params['normalise_data'] = 4 # adding more values to params, what does this value signify? # params['reset_state']=res # adding more values to params, what does this value signify? # params['seq_length']=seq # adding more values to params, what does this value signify? # params["reload_data"] = 0 # adding more values to params, what does this value signify? # params = config.update_params(params) # New param values are updated# #params["model"] = "kfl_QRf" if params["model"] == "lstm": Model = lstm(params=params) elif params["model"] == "kfl_QRf": Model = kfl_QRf(params=params) elif params["model"] == "kfl_Rf": Model = kfl_Rf(params=params) elif params["model"] == "kfl_QRFf": Model = kfl_QRFf(params=params) elif params["model"] == "kfl_K": Model = kfl_K(params=params) params["rn_id"]="dobuleloss081500_nrm4_seq%i_res%i_keep%f_lr%f"%(seq,res,rnn_keep_prob,params["lr"]) # adding more values to params, what does this value signify? # params=config.update_params(params) # New param values are updated# # (params, X_train, Y_train, F_list_train, G_list_train, S_Train_list, R_L_Train_list, # X_test, Y_test, F_list_test, G_list_test, S_Test_list, R_L_Test_list) = \ # dut.prepare_training_set(params) # This is where the observation set will be given# show_every = 1 #(index_train_list, S_Train_list) = dut.get_seq_indexes(params, S_Train_list) #(index_test_list, S_Test_list) = dut.get_seq_indexes(params, S_Test_list) batch_size = params['batch_size'] #n_train_batches = len(index_train_list) #n_train_batches /= batch_size #n_test_batches = len(index_test_list) #n_test_batches /= batch_size #params['training_size'] = len(X_train) * params['seq_length']
params['Qn_hidden'] = h params['Rn_hidden'] = h params['Kn_hidden'] = h if params["model"] == "lstm": tracker = lstm(params=params) elif params["model"] == "kfl_QRf": tracker = kfl_QRf(params=params) elif params["model"] == "kfl_Rf": tracker = kfl_Rf(params=params) elif params["model"] == "kfl_f": tracker = kfl_f(params=params) elif params["model"] == "kfl_QRFf": tracker = kfl_QRFf(params=params) elif params["model"] == "kfl_K": tracker = kfl_K(params=params) params = config.update_params(params) show_every = 1 ut.start_log(params) ut.log_write("Model training started", params) median_result_lst, mean_result_lst = train(tracker, params) # if params["data_mode"]=="David": # np.savetxt('/home/coskun/PycharmProjects/poseft/trials/res/'+params["sequence"]+'/'+msg,mean_result_lst) # else: # np.savetxt('/home/coskun/PycharmProjects/poseft/trials/res/klstm/'+params["sequence"]+'/'+msg,median_result_lst) # np.min(median_result_lst,axis=1) # print median_result_lst # print mean_result_lst # print 'Min...'