Spaces:
Running
Running
⚡️ improve performance, enable longer text
Browse filesSigned-off-by: peter szemraj <peterszemraj@gmail.com>
- app.py +3 -3
- summarize.py +11 -0
- utils.py +12 -3
app.py
CHANGED
@@ -73,11 +73,11 @@ def predict(
|
|
73 |
batch_length=token_batch_length,
|
74 |
**settings,
|
75 |
)
|
76 |
-
|
77 |
del model
|
78 |
del tokenizer
|
79 |
gc.collect()
|
80 |
-
|
81 |
return summaries
|
82 |
|
83 |
|
@@ -89,7 +89,7 @@ def proc_submission(
|
|
89 |
length_penalty: float,
|
90 |
repetition_penalty: float,
|
91 |
no_repeat_ngram_size: int,
|
92 |
-
max_input_length: int =
|
93 |
):
|
94 |
"""
|
95 |
proc_submission - a helper function for the gradio module to process submissions
|
|
|
73 |
batch_length=token_batch_length,
|
74 |
**settings,
|
75 |
)
|
76 |
+
|
77 |
del model
|
78 |
del tokenizer
|
79 |
gc.collect()
|
80 |
+
|
81 |
return summaries
|
82 |
|
83 |
|
|
|
89 |
length_penalty: float,
|
90 |
repetition_penalty: float,
|
91 |
no_repeat_ngram_size: int,
|
92 |
+
max_input_length: int = 4096,
|
93 |
):
|
94 |
"""
|
95 |
proc_submission - a helper function for the gradio module to process submissions
|
summarize.py
CHANGED
@@ -6,6 +6,8 @@ import torch
|
|
6 |
from tqdm.auto import tqdm
|
7 |
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
8 |
|
|
|
|
|
9 |
|
10 |
def load_model_and_tokenizer(model_name: str) -> tuple:
|
11 |
"""
|
@@ -24,6 +26,15 @@ def load_model_and_tokenizer(model_name: str) -> tuple:
|
|
24 |
|
25 |
logging.info(f"Loaded model {model_name} to {device}")
|
26 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
return model, tokenizer
|
28 |
|
29 |
|
|
|
6 |
from tqdm.auto import tqdm
|
7 |
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
8 |
|
9 |
+
from utils import validate_pytorch2
|
10 |
+
|
11 |
|
12 |
def load_model_and_tokenizer(model_name: str) -> tuple:
|
13 |
"""
|
|
|
26 |
|
27 |
logging.info(f"Loaded model {model_name} to {device}")
|
28 |
|
29 |
+
if validate_pytorch2():
|
30 |
+
try:
|
31 |
+
logging.info("Compiling model with Torch 2.0")
|
32 |
+
model = torch.compile(model)
|
33 |
+
except Exception as e:
|
34 |
+
logging.warning(f"Could not compile model with Torch 2.0: {e}")
|
35 |
+
else:
|
36 |
+
logging.info("Torch 2.0 not detected, skipping compilation")
|
37 |
+
|
38 |
return model, tokenizer
|
39 |
|
40 |
|
utils.py
CHANGED
@@ -3,10 +3,20 @@
|
|
3 |
"""
|
4 |
|
5 |
import re
|
6 |
-
|
7 |
from datetime import datetime
|
|
|
|
|
|
|
8 |
from natsort import natsorted
|
9 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
|
11 |
|
12 |
def get_timestamp() -> str:
|
@@ -114,7 +124,6 @@ def saves_summary(summarize_output, outpath: str or Path = None, add_signature=T
|
|
114 |
outpath,
|
115 |
"a",
|
116 |
) as fo:
|
117 |
-
|
118 |
fo.write("\n" * 3)
|
119 |
fo.write(f"\n\nSection Scores:\n")
|
120 |
fo.writelines(scores_text)
|
|
|
3 |
"""
|
4 |
|
5 |
import re
|
6 |
+
import subprocess
|
7 |
from datetime import datetime
|
8 |
+
from pathlib import Path
|
9 |
+
|
10 |
+
import torch
|
11 |
from natsort import natsorted
|
12 |
+
|
13 |
+
|
14 |
+
def validate_pytorch2(torch_version: str = None):
|
15 |
+
torch_version = torch.__version__ if torch_version is None else torch_version
|
16 |
+
|
17 |
+
pattern = r"^2\.\d+(\.\d+)*"
|
18 |
+
|
19 |
+
return True if re.match(pattern, torch_version) else False
|
20 |
|
21 |
|
22 |
def get_timestamp() -> str:
|
|
|
124 |
outpath,
|
125 |
"a",
|
126 |
) as fo:
|
|
|
127 |
fo.write("\n" * 3)
|
128 |
fo.write(f"\n\nSection Scores:\n")
|
129 |
fo.writelines(scores_text)
|