Atharva Maskar
@atharvamaskar.bsky.social
Join me in my journey learning ML, DL & LLMs 🚀
What I love about the OpenAI SDK is how easily it generates structured output exactly the way I need. Here’s the prompt I used to generate relevant and a sneak peek at what the final metadata for a verse looks like:
May 1, 2025 at 7:32 AM
What I love about the OpenAI SDK is how easily it generates structured output exactly the way I need. Here’s the prompt I used to generate relevant and a sneak peek at what the final metadata for a verse looks like:
Used GPT-4o to generate interpretations for each verse. Each entry looks pretty solid so far. Could’ve experimented with models like Claude or LLaMA, but I’ll save that rabbit hole for after the basic version is done.
May 1, 2025 at 5:51 AM
Used GPT-4o to generate interpretations for each verse. Each entry looks pretty solid so far. Could’ve experimented with models like Claude or LLaMA, but I’ll save that rabbit hole for after the basic version is done.
Next, I plan to write an interpretation for each verse and relate it to modern-day challenges people face. This would make it easier for the RAG system to retrieve and rank verses. This is the prompt I am planning to go with.
May 1, 2025 at 5:46 AM
Next, I plan to write an interpretation for each verse and relate it to modern-day challenges people face. This would make it easier for the RAG system to retrieve and rank verses. This is the prompt I am planning to go with.
3/n) We can calculate the mean and standard deviation of the trees planted by scaling the mean and standard deviation of the centuries scored accordingly.
This gives us a distribution having mean as 455.0 and standard deviation as 92.0
This gives us a distribution having mean as 455.0 and standard deviation as 92.0
November 24, 2024 at 1:07 PM
3/n) We can calculate the mean and standard deviation of the trees planted by scaling the mean and standard deviation of the centuries scored accordingly.
This gives us a distribution having mean as 455.0 and standard deviation as 92.0
This gives us a distribution having mean as 455.0 and standard deviation as 92.0
1/n) Imagine a 3-match ODI series between India and South Africa. The probability distribution of Rohit Sharma's centuries in the series looks something like this:
November 24, 2024 at 1:07 PM
1/n) Imagine a 3-match ODI series between India and South Africa. The probability distribution of Rohit Sharma's centuries in the series looks something like this:
Polishing my mathematics fundamentals to dive deeper into more complex generative and GAN-based models. It's a bit overwhelming to work on, but I'm sure I'll get the hang of it with practice. Here's a basic overview of Transforming Probability Densities:
- A Thread 🧵
- A Thread 🧵
November 24, 2024 at 1:07 PM
Polishing my mathematics fundamentals to dive deeper into more complex generative and GAN-based models. It's a bit overwhelming to work on, but I'm sure I'll get the hang of it with practice. Here's a basic overview of Transforming Probability Densities:
- A Thread 🧵
- A Thread 🧵
2/n) Based on students' profiles, we can generate personalized examples and questions of varying difficulty levels on topics previously taught by the professor. This will increase cognitive engagement helping students absorb more knowledge and retain it longer.
November 24, 2024 at 1:05 PM
2/n) Based on students' profiles, we can generate personalized examples and questions of varying difficulty levels on topics previously taught by the professor. This will increase cognitive engagement helping students absorb more knowledge and retain it longer.
1/n) Hyper-Personalized Student Profiles
Creating a detailed profile of each student, capturing their preferences, strengths, weaknesses, and interests through data on courses, extracurricular achievements, hobbies, and personal reflections from students about themselves.
Creating a detailed profile of each student, capturing their preferences, strengths, weaknesses, and interests through data on courses, extracurricular achievements, hobbies, and personal reflections from students about themselves.
November 24, 2024 at 1:05 PM
1/n) Hyper-Personalized Student Profiles
Creating a detailed profile of each student, capturing their preferences, strengths, weaknesses, and interests through data on courses, extracurricular achievements, hobbies, and personal reflections from students about themselves.
Creating a detailed profile of each student, capturing their preferences, strengths, weaknesses, and interests through data on courses, extracurricular achievements, hobbies, and personal reflections from students about themselves.
Recently got a chance to read this paper by some amazing folks at KAIST. It introduces a framework for building and evaluating personalized Pedagogical Conversational Agents (PCAs) to align with students' personality traits to enhance learning.
-🧵
-🧵
November 24, 2024 at 1:05 PM
Recently got a chance to read this paper by some amazing folks at KAIST. It introduces a framework for building and evaluating personalized Pedagogical Conversational Agents (PCAs) to align with students' personality traits to enhance learning.
-🧵
-🧵
2/n) Next, we loaded the metadata, which stores information about all audio files, and checked if the dataset is balanced or not. We found that the dataset was fairly balanced.
April 13, 2024 at 6:16 AM
2/n) Next, we loaded the metadata, which stores information about all audio files, and checked if the dataset is balanced or not. We found that the dataset was fairly balanced.
1/n) I used librosa to extract waveform data and the sample rate from a wave file. By default, librosa reads a file at a sample rate of 22050. Then, I used the `waveshow` method to visualize the audio file. Here's what it looks like:
April 13, 2024 at 6:15 AM
1/n) I used librosa to extract waveform data and the sample rate from a wave file. By default, librosa reads a file at a sample rate of 22050. Then, I used the `waveshow` method to visualize the audio file. Here's what it looks like:
As we have 38 unique symbols/notations in our song that we mapped to an integer vocabulary we want our neural network to have 38 output nodes. I left the model to train overnight and saved it so we can use it later.
April 9, 2024 at 6:03 AM
As we have 38 unique symbols/notations in our song that we mapped to an integer vocabulary we want our neural network to have 38 output nodes. I left the model to train overnight and saved it so we can use it later.
In my last post, I explained how I created a training sequence using an integer vocabulary. Now that our dataset is ready for operations, we can now train our LSTM Neural Network on it.
April 9, 2024 at 6:02 AM
In my last post, I explained how I created a training sequence using an integer vocabulary. Now that our dataset is ready for operations, we can now train our LSTM Neural Network on it.
4/) Finally, we implemented all our functionalities into our preprocess function so we can use it later.
April 7, 2024 at 6:23 AM
4/) Finally, we implemented all our functionalities into our preprocess function so we can use it later.
3/) Afterwards, we created a vocabulary that maps symbols to integers and saved the vocabulary in a JSON file.
April 7, 2024 at 6:23 AM
3/) Afterwards, we created a vocabulary that maps symbols to integers and saved the vocabulary in a JSON file.
2/) Next, we collected all our songs and put them into a single file. We added a delimiter "/" between songs to indicate when one song ends.
April 7, 2024 at 6:22 AM
2/) Next, we collected all our songs and put them into a single file. We added a delimiter "/" between songs to indicate when one song ends.
1/) Each song has been encoded into its time series representation. This allows the model to understand our data and make sense out of it.
April 7, 2024 at 6:21 AM
1/) Each song has been encoded into its time series representation. This allows the model to understand our data and make sense out of it.
5/n) For now I wrote a small script that implements all my features and opens my song in the musescore app.
April 3, 2024 at 3:27 PM
5/n) For now I wrote a small script that implements all my features and opens my song in the musescore app.
4/n) Lastly put it all together in my preprocess function, so I can use it later.
April 3, 2024 at 3:27 PM
4/n) Lastly put it all together in my preprocess function, so I can use it later.
3/n) Used the Music21 in built functions to determine the key of a song:
if key is not found, we use the analyze method.
We then transpose all songs in major keys to C Major and those in minor keys to A Minor.
if key is not found, we use the analyze method.
We then transpose all songs in major keys to C Major and those in minor keys to A Minor.
April 3, 2024 at 3:26 PM
3/n) Used the Music21 in built functions to determine the key of a song:
if key is not found, we use the analyze method.
We then transpose all songs in major keys to C Major and those in minor keys to A Minor.
if key is not found, we use the analyze method.
We then transpose all songs in major keys to C Major and those in minor keys to A Minor.
2/n) Implemented the has_acceptable_durations function, which checks if the loaded song has a desirable length for further processing.
April 3, 2024 at 3:25 PM
2/n) Implemented the has_acceptable_durations function, which checks if the loaded song has a desirable length for further processing.
1/n) Loaded songs from my dataset directory and converted them from kern format to Music21 stream class format.
April 3, 2024 at 3:24 PM
1/n) Loaded songs from my dataset directory and converted them from kern format to Music21 stream class format.
How many comments are too many comments? 👀
April 3, 2024 at 2:39 PM
How many comments are too many comments? 👀