Advertisement · 728 × 90

Posts by rajistics.bsky.socia

Post image

If you want to really assess your RAG system —
you need to go deeper.

Ask:
✅ Is your retriever surfacing the right chunks?
✅ Is your generator actually using them — or just hallucinating? 🤔

Here is how I like to think about my RAG metrics (inspired by RAGAS)

1 year ago 0 0 0 0
Post image Post image Post image

These are common patterns from Hugging Face and Anthropic
Hugging Face - SmolAgents: huggingface.co/blog/smolage...

Anthropic - Building effective agents: www.anthropic.com/research/bui...

1 year ago 1 0 0 0
Post image Post image Post image Post image

Are you going to chat all day with LLMs? 🤐
Here are the essential agentic workflows. 👇

1 year ago 1 0 1 0
Why Parquet? Introducing the the file format for data engineering and machine learning
Why Parquet? Introducing the the file format for data engineering and machine learning YouTube video by Rajistics - data science, AI, and machine learning

Dive into Parquet

It's a leading format for data engineering, data science, and machine learning.

youtube.com/shorts/_CnEK...

1 year ago 1 0 0 0
PyPI Name Squatting - How to reserve your python package name
PyPI Name Squatting - How to reserve your python package name YouTube video by Rajistics - data science, AI, and machine learning

PyPI Name Squatting

This didn't happen to me recently :)

To learn more:
An Empirical Analysis of the Python Package Index (PyPI) - arxiv.org/pdf/1907.11073

blog.checkpoint.com/securing-the...

blog.orsinium.dev/posts/py/pyp...

My Video:
youtube.com/shorts/H1Uja...

1 year ago 1 0 0 0
Using Mechanistic Interpretability to Steer a Model's Predictions (Transluce's Monitor)
Using Mechanistic Interpretability to Steer a Model's Predictions (Transluce's Monitor) YouTube video by Rajistics - data science, AI, and machine learning

Why do language models think 9.11 is greater than 9.9? 🤔
Mechanistic Interpretability is a useful tool for investigation and fixing the issue.
I am using Transluce's Monitor here:
My video summary: www.youtube.com/shorts/Kuh-i...
Try Monitor: monitor.transluce.org/dashboard

1 year ago 0 0 0 0
Top 26 Data Science Algorithms
Top 26 Data Science Algorithms YouTube video by Rajistics - data science, AI, and machine learning

My ranking of the top 26 algorithms for practical data science, breaking down their strengths, quirks, and when (or if) you should use them.

youtube.com/shorts/dt4uX...

1 year ago 1 1 0 0
Pandas versus Polars: A Quick Comparison of Single Node DataFrame Alternatives
Pandas versus Polars: A Quick Comparison of Single Node DataFrame Alternatives YouTube video by Rajistics - data science, AI, and machine learning

Polars verus Pandas
What is the best single node dataframe?

For Polars check out:
github.com/pola-rs/polars

Polars vs. pandas: What’s the Difference?
blog.jetbrains.com/pycharm/2024...

Database-like ops benchmark - duckdblabs.github.io/db-benchmark/

Short Video:
youtube.com/shorts/8DkIR...

1 year ago 0 0 0 0
Preview
bluesky-community/one-million-bluesky-posts · Datasets at Hugging Face We’re on a journey to advance and democratize artificial intelligence through open source and open science.

So I can expect a bunch of projects looking at the sentiment of bluesky posts 😂

1 year ago 1 0 1 0
Advertisement
Physics of Language Models - Extracting Knowledge
Physics of Language Models - Extracting Knowledge YouTube video by Rajistics - data science, AI, and machine learning

The Physics of Language Models
Check out a scientific approach that experiments with model architecture, synthetic datasets, and tasks to understand how language models work.

My short intro: youtu.be/9saXkwHKaLs

Longer Video: ICML 2024 Tutorial by Zeyuan Allen-Zhu - youtu.be/yBL7J0kgldU

1 year ago 0 0 0 0

Communications here 🙋‍♂️

1 year ago 1 0 1 0
Improving Data Quality in MultiModal Models (Molmo and PixMO)
Improving Data Quality in MultiModal Models (Molmo and PixMO) YouTube video by Rajistics - data science, AI, and machine learning

Ai2's 700k examples > Meta's 6B examples
the importance of data quality

My video: youtube.com/shorts/-_DGp...

Background:
Hannaneh Hajishirzi - OLMo: Accelerating the Science of Language Modeling (COLM)
www.youtube.com/watch?v=qMTz...
Molmo and PixMo paper -
arxiv.org/pdf/2409.17146

1 year ago 0 0 0 0
Are you Smarter than a AI Language Model like GPT4?
Are you Smarter than a AI Language Model like GPT4? YouTube video by Rajistics - data science, AI, and machine learning

Are you smarter than GPT-3 (you don't have a chance against GPT-4)
Test yourself:

Are you smarter than a language model? -
joel.tools/smarter/
Language modeling game!
rr-lm-game.herokuapp.com
Are You Smarter Than An LLM?
d.erenrich.net/are-you-smar...

My video on the topic:
youtu.be/kXQGivEAF1U

1 year ago 1 0 0 0
Why Logloss is a better loss function than Mean Squared Error
Why Logloss is a better loss function than Mean Squared Error YouTube video by Rajistics - data science, AI, and machine learning

Why do we use LogLoss as an error metric?
Exploring Mean Error, Mean Squared Error, and Log Loss

youtu.be/S_zxVfKI55c

1 year ago 1 0 0 0

Wow. I had no idea on BlueSky you can enable external media so you can watch YouTube videos on this platform. It overcomes the problem of only being able to upload 60 seconds of video here. It’s gets better & better

1 year ago 18238 1973 260 150

🔹 5th Place:
LightGBM + Time Series Foundation Model (TFM)

🔹 4th Place:
Temporal Fusion Transformer (TFT) from Neural Forecast

🔹 3rd:
🚀 LightGBM with recursive & direct forecasting

🔹 2nd:
🌠 LightGBM + Seasonal Theta

🔹 1st:
🧮 Anchored Multiplicative Seasonal Index + Seasonal ARIMA + LightGBM

1 year ago 0 0 0 0
Post image

What are the cutting-edge time series approaches? 📈✨
The VN1 Forecasting Competition showed winning techniques including time series foundation models, deep learning, statistical methods, machine learning, and ensembling.
Check out the techniques of the top 5 teams.
www.youtube.com/watch?v=CRGA...

1 year ago 1 0 1 0
Advertisement
4 Techniques for Dimensionality Reduction: PCA, AutoEncoder, TSNE, and UMAP
4 Techniques for Dimensionality Reduction: PCA, AutoEncoder, TSNE, and UMAP YouTube video by Rajistics - data science, AI, and machine learning

4 Techniques for Dimensionality Reduction: PCA, AutoEncoder, TSNE, and UMAP

youtu.be/EHWBP-OQwHk

1 year ago 3 0 0 0