Comments (10)
I just checked. There is no problem with the MD5 sum value.
On my mac:
$ curl http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz
$ md5 t10k-images-idx3-ubyte.gz
MD5 (t10k-images-idx3-ubyte.gz) = bef4ecab320f06d8554ea6380940ec79
$ ls -lth t10k-images-idx3-ubyte.gz
-rw-r--r-- 1 hanhxiao 4.2M Mar 12 20:37 t10k-images-idx3-ubyte.gz
from fashion-mnist.
@hanxiao , you could check it on Ubuntu16.04 and Windows, I also have checked it on mac, it is right, but on Ubuntu and Windows it is wrong, I have asked this problem on PyTorch torchvision Github issues.
from fashion-mnist.
@hanxiao I also tested on my new MacBook Pro, you could found the download link is wrong, it redirects to MNIST!
from fashion-mnist.
I've tested on Linux machine (in Shenzhen China), it gives the correct result as follows:
hanhxiao:~# wget http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz
--2018-03-13 11:21:08-- http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz
Connecting to 10.223.133.20:52107... connected.
Proxy request sent, awaiting response... 200 OK
Length: 4422102 (4.2M) [binary/octet-stream]
Saving to: ‘t10k-images-idx3-ubyte.gz’
100%[===================================================================================================>] 4,422,102 605KB/s in 19s
2018-03-13 11:21:29 (223 KB/s) - ‘t10k-images-idx3-ubyte.gz’ saved [4422102/4422102]
hanhxiao:~# md5sum t10k-images-idx3-ubyte.gz
bef4ecab320f06d8554ea6380940ec79 t10k-images-idx3-ubyte.gz
I saw a 302 redirect
response from your screenshot. This is really suspicious. We are hosting the data file as static file on Amazon S3 bucket, we are not setting any redirect there.
from fashion-mnist.
@leftthomas Can you do following in bash and check where 302 happens?
curl -v -L http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz -IX GET
or,
wget http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz -qSO /dev/null
For example, on my linux the second method gives:
hanhxiao:~# wget http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz -qSO /dev/null
HTTP/1.1 200 OK
x-amz-id-2: DK6MTacPtZm8r6WJMeSzrmG9DFQRTjnLc7irkMSCHcIF8UkZE5H2K2LSLwgTJs31AaEP5bZSoTk=
x-amz-request-id: 2C990AB5ADEAA5D1
Date: Tue, 13 Mar 2018 03:41:48 GMT
Last-Modified: Thu, 31 Aug 2017 12:17:53 GMT
ETag: "bef4ecab320f06d8554ea6380940ec79"
Content-Type: binary/octet-stream
Content-Length: 4422102
Server: AmazonS3
X-Cache: MISS from TENCENT
X-Cache-Lookup: MISS from TENCENT
Via: 1.1 TENCENT (squid/3.3.8)
Connection: keep-alive
Notice that ETag
gives the correct MD5 value. And there is no redirect.
Fyi, here is an example where a 302 happens on purpose.
hanhxiao:~# wget https://httpstat.us/302 -qSO /dev/null
HTTP/1.1 302 Found
Cache-Control: private
Content-Length: 9
Content-Type: text/plain; charset=utf-8
Location: https://httpstat.us
Server: Microsoft-IIS/10.0
X-AspNetMvc-Version: 5.1
Access-Control-Allow-Origin: *
X-AspNet-Version: 4.0.30319
X-Powered-By: ASP.NET
Set-Cookie: ARRAffinity=7943fc4fbbc26574a46521f2ec212be4ee889bc2a88a9742ab65676f5bd7e9e6;Path=/;HttpOnly;Domain=httpstat.us
Date: Tue, 13 Mar 2018 03:41:12 GMT
Connection: close
HTTP/1.1 200 OK
Cache-Control: private
Content-Length: 7958
Content-Type: text/html; charset=utf-8
Server: Microsoft-IIS/10.0
X-AspNetMvc-Version: 5.1
Access-Control-Allow-Origin: *
X-AspNet-Version: 4.0.30319
X-Powered-By: ASP.NET
Date: Tue, 13 Mar 2018 03:41:12 GMT
Connection: close
from fashion-mnist.
@hanxiao I just tried that two command on my Mac, and here are the results
from fashion-mnist.
@hanxiao Further more, I found not only the t10k-images-idx3-ubyte.gz
is wrong, but also the train-images-idx3-ubyte.gz
is wrong by using the curl
command to download the datas on my Mac:
And if I use wget
to download the datas on my Mac and Ubuntu, it just only train-images-idx3-ubyte.gz
is wrong:
from fashion-mnist.
@leftthomas First of all, no panic, all links are correct. I've checked on multiple devices with different ISPs. Everything is fine.
Now to your specific problem, it looks like someone redirects your request to 10.6.0.123. You can see that from your curl
log. The redirection is definitely not from us. It could be made by your ISP or your smart router, especially due to the following reasons:
- 10.6.0.123 is an internal IP address. It's not a public IP host by any public service. So your dataset must be downloaded from the intranet, not the internet!
- It looks like either your ISP or your smart router are doing some name-based caching. As you previously downloaded LeCun's data (which has the exact same name), it automatically redirects you to the cached files, not the real files. That also explains the high download speed.
Here are my suggestions:
- If you are using smart router, reset it or clear the cache might be helpful. Make a backup, no guarantee from my side.
- Meanwhile, please try visiting http://10.6.0.123 directly and see what it is? who is it? Is it a router? is it your ISP? Is it a proxy? As I said, 10.6.0.123 is an internal IP address specific to your network, I can not see it.
from fashion-mnist.
@hanxiao Thanks for your help, I'll convey this problem to our informatization office.
from fashion-mnist.
I have downloaded the
t10k-images-idx3-ubyte.gz
from thereadme
provided download link, but I checked the md5sum value is9fb629c4189551a2d022fa330f9573f3
, it not the same asreadme
given, I have deleted it and redownload again for five times, it is wrong.
I get same issue for downloading t10k-images-idx3-ubyte.gz with CMCC ISP, the md5sum is 9fb629c4189551a2d022fa330f9573f3, it is changed to minist testset. But when I use another server with different ISP (China Telecom), download the t10k-images-idx3-ubyte.gz again, the file is OK, md5sum is bef4ecab320f06d8554ea6380940ec79.
from fashion-mnist.
Related Issues (20)
- why running the benchmark/runner.py very slow? very slow !!!
- Benchmark - Free Pascal Implementation - 94% with ~174K parameters
- I NEED HELP WITH THE GRAPHICS
- Fashion-MNIST converted to png
- Benchmark : CNN-Fashion-MNIST
- Cant find the csv file of the data set HOT 2
- How to create clustering visulization? label:"help wanted"
- optional native extra labels.
- What's the mean and std of FMNIST? HOT 1
- Benchmark: E2E-3M Accuracy Result of 95.92% for Fashion-MNIST
- Interactive Visualization of the dataset
- Mislabeled Instances Found HOT 5
- Loading the dataset from the local path using tensorflow 2.0 HOT 6
- Benchmark: Ensemble 4 CNNs on censored images
- Link to Pytorch Dataset broken HOT 2
- convolution network mean acc achieve 0.9765
- Incremental learning HOT 1
- Direct download links not working HOT 3
- Invalid Tensor size in TF.js
- Image of full body in trainingset HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from fashion-mnist.