iiinova — where your digital self keeps evolving

Reposted by Christopher Akiki

Marco @mcognetta.bsky.social · 30/09/2026

🚨 [Token][ization] Paper Alert 🚨 Tokenization is a wildly understudied area of language modeling despite it having effects across all of NLP. Over the past ~8 months, 32 (!) tokenizer researchers put together the most comprehensive survey of the field. Check it out!

112333

Reposted by Christopher Akiki

Lichess @lichess.org · 20/07/2026

Were you watching the World Cup or playing chess? Well, we have the stats for it... (1/2)

A visible dip in a line chart of online users for the World Cup Final, and temporary rebound during the halftime show

1357

Christopher Akiki @cakiki.bsky.social · 17/07/2026

t-SNE on SigLIP embeddings of 1,360 @oreilly.bsky.social book cover animals. The penguin cluster is my favorite.

020

Christopher Akiki @cakiki.bsky.social · 14/07/2026

This version has 1360 @oreilly.bsky.social book covers, sorted by color.

image quilt of all o'reilly media books, sorted by hue.

020

Reposted by Christopher Akiki

Lichess @lichess.org · 26/06/2026

We recently crossed 6 million chess puzzles in our open database. Like the platform itself, Lichess data is free and open source. Go and build something cool with it, Available wherever you get your datasets!

5457

Christopher Akiki @cakiki.bsky.social · 22/06/2026

work in progress: 1 million synthetic personas from NVIDIA's Nemotron-Personas-USA dataset.

scatter plot of 1 million personas from https://huggingface.co/datasets/nvidia/Nemotron-Personas-USA

020

Christopher Akiki @cakiki.bsky.social · 04/06/2026

Been trying to reproduce datashader functionality using only Apache Arrow and Acero as a learning exercise. This is a 1-billion point Clifford attractor rendered in ~8s from a 9GB parquet file. (No JIT)

010

Reposted by Christopher Akiki

Marco @mcognetta.bsky.social · 11/02/2026

I'm looking for 5-10 #chess players to test out a tool I'm building. Preferably who play on @lichess.org and are 1200+ in rapid or blitz. And if you coach chess at all, I'd be extra grateful to have you test it! NOTE: it is _NOT_ an "LLM chess coach" tool, I promise! 🙏

264

Reposted by Christopher Akiki

Lichess @lichess.org · 07/02/2026

Any chess position with 8 pieces on the board and at least one pair of opposed pawns has been solved! Lichess can now tell you definitively if it's a win, loss or draw with no engine required. Read about the massive technological undertaking to accomplish this partial 8 piece tablebase on our blog:

24010

Christopher Akiki @cakiki.bsky.social · 20/01/2026

UMAP connectivity plots of 3,627 chess openings from the @lichess.org datasets (huggingface.co/datasets/Lic...)

3 by 3 grid of networks color-mapped with "plasma" and with edge-bundling. Lung looking structure.

161

Christopher Akiki @cakiki.bsky.social · 18/01/2026

Which colormap do you think looks the nicest? I'm leaning toward plasma.

3 by 3 grid of the same plot in different colormaps. Plot is an edge-bundled connectivity graph. Color maps are, left-to-right and top-to-bottom: viridis, plasma, inferno, magma, cividis, binary, copper, coolwarm, managua.

300

Christopher Akiki @cakiki.bsky.social · 16/01/2026

Scatterplot of 4 million computer science authors, laid out according to co-authorship connections. Large blob in the bottom left are all single authors; removing them lets the plot breathe more somehow. The source of the data is the @dblp.org bibliography.

Organic looking scatterplot. Blob in the bottom left. Hedgehog looking.

Organic looking scatterplot. Sort of looks like Romanesco broccoli.

142

Reposted by Christopher Akiki

Shayne Longpre @shaynelongpre.bsky.social · 26/11/2025

Who is winning the open AI race? Our new study Economies of Open Intelligence maps @hf.co 851k models' downloads 2020→2025. 1) Power rebalance: US tech ↓; China + community ↑ 2) Models size & efficient ↑ (MoE, quant, multimodal) 3) Intermediary layers ↑ (adapters/quantizers) 4) Transparency ↓ /🧵

293

Reposted by Christopher Akiki

Lichess @lichess.org · 11/11/2025

Researchers at Google DeepMind used our free puzzle database and reinforcement learning to train a model to generate creative chess puzzles. ➡️ Read more on this by Tom Zahavy from the DeepMind discovery team: lichess.org/@/tomas135/b...

lichess.org

AI-Generated Chess Puzzles

A new research by the Discovery team at @GoogleDeepMind using RL and generative models to discover creative chess puzzles

0123

Christopher Akiki @cakiki.bsky.social · 04/11/2025

Three different ways to represent colo(u)r. Work in progress, inspired by an old post by Kat Zhang / The Poet Engineer.

Three scatterplots of colorful points.
titles = ['Color Space', 'Text Space', 'Image Space']
subtitles = ['Embeddings of color features', 'Text embedding of color names', 'Image embeddings of color swatches']

151

Christopher Akiki @cakiki.bsky.social · 31/10/2025

I made this annotated scatter plot of 1 million FineWeb-Edu documents for @sashamtl.bsky.social's new TED talk.

Scatterplot of a document corpus with cluster information in the form of colors and cluster labels. Includes labels like computing, mental health, religion, etc.

141

Reposted by Christopher Akiki

Xiaoyi @cleefouti.bsky.social · 28/10/2025

When the fish left the river:

A four frame comic about the Cambodian genocide. In the first frame, a man can be seen from the back, looking at a window. The text reads: "I never ask my father about Cambodia." In the second frame, a silhouette of the father can be seen, screaming in his sleep, arms reaching into nothingness. The text reads: "What could he say, really, that he didn't already scream in his nightmares?". The third frame shows the father pondering on something, expression neutral. The text reads: But sometimes he would say "It's funny, when it happened, all the fish left the river. They never came back." . The fourth frame shows the father lowering his head, thinking. He says: "I guess they knew."

The second page of the comic on the Cambodian genocide, and everything to come. The first frame shows the daughter, young, silent and unsure of what to say. The text reads: "I didn't say anything then." In the present, the daughter, now an adult, looks at the Seine in Paris. The text reads "But I often think of it now.". The daughter then crouches near the Seine to see if there are any fish left. She says: "Are you still here?".

215046

Christopher Akiki @cakiki.bsky.social · 27/10/2025

Also really love how organic the plot looks with "inferno" (left) and "viridis" (right).

organic looking network representing the internet BGP nodes. Inferno colormap.

organic looking network representing the internet BGP nodes. Viridis colormap.

041

Christopher Akiki @cakiki.bsky.social · 26/10/2025

Thanks to @jamesabednar.bsky.social I realized I had used the wrong background color for the colormap I had chosen. This is another version of the plot (different embeddings) with the corrected background.

Correct version of the plot in the quoted post. Graph with an organic like structure of the BGP network.

110

Christopher Akiki @cakiki.bsky.social · 26/10/2025

Map of the internet: 1.3M nodes (BGP)

organic looking graph of the BGP nodes of the internet. black and white

4306

Reposted by Christopher Akiki

Lichess @lichess.org · 07/10/2025

We're cooking.. 👀

4191

Christopher Akiki @cakiki.bsky.social · 28/09/2025

526.9 million player deaths in 24.7 million levels of Super Mario Maker 2. Data by @tgr.bsky.social

very colorful scatterplot of player deaths in mario maker levels. "inferno" colormap. some level elements highlighted.

050

Christopher Akiki @cakiki.bsky.social · 11/07/2025

Really cool new embeddings exploration tool by @domoritz.de and colleagues from Apple. Can't wait to build with this. Also includes a streamlit component and a Jupyter widget.

screenshot from the embeddings atlas repo

example density plot exploring a wine dataset.

120

Christopher Akiki @cakiki.bsky.social · 28/02/2025

Woah! EA just open sourced "Command and Conquer: Red Alert" and a bunch of other CnC games! github.com/electronicar...

box art from 1996 version of CNC RED ALERT

121

Reposted by Christopher Akiki

Lichess @lichess.org · 02/02/2025

Lichess is now on @kaggle.com! Use our puzzles, openings, and engine evaluation datasets directly in your kaggle notebooks: www.kaggle.com/organizations/lichess ♟️

Screenshot of Lichess organization overview page on Kaggle

1524

Christopher Akiki @cakiki.bsky.social · 08/12/2024

The folks at Foursquare released a @hf.co dataset of 104.5 million places of interest and here's all of them plotted using datashader

2173

Christopher Akiki @cakiki.bsky.social · 06/12/2024

I recently used the @lichess.org puzzles dataset to experiment with chess position embeddings and visualize 4.5M starting positions. (hf.co/datasets/Lic...)

Scatterplot with 4.5M data points. Inferno colormap. Looks somewhat organic with neat clusters.

2284

Reposted by Christopher Akiki

Lichess @lichess.org · 06/12/2024

The Lichess database of games, puzzles, and engine evaluations is now on @hf.co - huggingface.co/Lichess. Billions of chess data points to download, query, and stream and we're excited to see what you'll build with it! ♟️ 🤗

39423

Christopher Akiki @cakiki.bsky.social · 13/02/2024

Early experiment visualizing of Cohere For AI's newly-released Aya dataset. Multilingual corpora are always so fun to play with.

Scatterplot of the Aya dataset. 250K points in a floral / broccoli like pattern. Inferno color map.

020

Christopher Akiki @cakiki.bsky.social · 17/11/2023

Clifford-inspired strange attractor.

Black and white image of a strange attractor.

171

Christopher Akiki @cakiki.bsky.social · 27/09/2023

10 million digits of Pi. Kind of.

scatter plot of a UMAP projection. magma colormap. quite clustered. looks almost celestial.

120

Christopher Akiki @cakiki.bsky.social · 26/09/2023

835 languages. 3.5 million bible verses. Work in progress.

Scatterplot. Starbursty and elongated clusters. Magma colormap.

140

Christopher Akiki @cakiki.bsky.social · 26/09/2023

UMAP connectivity graphs—with edgehammer bundling—are always something to gaze at.

connectivity graph of the previous scatter plot. grayscale. edges bundled.

030

Christopher Akiki @cakiki.bsky.social · 25/09/2023

Revisiting John Williamson's prime factors plot with a few differences in implementation. I am using UMAP and Datashader to visualize the first million integers. Not quite there yet.

2d scatterplot in the viridis palette. Different blobs and clusters. Looks very starbursty in some places and very organic in others.

030

Christopher Akiki @cakiki.bsky.social · 06/06/2023

Multilingual text corpus or Petri dish?

Scatterplot of the BigScience ROOTS corpus

191

Christopher Akiki @cakiki.bsky.social · 05/06/2023

Code Dataset Visualization—11.66 million files from the Stack, a dataset sourced from permissively-licensed GitHub repositories spanning 86 programming languages (StarCoder languages subset).

Scatterplot of many points using a fire-theme palette on a black background. The points cluster in small groups. The general shape is circular, as is common of TSNE embeddings.

1143