6. Applications of TinyML You will see examples of TinyML applications, and learn first-hand how to train these models for tiny applications such as keyword spotting, visual wake words, and gesture recognition. https://
pll.harvard.edu/course/applica
tions-tinyml?delta=0
…
RESEARCH
-
TinyML Applications: Keyword Spotting, Wake Words, Gesture Recognition
By
–
-
Reproducible Data Science: Statistical and Computational Tools
By
–
4. Principles, Statistical and Computational Tools for Reproducible Data Science Learn skills to ensure you can trust your own research results, reproduce them yourself, and communicate them to others. https://
pll.harvard.edu/course/princip
les-statistical-and-computational-tools-reproducible-data-science?delta=3
… -
Harvard Probability Course: Essential Data Science Foundations
By
–
3. Introduction to Probability Learn probability, an essential language and set of tools for understanding data, randomness, and uncertainty. https://
pll.harvard.edu/course/introdu
ction-probability-edx?delta=2
… -
Free Harvard Resources for Programming Data Science Machine Learning
By
–
Don't pay ridiculous amounts of money to study Programming, Data Science, and Machine Learning. Learn these for FREE from the experts at Harvard university. (A thread)
-
Artificial Intelligences Can Even Have Humor
By
–
Les Intelligences Artificielles peuvent même avoir de l’humour ! https://
x.com/ribodanslasauc
/ribodanslasauce/status/1609991475429011458
… -
Difficulty without model release
By
–
well its kinda hard to do that since they didnt even release the model
-
AI Limitations: Humans Still Essential for Quality Output
By
–
Here are some key points from the space: 1. AI can't completely replace humans in its current implementation. 2. AI can give you a skeleton to solve a problem but specifics need to filled by humans. 3. Quality of output for these tools totally depends on the quality of input.
-
Prompt Injection Discovery and Attribution: A History Correction
By
–
A history correction: I am not the first to discover prompt injection. I was merely the first to do so and discuss it publicly. PI was discovered independently by multiple teams. The first was Preamble, an LLM security company, whose find predates mine by several months.
-
Curriculum Learning and Dropout Improve Model Performance
By
–
Thanks to using a form of curriculum learning (5 epochs) and better dropout (20 & 50 epochs) I've now been knocked off the leaderboard entirely! Here's the latest:
-
LLM Tokenization: Not Vectors or Letters, but Token Sequences
By
–
It doesn’t see words as embedded vectors or as letters, it sees them as token sequences — chunks of about 4 letters from a fixed mapping of strings to tokens. It does better on letter-based tasks if you ask it to first rewrite words l-i-k-e t-h-i-s.