Sebastian Raschka

Blue-verified on X
@rasbt
  • 100K followers

ML/AI research engineer. Ex stats professor. Author of "Build a Large Language Model From Scratch" (https://amzn.to/4fqvn0D) & reasoning (https://mng.bz/lZ5B)

AIUnited Statessebastianraschka.comJoined October 2012Listing updated 2026-10-29
1.2K Following532K Followers20K Posts

Who is Sebastian Raschka?

Sebastian Raschka (@rasbt) is a creator on X, focused on AI. Sebastian Raschka is an ML/AI research engineer and former statistics professor. He shares educational content, video lectures, and write-ups focused on LLMs, reasoning models, reinforcement learning, and open-source AI development. Sebastian Raschka has 532K followers and posts about 1.3 times a week, with videos doing best. By Growth Score, Sebastian Raschka ranks #19 of 102 AI creators on the X Growth Directory. On X since 2012, with a website at sebastianraschka.com.

Growth Score 68

Same formula for everyone, from public posts. Week of 2026-10-26.

How it works

#19 of 102 in AIby Growth Score among 102 scored AI creators

Engagement79
Consistency25
Reach96
Posts / week
1.3
Last 28 days
Views / post
48K
Median
Engagement
2.0%
Reactions per view
Reply rate
0.09%
Replies per view
Save rate
0.66%
Bookmarks per view
Views trend
—
Last 14 days vs prior 14

Top posts

Their most-viewed recent posts, summarised by us. Open on X to read them in full.

Explains how encoder-style classification models have evolved and why Jev achieves strong generalization.
QuoteBreakout · 5× usual views1.1% engagement
135 205 2.2K 242K 1.1KOpen on X ↗
Publishes a massive visual guide and write-up on language models for text classification and architectures.
Image2.4% engagement
78 324 2.4K 117K 2.1KOpen on X ↗
Highlights MiMo-V2.6 as the top open-weight benchmark model despite its simple grouped-query attention architecture.
Image1.8% engagement
77 173 1.5K 103K 679Open on X ↗
Announces the complete reasoning-from-scratch video series and links the magazine write-up before a short break.
Link2.1% engagement
77 252 1.7K 99K 1.8KOpen on X ↗
Introduces reinforcement learning with verifiable rewards and group relative policy optimization in part six.
Video2.5% engagement
82 199 1.6K 75K 1.5KOpen on X ↗

Writing style

How Sebastian Raschka writes, read from their recent posts.

Tone
Educational, technical, accessible, and enthusiastic.
Hooks
Directly announces new educational modules, video releases, or striking insights about AI architectures.
Structure
Combines concise explanatory text with video timestamps, code concepts, and links to detailed guides.
Length
Medium length with structured lists or descriptions.
Emoji
Occasional use of emojis to highlight sections.
Endings
Directs users to full newsletters, videos, or code repositories.

Signature moves

  • Sharing comprehensive from-scratch video lecture series
  • Breaking down complex AI training techniques and metrics
  • Discussing open-source models and practical implementations

What they post about

Reasoning Models and RLVR45%
LLM Architecture and Training35%
AI Evaluation and Benchmarks20%

What works for Sebastian Raschka

Computed from their own posts: reactions per view, compared with their median.

Best formats
Video (1.2×) · Text (1.0×) · Image (0.9×)
Best day
Saturday (1.2× their usual engagement, 6 posts)
Top posts went out
around 13:00 UTC (6 of their 10 most-viewed)
Posts read
15 recent original posts

Compare with every listed creator: Twitter (X) engagement rate benchmarks by niche · Best time to post on Twitter (X), by niche · How top X creators write: hooks, length and style

Tactics they repeat

Recent posts

Their latest original posts with live numbers.

Announces the complete reasoning-from-scratch video series and links the magazine write-up before a short break.
Link2.1% engagement
77 252 1.7K 99K 1.8KOpen on X ↗
Releases the finale of the reasoning series covering model distillation as an efficient alternative to RL.
Video2.6% engagement
65 91 806 37K 520Open on X ↗
Covers reinforcement learning with verifiable rewards, policy ratios, and GRPO training tips in a new video.
Video2.5% engagement
83 138 1.3K 59K 1.2KOpen on X ↗
Shares a spontaneous morning chat philosophizing about AI with Juergen Schmidhuber and Jake Newfield.
Quote0.5% engagement
15 15 151 36K 52Open on X ↗
Introduces reinforcement learning with verifiable rewards and group relative policy optimization in part six.
Video2.5% engagement
82 199 1.6K 75K 1.5KOpen on X ↗
Publishes a massive visual guide and write-up on language models for text classification and architectures.
Image2.4% engagement
78 324 2.4K 117K 2.1KOpen on X ↗
Open on X to read this post.
Image1.1% engagement
52 43 436 48K 242Open on X ↗
Discusses log-probability scoring, loss functions, and self-refinement in part five of the reasoning series.
Video2.0% engagement
43 129 1K 60K 908Open on X ↗

Growth

Public counts only. Weekly history builds up from the first week we listed them.

Average pace
About 38K followers a year since joining in 2012
Biggest post
242K views, 5× their median

Views per post, oldest to newest

Questions about Sebastian Raschka

What does Sebastian Raschka post about on X?

Main topics: Reasoning Models and RLVR, LLM Architecture and Training and AI Evaluation and Benchmarks. Videos get the most reactions per view. A typical post gets about 48K views.

How many followers does Sebastian Raschka have on X?

532,264 followers as of 27 Oct 2026.

What is Sebastian Raschka's website?

sebastianraschka.com, linked at the top of this page.

What is Sebastian Raschka's Growth Score?

68 out of 100, #19 of 102 AI creators. The Growth Score is one public formula for every account: engagement, consistency and reach from the last 28 days of original posts. Nobody can pay to change it.