Update handler.py
Browse files- handler.py +29 -22
handler.py
CHANGED
@@ -3,23 +3,44 @@ import torch
|
|
3 |
from transformers import AutoModelForCausalLM, AutoTokenizer
|
4 |
from typing import Any, Dict
|
5 |
|
6 |
-
class
|
7 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8 |
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
self.model = AutoModelForCausalLM.from_pretrained(
|
10 |
path,
|
11 |
torch_dtype=torch_dtype,
|
12 |
trust_remote_code=trust_remote_code
|
13 |
)
|
14 |
-
tokenizer = AutoTokenizer.from_pretrained(
|
15 |
-
|
|
|
|
|
16 |
if tokenizer.pad_token_id is None:
|
17 |
warnings.warn(
|
18 |
"pad_token_id is not set for the tokenizer. Using eos_token_id as pad_token_id."
|
19 |
)
|
20 |
tokenizer.pad_token = tokenizer.eos_token
|
21 |
-
|
22 |
-
tokenizer.padding_side = "right"
|
23 |
self.tokenizer = tokenizer
|
24 |
|
25 |
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
@@ -39,21 +60,7 @@ class EndpointHandler:
|
|
39 |
}
|
40 |
|
41 |
def format_instruction(self, instruction):
|
42 |
-
|
43 |
-
RESPONSE_KEY = "### Response:"
|
44 |
-
END_KEY = "### End"
|
45 |
-
INTRO_BLURB = "Below is an instruction that describes a task. Write a response that appropriately completes the request."
|
46 |
-
PROMPT_FOR_GENERATION_FORMAT = """{intro}
|
47 |
-
{instruction_key}
|
48 |
-
{instruction}
|
49 |
-
{response_key}
|
50 |
-
""".format(
|
51 |
-
intro=INTRO_BLURB,
|
52 |
-
instruction_key=INSTRUCTION_KEY,
|
53 |
-
instruction="{instruction}",
|
54 |
-
response_key=RESPONSE_KEY,
|
55 |
-
)
|
56 |
-
return PROMPT_FOR_GENERATION_FORMAT.format(instruction=instruction)
|
57 |
|
58 |
def __call__(self, data: Dict[str, Any]) -> Dict[str, str]:
|
59 |
# process input
|
@@ -61,7 +68,7 @@ class EndpointHandler:
|
|
61 |
parameters = data.pop("parameters", None)
|
62 |
|
63 |
# preprocess
|
64 |
-
s =
|
65 |
input_ids = self.tokenizer(s, return_tensors="pt").input_ids.to(self.device)
|
66 |
gkw = {**self.generate_kwargs, **parameters}
|
67 |
# pass inputs with all kwargs in data
|
|
|
3 |
from transformers import AutoModelForCausalLM, AutoTokenizer
|
4 |
from typing import Any, Dict
|
5 |
|
6 |
+
class InstructionTextGenerationPipeline:
|
7 |
+
INSTRUCTION_KEY = "### Instruction:"
|
8 |
+
RESPONSE_KEY = "### Response:"
|
9 |
+
END_KEY = "### End"
|
10 |
+
INTRO_BLURB = "Below is an instruction that describes a task. Write a response that appropriately completes the request."
|
11 |
+
PROMPT_FOR_GENERATION_FORMAT = """{intro}
|
12 |
+
{instruction_key}
|
13 |
+
{instruction}
|
14 |
+
{response_key}
|
15 |
+
""".format(
|
16 |
+
intro=INTRO_BLURB,
|
17 |
+
instruction_key=INSTRUCTION_KEY,
|
18 |
+
instruction="{instruction}",
|
19 |
+
response_key=RESPONSE_KEY,
|
20 |
+
)
|
21 |
|
22 |
+
def __init__(
|
23 |
+
self,
|
24 |
+
path,
|
25 |
+
torch_dtype=torch.bfloat16,
|
26 |
+
trust_remote_code=True,
|
27 |
+
) -> None:
|
28 |
self.model = AutoModelForCausalLM.from_pretrained(
|
29 |
path,
|
30 |
torch_dtype=torch_dtype,
|
31 |
trust_remote_code=trust_remote_code
|
32 |
)
|
33 |
+
tokenizer = AutoTokenizer.from_pretrained(
|
34 |
+
"mosaicml/mpt-7b-instruct",
|
35 |
+
trust_remote_code=trust_remote_code
|
36 |
+
)
|
37 |
if tokenizer.pad_token_id is None:
|
38 |
warnings.warn(
|
39 |
"pad_token_id is not set for the tokenizer. Using eos_token_id as pad_token_id."
|
40 |
)
|
41 |
tokenizer.pad_token = tokenizer.eos_token
|
42 |
+
|
43 |
+
tokenizer.padding_side = "right"
|
44 |
self.tokenizer = tokenizer
|
45 |
|
46 |
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
|
|
60 |
}
|
61 |
|
62 |
def format_instruction(self, instruction):
|
63 |
+
return self.PROMPT_FOR_GENERATION_FORMAT.format(instruction=instruction)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
64 |
|
65 |
def __call__(self, data: Dict[str, Any]) -> Dict[str, str]:
|
66 |
# process input
|
|
|
68 |
parameters = data.pop("parameters", None)
|
69 |
|
70 |
# preprocess
|
71 |
+
s = self.format_instruction(instruction=inputs)
|
72 |
input_ids = self.tokenizer(s, return_tensors="pt").input_ids.to(self.device)
|
73 |
gkw = {**self.generate_kwargs, **parameters}
|
74 |
# pass inputs with all kwargs in data
|