The events which are happening on day to day basis and the events that have happened earlier around us can be considered as real-world events. And to detect such events is a crucial process from a stream of text which is available on the web. Event detection is research which going on for the last decade and it is still in a developing phase and it is yet to mature fully. In this research paper, a novel model for detecting events from the stream of Assamese text using Event Detection Long Short Term Memory (EDLSTM) is proposed. The model is developed to detect real-world events. The EDLSTM can capture the context of the sentences and detect the events along with temporal events corresponding to them. The model's accuracy is put to test using different optimizers that are used to minimize the losses during the training phase. For better results, the researchers have compared three optimizers Adam, SGD, and RMSprop to find the one which causes less loss while training the model for event detection. The model is trained using SGD optimizer with the Assamese Wikipedia to have a broader knowledge of the general information of events.