Pierce Maloney
commited on
Commit
•
a36be93
1
Parent(s):
b873ed7
pipeline trial
Browse files- handler.py +75 -37
handler.py
CHANGED
@@ -1,59 +1,97 @@
|
|
1 |
from typing import Dict, List, Any
|
2 |
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline, StoppingCriteria, StoppingCriteriaList
|
3 |
-
import torch
|
4 |
|
5 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
class EndpointHandler():
|
7 |
def __init__(self, path=""):
|
|
|
8 |
tokenizer = AutoTokenizer.from_pretrained(path)
|
9 |
-
tokenizer.pad_token = tokenizer.eos_token
|
10 |
-
self.model = AutoModelForCausalLM.from_pretrained(path).to('cuda')
|
11 |
self.tokenizer = tokenizer
|
12 |
-
|
|
|
13 |
|
14 |
def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
|
15 |
"""
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
"""
|
22 |
-
torch.cuda.empty_cache()
|
23 |
inputs = data.pop("inputs", data)
|
24 |
additional_bad_words_ids = data.pop("additional_bad_words_ids", [])
|
25 |
|
26 |
-
|
27 |
-
# 3070, 10456, [313, 334] corresponds to "(*", and we do not want to output a comment
|
28 |
-
# 13 is a newline character
|
29 |
-
# [1976, 441, 29889], [4920, 441, 29889] is "Abort." [4920, 18054, 29889] is "Aborted."
|
30 |
-
# [2087, 29885, 4430, 29889] is "Admitted."
|
31 |
bad_words_ids = [[3070], [313, 334], [10456], [13], [1976, 441, 29889], [2087, 29885, 4430, 29889], [4920, 441], [4920, 441, 29889], [4920, 18054, 29889]]
|
32 |
bad_words_ids.extend(additional_bad_words_ids)
|
33 |
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
|
|
|
|
41 |
|
42 |
-
|
43 |
-
|
44 |
-
|
45 |
-
input_ids,
|
46 |
-
max_length=max_length, # 50 new tokens
|
47 |
-
bad_words_ids=bad_words_ids,
|
48 |
-
temperature=1,
|
49 |
-
top_k=40,
|
50 |
-
do_sample=True,
|
51 |
-
stopping_criteria=self.stopping_criteria,
|
52 |
-
)
|
53 |
-
|
54 |
-
generated_text = self.tokenizer.decode(generated_ids[0][input_ids.shape[1]:], skip_special_tokens=True)
|
55 |
-
prediction = [{"generated_text": generated_text, "generated_ids": generated_ids[0][input_ids.shape[1]:].tolist()}]
|
56 |
-
return prediction
|
57 |
|
58 |
|
59 |
class StopAtPeriodCriteria(StoppingCriteria):
|
|
|
1 |
from typing import Dict, List, Any
|
2 |
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline, StoppingCriteria, StoppingCriteriaList
|
|
|
3 |
|
4 |
|
5 |
+
# class EndpointHandler():
|
6 |
+
# def __init__(self, path=""):
|
7 |
+
# tokenizer = AutoTokenizer.from_pretrained(path)
|
8 |
+
# tokenizer.pad_token = tokenizer.eos_token
|
9 |
+
# self.model = AutoModelForCausalLM.from_pretrained(path).to('cuda')
|
10 |
+
# self.tokenizer = tokenizer
|
11 |
+
# self.stopping_criteria = StoppingCriteriaList([StopAtPeriodCriteria(tokenizer)])
|
12 |
+
|
13 |
+
# def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
|
14 |
+
# """
|
15 |
+
# data args:
|
16 |
+
# inputs (:obj: `str`)
|
17 |
+
# kwargs
|
18 |
+
# Return:
|
19 |
+
# A :obj:`list` | `dict`: will be serialized and returned
|
20 |
+
# """
|
21 |
+
# inputs = data.pop("inputs", data)
|
22 |
+
# additional_bad_words_ids = data.pop("additional_bad_words_ids", [])
|
23 |
+
|
24 |
+
|
25 |
+
# # 3070, 10456, [313, 334] corresponds to "(*", and we do not want to output a comment
|
26 |
+
# # 13 is a newline character
|
27 |
+
# # [1976, 441, 29889], [4920, 441, 29889] is "Abort." [4920, 18054, 29889] is "Aborted."
|
28 |
+
# # [2087, 29885, 4430, 29889] is "Admitted."
|
29 |
+
# bad_words_ids = [[3070], [313, 334], [10456], [13], [1976, 441, 29889], [2087, 29885, 4430, 29889], [4920, 441], [4920, 441, 29889], [4920, 18054, 29889]]
|
30 |
+
# bad_words_ids.extend(additional_bad_words_ids)
|
31 |
+
|
32 |
+
# input_ids = self.tokenizer.encode(inputs, return_tensors="pt").to('cuda')
|
33 |
+
# max_generation_length = 75 # Desired number of tokens to generate
|
34 |
+
# # max_input_length = 4092 - max_generation_length # Maximum input length to allow space for generation
|
35 |
+
|
36 |
+
# # # Truncate input_ids to the most recent tokens that fit within the max_input_length
|
37 |
+
# # if input_ids.shape[1] > max_input_length:
|
38 |
+
# # input_ids = input_ids[:, -max_input_length:]
|
39 |
+
|
40 |
+
# max_length = input_ids.shape[1] + max_generation_length
|
41 |
+
|
42 |
+
# generated_ids = self.model.generate(
|
43 |
+
# input_ids,
|
44 |
+
# max_length=max_length, # 50 new tokens
|
45 |
+
# bad_words_ids=bad_words_ids,
|
46 |
+
# temperature=1,
|
47 |
+
# top_k=40,
|
48 |
+
# do_sample=True,
|
49 |
+
# stopping_criteria=self.stopping_criteria,
|
50 |
+
# )
|
51 |
+
|
52 |
+
# generated_text = self.tokenizer.decode(generated_ids[0][input_ids.shape[1]:], skip_special_tokens=True)
|
53 |
+
# prediction = [{"generated_text": generated_text, "generated_ids": generated_ids[0][input_ids.shape[1]:].tolist()}]
|
54 |
+
# return prediction
|
55 |
+
|
56 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
|
57 |
+
|
58 |
class EndpointHandler():
|
59 |
def __init__(self, path=""):
|
60 |
+
self.model_path = path
|
61 |
tokenizer = AutoTokenizer.from_pretrained(path)
|
62 |
+
tokenizer.pad_token = self.tokenizer.eos_token
|
|
|
63 |
self.tokenizer = tokenizer
|
64 |
+
# Initialize the pipeline for text generation
|
65 |
+
self.text_generation_pipeline = pipeline("text-generation", model=path, tokenizer=self.tokenizer, device=0) # device=0 for CUDA
|
66 |
|
67 |
def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
|
68 |
"""
|
69 |
+
data args:
|
70 |
+
inputs (:obj: `str`)
|
71 |
+
kwargs
|
72 |
+
Return:
|
73 |
+
A :obj:`list` | `dict`: will be serialized and returned
|
74 |
"""
|
|
|
75 |
inputs = data.pop("inputs", data)
|
76 |
additional_bad_words_ids = data.pop("additional_bad_words_ids", [])
|
77 |
|
78 |
+
# Define bad words to avoid in the output
|
|
|
|
|
|
|
|
|
79 |
bad_words_ids = [[3070], [313, 334], [10456], [13], [1976, 441, 29889], [2087, 29885, 4430, 29889], [4920, 441], [4920, 441, 29889], [4920, 18054, 29889]]
|
80 |
bad_words_ids.extend(additional_bad_words_ids)
|
81 |
|
82 |
+
# Generate text using the pipeline
|
83 |
+
generation_kwargs = {
|
84 |
+
"max_length": 75, # Adjust as needed
|
85 |
+
"temperature": 1,
|
86 |
+
"top_k": 40,
|
87 |
+
"bad_words_ids": bad_words_ids,
|
88 |
+
"pad_token_id": self.tokenizer.eos_token_id # Ensure padding with EOS token
|
89 |
+
}
|
90 |
+
generated_outputs = self.text_generation_pipeline(inputs, **generation_kwargs)
|
91 |
|
92 |
+
# Format the output
|
93 |
+
predictions = [{"generated_text": output["generated_text"]} for output in generated_outputs]
|
94 |
+
return predictions
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
95 |
|
96 |
|
97 |
class StopAtPeriodCriteria(StoppingCriteria):
|