machinelearning
/machinelearning47
For those interested in the intersection of AI and statistics, I have written a blog post how to build bayesian attention:
https://medium.com/data-science-collective/exploiting-the-structured-state-space-duality-to-build-bayesian-attention-3883ab8bacd4
https://medium.com/data-science-collective/exploiting-the-structured-state-space-duality-to-build-bayesian-attention-3883ab8bacd4
Is anyone aware of a machine learning/AI model that can generate titles based on Video content?
Where can I find any thing machine learning engineering?
State space models can be used as drop in replacements for attention, but with more favourable sequence length scaling. This video may well be the most lucid intro to state space models I've come across:
https://youtu.be/QJHA-PY8zDc?si=J5kGW87Yg0SAFdpR
https://youtu.be/QJHA-PY8zDc?si=J5kGW87Yg0SAFdpR
Interesting to learn how linear regression training works. It finds optimal parameter values by minimizing the Mean Squared Error (MSE).
The gradient descent algorithm adjusts the parameters iteratively to achieve the lowest cost.
The gradient descent algorithm adjusts the parameters iteratively to achieve the lowest cost.
Taking a course on machine learning.
I learned about how a machine actually learns, which is by minimising the cost function. In sense is making random guesses to get aline that minimises the RSS value.
Gradient descent algorithm is one way to achieve the above, giving it the right initial guess and learning rate.
I learned about how a machine actually learns, which is by minimising the cost function. In sense is making random guesses to get aline that minimises the RSS value.
Gradient descent algorithm is one way to achieve the above, giving it the right initial guess and learning rate.
has anyone else tried using this model for time series forecasting?
https://docs.nixtla.io/docs/getting-started-about_timegpt
TimeGPT
https://docs.nixtla.io/docs/getting-started-about_timegpt
TimeGPT
Hello guys, Am a web developer trying to transition into Data science and Machine Learning. Nice to meet you
Training my XGBoost model for a simple classification task.
Optimizing hyperparameters using optuna
Optimizing hyperparameters using optuna
Hello machine learning group!
This paper looks interesting. Gonna study it sometime soon.
https://ar5iv.labs.arxiv.org/html/2310.04948
Anyone else interested in deep learning based time series analysis?
Let's be friends #F4F #followforfollow
This paper looks interesting. Gonna study it sometime soon.
https://ar5iv.labs.arxiv.org/html/2310.04948
Anyone else interested in deep learning based time series analysis?
Let's be friends #F4F #followforfollow
what's going on with q-learning?
https://github.com/mttga/purejaxql
https://github.com/younggyoseo/CQN
https://github.com/mttga/purejaxql
https://github.com/younggyoseo/CQN
GitHub - mttga/purejaxql: Simple single-file baselines for Q-Learning in pure-GPU setting
Simple single-file baselines for Q-Learning in pure-GPU setting - mttga/purejaxql
github.com
GitHub - younggyoseo/CQN: Coarse-to-fine Q-Network
Coarse-to-fine Q-Network. Contribute to younggyoseo/CQN development by creating an account on GitHub.
github.com
https://arxiv.org/abs/2407.08447
3D gaussian splatting is for all intents & purposes a realtime 3D scene.
3D gaussian splatting is for all intents & purposes a realtime 3D scene.
https://arxiv.org/abs/2404.07647
"We measure the effect of the softmax bottleneck in various settings and find that models based on less than 1000 hidden dimensions tend to adopt degenerate latent representations in late pretraining, which leads to reduced evaluation performance."
"We measure the effect of the softmax bottleneck in various settings and find that models based on less than 1000 hidden dimensions tend to adopt degenerate latent representations in late pretraining, which leads to reduced evaluation performance."
Debugging neural nets is always a pain, but maybe penzai may bring some relief?
https://github.com/google-deepmind/penzai
https://github.com/google-deepmind/penzai
The development of machine learning is expected to be increasingly automated. Some sectors that best exemplify this fun fact about technology are agriculture, cybersecurity, fintech, manufacturing, and many more.
DeepMind announced AlphaFold 3, the latest iteration of its protein folding project.
AlphaFold 3, like its predecessors, primarily predicts how proteins fold based on their amino acid sequences.
AlphaFold uses machine learning to simulate the likely 3D structure a protein will adopt through folding.
AlphaFold 3, like its predecessors, primarily predicts how proteins fold based on their amino acid sequences.
AlphaFold uses machine learning to simulate the likely 3D structure a protein will adopt through folding.
The programming language to develop AI has changed a lot, in its beginnings Prolog was used as the main language, nowadays Python and C++ have that role, due to their applicability in this field.
Is the plot of The Matrix an example of supervised or unsupervised AI training? I feel like the Architect does unsupervised learning and The Oracle explains the process of unsupervised learning to the algorithm.
The 99.99% effectiveness mentioned by the Architect in his speech tells us the error is .01. Thoughts?
The 99.99% effectiveness mentioned by the Architect in his speech tells us the error is .01. Thoughts?
I had been training my models on Runway ML lab since 2020. Now Runway deprecated the lab and left me with a pile of .pkl files of my trained models. Where can I host them and continue training?
@scizors.eth this might be handy for u too
@scizors.eth this might be handy for u too
Anyone working on anything cool?
