy", num_beams=2, length_penalty=1, no_repeat_ngram_size=3, temperature=0.1, topk=1, topp=0.1):
|
15 |
+
url = 'https://pretrain.aminer.cn/api/v2/completions'
|
16 |
+
prompt = question
|
17 |
+
payload = json.dumps({
|
18 |
+
"apikey": os.environ.get("WINNIE_APIKEY"),
|
19 |
+
"apisecret": os.environ.get("WINNIE_APISECRET"),
|
20 |
+
"model": "glm",
|
21 |
+
"language": "zh-CN",
|
22 |
+
"prompt": prompt,
|
23 |
+
"temperature": temperature,
|
24 |
+
"top_k": topk,
|
25 |
+
"top_p": topp,
|
26 |
+
"max_tokens": out_seq_length,
|
27 |
+
"stop": ["\n"],
|
28 |
+
"presence_penalty": 2,
|
29 |
+
"frequency_penalty": 2
|
30 |
+
})
|
31 |
+
|
32 |
+
headers = {
|
33 |
+
'Content-Type': 'application/json'
|
34 |
+
}
|
35 |
+
response = requests.request("POST", url, headers=headers, data=payload).json()
|
36 |
+
if "output" in response["result"]:
|
37 |
+
answer = response["result"]["output"]["raw"].split("<|startofpiece|>")[-1]
|
38 |
+
return answer
|
39 |
+
|
40 |
+
def process(self:object, input_txt:str, history_list:list, role_card:dict):
|
41 |
+
output_text:str = None
|
42 |
+
if input_txt.find("[MASK]") != -1:
|
43 |
+
history_list.append( self.tokenizer.encode(input_txt, add_special_tokens=False) )
|
44 |
+
output_text = self.predict(input_txt)
|
45 |
+
history_list.append( self.tokenizer.encode(output_text, add_special_tokens=False) )
|
46 |
+
return output_text, history_list, role_card
|