Spaces:
Sleeping
Sleeping
Commit
•
ce1fac2
1
Parent(s):
b9dee86
refactoring...
Browse files- .DS_Store +0 -0
- .gitignore +2 -0
- app.py +45 -118
- assets/.DS_Store +0 -0
- banner.png +0 -0
- constants.py +16 -12
- init.py +0 -93
- leaderboard.csv +3 -0
- requested_models.txt +3 -0
- utils/__init__.py +0 -0
- utils/display.py +10 -0
- utils/eval_request.py +62 -0
- utils_display.py +0 -40
.DS_Store
DELETED
Binary file (6.15 kB)
|
|
.gitignore
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
__pycache__
|
2 |
+
.DS_Store
|
app.py
CHANGED
@@ -1,148 +1,75 @@
|
|
1 |
import gradio as gr
|
2 |
import pandas as pd
|
3 |
-
import json
|
4 |
-
from constants import BANNER, INTRODUCTION_TEXT, CITATION_TEXT, METRICS_TAB_TEXT, DIR_OUTPUT_REQUESTS
|
5 |
-
from init import is_model_on_hub, upload_file, load_all_info_from_dataset_hub
|
6 |
-
from utils_display import AutoEvalColumn, fields, make_clickable_model, styled_error, styled_message
|
7 |
-
from datetime import datetime, timezone
|
8 |
|
9 |
-
|
|
|
10 |
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
"Earnings22 WER": "Earnings22",
|
17 |
-
"Gigaspeech WER": "Gigaspeech",
|
18 |
-
"LS Clean WER": "LS Clean",
|
19 |
-
"LS Other WER": "LS Other",
|
20 |
-
"SPGISpeech WER": "SPGISpeech",
|
21 |
-
"Tedlium WER": "Tedlium",
|
22 |
-
"Voxpopuli WER": "Voxpopuli",
|
23 |
-
"Common Voice WER": "Common Voice 9"}
|
24 |
|
25 |
-
|
26 |
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
# Get csv with data and parse columns
|
31 |
-
original_df = pd.read_csv(csv_results)
|
32 |
-
|
33 |
-
# Formats the columns
|
34 |
-
def formatter(x):
|
35 |
-
if type(x) is str:
|
36 |
-
x = x
|
37 |
-
else:
|
38 |
-
x = round(x, 2)
|
39 |
-
return x
|
40 |
-
|
41 |
-
for col in original_df.columns:
|
42 |
if col == "model":
|
43 |
-
|
44 |
else:
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
TYPES = [c.type for c in fields(AutoEvalColumn)]
|
52 |
-
|
53 |
-
|
54 |
-
def request_model(model_text, chbcoco2017):
|
55 |
-
|
56 |
-
# Determine the selected checkboxes
|
57 |
-
dataset_selection = []
|
58 |
-
if chbcoco2017:
|
59 |
-
dataset_selection.append("ESB Datasets tests only")
|
60 |
-
|
61 |
-
if len(dataset_selection) == 0:
|
62 |
-
return styled_error("You need to select at least one dataset")
|
63 |
-
|
64 |
-
base_model_on_hub, error_msg = is_model_on_hub(model_text)
|
65 |
|
66 |
-
|
67 |
-
return styled_error(f"Base model '{model_text}' {error_msg}")
|
68 |
-
|
69 |
-
# Construct the output dictionary
|
70 |
-
current_time = datetime.now(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ")
|
71 |
-
required_datasets = ', '.join(dataset_selection)
|
72 |
-
eval_entry = {
|
73 |
-
"date": current_time,
|
74 |
-
"model": model_text,
|
75 |
-
"datasets_selected": required_datasets
|
76 |
-
}
|
77 |
-
|
78 |
-
# Prepare file path
|
79 |
-
DIR_OUTPUT_REQUESTS.mkdir(parents=True, exist_ok=True)
|
80 |
-
|
81 |
-
fn_datasets = '@ '.join(dataset_selection)
|
82 |
-
filename = model_text.replace("/","@") + "@@" + fn_datasets
|
83 |
-
if filename in requested_models:
|
84 |
-
return styled_error(f"A request for this model '{model_text}' and dataset(s) was already made.")
|
85 |
-
try:
|
86 |
-
filename_ext = filename + ".txt"
|
87 |
-
out_filepath = DIR_OUTPUT_REQUESTS / filename_ext
|
88 |
-
|
89 |
-
# Write the results to a text file
|
90 |
-
with open(out_filepath, "w") as f:
|
91 |
-
f.write(json.dumps(eval_entry))
|
92 |
-
|
93 |
-
upload_file(filename, out_filepath)
|
94 |
-
|
95 |
-
# Include file in the list of uploaded files
|
96 |
-
requested_models.append(filename)
|
97 |
-
|
98 |
-
# Remove the local file
|
99 |
-
out_filepath.unlink()
|
100 |
|
101 |
-
|
102 |
-
|
103 |
-
return styled_error(f"Error submitting request!")
|
104 |
-
|
105 |
-
with gr.Blocks() as demo:
|
106 |
-
gr.HTML(BANNER, elem_id="banner")
|
107 |
-
gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text")
|
108 |
|
109 |
with gr.Tabs(elem_classes="tab-buttons") as tabs:
|
|
|
110 |
with gr.TabItem("🏅 Leaderboard", elem_id="od-benchmark-tab-table", id=0):
|
111 |
leaderboard_table = gr.components.Dataframe(
|
112 |
-
|
113 |
-
|
114 |
-
|
115 |
-
|
116 |
-
|
117 |
-
|
118 |
-
)
|
119 |
|
120 |
with gr.TabItem("📈 Metrics", elem_id="od-benchmark-tab-table", id=1):
|
121 |
-
gr.Markdown(METRICS_TAB_TEXT, elem_classes="markdown-text")
|
122 |
|
123 |
-
with gr.TabItem("
|
124 |
with gr.Column():
|
125 |
-
gr.Markdown("#
|
126 |
with gr.Column():
|
127 |
-
gr.Markdown("Select a dataset:", elem_classes="markdown-text")
|
128 |
with gr.Column():
|
129 |
-
|
130 |
-
chb_coco2017 = gr.Checkbox(label="COCO validation 2017 dataset", visible=False, value=True, interactive=False)
|
131 |
with gr.Column():
|
132 |
-
|
133 |
btn_submitt = gr.Button(value="🚀 Request")
|
134 |
-
btn_submitt.click(request_model,
|
135 |
-
[model_name_textbox, chb_coco2017],
|
136 |
-
mdw_submission_result)
|
137 |
|
138 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
139 |
|
140 |
with gr.Row():
|
141 |
with gr.Accordion("📙 Citation", open=False):
|
142 |
gr.Textbox(
|
143 |
-
value=CITATION_TEXT, lines=7,
|
144 |
label="Copy the BibTeX snippet to cite this source",
|
145 |
elem_id="citation-button",
|
146 |
-
|
|
|
|
|
147 |
|
148 |
-
|
|
|
1 |
import gradio as gr
|
2 |
import pandas as pd
|
|
|
|
|
|
|
|
|
|
|
3 |
|
4 |
+
import constants
|
5 |
+
from utils import display, eval_request
|
6 |
|
7 |
+
#
|
8 |
+
##
|
9 |
+
###
|
10 |
+
##
|
11 |
+
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
|
13 |
+
leaderboard_df = pd.read_csv("leaderboard.csv")
|
14 |
|
15 |
+
# Format the dataframe
|
16 |
+
for col in leaderboard_df.columns:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
17 |
if col == "model":
|
18 |
+
leaderboard_df[col] = leaderboard_df[col].apply(lambda x: x.replace(x, display.make_clickable_model(x)))
|
19 |
else:
|
20 |
+
leaderboard_df[col] = leaderboard_df[col].apply(display.round_numbers)
|
21 |
+
leaderboard_df.rename(
|
22 |
+
columns={"Average WER": "Average WER ⬇️", "RTF (1e-3)": "RTF (1e-3) ⬇️"},
|
23 |
+
inplace=True,
|
24 |
+
)
|
25 |
+
leaderboard_df.sort_values(by='Average WER ⬇️', inplace=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
26 |
|
27 |
+
with gr.Blocks() as leaderboard_app:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
|
29 |
+
gr.HTML(constants.BANNER, elem_id="constants.banner")
|
30 |
+
gr.Markdown(constants.INTRODUCTION_TEXT, elem_classes="markdown-text")
|
|
|
|
|
|
|
|
|
|
|
31 |
|
32 |
with gr.Tabs(elem_classes="tab-buttons") as tabs:
|
33 |
+
|
34 |
with gr.TabItem("🏅 Leaderboard", elem_id="od-benchmark-tab-table", id=0):
|
35 |
leaderboard_table = gr.components.Dataframe(
|
36 |
+
value=leaderboard_df,
|
37 |
+
datatype=constants.COLUMN_DTYPES_LIST,
|
38 |
+
elem_id="leaderboard-table",
|
39 |
+
interactive=False,
|
40 |
+
visible=True,
|
41 |
+
)
|
|
|
42 |
|
43 |
with gr.TabItem("📈 Metrics", elem_id="od-benchmark-tab-table", id=1):
|
44 |
+
gr.Markdown(constants.METRICS_TAB_TEXT, elem_classes="markdown-text")
|
45 |
|
46 |
+
with gr.TabItem("✉️ Request a model here!", elem_id="od-benchmark-tab-table", id=2):
|
47 |
with gr.Column():
|
48 |
+
gr.Markdown("# ✉️ Request results for a new model here!", elem_classes="markdown-text")
|
49 |
with gr.Column():
|
|
|
50 |
with gr.Column():
|
51 |
+
model_id = gr.Textbox(label="Model ID (user_name/model_name)")
|
|
|
52 |
with gr.Column():
|
53 |
+
md_submission_result = gr.Markdown()
|
54 |
btn_submitt = gr.Button(value="🚀 Request")
|
55 |
+
btn_submitt.click(eval_request.request_model, [model_id], md_submission_result)
|
|
|
|
|
56 |
|
57 |
+
with gr.TabItem("☢️ Evaluate", elem_id="od-benchmark-tab-table", id=3):
|
58 |
+
with gr.Column():
|
59 |
+
gr.Markdown("For admins only.", elem_classes="markdown-text")
|
60 |
+
with gr.Column():
|
61 |
+
md_submission_result = gr.Markdown()
|
62 |
+
btn_submitt = gr.Button(value="RUN EVALUATION")
|
63 |
+
# btn_submitt.click(eval_request.request_model, [model_id], md_submission_result)
|
64 |
|
65 |
with gr.Row():
|
66 |
with gr.Accordion("📙 Citation", open=False):
|
67 |
gr.Textbox(
|
68 |
+
value=constants.CITATION_TEXT, lines=7,
|
69 |
label="Copy the BibTeX snippet to cite this source",
|
70 |
elem_id="citation-button",
|
71 |
+
show_label=True,
|
72 |
+
show_copy_button=True,
|
73 |
+
)
|
74 |
|
75 |
+
leaderboard_app.launch(allowed_paths=["banner.png"])
|
assets/.DS_Store
DELETED
Binary file (6.15 kB)
|
|
banner.png
ADDED
![]() |
constants.py
CHANGED
@@ -1,15 +1,19 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
|
|
|
|
|
|
|
|
13 |
|
14 |
TITLE = "<html> <head> <style> h1 {text-align: center;} </style> </head> <body> <h1> 🤗 Open Automatic Speech Recognition Leaderboard </b> </body> </html>"
|
15 |
|
|
|
1 |
+
#
|
2 |
+
##
|
3 |
+
### CONSTANTS
|
4 |
+
##
|
5 |
+
#
|
6 |
+
|
7 |
+
COLUMN_DTYPES_LIST = ["markdown", "number", "number", "number"] # in accordance with ./leaderboard.csv
|
8 |
+
|
9 |
+
#
|
10 |
+
##
|
11 |
+
### TEXT WALLS
|
12 |
+
##
|
13 |
+
#
|
14 |
+
|
15 |
+
banner_url = "/file=banner.png" # gradio file path
|
16 |
+
BANNER = f'<div style="display: flex; justify-content: space-around;"><img src="{banner_url}" alt="Banner" style="width: 70%;"> </div>'
|
17 |
|
18 |
TITLE = "<html> <head> <style> h1 {text-align: center;} </style> </head> <body> <h1> 🤗 Open Automatic Speech Recognition Leaderboard </b> </body> </html>"
|
19 |
|
init.py
DELETED
@@ -1,93 +0,0 @@
|
|
1 |
-
import os
|
2 |
-
from constants import EVAL_REQUESTS_PATH
|
3 |
-
from pathlib import Path
|
4 |
-
from huggingface_hub import HfApi, Repository
|
5 |
-
|
6 |
-
TOKEN_HUB = os.environ.get("TOKEN_HUB", None)
|
7 |
-
QUEUE_REPO = os.environ.get("QUEUE_REPO", None)
|
8 |
-
QUEUE_PATH = os.environ.get("QUEUE_PATH", None)
|
9 |
-
|
10 |
-
hf_api = HfApi(
|
11 |
-
endpoint="https://huggingface.co",
|
12 |
-
token=TOKEN_HUB,
|
13 |
-
)
|
14 |
-
|
15 |
-
def load_all_info_from_dataset_hub():
|
16 |
-
eval_queue_repo = None
|
17 |
-
results_csv_path = None
|
18 |
-
requested_models = None
|
19 |
-
|
20 |
-
passed = True
|
21 |
-
if TOKEN_HUB is None:
|
22 |
-
passed = False
|
23 |
-
else:
|
24 |
-
print("Pulling evaluation requests and results.")
|
25 |
-
|
26 |
-
eval_queue_repo = Repository(
|
27 |
-
local_dir=QUEUE_PATH,
|
28 |
-
clone_from=QUEUE_REPO,
|
29 |
-
use_auth_token=TOKEN_HUB,
|
30 |
-
repo_type="dataset",
|
31 |
-
)
|
32 |
-
eval_queue_repo.git_pull()
|
33 |
-
|
34 |
-
# Local directory where dataset repo is cloned + folder with eval requests
|
35 |
-
directory = QUEUE_PATH / EVAL_REQUESTS_PATH
|
36 |
-
requested_models = get_all_requested_models(directory)
|
37 |
-
requested_models = [p.stem for p in requested_models]
|
38 |
-
# Local directory where dataset repo is cloned
|
39 |
-
csv_results = get_csv_with_results(QUEUE_PATH)
|
40 |
-
if csv_results is None:
|
41 |
-
passed = False
|
42 |
-
if not passed:
|
43 |
-
print("No HuggingFace token provided. Skipping evaluation requests and results.")
|
44 |
-
|
45 |
-
return eval_queue_repo, requested_models, csv_results
|
46 |
-
|
47 |
-
|
48 |
-
def upload_file(requested_model_name, path_or_fileobj):
|
49 |
-
dest_repo_file = Path(EVAL_REQUESTS_PATH) / path_or_fileobj.name
|
50 |
-
dest_repo_file = str(dest_repo_file)
|
51 |
-
hf_api.upload_file(
|
52 |
-
path_or_fileobj=path_or_fileobj,
|
53 |
-
path_in_repo=str(dest_repo_file),
|
54 |
-
repo_id=QUEUE_REPO,
|
55 |
-
token=TOKEN_HUB,
|
56 |
-
repo_type="dataset",
|
57 |
-
commit_message=f"Add {requested_model_name} to eval queue")
|
58 |
-
|
59 |
-
def get_all_requested_models(directory):
|
60 |
-
directory = Path(directory)
|
61 |
-
all_requested_models = list(directory.glob("*.txt"))
|
62 |
-
return all_requested_models
|
63 |
-
|
64 |
-
def get_csv_with_results(directory):
|
65 |
-
directory = Path(directory)
|
66 |
-
all_csv_files = list(directory.glob("*.csv"))
|
67 |
-
latest = [f for f in all_csv_files if f.stem.endswith("latest")]
|
68 |
-
if len(latest) != 1:
|
69 |
-
return None
|
70 |
-
return latest[0]
|
71 |
-
|
72 |
-
|
73 |
-
|
74 |
-
def is_model_on_hub(model_name, revision="main") -> bool:
|
75 |
-
try:
|
76 |
-
model_name = model_name.replace(" ","")
|
77 |
-
author = model_name.split("/")[0]
|
78 |
-
model_id = model_name.split("/")[1]
|
79 |
-
if len(author) == 0 or len(model_id) == 0:
|
80 |
-
return False, "is not a valid model name. Please use the format `author/model_name`."
|
81 |
-
except Exception as e:
|
82 |
-
return False, "is not a valid model name. Please use the format `author/model_name`."
|
83 |
-
|
84 |
-
try:
|
85 |
-
models = list(hf_api.list_models(author=author, search=model_id))
|
86 |
-
matched = [model_name for m in models if m.modelId == model_name]
|
87 |
-
if len(matched) != 1:
|
88 |
-
return False, "was not found on the hub!"
|
89 |
-
else:
|
90 |
-
return True, None
|
91 |
-
except Exception as e:
|
92 |
-
print(f"Could not get the model from the hub.: {e}")
|
93 |
-
return False, "was not found on hub!"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
leaderboard.csv
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
model,Average WER,RTF (1e-3),AMI WER,Earnings22 WER,Gigaspeech WER,LS Clean WER,LS Other WER,SPGISpeech WER,Tedlium WER,Voxpopuli WER,Common Voice WER
|
2 |
+
openai/whisper-large-v3,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9,1.0,1.1,1.2
|
3 |
+
nvidia/canary-1b,0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9,1.0,1.1
|
requested_models.txt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
nvidia/parakeet-ctc-0.6b
|
2 |
+
openai/whisper-large
|
3 |
+
distil-whisper/distil-medium.en
|
utils/__init__.py
ADDED
File without changes
|
utils/display.py
ADDED
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
def make_clickable_model(model_name):
|
2 |
+
link = f"https://huggingface.co/{model_name}"
|
3 |
+
return f'<a target="_blank" href="{link}" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">{model_name}</a>'
|
4 |
+
|
5 |
+
def round_numbers(x):
|
6 |
+
if type(x) is str:
|
7 |
+
x = x
|
8 |
+
else:
|
9 |
+
x = round(x, 2)
|
10 |
+
return x
|
utils/eval_request.py
ADDED
@@ -0,0 +1,62 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import huggingface_hub as hf_hub
|
3 |
+
|
4 |
+
TOKEN_HUB = os.environ.get("TOKEN_HUB", None)
|
5 |
+
hf_api = hf_hub.HfApi(
|
6 |
+
endpoint="https://huggingface.co",
|
7 |
+
token=TOKEN_HUB,
|
8 |
+
)
|
9 |
+
|
10 |
+
|
11 |
+
#
|
12 |
+
##
|
13 |
+
###
|
14 |
+
##
|
15 |
+
#
|
16 |
+
|
17 |
+
|
18 |
+
def is_model_on_hub(model_id):
|
19 |
+
"""Check if a model is on the hub and return a failure message if not."""
|
20 |
+
|
21 |
+
try:
|
22 |
+
author = model_id.split("/")[0]
|
23 |
+
model_name = model_id.split("/")[1]
|
24 |
+
if len(author) == 0 or len(model_name) == 0:
|
25 |
+
return "is not a valid model name. Please use the format `author/model_name`."
|
26 |
+
except:
|
27 |
+
return "is not a valid model name. Please use the format `author/model_name`."
|
28 |
+
|
29 |
+
try:
|
30 |
+
models = list(hf_api.list_models(author=author, search=model_name))
|
31 |
+
matched = [model_id for m in models if m.modelId == model_id]
|
32 |
+
if len(matched) != 1:
|
33 |
+
return "was not found on the hub!"
|
34 |
+
else:
|
35 |
+
return None
|
36 |
+
except:
|
37 |
+
return "was not found on hub!"
|
38 |
+
|
39 |
+
def request_model(model_id):
|
40 |
+
|
41 |
+
def styled_error(error):
|
42 |
+
return f"<p style='color: red; font-size: 20px; text-align: center;'>{error}</p>"
|
43 |
+
|
44 |
+
def styled_message(message):
|
45 |
+
return f"<p style='color: green; font-size: 20px; text-align: center;'>{message}</p>"
|
46 |
+
|
47 |
+
# Check if the model is on the hub
|
48 |
+
error_msg = is_model_on_hub(model_id)
|
49 |
+
if error_msg != None:
|
50 |
+
return styled_error(f"{model_id} {error_msg}")
|
51 |
+
|
52 |
+
# Check if the model was already requested
|
53 |
+
with open("requested_models.txt", "r") as f:
|
54 |
+
requested_models_list = f.read().splitlines()
|
55 |
+
if model_id in requested_models_list:
|
56 |
+
return styled_error(f"A request for {model_id} was already made.")
|
57 |
+
|
58 |
+
# Add the model to the evaluation queue
|
59 |
+
with open("requested_models.txt", "a") as f:
|
60 |
+
f.write(model_id + "\n")
|
61 |
+
|
62 |
+
return styled_message("🤗 Your request has been submitted and will be evaluated as soon as possible!")
|
utils_display.py
DELETED
@@ -1,40 +0,0 @@
|
|
1 |
-
from dataclasses import dataclass
|
2 |
-
|
3 |
-
# These classes are for user facing column names, to avoid having to change them
|
4 |
-
# all around the code when a modif is needed
|
5 |
-
@dataclass
|
6 |
-
class ColumnContent:
|
7 |
-
name: str
|
8 |
-
type: str
|
9 |
-
|
10 |
-
def fields(raw_class):
|
11 |
-
return [v for k, v in raw_class.__dict__.items() if k[:2] != "__" and k[-2:] != "__"]
|
12 |
-
|
13 |
-
@dataclass(frozen=True)
|
14 |
-
class AutoEvalColumn: # Auto evals column
|
15 |
-
model = ColumnContent("Model", "markdown")
|
16 |
-
avg_wer = ColumnContent("Average WER ⬇️", "number")
|
17 |
-
rtf = ColumnContent("RTF (1e-3) ⬇️", "number")
|
18 |
-
ami_wer = ColumnContent("AMI", "number")
|
19 |
-
e22_wer = ColumnContent("Earnings22", "number")
|
20 |
-
gs_wer = ColumnContent("Gigaspeech", "number")
|
21 |
-
lsc_wer = ColumnContent("LS Clean", "number")
|
22 |
-
lso_wer = ColumnContent("LS Other", "number")
|
23 |
-
ss_wer = ColumnContent("SPGISpeech", "number")
|
24 |
-
tl_wer = ColumnContent("Tedlium", "number")
|
25 |
-
vp_wer = ColumnContent("Voxpopuli", "number")
|
26 |
-
cv_wer = ColumnContent("Common Voice", "number")
|
27 |
-
|
28 |
-
|
29 |
-
def make_clickable_model(model_name):
|
30 |
-
link = f"https://huggingface.co/{model_name}"
|
31 |
-
return f'<a target="_blank" href="{link}" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">{model_name}</a>'
|
32 |
-
|
33 |
-
def styled_error(error):
|
34 |
-
return f"<p style='color: red; font-size: 20px; text-align: center;'>{error}</p>"
|
35 |
-
|
36 |
-
def styled_warning(warn):
|
37 |
-
return f"<p style='color: orange; font-size: 20px; text-align: center;'>{warn}</p>"
|
38 |
-
|
39 |
-
def styled_message(message):
|
40 |
-
return f"<p style='color: green; font-size: 20px; text-align: center;'>{message}</p>"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|