Examples: query, "exact match", wildcard*, wild?ard, wild*rd
Fuzzy search: cake~ (finds cakes, bake)
Term boost: "red velvet"^4, chocolate^2
Field grouping: tags:(+work -"fun-stuff")
Escaping: Escape characters +-&|!(){}[]^"~*?:\ with \, e.g. \+
Range search: properties.timestamp:[1587729413488 TO *] (inclusive), properties.title:{A TO Z}(excluding A and Z)
Combinations: chocolate AND vanilla, chocolate OR vanilla, (chocolate OR vanilla) NOT "vanilla pudding"
Field search: properties.title:"The Title" AND text
Answered
Hello, I Work On A Ml Project (Using Frameworks Like Pytorch And Tensorflow), I Use S3 To Save Model Checkpoints. Recently We Adopted Clearml To Track Our Experiments Which Was Indeed A Good Idea. So I Have A Small Issue With S3 And Clearml:

Hello, I work on a ML Project (Using Frameworks like Pytorch and Tensorflow), I use s3 to save model checkpoints. Recently we adopted Clearml to track our experiments which was indeed a good idea.
So I have a small issue With S3 and ClearML:
For instance i don't use AWS but rather has my own s3 endpoint:export AWS_ACCESS_KEY_ID=xxxx export AWS_SECRET_ACCESS_KEY=xxxx export S3_ENDPOINT=xxxx export S3_VERIFY_SSL=0 export S3_USE_HTTPS=1With that, i can actually save model checkpoints and load them (for both tf and pytorch) even use and access tensorboard. When i added ClearML " output_uri " I get this error:
2021-05-14 19:05:47,668 - clearml.storage - ERROR - Failed uploading: An error occurred (InvalidAccessKeyId) when calling the PutObject operation: The AWS Access Key Id you provided does not exist in our records.Which seems Like it trying to use AWS and not my own endpoint. Can anyone guide me through this. THANKS

  
  
Posted 2 years ago
Votes Newest

Answers 8


Ah Many thanks it worked 🙂

  
  
Posted 2 years ago

I am basically saving checkpoints ( so thing like totch.save( http://foo.pt )) and the output_ur= s3://bucket/folder

  
  
Posted 2 years ago

Did you update the required credentials in the clearml.conf file?

  
  
Posted 2 years ago

SuccessfulKoala55 , Yes absolutely, i have this section:
` "sdk":{

       aws: {
             s3: {

           credentials: [

                {
                    bucket: xxxx,
                    key: "xxx",
                    secret: xxxx,
                    host: "xxxxx",
                    secure: true,
                },

           ]
       }
}

} `

  
  
Posted 2 years ago

hostname and port, that is, in your case, so output_uri=s3://<hostname>:port/bucket/folder

  
  
Posted 2 years ago

Oh, I see your issue - if you're not using the AWS S3, the output_uri must contain your hostname

  
  
Posted 2 years ago

Hi AgitatedTurtle16 , what exactly did you use for output_uri ?

  
  
Posted 2 years ago

Otherwise ClearML will think you wish to upload to AWS...

  
  
Posted 2 years ago
678 Views
8 Answers
2 years ago
one year ago
Tags
tf