Could not broadcast input array from shape (300,) into shape when mapping FastText word vectors with word in the dataset

I am working on creating a NLP model for my final year project. I am currently using Tensorflow Keras LSTM model to train the model. I have found a online guide for this as my dataset is in Sinhala but the code in that tutorial is old or has some issues. Currently when ever I runs

import fasttext
import fasttext.util

ft = fasttext.load_model("cc.si.300.bin")
ft.get_dimension()

# Mapping FastText word vectors with word in the dataset 
embeddings_matrix = np.zeros((vocab_size+1, embedding_dim));
for word, i in word_index.items():
    embedding_vector = ft.get_word_vector(word)
    print(word)
    if embedding_vector is not None:
        embeddings_matrix[i] = embedding_vector;

I get the error below.

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-44-797ce5392360> in <cell line: 9>()
     11     print(word)
     12     if embedding_vector is not None:
---> 13         embeddings_matrix[i] = embedding_vector;

ValueError: could not broadcast input array from shape (300,) into shape (16,)

As I am new to this, I don't know how to fix this issue exactly and I was not able to find a proper answer in the site too.

You're not showing the values of vocab_size & embedding_dim that you're using to create your embeddings_matrix. (Also, at the time you're operating on it, word_index hasn't yet been established to hold any value.)

But, that seems like an error one might get trying to put a 300-dimensional vector into another array where it doesn't fit.

What are the values of vocab_size & embedding_dim – or equivalently what's embeddings_matrix.shape just before you get the error? If its 2nd dimension isn't 300, it should be, and you should change your (unshown) code that sets up the controlling values to ensure that embedding_matrix is the right size.

(More foundationally, are you sure you need to copy each word's vector out of the ft model into your own matrix? You might not, depending on what you're planning to do next.)

I have found out the mistake i made. I am not knowledge enough to explain this but it seems the embedding_dim I defied earlier in the code doesn't support when it comes to the implementing the fasttext functionalities.

# Mapping FastText word vectors with word in the dataset 
embedding_dim = ft.get_dimension()
embeddings_matrix = np.zeros((vocab_size+1, embedding_dim))
for word, i in word_index.items():
    embedding_vector = ft.get_word_vector(word)
    if embedding_vector is not None:
        embeddings_matrix[i] = embedding_vector

Thank you for your help