Open antoajayraj opened 8 years ago
Running Torch/Googlenet using the following parameters does not go beyond 0% accuracy.
nGPU 4 1 -- ignore option optimState nDonkeys 16 2 -- ignore option netType -- ignore option data nEpochs 50 55 -- ignore option cache -- ignore option retrain => Creating model from file: models/googlenet.lua converting module to nn.DataParallelTable
Even changing the batch sizes does not get the accuracy beyond 0%. Tried running the code for more than 10 hours.
Running Torch/Googlenet using the following parameters does not go beyond 0% accuracy.
nGPU 4 1 -- ignore option optimState nDonkeys 16 2 -- ignore option netType -- ignore option data nEpochs 50 55 -- ignore option cache -- ignore option retrain => Creating model from file: models/googlenet.lua converting module to nn.DataParallelTable
Even changing the batch sizes does not get the accuracy beyond 0%. Tried running the code for more than 10 hours.