These are chat archives for beniz/deepdetect

23rd
May 2018
Preet Singh Khalsa
@preetskhalsa97
May 23 2018 11:56

@beniz Thanks a lot for making this amazing tool for deep learning applications.
I haven't been able to get around training a transfer learning model, I was wondering if you could throw some light here.
While training, I am getting the following docker log:
[2018-05-23 11:30:41.250] [api] [info] 172.17.0.1 "PUT /services/greendeck" 201 1
[2018-05-23 11:30:47.332] [api] [info] 172.17.0.1 "POST /train" 201 0
[2018-05-23 11:30:47.335] [greendeck] [error] training status call failed
[2018-05-23 11:30:47.335] [api] [error] {"code":400,"msg":"BadRequest","dd_code":1006,"dd_msg":"Service Bad Request Error"}
[2018-05-23 11:30:47.335] [api] [info] 172.17.0.1 "GET /train?job=1&timeout=10&service=greendeck" 200 0
I have taken care of putting the data folder in the docker with simplified names of the sub- folder file, removing template from the service file and other such things that I could comprehend from reading external issues.
Also, I decreased the batch size to 3 and number of iterations to 10,000 to solve the issue of RAM, if present.

Could you please help me with getting around this?
Thanks!

Emmanuel Benazera
@beniz
May 23 2018 12:16
Hi, you need to look as the error server-side, using the docker log, look at the docker readme in the dd repository
Preet Singh Khalsa
@preetskhalsa97
May 23 2018 13:01
Thanks for the prompt reply!
The error posted in the above message is from the docker log itself (obtained from docker logs -f <container-id>
Emmanuel Benazera
@beniz
May 23 2018 14:20
then this only means your input API call is wrong, you may want to post it here
Preet Singh Khalsa
@preetskhalsa97
May 23 2018 15:46
I am posting the service as well as train file:
I am running these two, training after the service file.
Can you please have a look?
Emmanuel Benazera
@beniz
May 23 2018 16:07
can you simply post the JSON input of your API calls ?
thanks!
Preet Singh Khalsa
@preetskhalsa97
May 23 2018 16:16

{

"service":"myserv",

"async":true,

"parameters":{

  "mllib":{  

     "gpu":False,

     "solver":{  

       "test_interval": 4000,

       "iterations": 50000,

       "snapshot": 10000,

       "base_lr": 0.01,

       "solver_type": "NESTEROV",

       "test_initialization": True,

     },

     "net":{  

        "batch_size":5,

        "test_batch_size": 5            

     }

  },

  "input":{  

     "shuffle":true,

     "test_split":0.1

  },

  "output":{  

     "measure":[  

        "acc-5",

        "mcll",

        "f1"

     ]

  }

},

"data":[

  "/home/dd/data/images"

]

}

this is the train json

Sending you service creation json as well

Preet Singh Khalsa
@preetskhalsa97
May 23 2018 16:31

{

"mllib":"caffe",

"description":"clothes classification",

"type":"supervised",

"parameters":{

  "input":{  

     "connector":"image",

     "height":224,

     "width":224

  },

  "mllib":{  

     "nclasses":46,

     "finetuning":True,

     "rotate":False,

     "mirror":True,

     "weights":"model_iter_300000.caffemodel"

  }

},

"model":{

  "repository":"/home/dd/models/clothing"

}

}