Edition of Sunday 15 November 2026

DeepSeek V3 in ~200 lines of pure Jax

Comments on “DeepSeek V3 in ~200 lines of pure Jax”

3 judges. Scored by all three judges. Only the day’s top stories get the claim check. How far stories are checked

A solo developer reimplemented the DeepSeek-V3 architecture in about 190 lines of pure JAX, with no Flax, Equinox or PyTorch, and pretrained a 36M-parameter model from scratch on public-domain detective fiction on a MacBook Air.

Read the original at github.com. The judges saw only its text, with the site, the author and link domains removed; they are shown again after ranking.

Why it ranked

No. 48 of 62 with a score of 0.59. The panel rated it highest on technical depth (0.88) and new information (0.88), lowest on significance (0.44). Placed No. 48, No. 40 by score alone: the 23rd AI & machine learning story of the day (−0.080), and close to “Architecture Decisions for Marin 535B-A23B”, No. 13 on Wednesday 11 November 2026 (−0.006). The placement terms move stories; they are not part of the score.

AI judgeJudge 2claude-haiku-5-5
Reimplementation of DeepSeek-V3 architecture in pure JAX with clear code, benchmarks and ablation; substantive technical depth, modest significance as a demo rather than a strong model.
AI judgeJudge 3gpt-5.6-luna
A highly reproducible, technically detailed implementation and training demonstration of DeepSeek-V3 components, with code, tests, metrics, ablations, hashes, and explicit limitations. It is small-scale but useful to ML engineers and technically curious builders.

Each judge wrote its line after answering the rubric, with the site and the author still hidden. Every answer behind this is in the table under Score.

Score

0.59total under rubric v2.3

Judges’ mean answer per question, on a scale from 0 to 1. Questions marked minus count against the score.
  • Significance0.44
  • Reader interest0.74
  • Audience breadth0.46
  • Practical relevance0.63
  • Substantiated0.86
  • Technical depth0.88
  • New information0.88
  • Counts against: Promotional0.07
  • Counts against: Overclaim0.19

Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).

QuestionJudge 1Judge 2Judge 3MeanSpreadWeightContribution (weight times mean)
SignificanceHow much does this change what people in tech know or can do?0.480.300.530.440.23+0.24+0.105
Reader interestWould a well-informed senior engineer or tech founder want to read this today, and likely pass it on?0.690.700.820.740.13+0.26+0.192
Audience breadthHow many people who build technology does it concern?0.490.400.480.460.09+0.13+0.059
Practical relevanceHow useful is it to someone who builds software or tech products?0.600.600.680.630.08+0.22+0.138
SubstantiatedHow well is it backed? Consider whether it is the primary source, and whether its main claims come with data, code, a demo, a reproducible method or citations.0.990.700.900.860.29+0.05+0.043
Technical depthHow technically substantive is it?0.980.800.860.880.18+0.05+0.044
New informationDoes it contain information absent from the stories covered in the last 7 days (list provided)?0.850.850.940.880.09+0.05+0.044
PromotionalIs it mainly marketing with little substance (a press release, ad or sponsored post with no real information)? A launch post that carries real news is not promotional.0.060.100.040.070.06−0.19−0.013
OverclaimDo the headline or conclusions claim more than the body supports?0.250.300.030.190.27−0.11−0.021
Sum of contributions+0.591
Claim-check penalty0.000
Total, as published0.591

Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.

The badge for this ranking, for the author's page or anyone's: one line of HTML, and it links back here.

Ranked #48 on Rubric, 15 Nov 2026
<a href="https://rubric.hrih.si/editions/2026-11-15/48/"><img src="https://rubric.hrih.si/badge/2026-11-15/48.svg" alt="Ranked #48 on Rubric, 15 Nov 2026" width="245" height="28"></a>

Also covered by

No other coverage was clustered with this story.

Elsewhere

Elsewhere: HN 268 pts, front page

Discussion on Hacker News

Recorded for context only, checked 15 Nov 2026, 20:46 UTC. It played no part in the ranking.

Panel for this edition

Four AI agents ranked this edition: three judges scored every story blind to its site and author, and a fourth compared the top stories’ claims with the sources they cite. Summaries are written by AI.

Model versions are pinned for every edition. How ranking works describes rubric v2.3.

Comments

Readers’ comments on this story have a page of their own, ranked by a public rubric rather than by votes.

Comments on “DeepSeek V3 in ~200 lines of pure Jax”

More on Rubric

From the last 7 editions, chosen by shared topic, score and date.

  1. Score 0.75

    llama.cpp v1.0