train_inductive.py

import random
import pandas as pd
from datetime import datetime
from collections import defaultdict,Counter
import matplotlib.pyplot as plt
import numpy as np
import random
import pickle
import argparse
import numpy as np
import pickle
import time
import torch
import torch.nn as nn
import torch.optim as optim
from tqdm import tqdm
import sklearn
from sklearn.model_selection import train_test_split
import csv
from torch.autograd import Variable
from torch.nn import functional as F
from tgg_utils import *
import argparse
from sklearn.decomposition import PCA
from sklearn.cluster import KMeans
from sklearn.metrics import silhouette_samples, silhouette_score
from sklearn.mixture import GaussianMixture
from sklearn.cluster import KMeans
import matplotlib.pyplot as plt
import matplotlib.cm as cm
import scann

from model_classes.inductive_model import EventClusterLSTM,get_event_prediction_rate,get_time_mse,get_topk_event_prediction_rate


parser = argparse.ArgumentParser()
parser.add_argument("--data_path", help="full path of dataset in csv format(start,end,time)",
                    type=str)
parser.add_argument("--gpu_num",help="GPU no. to use, -1 in case of no gpu", type=int)
parser.add_argument("--config_path",help="full path of the folder where models and related data need to be saved", type=str)
parser.add_argument("--num_epochs",help="Number of epochs for training", type=int)
parser.add_argument("--graph_sage_embedding_path",help="GraphSage embedding path",type=str)
parser.add_argument("--num_clusters",default=500,help="Cluster size for the MND",type=int)
parser.add_argument("--window_interactions",default=6,help="Interaction window", type=int)
parser.add_argument("--l_w",default=20,help="lw", type=int)
parser.add_argument("--filter_walk",default=2,help="filter_walk", type=int)


args = parser.parse_args()




print(args)
data_path = args.data_path
gpu_num = args.gpu_num
config_path = args.config_path
num_epochs = args.num_epochs
graphsage_embeddings_path = args.graph_sage_embedding_path
num_clusters = args.num_clusters
window_interactions = args.window_interactions
l_w = args.l_w
filter_walk = args.filter_walk
data= pd.read_csv(data_path)
data = data[['start','end','days']]
#data = data.drop_duplicates(['start','end','days'])

node_set = set(data['start']).union(set(data['end']))
print("number of nodes,",len(node_set))
node_set.update('end_node')
max_days = max(data['days'])
print("Minimum, maximum timestamps",min(data['days']),max_days)
data = data.sort_values(by='days',inplace=False)
print("number of interactions," ,data.shape[0])

### configurations 

strictly_increasing_walks = True
num_next_edges_to_be_stored = 100

edges = []
node_id_to_object = {}
undirected=True
for start,end,days in data[['start','end','days']].values:
    if start not in node_id_to_object:
        node_id_to_object[start] = Node(id=start,as_start_node= [],as_end_node=[])
    if end not in node_id_to_object:
        node_id_to_object[end] = Node(id=end,as_start_node= [],as_end_node=[])
        
    edge= Edge(start=start,end=end,time=days,outgoing_edges = [],incoming_edges=[])
    edges.append(edge) 
    node_id_to_object[start].as_start_node.append(edge)
    node_id_to_object[end].as_end_node.append(edge)
    if undirected:
        edge= Edge(start=end,end=start,time=days,outgoing_edges = [],incoming_edges=[])
        edges.append(edge) 
        node_id_to_object[end].as_start_node.append(edge)
        node_id_to_object[start].as_end_node.append(edge)        
print("length of edges,", len(edges), " length of nodes,", len(node_id_to_object))

ct = 0
#for edge in tqdm(edges): 
for edge in edges:
        
    end_node_edges = node_id_to_object[edge.end].as_start_node
    index = binary_search_find_time_greater_equal(end_node_edges,edge.time,strictly=strictly_increasing_walks)
    if index != -1:
        if strictly_increasing_walks:
            edge.outgoing_edges = end_node_edges[index:index+num_next_edges_to_be_stored]
        else:
            edge.outgoing_edges = [item for item in end_node_edges[index:index+num_next_edges_to_be_stored] if item.end != edge.start]

    start_node_edges = node_id_to_object[edge.start].as_end_node
    index = binary_search_find_time_lesser_equal(start_node_edges,edge.time,strictly=strictly_increasing_walks)
    if index != -1:
        if strictly_increasing_walks:
            edge.incoming_edges = start_node_edges[max(0,index-num_next_edges_to_be_stored):index+1]
        else:
            edge.incoming_edges = [item for item in start_node_edges[max(0,index-num_next_edges_to_be_stored):index+1] if item.start != edge.end]
        edge.incoming_edges.reverse()

    ct += 1

for edge in tqdm(edges):
    edge.out_nbr_sample_probs = []
    edge.in_nbr_sample_probs = []

    if len(edge.outgoing_edges) >= 1:
        edge.out_nbr_sample_probs,edge.outJ, edge.outq =  prepare_alias_table_for_edge(edge,incoming=False,window_interactions=window_interactions) ### Gaussian Time Sampling
        
    if len(edge.incoming_edges) >= 1:
        edge.in_nbr_sample_probs,edge.inJ, edge.inq =  prepare_alias_table_for_edge(edge,incoming=True,window_interactions=2)

temporal_graph_original = defaultdict(lambda: defaultdict(lambda:defaultdict(lambda: 0)))
for start,end,day in data[['start','end','days']].values:
    temporal_graph_original[day][start][end] += 1
    if undirected:
        temporal_graph_original[day][end][start] += 1
vocab = {'<PAD>': 0,'end_node':1}
for node in data['start']:
    if node not in vocab:
        vocab[node] = len(vocab)
for node in data['end']:
    if node not in vocab:
        vocab[node] = len(vocab)
print("Length of vocab, ", len(vocab))
print("Id of end node , ",vocab['end_node'])
pad_token = vocab['<PAD>']


def sample_random_Walks():
    print("Running Random Walk, Length of edges, ", len(edges))
    random_walks = []
    for edge in edges:
        random_walks.append(run_random_walk_without_temporal_constraints(edge,l_w,1))
    print("length of collected random walks,", len(random_walks))
    random_walks = [item for item in random_walks if item is not None]
    print("length of collected random walks after removing None,", len(random_walks))
    random_walks = [clean_random_walk(item) for item in random_walks]
    random_walks = [item for item in random_walks if filter_rw(item,filter_walk)]
    print("Length of random walks after removing short ranadom walks", len(random_walks))
    return random_walks
def get_sequences_from_random_walk(random_walks):
    sequences = [convert_walk_to_seq(item) for item in random_walks]
    sequences = [convert_seq_to_id(vocab, item) for item in sequences]
    sequences = [get_time_delta(item,0) for item in sequences]
    return sequences
random_walks = sample_random_Walks()
sequences = get_sequences_from_random_walk(random_walks)
print("Average length of random walks")
lengths = []
for wk in random_walks:
    lengths.append(len(wk))
print("Mean length {} and Std deviation {}".format(str(np.mean(lengths)),str(np.std(lengths))))

inter_times = []
for seq in sequences:
    for item in seq:
        inter_times.append(np.log(item[2]))
mean_log_inter_time = np.mean(inter_times)
std_log_inter_time = np.std(inter_times)
print("mean log inter time and std log inter time ", mean_log_inter_time,std_log_inter_time)


node_embeddings_feature = pickle.load(open(graphsage_embeddings_path,"rb"))  ### We tried deep walk or node2vec embeddings

node_emb_size = 128
node_embedding_matrix = np.zeros((len(vocab),node_emb_size))
for item in vocab:
    if item == '<PAD>':
        arr = np.zeros(node_emb_size)
    elif item == 'end_node':
        arr = np.random.uniform(-0.1,0.1,node_emb_size)
    else:
        arr = node_embeddings_feature[item]
    index= vocab[item]
    node_embedding_matrix[index] = arr
print("Node embedding matrix, shape,", node_embedding_matrix.shape)
normalized_dataset = node_embedding_matrix[1:] / np.linalg.norm(node_embedding_matrix[1:], axis=1)[:, np.newaxis]

searcher = scann.scann_ops_pybind.builder(normalized_dataset, 20, "dot_product").tree(
    num_leaves=200, num_leaves_to_search=1000, training_sample_size=250000).score_ah(
    2, anisotropic_quantization_threshold=0.2).reorder(100).build()
searcher_1 = scann.scann_ops_pybind.builder(normalized_dataset, 1, "dot_product").tree(
    num_leaves=1000, num_leaves_to_search=1000, training_sample_size=3000).score_ah(
    2, anisotropic_quantization_threshold=0.2).reorder(100).build()

pca = PCA(n_components=110)
node_embedding_matrix_pca = pca.fit_transform(node_embedding_matrix[2:]) ### since 0th and 1st index is of padding and end node
print("PCA variance explained",np.sum(pca.explained_variance_ratio_))
pca_2 = PCA(n_components=2)
node_embedding_matrix_pca_2 = pca_2.fit_transform(node_embedding_matrix_pca)
print(np.sum(pca_2.explained_variance_ratio_))



kmeans = KMeans(n_clusters=num_clusters, random_state=0,max_iter=10000).fit(node_embedding_matrix_pca)
labels = kmeans.labels_
label_freq = Counter(labels)
print("max size cluster, ", max(label_freq.values()))
cluster_labels = [0,1]+[item+2 for item in labels]
#print(cluster_labels)
print(len(cluster_labels))
max_label = np.max(cluster_labels)
print("Max cluster label",max_label)

pad_cluster_id = cluster_labels[0]
print("Pad cluster id, ", pad_cluster_id)

num_components = np.max(cluster_labels) + 1
print("num_components ",num_components)

reverse_vocab= {value:key for key,value in vocab.items() }

def add_cluster_id(sequence,labels):
    return [(a,b,c,labels[a]) for a,b,c in sequence]
sequences = [add_cluster_id(sequence,cluster_labels) for sequence in sequences]


import os
import json
import pickle
config_dir = config_path ### Change in random walks
isdir = os.path.isdir(config_dir) 
if not isdir:
    os.mkdir(config_dir)
isdir = os.path.isdir(config_dir+"/models") 
if not isdir:
    os.mkdir(config_dir+"/models")

start_node_and_times = [(seq[0][0],seq[0][1],seq[0][2],seq[0][3]) for seq in sequences ]

pickle.dump(vocab,open(config_dir+"/vocab.pkl","wb"))
pickle.dump(cluster_labels,open(config_dir+"/cluster_labels.pkl","wb"))
pickle.dump(pca,open(config_dir+"/pca.pkl","wb"))
pickle.dump(kmeans,open(config_dir+"/kmeans.pkl","wb"))
pickle.dump({"mean_log_inter_time":mean_log_inter_time,"std_log_inter_time":std_log_inter_time},open(config_dir+"/time_stats.pkl","wb"))
np.save(open(config_dir+"/node_embedding_matrix.npy","wb"),node_embedding_matrix)

def get_X_Y_T_CID_from_sequences(sequences):  ### This also need to provide the cluster id of the 
    seq_X = []
    seq_Y = []
    seq_Xt = []
    seq_Yt = []
    seq_XDelta = []
    seq_YDelta = []
    seq_XCID = []
    seq_YCID = []
    for seq in sequences:
        seq_X.append([item[0] for item in seq[:-1]])  ## O contain node id
        seq_Y.append([item[0] for item in seq[1:]])
        seq_Xt.append([item[1] for item in seq[:-1]])   ## 1 contain timestamp
        seq_Yt.append([item[1] for item in seq[1:]])
        seq_XDelta.append([item[2] for item in seq[:-1]])   ## 2 contain delta from previous event
        seq_YDelta.append([item[2] for item in seq[1:]])
        seq_XCID.append([item[3] for item in seq[:-1]])   ## 3 contains the cluster id
        seq_YCID.append([item[3] for item in seq[1:]])
    X_lengths = [len(sentence) for sentence in seq_X]
    Y_lengths = [len(sentence) for sentence in seq_Y]
    max_len = max(X_lengths)
    return seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,max_len,seq_XCID,seq_YCID
#seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,max_len,seq_XCID,seq_YCID = get_X_Y_T_CID_from_sequences(sequences)

def get_batch(start_index,batch_size,seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,seq_XCID,seq_YCID):
    batch_X = seq_X[start_index:start_index+batch_size]
    batch_Y = seq_Y[start_index:start_index+batch_size]
    batch_Xt = seq_Xt[start_index:start_index+batch_size]
    batch_Yt = seq_Yt[start_index:start_index+batch_size] 
    batch_XDelta = seq_XDelta[start_index:start_index+batch_size]
    batch_YDelta = seq_YDelta[start_index:start_index+batch_size]   
    batch_X_len = X_lengths[start_index:start_index+batch_size]
    batch_Y_len = Y_lengths[start_index:start_index+batch_size]
    batch_XCID = seq_XCID[start_index:start_index+batch_size]
    batch_YCID = seq_YCID[start_index:start_index+batch_size] 
    max_len = max(batch_X_len)
    #print(max_len)
    pad_batch_X = np.ones((batch_size, max_len),dtype=np.int64)*pad_token
    pad_batch_Y = np.ones((batch_size, max_len),dtype=np.int64)*pad_token
    pad_batch_Xt = np.ones((batch_size, max_len),dtype=np.float32)*pad_token
    pad_batch_Yt = np.ones((batch_size, max_len),dtype=np.float32)*pad_token
    pad_batch_XDelta = np.ones((batch_size, max_len),dtype=np.float32)*pad_token
    pad_batch_YDelta = np.ones((batch_size, max_len),dtype=np.float32)*pad_token
    pad_batch_XCID = np.ones((batch_size, max_len),dtype=np.int64)*pad_cluster_id
    pad_batch_YCID = np.ones((batch_size, max_len),dtype=np.int64)*pad_cluster_id
    for i, x_len in enumerate(batch_X_len):
        #print(i,x_len,len(batch_X[i][:x_len]),len(pad_batch_X[i, 0:x_len]))
        pad_batch_X[i, 0:x_len] = batch_X[i][:x_len]
        pad_batch_Y[i, 0:x_len] = batch_Y[i][:x_len]
        pad_batch_Xt[i, 0:x_len] = batch_Xt[i][:x_len]
        pad_batch_Yt[i, 0:x_len] = batch_Yt[i][:x_len]
        pad_batch_XDelta[i, 0:x_len] = batch_XDelta[i][:x_len]
        pad_batch_YDelta[i, 0:x_len] = batch_YDelta[i][:x_len]
        pad_batch_XCID[i, 0:x_len] = batch_XCID[i][:x_len]
        pad_batch_YCID[i, 0:x_len] = batch_YCID[i][:x_len]
    pad_batch_X =  torch.LongTensor(pad_batch_X).to(device)
    pad_batch_Y =  torch.LongTensor(pad_batch_Y).to(device)
    pad_batch_Xt =  torch.Tensor(pad_batch_Xt).to(device)
    pad_batch_Yt =  torch.Tensor(pad_batch_Yt).to(device)
    pad_batch_XDelta =  torch.Tensor(pad_batch_XDelta).to(device)
    pad_batch_YDelta =  torch.Tensor(pad_batch_YDelta).to(device)
    batch_X_len = torch.LongTensor(batch_X_len).to(device)
    batch_Y_len = torch.LongTensor(batch_Y_len).to(device)
    pad_batch_XCID =  torch.LongTensor(pad_batch_XCID).to(device)
    pad_batch_YCID =  torch.LongTensor(pad_batch_YCID).to(device)
    return pad_batch_X,pad_batch_Y,pad_batch_Xt,pad_batch_Yt,pad_batch_XDelta,pad_batch_YDelta,batch_X_len,batch_Y_len,pad_batch_XCID,pad_batch_YCID

def data_shuffle(seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,seq_XCID,seq_YCID):
    indices = list(range(0, len(seq_X)))
    random.shuffle(indices)
    #### Data Shuffling
    seq_X = [seq_X[i] for i in indices]   #### 
    seq_Y = [seq_Y[i] for i in indices]
    seq_Xt = [seq_Xt[i] for i in indices]
    seq_Yt = [seq_Yt[i] for i in indices]    
    seq_XDelta = [seq_XDelta[i] for i in indices]
    seq_YDelta = [seq_YDelta[i] for i in indices]
    X_lengths = [X_lengths[i] for i in indices]
    Y_lengths = [Y_lengths[i] for i in indices]
    seq_XCID = [seq_XCID[i] for i in indices]
    seq_YCID = [seq_YCID[i] for i in indices]
    return seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,seq_XCID,seq_YCID
seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,max_len,seq_XCID,seq_YCID = get_X_Y_T_CID_from_sequences(sequences)
print("Max lengths of walks", max_len)
seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,seq_XCID,seq_YCID = data_shuffle(seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,seq_XCID,seq_YCID)

def evaluate_model(elstm,batch_size=128):
    elstm.eval()
    running_loss= []
    running_event_loss = []
    running_time_loss= []
    event_prediction_rates = []
    topk_event_prediction_rates = []
    topk10_event_prediction_rates = []
    topk20_event_prediction_rates = []

    random_walks = sample_random_Walks()
    sequences = get_sequences_from_random_walk(random_walks)
    sequences = [add_cluster_id(sequence,cluster_labels) for sequence in sequences]
    seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,max_len,seq_XCID,seq_YCID= get_X_Y_T_CID_from_sequences(sequences)

    seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,seq_XCID,seq_YCID = data_shuffle(seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,seq_XCID,seq_YCID)

    for start_index in range(0, len(seq_X),batch_size):

        if start_index+batch_size < len(seq_X) and start_index + batch_size < 10000:
            #try:
                pad_batch_X,pad_batch_Y,pad_batch_Xt,pad_batch_Yt,pad_batch_XDelta,pad_batch_YDelta,batch_X_len,batch_Y_len,pad_batch_XCID,pad_batch_YCID = get_batch(start_index,batch_size,seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,seq_XCID,seq_YCID)
                mask_distribution = pad_batch_Y!=0
                mask_distribution = mask_distribution.to(device)
                num_events_time_loss = mask_distribution.sum().item()

                Y_hat,inter_time_log_loss,elbo,log_dict,Y_clusterid = elstm(X=pad_batch_X,Y=pad_batch_Y,
                            Xt=pad_batch_Xt,Yt = pad_batch_Yt,
                            XDelta = pad_batch_XDelta,YDelta = pad_batch_YDelta,
                            X_lengths= batch_X_len,mask=mask_distribution,XCID=pad_batch_XCID,YCID=pad_batch_YCID,epoch=10,kl_weight=kl_weight)   
                Y = pad_batch_Y
                Y = Y.view(-1)
                Y_hat = Y_hat.view(-1, Y_hat.shape[-1])   
                mask_distribution = mask_distribution.view(-1)
                Y_hat = Y_hat.detach().cpu().numpy()
                Y = Y.detach().cpu().numpy()
                # neighbors, distances = searcher.search_batched(node_embedding_matrix[:100], leaves_to_search=1000, pre_reorder_num_neighbors=1000)
                Y_hat, distances = searcher.search_batched(Y_hat, leaves_to_search=1000, pre_reorder_num_neighbors=1000)
                Y_hat = Y_hat+1
                topk_event_prediction_rates.append(get_topk_event_prediction_rate(Y,Y_hat,k=5,ignore_Y_value=0))
                topk10_event_prediction_rates.append(get_topk_event_prediction_rate(Y,Y_hat,k=10,ignore_Y_value=0))
                topk20_event_prediction_rates.append(get_topk_event_prediction_rate(Y,Y_hat,k=20,ignore_Y_value=0))

    print("Event prediction rate@top5:,", np.mean(topk_event_prediction_rates))
    print("Event prediction rate@top10:,", np.mean(topk10_event_prediction_rates))
    print("Event prediction rate@top20:,", np.mean(topk20_event_prediction_rates))



if gpu_num == -1:
    device = torch.device("cpu")
else:
    device = torch.device("cuda:{}".format(str(gpu_num)) if torch.cuda.is_available() else "cpu")

print("Computation device, ", device)



batch_size = 128  ### Experiment wit 



import copy

wt_update_ct = 0
debug = False
print_ct = 10000000
best_loss = 1000000000
best_model = None
celoss = nn.CrossEntropyLoss(ignore_index=-1) #### -1 is padded     
celoss_cluster = nn.CrossEntropyLoss(ignore_index=pad_cluster_id)
epoch_wise_loss = []
kl_weight = 0.00001


elstm = EventClusterLSTM(vocab=vocab,node_pretrained_embedding = node_embedding_matrix,nb_layers=2, nb_lstm_units=128,time_emb_dim= 64,
        embedding_dim=128, batch_size= batch_size,device=device,
        mean_log_inter_time=mean_log_inter_time,
        std_log_inter_time=std_log_inter_time,num_components=num_components)
elstm = elstm.to(device)
optimizer = optim.Adam(elstm.parameters(), lr=.001)
num_params = sum(p.numel() for p in elstm.parameters() if p.requires_grad)
print(" ##### Number of parameters#### " ,num_params)

    

for epoch in range(0,num_epochs+1):
    print("KL weight is , ", kl_weight)
    elstm.train()
    try:
        print("Epoch and num of batches, :",epoch , len(seq_X)/batch_size)
    except:
        print("Seqx not defined")
    running_loss= []
    running_event_loss = []
    running_time_loss= []
    running_recon_loss = []
    running_kl_loss = []
    running_cluster_loss = []
    
    event_prediction_rates = []
    topk_event_prediction_rates = []
    topk10_event_prediction_rates = []

    random_walks = sample_random_Walks()
    sequences = get_sequences_from_random_walk(random_walks)
    sequences = [add_cluster_id(sequence,cluster_labels) for sequence in sequences]
    start_node_and_times += [(seq[0][0],seq[0][1],seq[0][2],seq[0][3]) for seq in sequences ]
    seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,max_len,seq_XCID,seq_YCID= get_X_Y_T_CID_from_sequences(sequences)

    seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,seq_XCID,seq_YCID = data_shuffle(seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,seq_XCID,seq_YCID)


    
    for start_index in range(0, len(seq_X),batch_size):
        if start_index+batch_size < len(seq_X):
            print("\r%d/%d" %(int(start_index),len(seq_X)),end="")
            pad_batch_X,pad_batch_Y,pad_batch_Xt,pad_batch_Yt,pad_batch_XDelta,pad_batch_YDelta,batch_X_len,batch_Y_len,pad_batch_XCID,pad_batch_YCID = get_batch(start_index,batch_size,seq_X,seq_Y,seq_Xt,seq_Yt,seq_XDelta,seq_YDelta,X_lengths,Y_lengths,seq_XCID,seq_YCID)
            elstm.zero_grad()
            mask_distribution = pad_batch_Y!=0
            mask_distribution = mask_distribution.to(device)
            num_events_time_loss = mask_distribution.sum().item()
            Y_hat,inter_time_log_loss,elbo,log_dict,Y_clusterid = elstm(X=pad_batch_X,Y=pad_batch_Y,
                            Xt=pad_batch_Xt,Yt = pad_batch_Yt,
                            XDelta = pad_batch_XDelta,YDelta = pad_batch_YDelta,
                            X_lengths= batch_X_len,mask=mask_distribution,XCID=pad_batch_XCID,YCID=pad_batch_YCID,epoch=10,kl_weight=kl_weight)
            inter_time_log_loss *= mask_distribution
            loss_time = (-1)*inter_time_log_loss.sum()*1.00/num_events_time_loss
            Y_clusterid = Y_clusterid.view(-1, Y_clusterid.shape[-1])
            pad_batch_YCID = pad_batch_YCID.view(-1)
            loss_cluster = celoss_cluster(Y_clusterid,pad_batch_YCID)
            loss = elbo+loss_time + loss_cluster
            
            loss.backward()
            torch.nn.utils.clip_grad_norm_(elstm.parameters(), 1)
            optimizer.step()
            running_loss.append(loss.item())
            running_event_loss.append(elbo.item())
            running_time_loss.append(loss_time.item())  
            running_cluster_loss.append(loss_cluster.item())
            running_recon_loss.append(log_dict['reconstruction'])
            running_kl_loss.append(log_dict['kl'])
            wt_update_ct += 1
            if wt_update_ct%print_ct == 0 and debug:
                print("Running Loss :, ",np.mean(running_loss[-print_ct:]) )
                print("Running event elbo loss: ", np.mean(running_event_loss[-print_ct:]),np.std(running_event_loss[-print_ct:]))
                print("Running time log loss: ", np.mean(running_time_loss[-print_ct:]),np.std(running_time_loss[-print_ct:]))
                print("Running cluster ce loss: ", np.mean(running_cluster_loss[-print_ct:]),np.std(running_cluster_loss[-print_ct:]))
                print("Running recon elbo loss: ", np.mean(running_recon_loss[-print_ct:]),np.std(running_recon_loss[-print_ct:]))
                print("Running kl elbo loss: ", np.mean(running_kl_loss[-print_ct:]),np.std(running_kl_loss[-print_ct:]))
                print()

    print("\nEpoch done")
    print_ct = len(running_loss)
    print("Running Loss :, ",np.mean(running_loss[-print_ct:]) )
    print("Running event elbo loss: ", np.mean(running_event_loss[-print_ct:]),np.std(running_event_loss[-print_ct:]))
    print("Running time log loss: ", np.mean(running_time_loss[-print_ct:]),np.std(running_time_loss[-print_ct:]))
    print("Running cluster ce loss: ", np.mean(running_cluster_loss[-print_ct:]),np.std(running_cluster_loss[-print_ct:]))
    print("Running recon elbo loss: ", np.mean(running_recon_loss[-print_ct:]),np.std(running_recon_loss[-print_ct:]))
    print("Running kl elbo loss: ", np.mean(running_kl_loss[-print_ct:]),np.std(running_kl_loss[-print_ct:]))
    #break
    if epoch%20 == 0:
        print("Running evaluation")
        evaluate_model(elstm)
    state = {
        'model':elstm.state_dict(),
        'optimizer': optimizer.state_dict(),
        'loss':np.mean(running_loss)
    }
    epoch_wise_loss.append(np.mean(running_loss))
    if epoch%20 == 0:
        torch.save(state, config_dir+"/models/{}.pth".format(str(epoch)))
        import h5py
        hf = h5py.File(config_dir+'/start_node_and_times.h5', 'w')
        hf.create_dataset('1', data=start_node_and_times)
        hf.close()
    if np.mean(running_loss) < best_loss:
        print("### Saving the best model ####")
        best_loss = np.mean(running_loss)
        best_elstm = copy.deepcopy(elstm.state_dict())
        torch.save(state, config_dir+"/models/best_model.pth".format(str(epoch)))