![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1939799%2F3e4c0a69dae8077131350fa3a2233324%2F.png?generation=1561881283692084&alt=media\")
hello,
\n\nI have \"submission error\". \nIf I pass the mouse over info it says \"No additional details provided for this error \"
\n\nI train my model at home and build de submission file at home. \nI upload this file inside the kernel editor using \"+ Add Dataset\". \nthe file goes to :
\n\n../input/submission/submission.csv
Then I load this file and save it again ... like This :
\n\n```\nimport numpy as np\nimport pandas as pd \nimport os
\n\nsub = pd.read_csv('../input/submission/submission.csv')\nsub.diagnosis=sub.diagnosis.astype('int') #just to make sure\nsub.to_csv('submission.csv',index=False)\n```
\n\nThat is all I have but on the submission, I have \"error\" (?)
\n\nAny suggestions?
\n\nThanks!
", "messages": [ { "id": 564763, "postDate": "2019-06-30T00:21:56.507Z", "content": "you cannot submit a submission.csv in this competition. your model must run on stage 2 dataset. take a look at the example kernel here to know more: https://www.kaggle.com/abhishek/pytorch-inference-kernel
", "rawMarkdown": "you cannot submit a submission.csv in this competition. your model must run on stage 2 dataset. take a look at the example kernel here to know more: https://www.kaggle.com/abhishek/pytorch-inference-kernel", "votes": 1 }, { "id": 564762, "postDate": "2019-06-30T00:19:34.563Z", "content": "hello,
\n\nI have \"submission error\". \nIf I pass the mouse over info it says \"No additional details provided for this error \"
\n\nI train my model at home and build de submission file at home. \nI upload this file inside the kernel editor using \"+ Add Dataset\". \nthe file goes to :
\n\n../input/submission/submission.csv
Then I load this file and save it again ... like This :
\n\n```\nimport numpy as np\nimport pandas as pd \nimport os
\n\nsub = pd.read_csv('../input/submission/submission.csv')\nsub.diagnosis=sub.diagnosis.astype('int') #just to make sure\nsub.to_csv('submission.csv',index=False)\n```
\n\nThat is all I have but on the submission, I have \"error\" (?)
\n\nAny suggestions?
\n\nThanks!
", "rawMarkdown": "hello,\n\nI have \"submission error\". \nIf I pass the mouse over info it says \"No additional details provided for this error \"\n\nI train my model at home and build de submission file at home. \nI upload this file inside the kernel editor using \"+ Add Dataset\". \nthe file goes to : \n\n`../input/submission/submission.csv`\n\nThen I load this file and save it again ... like This : \n\n```\nimport numpy as np\nimport pandas as pd \nimport os\n\nsub = pd.read_csv('../input/submission/submission.csv')\nsub.diagnosis=sub.diagnosis.astype('int') #just to make sure\nsub.to_csv('submission.csv',index=False)\n```\n\n\nThat is all I have but on the submission, I have \"error\" (?)\n\nAny suggestions?\n\nThanks! " }, { "id": 565655, "postDate": "2019-07-01T08:38:21.930Z", "content": "Did you find reducing the batch size solved your problem?
\n\nIf so please comment here.
", "rawMarkdown": "Did you find reducing the batch size solved your problem?\n\nIf so please comment [here](https://www.kaggle.com/c/aptos2019-blindness-detection/discussion/98087).", "replies": [ { "id": 565992, "postDate": "2019-07-01T17:05:12.517Z", "content": "yes I load the test by parts, do the predictions for one part, call de garbage collector, load another part do the prediction and in the end concatenate all predictions. My memory usage goes from 12GB to 3.6GB... submit and work !
", "rawMarkdown": "yes I load the test by parts, do the predictions for one part, call de garbage collector, load another part do the prediction and in the end concatenate all predictions. My memory usage goes from 12GB to 3.6GB... submit and work ! " } ] }, { "id": 565067, "postDate": "2019-06-30T11:42:11.760Z", "content": "Try reducing batch size. More batch size increases RAM and GPU mem
", "rawMarkdown": "Try reducing batch size. More batch size increases RAM and GPU mem", "replies": [ { "id": 565127, "postDate": "2019-06-30T13:25:08.433Z", "content": "I only execute predictons with the Uploaded model on the kaggle kernel. The predict method uses batch size of 32 by default (https://keras.io/models/model/) ... Anyway I Will try tomorrow , Change this parameter.
", "rawMarkdown": "I only execute predictons with the Uploaded model on the kaggle kernel. The predict method uses batch size of 32 by default (https://keras.io/models/model/) ... Anyway I Will try tomorrow , Change this parameter. " }, { "id": 573331, "postDate": "2019-07-12T05:58:17.973Z", "content": "Does it work?
", "rawMarkdown": "Does it work?" } ] }, { "id": 564794, "postDate": "2019-06-30T01:40:56.080Z", "content": "upload the model run it inside but I have the error on the submission :
\n\n\"Kernel Out of Resources\"
\n\nEDIT: Again after I stop all kernels I was using... I think I have spent all of my 5 tries for today ...\nThere is no logic in analyzing images less than 1/4 of the size for this problem ...
", "rawMarkdown": "upload the model run it inside but I have the error on the submission : \n\n\"Kernel Out of Resources\"\n\nEDIT: Again after I stop all kernels I was using... I think I have spent all of my 5 tries for today ...\nThere is no logic in analyzing images less than 1/4 of the size for this problem ...", "replies": [ { "id": 564947, "postDate": "2019-06-30T07:54:50.057Z", "content": "
I submitted successfully only one time.
", "rawMarkdown": "\n\nI submitted successfully only one time." }, { "id": 565036, "postDate": "2019-06-30T10:34:15.933Z", "content": "ZERO Times successfully for me . \"Kernel Out of Resources\"
", "rawMarkdown": "ZERO Times successfully for me . \"Kernel Out of Resources\"", "votes": 1 }, { "id": 565037, "postDate": "2019-06-30T10:35:43.763Z", "content": "Failed submissions in this competition count towards your daily limit! Its on the first page of competition.
", "rawMarkdown": "Failed submissions in this competition count towards your daily limit! Its on the first page of competition." }, { "id": 565043, "postDate": "2019-06-30T10:48:54.337Z", "content": "Yes. My problema is to know the Resources limits on the submission stage since my kernel runs ok before that.
", "rawMarkdown": "Yes. My problema is to know the Resources limits on the submission stage since my kernel runs ok before that. " } ] }, { "id": 564783, "postDate": "2019-06-30T01:11:27.937Z", "content": "I have upload my model in keras format (model.h5).
\n\nI can load the model :
\n\nmodel = load_model('../input/modeltoload/model.h5')
but when I try to execute predictions:
\n\n```\npred_test=model.predict(X_test)
\n\n```\n I have error :
\n\nFailedPreconditionError Traceback (most recent call last)\n in \n----> 1 pred_test=model.predict(X_test)
\n\n/opt/conda/lib/python3.6/site-packages/keras/engine/training.py in predict(self, x, batch_size, verbose, steps)\n 1167 batch_size=batch_size,\n 1168 verbose=verbose,\n-> 1169 steps=steps)\n 1170 \n 1171 def train_on_batch(self, x, y,
\n\n/opt/conda/lib/python3.6/site-packages/keras/engine/training_arrays.py in predict_loop(model, f, ins, batch_size, verbose, steps)\n 292 ins_batch[i] = ins_batch[i].toarray()\n 293 \n--> 294 batch_outs = f(ins_batch)\n 295 batch_outs = to_list(batch_outs)\n 296 if batch_index == 0:
\n\n/opt/conda/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py in call(self, inputs)\n 2713 return self._legacy_call(inputs)\n 2714 \n-> 2715 return self._call(inputs)\n 2716 else:\n 2717 if py_any(is_tensor(x) for x in inputs):
\n\n/opt/conda/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py in _call(self, inputs)\n 2673 fetched = self._callable_fn(*array_vals, run_metadata=self.run_metadata)\n 2674 else:\n-> 2675 fetched = self._callable_fn(*array_vals)\n 2676 return fetched[:len(self.outputs)]\n 2677
\n\n/opt/conda/lib/python3.6/site-packages/tensorflow/python/client/session.py in call(self, *args, **kwargs)\n 1437 ret = tf_session.TF_SessionRunCallable(\n 1438 self._session._session, self._handle, args, status,\n-> 1439 run_metadata_ptr)\n 1440 if run_metadata:\n 1441 proto_data = tf_session.TF_GetBuffer(run_metadata_ptr)
\n\n/opt/conda/lib/python3.6/site-packages/tensorflow/python/framework/errors_impl.py in exit(self, type_arg, value_arg, traceback_arg)\n 526 None, None,\n 527 compat.as_text(c_api.TF_Message(self.status.status)),\n--> 528 c_api.TF_GetCode(self.status.status))\n 529 # Delete the underlying status object from memory otherwise it stays alive\n 530 # as there is a reference to status from this from the traceback due to
\n\nFailedPreconditionError: Attempting to use uninitialized value dense_9/kernel\n [[{{node dense_9/kernel/read}}]]
", "rawMarkdown": "I have upload my model in keras format (model.h5). \n\nI can load the model : \n\n`model = load_model('../input/modeltoload/model.h5')`\n\nbut when I try to execute predictions: \n\n```\npred_test=model.predict(X_test)\n\n```\n I have error : \n\n---------------------------------------------------------------------------\nFailedPreconditionError Traceback (most recent call last)\n", "replies": [ { "id": 564788, "postDate": "2019-06-30T01:32:35.803Z", "content": "Solved it was the memory over ... it is processing the submission .. let's see ...
", "rawMarkdown": "Solved it was the memory over ... it is processing the submission .. let's see ..." } ] }, { "id": 564765, "postDate": "2019-06-30T00:29:20.443Z", "content": "Sorry, I am new to this. What is stage 2 dataset?
\n\nDo you mean there is no way to build the model at home and upload the results only?
\n\nIf that is so I don't think running the code here will work for me because I want to use the full resolution of the images ... (no memory or CPU/GPU/TPU enough here)
", "rawMarkdown": "Sorry, I am new to this. What is stage 2 dataset?\n\nDo you mean there is no way to build the model at home and upload the results only? \n\nIf that is so I don't think running the code here will work for me because I want to use the full resolution of the images ... (no memory or CPU/GPU/TPU enough here)\n", "replies": [ { "id": 564766, "postDate": "2019-06-30T00:31:03.977Z", "content": "Its a hidden test set. you have to make an inference kernel and for that instead of uploading csv, you been to upload the trained model. if you take a look at the kernel i mentioned, you will understand how it works :)
", "rawMarkdown": "Its a hidden test set. you have to make an inference kernel and for that instead of uploading csv, you been to upload the trained model. if you take a look at the kernel i mentioned, you will understand how it works :)", "votes": 2 }, { "id": 564767, "postDate": "2019-06-30T00:34:52.440Z", "content": "Okay I see ... I upload the model and build the submission file there . Ok, Thanks !
\n\nPS - I was thinkg about contract an Ophthalmologist to build my submission :-) lol
", "rawMarkdown": "Okay I see ... I upload the model and build the submission file there . Ok, Thanks ! \n\nPS - I was thinkg about contract an Ophthalmologist to build my submission :-) lol " } ] }, { "id": 565936, "postDate": "2019-07-01T15:28:10.573Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 564763, "author_name": "Abhishek Thakur", "author_url": "", "post_date": "2019-06-30T00:21:56.507000", "content": "you cannot submit a submission.csv in this competition. your model must run on stage 2 dataset. take a look at the example kernel here to know more: https://www.kaggle.com/abhishek/pytorch-inference-kernel
", "votes": 1, "replies": [] }, { "id": 565655, "author_name": "Karl Hornlund", "author_url": "", "post_date": "2019-07-01T08:38:21.930000", "content": "Did you find reducing the batch size solved your problem?
\n\nIf so please comment here.
", "votes": 0, "replies": [ { "id": 565992, "author_name": "birinhos", "author_url": "", "post_date": "2019-07-01T17:05:12.517000", "content": "yes I load the test by parts, do the predictions for one part, call de garbage collector, load another part do the prediction and in the end concatenate all predictions. My memory usage goes from 12GB to 3.6GB... submit and work !
", "votes": 0, "replies": [] } ] }, { "id": 565067, "author_name": "Murlikrishnan Tripathi", "author_url": "", "post_date": "2019-06-30T11:42:11.760000", "content": "Try reducing batch size. More batch size increases RAM and GPU mem
", "votes": 0, "replies": [ { "id": 565127, "author_name": "birinhos", "author_url": "", "post_date": "2019-06-30T13:25:08.433000", "content": "I only execute predictons with the Uploaded model on the kaggle kernel. The predict method uses batch size of 32 by default (https://keras.io/models/model/) ... Anyway I Will try tomorrow , Change this parameter.
", "votes": 0, "replies": [] }, { "id": 573331, "author_name": "Lisheng", "author_url": "", "post_date": "2019-07-12T05:58:17.973000", "content": "Does it work?
", "votes": 0, "replies": [] } ] }, { "id": 564794, "author_name": "birinhos", "author_url": "", "post_date": "2019-06-30T01:40:56.080000", "content": "upload the model run it inside but I have the error on the submission :
\n\n\"Kernel Out of Resources\"
\n\nEDIT: Again after I stop all kernels I was using... I think I have spent all of my 5 tries for today ...\nThere is no logic in analyzing images less than 1/4 of the size for this problem ...
", "votes": 0, "replies": [ { "id": 564947, "author_name": "Pooh", "author_url": "", "post_date": "2019-06-30T07:54:50.057000", "content": "
I submitted successfully only one time.
", "votes": 0, "replies": [] }, { "id": 565036, "author_name": "birinhos", "author_url": "", "post_date": "2019-06-30T10:34:15.933000", "content": "ZERO Times successfully for me . \"Kernel Out of Resources\"
", "votes": 1, "replies": [] }, { "id": 565037, "author_name": "Abhishek Thakur", "author_url": "", "post_date": "2019-06-30T10:35:43.763000", "content": "Failed submissions in this competition count towards your daily limit! Its on the first page of competition.
", "votes": 0, "replies": [] }, { "id": 565043, "author_name": "birinhos", "author_url": "", "post_date": "2019-06-30T10:48:54.337000", "content": "Yes. My problema is to know the Resources limits on the submission stage since my kernel runs ok before that.
", "votes": 0, "replies": [] } ] }, { "id": 564783, "author_name": "birinhos", "author_url": "", "post_date": "2019-06-30T01:11:27.937000", "content": "I have upload my model in keras format (model.h5).
\n\nI can load the model :
\n\nmodel = load_model('../input/modeltoload/model.h5')
but when I try to execute predictions:
\n\n```\npred_test=model.predict(X_test)
\n\n```\n I have error :
\n\nFailedPreconditionError Traceback (most recent call last)\n in \n----> 1 pred_test=model.predict(X_test)
\n\n/opt/conda/lib/python3.6/site-packages/keras/engine/training.py in predict(self, x, batch_size, verbose, steps)\n 1167 batch_size=batch_size,\n 1168 verbose=verbose,\n-> 1169 steps=steps)\n 1170 \n 1171 def train_on_batch(self, x, y,
\n\n/opt/conda/lib/python3.6/site-packages/keras/engine/training_arrays.py in predict_loop(model, f, ins, batch_size, verbose, steps)\n 292 ins_batch[i] = ins_batch[i].toarray()\n 293 \n--> 294 batch_outs = f(ins_batch)\n 295 batch_outs = to_list(batch_outs)\n 296 if batch_index == 0:
\n\n/opt/conda/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py in call(self, inputs)\n 2713 return self._legacy_call(inputs)\n 2714 \n-> 2715 return self._call(inputs)\n 2716 else:\n 2717 if py_any(is_tensor(x) for x in inputs):
\n\n/opt/conda/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py in _call(self, inputs)\n 2673 fetched = self._callable_fn(*array_vals, run_metadata=self.run_metadata)\n 2674 else:\n-> 2675 fetched = self._callable_fn(*array_vals)\n 2676 return fetched[:len(self.outputs)]\n 2677
\n\n/opt/conda/lib/python3.6/site-packages/tensorflow/python/client/session.py in call(self, *args, **kwargs)\n 1437 ret = tf_session.TF_SessionRunCallable(\n 1438 self._session._session, self._handle, args, status,\n-> 1439 run_metadata_ptr)\n 1440 if run_metadata:\n 1441 proto_data = tf_session.TF_GetBuffer(run_metadata_ptr)
\n\n/opt/conda/lib/python3.6/site-packages/tensorflow/python/framework/errors_impl.py in exit(self, type_arg, value_arg, traceback_arg)\n 526 None, None,\n 527 compat.as_text(c_api.TF_Message(self.status.status)),\n--> 528 c_api.TF_GetCode(self.status.status))\n 529 # Delete the underlying status object from memory otherwise it stays alive\n 530 # as there is a reference to status from this from the traceback due to
\n\nFailedPreconditionError: Attempting to use uninitialized value dense_9/kernel\n [[{{node dense_9/kernel/read}}]]
", "votes": 0, "replies": [ { "id": 564788, "author_name": "birinhos", "author_url": "", "post_date": "2019-06-30T01:32:35.803000", "content": "Solved it was the memory over ... it is processing the submission .. let's see ...
", "votes": 0, "replies": [] } ] }, { "id": 564765, "author_name": "birinhos", "author_url": "", "post_date": "2019-06-30T00:29:20.443000", "content": "Sorry, I am new to this. What is stage 2 dataset?
\n\nDo you mean there is no way to build the model at home and upload the results only?
\n\nIf that is so I don't think running the code here will work for me because I want to use the full resolution of the images ... (no memory or CPU/GPU/TPU enough here)
", "votes": 0, "replies": [ { "id": 564766, "author_name": "Abhishek Thakur", "author_url": "", "post_date": "2019-06-30T00:31:03.977000", "content": "Its a hidden test set. you have to make an inference kernel and for that instead of uploading csv, you been to upload the trained model. if you take a look at the kernel i mentioned, you will understand how it works :)
", "votes": 2, "replies": [] }, { "id": 564767, "author_name": "birinhos", "author_url": "", "post_date": "2019-06-30T00:34:52.440000", "content": "Okay I see ... I upload the model and build the submission file there . Ok, Thanks !
\n\nPS - I was thinkg about contract an Ophthalmologist to build my submission :-) lol
", "votes": 0, "replies": [] } ] }, { "id": 565936, "author_name": "", "author_url": "", "post_date": "2019-07-01T15:28:10.573000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "564763": "you cannot submit a submission.csv in this competition. your model must run on stage 2 dataset. take a look at the example kernel here to know more: https://www.kaggle.com/abhishek/pytorch-inference-kernel", "564762": "hello,\n\nI have \"submission error\". \nIf I pass the mouse over info it says \"No additional details provided for this error \"\n\nI train my model at home and build de submission file at home. \nI upload this file inside the kernel editor using \"+ Add Dataset\". \nthe file goes to : \n\n`../input/submission/submission.csv`\n\nThen I load this file and save it again ... like This : \n\n```\nimport numpy as np\nimport pandas as pd \nimport os\n\nsub = pd.read_csv('../input/submission/submission.csv')\nsub.diagnosis=sub.diagnosis.astype('int') #just to make sure\nsub.to_csv('submission.csv',index=False)\n```\n\n\nThat is all I have but on the submission, I have \"error\" (?)\n\nAny suggestions?\n\nThanks! ", "565655": "Did you find reducing the batch size solved your problem?\n\nIf so please comment [here](https://www.kaggle.com/c/aptos2019-blindness-detection/discussion/98087).", "565067": "Try reducing batch size. More batch size increases RAM and GPU mem", "564794": "upload the model run it inside but I have the error on the submission : \n\n\"Kernel Out of Resources\"\n\nEDIT: Again after I stop all kernels I was using... I think I have spent all of my 5 tries for today ...\nThere is no logic in analyzing images less than 1/4 of the size for this problem ...", "564783": "I have upload my model in keras format (model.h5). \n\nI can load the model : \n\n`model = load_model('../input/modeltoload/model.h5')`\n\nbut when I try to execute predictions: \n\n```\npred_test=model.predict(X_test)\n\n```\n I have error : \n\n---------------------------------------------------------------------------\nFailedPreconditionError Traceback (most recent call last)\n", "564765": "Sorry, I am new to this. What is stage 2 dataset?\n\nDo you mean there is no way to build the model at home and upload the results only? \n\nIf that is so I don't think running the code here will work for me because I want to use the full resolution of the images ... (no memory or CPU/GPU/TPU enough here)\n", "565936": "" } }