Conversations with Hugging Face CTO

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19

import torch
import pandas as pd
from bs4 import BeautifulSoup
import requests
url                = "https://huggingface.co/models"
response           = requests.get(url)
html_soup          = BeautifulSoup(response.text, 'html.parser')
models             = html_soup.find(class_ = 'models-list')
model_names=[]
for a in models.find_all('li'):
    model_names.append(a.a['href'])
model_names  = [m[1:] for m in model_names]
download_counts = []
for a in models.find_all('li'):
count = a.find(class_='tooltip').text.strip().split("downloads")[0].strip()
download_counts.append(count)
model_stats = pd.DataFrame({'model':model_names,'downloads_l30days':download_counts})
model_stats.head(20)

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

number                                          model downloads_l30days
0                             distilbert-base-uncased        44,291,279
1                                   bert-base-uncased        43,808,273
2                                     bert-base-cased           830,551
3                                                gpt2           657,636
4                                        roberta-base           619,027
5                                    xlnet-base-cased           474,964
6                        bert-base-multilingual-cased           433,170
7                                       roberta-large           355,542
8          xlm-roberta-large-finetuned-conll03-german           337,872
9                                   xlm-roberta-large           280,967
10                                     albert-base-v2           222,233
11                                         openai-gpt           198,678
12                                  bert-base-chinese           196,817
13                                 bert-large-uncased           158,196
14                                         gpt2-large           155,298
15    cl-tohoku/bert-base-japanese-whole-word-masking           151,115
16                     bert-base-multilingual-uncased           145,300
17                                            gpt2-xl           134,733
18                                 distilroberta-base           126,489
19                 distilbert-base-multilingual-cased           125,623
20                                         distilgpt2           123,623
21                                            t5-base           121,663
22                                           t5-small           102,185
23  deepset/bert-large-uncased-whole-word-masking-...            91,045
24              mrm8488/bert-tiny-5-finetuned-squadv2            85,570