Real world applications of these like sentence transformers embeddings model, which we mostly use for embedding generation and semantic search, this tutorial is great and helps us to grasp and picture the workings behind a word embeddings models. This is a great demonstration , really great. Thank you for your time and making such content, especially these scratch ones. Thanks Man, please keep going.
@@MachineLearningWithJayHello, I found your video very informative and clear fundamentals. Just a quick question, what if I have a corpus of total 37million sentences and 150 thousand unique words. How do I train my model without going into complexity of having 150thousand nodes in input for one_hot encoding?
When I see 13k subscribers; 2.5k views and only 85 likes I see why good programmers and good data scientists are rare. You can't get a more concise hands on presentation that teaches you how to create a language model.
good and understandable explanation. It would be great if you may upload a video wherein character level embeddings are also learnt where we may split a word in characters- and do the process.
When creating the bigrams (for each of two adjacent words), the order should matter. But why you insert all the possible combinations in the bigrams list? I think the order of the words as it appear in the corpus is important to capture the relationship between adjacent words.
Hi Pavan, sorry for not uploading videos. I will upload the next video very soon in the upcoming few days. But might take some time to upload more other videos. Thanks for supporting the channel. And I am glad you find my channel useful 🙂
Hi Jay "Don't laugh at me, I have always been programing of a external compiler (Java / Matlab) - After launching your github template, the jupyter blank code is not in edit mode, what must I do
Hello, sir, and same to you. Thank you for this video, I show you the same. India is now Bhaarat Aatmanirbhar and have popular searching engines looking for talent to engage! Indian is in need of popular searching engines and any help is good.
You never stated what Python packages are needed to do the creation of the bigrams, the tokenization, and the removal of stopwords. Are you using NLTK?
You are using wrong weights for plotting. weights[0] refers to the weights of the first layer, weights[1] refers to the biases of the first layer, weights[2] refers to the weights of the second layer, and weights[3] refers to the biases of the second layer. You can confirm this by running the following code: for layer in model.layers: print(layer.get_config(), layer.get_weights()) Alternatively, you can also use the following where the index [1] refers to the second layer (layer index starts from 0) and the second index [0] refers to weights. weights = model.layers[1].get_weights()[0]
Real world applications of these like sentence transformers embeddings model, which we mostly use for embedding generation and semantic search, this tutorial is great and helps us to grasp and picture the workings behind a word embeddings models. This is a great demonstration , really great. Thank you for your time and making such content, especially these scratch ones. Thanks Man, please keep going.
Glad to help! I have started uploading videos on transformers, and more coming soon! Let me know how they are, would appreciate the feedback.
Am happy that I found this channel. You have a gift of teaching.
Thank you! It means a lot to me 😇
@@MachineLearningWithJayHello, I found your video very informative and clear fundamentals. Just a quick question, what if I have a corpus of total 37million sentences and 150 thousand unique words. How do I train my model without going into complexity of having 150thousand nodes in input for one_hot encoding?
Thanks jay, the whole playlist is awesome.. Thank you so much for the creating these wonderful videos and educating us...
You’re welcome
When I see 13k subscribers; 2.5k views and only 85 likes I see why good programmers and good data scientists are rare. You can't get a more concise hands on presentation that teaches you how to create a language model.
Hehe… really appreciate your comment 🥺
Super good explanation. Very indepth insights given. Thank you for taking time and explaining it
Very Informative. Thanks for doing this.
Glad to help 😇
Thank you for the video as well as the whole code 😊 You could set a random seed to obtain always the same randomly initialized weights.
Crisp n Clear. Thanks
You’re welcome! Glad it was helpful
good and understandable explanation. It would be great if you may upload a video wherein character level embeddings are also learnt where we may split a word in characters- and do the process.
When creating the bigrams (for each of two adjacent words), the order should matter. But why you insert all the possible combinations in the bigrams list?
I think the order of the words as it appear in the corpus is important to capture the relationship between adjacent words.
I really liked. Wonderful tutorial.
thanks i really enjoy the playlist
Thank you!
You're welcome!
Thank you for your teaching
thanks , that was very useful
You're welcome!
thank you
@@maryamyuusufi7703 you’re welcome
man you are doing really well. you stopped uploading videos on machine learning algorithms. when will u resume bro.
Hi Pavan, sorry for not uploading videos. I will upload the next video very soon in the upcoming few days. But might take some time to upload more other videos. Thanks for supporting the channel. And I am glad you find my channel useful 🙂
wonderful explanation, please attach your linkedin profile in YT about section
hello bro, when we apply cosine similarity function?
Is this feasible for a very large size vocabulary?
unique_words = set(filtered_data.flat)
How do you make a prediction after u trained your own model ?
Hi Jay "Don't laugh at me, I have always been programing of a external compiler (Java / Matlab) - After launching your github template, the jupyter blank code is not in edit mode, what must I do
Hello, sir, and same to you. Thank you for this video, I show you the same. India is now Bhaarat Aatmanirbhar and have popular searching engines looking for talent to engage! Indian is in need of popular searching engines and any help is good.
Glad to help 😇
King is appearing as a target word twice so two vectors will be created for king ? And then we take average????
You never stated what Python packages are needed to do the creation of the bigrams, the tokenization, and the removal of stopwords. Are you using NLTK?
He coded his own function. See the whole code in Jupiter notebook provided in the description
You are using wrong weights for plotting. weights[0] refers to the weights of the first layer, weights[1] refers to the biases of the first layer, weights[2] refers to the weights of the second layer, and weights[3] refers to the biases of the second layer. You can confirm this by running the following code:
for layer in model.layers:
print(layer.get_config(), layer.get_weights())
Alternatively, you can also use the following where the index [1] refers to the second layer (layer index starts from 0) and the second index [0] refers to weights.
weights = model.layers[1].get_weights()[0]
is this video word2vec from scratch
Yes
Can you please give a private mentor support? Paid service
Hi Ali, I would love to… but currently I am jamm packed in my schedule, so won’t have time to give private mentorship.
@@MachineLearningWithJay it is just one our, questions about multivariate linear regression.
@@alidakhil3554 Alright, can you msg me on whatsapp/mail?
@@MachineLearningWithJay could you please share your contact info? Thanks
title is wrong you should write that how to make pytorch from 🤣🤣🤣
Well done!
@@Saed7630 thanks!