id
type
5 (blog/news article)
status
21 (imported old-v2, waiting for another import)
review version
0
cleanup version
0
pending deletion
0 (-)
created at
2025-11-08 12:05:45
updated at
2025-11-08 12:05:46
url
http://blog.so8848.com/2008/04/sgd-special-ingredient-reckless-driving.html
url length
75
url crc
1242
url crc32
1329464538
location type
1 (url matches target location, page_location is empty)
canonical status
2 (missing canonical tag in html)
canonical page id
-
domain id
domain tld
2211
domain parts
0
originating warc id
-
originating url
https://data.commoncrawl.org/crawl-data/CC-MAIN-2025-33/segments/1754151280076.69/warc/CC-MAIN-20250809045158-20250809075158-00963.warc.gz
source type
11 (CommonCrawl)
server ip
Publication date
2025-08-09 05:11:01
Fetch attempts
0
Original html size
60922
Normalized and saved size
22447
title
SGD Special Ingredient: Reckless Driving | Information Retrieval Blog
excerpt
content
I’ve been playing around with convergence for my SGD implementation for the upcoming LingPipe 3.5, in the context of the 2008 i2b2 Obesity Challenge,the full title of which is "Second i2b2 Shared-Task and Workshop;Challenges in Natural Language Processing for Clinical Data; ObesityChallenge (A Shared-Task on Obesity): Who’s obese and whatco-morbidities do they (definitely/likely) have?". Participants haveuntil April 15, 2008 to register to participate. Slide 37 of Léon Bottou’s NIPS tutorial Learning with Large Datasetsreveals the "secret ingredient" behind (his) successful stochasticgradient search (where η is the learning rate, which he calls a "gain"):At any moment during training, we can: Select a small subsample of examples.Try various gains η on the subsample.Pick the gain η that most reduces the cost.Use it for the next 100000 iterations on the full dataset.This is a kind of meta-descent algorithm, the most well known of which is Nicolas Shraudolph’s Stochastic Meta-Descent.I’m...
author
updated
1764667526
block type
0
extracted fields
104
extracted bits
title
full content
content was extracted heuristically
detected location
0
detected language
1 (English)
category id
index version
2025110801
paywall score
0
spam phrases
0
text nonlatin
3
text cyrillic
0
text characters
3430
text words
650
text unique words
368
text lines
1
text sentences
24
text paragraphs
1
text words per sentence
27
text matched phrases
1
text matched dictionaries
2
links self subdomains
0
links other subdomains
5
links other domains
7
links spam adult
0
links spam random
3
links spam expired
0
links ext activities
1
links ext ecommerce
0
links ext finance
0
links ext crypto
0
links ext booking
0
links ext news
0
links ext leaks
0
links ext ugc
13
links ext klim
0
links ext generic
0
image author