osanseviero
commited on
Commit
•
4fec958
1
Parent(s):
6979234
Add initial model
Browse files- .gitattributes +2 -0
- hubert_sd.ckpt +3 -0
- model.py +47 -0
.gitattributes
CHANGED
@@ -15,3 +15,5 @@
|
|
15 |
*.pt filter=lfs diff=lfs merge=lfs -text
|
16 |
*.pth filter=lfs diff=lfs merge=lfs -text
|
17 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
15 |
*.pt filter=lfs diff=lfs merge=lfs -text
|
16 |
*.pth filter=lfs diff=lfs merge=lfs -text
|
17 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
18 |
+
*ckpt* filter=lfs diff=lfs merge=lfs -text
|
19 |
+
|
hubert_sd.ckpt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f6ae7e9609674c3aa6104a2a037e85789f0b35e47d232e6754c2012b6bf8ca6d
|
3 |
+
size 31527386
|
model.py
ADDED
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
"""
|
2 |
+
This is just an example of what people would submit for inference.
|
3 |
+
"""
|
4 |
+
import os
|
5 |
+
from typing import Dict
|
6 |
+
|
7 |
+
import torch
|
8 |
+
from s3prl.downstream.runner import Runner
|
9 |
+
|
10 |
+
class PreTrainedModel(Runner):
|
11 |
+
def __init__(self, path=""):
|
12 |
+
"""
|
13 |
+
Initialize downstream model.
|
14 |
+
"""
|
15 |
+
ckp_file = os.path.join(path, "hubert_sd.ckpt")
|
16 |
+
ckp = torch.load(ckp_file, map_location="cpu")
|
17 |
+
ckp["Args"].init_ckpt = ckp_file
|
18 |
+
ckp["Args"].mode = "inference"
|
19 |
+
ckp["Args"].device = "cpu" # Just to try in my computer
|
20 |
+
Runner.__init__(self, ckp["Args"], ckp["Config"])
|
21 |
+
|
22 |
+
def __call__(self, inputs) -> list[int]:
|
23 |
+
"""
|
24 |
+
Args: inputs (:obj:`np.array`): The raw waveform of audio received. By
|
25 |
+
default at 16KHz.
|
26 |
+
Return: A list with logits.
|
27 |
+
"""
|
28 |
+
for entry in self.all_entries:
|
29 |
+
entry.model.eval()
|
30 |
+
|
31 |
+
inputs = [torch.FloatTensor(inputs)]
|
32 |
+
|
33 |
+
with torch.no_grad():
|
34 |
+
features = self.upstream.model(inputs)
|
35 |
+
features = self.featurizer.model(inputs, features)
|
36 |
+
preds = self.downstream.model.inference(features, [])
|
37 |
+
return preds[0]
|
38 |
+
|
39 |
+
"""
|
40 |
+
import io
|
41 |
+
import soundfile as sf
|
42 |
+
from urllib.request import urlopen
|
43 |
+
model = PreTrainedModel()
|
44 |
+
url = "https://huggingface.co/datasets/lewtun/s3prl-sd-dummy/raw/main/audio.wav"
|
45 |
+
data, samplerate = sf.read(io.BytesIO(urlopen(url).read()))
|
46 |
+
print(model(data))
|
47 |
+
"""
|