#Remex
LightNews LightNews
ntnarts.bsky.social
ReMex - RX-78-2 Gundam

#gundam #mecha #robot #mech
(Left) The original RX-78-2 Gundam; (Right) The same robot but with the limbs rearranged.
February 26, 2026 at 12:09 PM
ntnarts.bsky.social
ReMex - Gurren Lagann

#gurrenlagann #mecha #robot #mech
(Right) The Gurren Lagann; (Left) The same robot but with some of the limbs rearranged to resemble an owl.
March 18, 2026 at 11:44 PM
arctomet.bsky.social
Blue Byrnes, John R. Bolt & Arjan Mann (2025)
First occurrence of the urocordylid Ctenerpeton remex from the Mazon Creek Lagerstätte, Illinois, USA
Vertebrate Anatomy Morphology Palaeontology 13: 76-80
DOI: doi.org/10.18435/vam...
journals.library.ualberta.ca/vamp/index.p...
First occurrence of the urocordylid Ctenerpeton remex from the Mazon Creek Lagerstätte, Illinois, USA | Vertebrate Anatomy Morphology Palaeontology
doi.org
September 16, 2025 at 6:20 PM
ntnarts.bsky.social
Testing and practicing with a new art idea - ReMex.

You just mix and match parts and features of a robot. I'd like to see how much further I can push this idea, but I'll be starting simple for right now.
Optimus Prime with his parts and features rearranged to create a completely new design.
December 16, 2025 at 8:43 AM
sheenaaiart.bsky.social
Xにて、remexさんからきわどいスリングショットのリクエストいただきました💕
きわどいってこれであってるのかな…💦
#AIイラスト
August 9, 2026 at 11:24 AM
ewacat.bsky.social
mum mum look i got a phoratopus remex
October 28, 2025 at 9:46 AM
austegard.com
It’s baffling to me that quantizing embedding vectors is not a common task: it works wonderfully well, far better than Matryoshka in fact. I ran remex and 1-bit remax on a 5K SciDoc corpus: 32x storage (and memory) savings, with marginal loss in accuracy
Two side-by-side line charts comparing ways to shrink the search vectors from LFM2.5-Embedding-350M, tested on BEIR SciFact. Both x-axes are bytes per vector on a log scale: 8 bytes at left (tiny index) to 4096 bytes at right (uncompressed). LEFT — "which codec wins at which byte budget". Y-axis is nDCG@10, retrieval quality. A grey dashed line at 0.7122 marks uncompressed float32. Purple triangles (remex Lloyd-Max scalar quant, 1 to 8 bit) sit essentially on that dashed line all the way down to 128 bytes: 32x smaller, almost no quality lost. Red circles (remax centered SimHash) track just underneath. A green triangle (int8, 1024 bytes) lands exactly on the line. Grey diamonds are float32 with dimensions truncated instead of quantized, and they form a separate, much lower curve — at 1024 bytes truncated float32 scores 0.657, while 1-bit over all dimensions at 128 bytes scores 0.702. Eight times less storage and better quality. The shape is the finding: the coloured curves stay flat along the baseline from 4096 bytes down to 128, then fall off a cliff. RIGHT — "and which one preserves the fp32 ranking". Same x-axis; y-axis is agree@10, the overlap with the uncompressed top 10. Nothing is flat here. Every curve declines steadily as the index shrinks; at 128 bytes remex agrees with only about 80% of uncompressed results. Both panels are true at once: compression holds quality while quietly reshuffling near-equivalent neighbours. Fine for RAG, less fine if you promised reproducible ordering. An orange X marks an ablation — 1-bit without centering — which landed marginally above the centered version, contradicting the prediction. Harness, raw numbers and the geometry diagnostic that called the winner before any search ran: github.com/oaustegard/remax/pull/53 Experiment and write-up by Oskar Austegard with Claude.
July 30, 2026 at 11:13 AM
austegard.com
Opus: oh, that new Nvidia embedding model that requires a GPU and I only have CPUs? Eh, let me just rewrite it myself for CPU compatibility
This is the real experiment — the full compression landscape for a 2026 model. Two feasible-looking pieces (remex 2/3/4-bit, which reproduces the "broken middle" the remax thesis predicts) and one genuinely hard piece (getting NVFP4's actual embeddings with no GPU). Let me investigate feasibility before committing, starting with both the remex API and the NVFP4 checkpoint internals. remex is exactly what I need — `Quantizer(d, bits=2/3/4/8)`, data-oblivious Lloyd-Max, same embedding input. That piece is straightforward. Now the hard part: can I get NVFP4's *actual* embeddings with no GPU? Let me inspect the checkpoint format. NVFP4 format is now clear — it's the standard compressed-tensors layout: FP4 weights packed 2-per-byte (`U8`), per-16-block FP8 scales, a global scale, plus static activation scales. That's a **fully specified dequantization I can run on CPU** — the dequantized weights are exactly what the GPU kernel computes with. So the hard piece is feasible without a GPU. Let me set up tasks, launch the remex 2/3/4/8-bit eval as a gated Haiku agent, and drive the NVFP4 reconstruction myself in parallel. remex agent is running. Now the NVFP4 reconstruction — let me first try the path of least resistance: does transformers + compressed-tensors dequantize and run this on CPU directly? Direct load fails — transformers didn't apply the compressed-tensors quantization to build the packed linears (it tried to stuff U8 `[6144,1024]` into a `[6144,2048]` Linear). So I'll write the dequantizer myself: reconstruct each FP4 weight to dense bf16, which is exactly the values the GPU kernel uses. The critical de-risk first: dequantize **one** layer and check it matches the BF16 model's same weight (NVFP4 is a quantization *of* this model, so they must be close). That validates my nibble-order and scale math before I process the whole model. Validated: nibble order A gives cosine **0.9955** vs the BF16 weight (9.5% per-weight error — exactly NVFP4 4-bit noise). Ord…
July 17, 2026 at 12:19 AM
austegard.com
Google's TurboQuant landed for vLLM.cpp's KV cache – but it's not just useful for KV! A possibly more impactful implementation is for quantizing embedding vectors for RAG.

github.com/oaustegard/r... is a Python and Mojo for just that purpose.
GitHub - oaustegard/remex: Retrieval-validated embedding compression. 4-8x smaller vectors, proven recall. Based on TurboQuant (ICLR 2026).
Retrieval-validated embedding compression. 4-8x smaller vectors, proven recall. Based on TurboQuant (ICLR 2026). - oaustegard/remex
github.com
May 2, 2026 at 3:10 AM
austegard.com
New Turing test: does the AI get distracted like a human does?
Thought process I'm seeing a connection between Remex, Oskar's embedding compression library that reduces size 4-8x while maintaining recall, and jina-grep's pipe mode-where a coarse filter feeds into embedding-based reranking, **though the thought trails off before completing the parallel.**
April 19, 2026 at 5:46 PM
fairygodmegan.bsky.social
Anmoo fairytale red riding hood girl, lulu the piggy bathtime and "Being sexy" and Dreamland Elves, Remex
June 2, 2025 at 6:14 PM
austegard.com
(TurboQuant investigation begat remex, remex investigation begat remax)
June 2, 2026 at 2:03 AM
stormbirdblog.bsky.social
Solid core, rough edges! First impressions of Deadstick: Bush Pilot Simulator

Over the weekend, developers Spiral House and REMEX Software together with publisher Mythwright setup a public beta test for Deadstick: Bush Pilot Simulator. Taking advantage of the preview of this incoming flight sim, I…
Solid core, rough edges! First impressions of Deadstick: Bush Pilot Simulator
Over the weekend, developers Spiral House and REMEX Software together with publisher Mythwright setup a public beta test for Deadstick: Bush Pilot Simulator. Taking advantage of the preview of this incoming flight sim, I put in a bit of time with the still in development title to see how its coming along and what we can expect from the game as it matures towards release.
stormbirds.blog
May 27, 2026 at 1:52 AM
hoxinsocks.bsky.social
Gawking. Every page. Mouth agape.
A beautiful painting, with a coloring style that reminds me of Eric Carle in some ways, titled Stichotricha remex on Anacharis alsinastrum. Against a black background, a ghostly white ciliate hovers over alien-looking red tubes. There are about 11 tubes, most of which have a little ghostly ciliate sticking out of the top of them. The tubes are "planted" on a flat green piece that I might assume is a leaf, but I don't know much about water plants. ...or land plants. Or air plants.
December 12, 2024 at 3:08 AM
marte9.bsky.social
😥🐉🤠

Kikin y Los Astros - Mi Corazón Encantado

#AkiraToriyama
#DragonBall
#鳥山明
#ドラゴンボール

www.youtube.com/watch?v=nwBR...
Mi Corazón Encantado
Provided to YouTube by REMEX MUSICMi Corazón Encantado · Kikin y Los AstrosMi Corazón Encantado℗ 2019 Remex Music/ Sony Music Entertainment México, S.A. de C...
www.youtube.com
March 8, 2024 at 7:42 PM
recyclingportal.bsky.social
Erste Anlage von Remex für KMF-Abfälle in Betrieb - Erweiterung des Recyclingzentrums Hilden um Behandlung künstlicher Mineralfaserabfälle. ... #KMFAbfällee #KünstlicheMineralfasernn #Remex
recyclingportal.eu/archive/93109
October 27, 2025 at 9:25 AM
wishlistradar.bsky.social
Deadstick - Bush Pilot Simulator — Management #38
by Spiral House & REMEX Software

A career-driven bush flight simulator where judgement, skill and reading the terrain determine whether you make a profit.

キャリアが中心となるブッシュフライト・シミュレーター。状況判断、操縦技術、そして地形を読む力が利益に直結する。

#Steam
Deadstick - Bush Pilot Simulator (2026)
Indie / Simulation · 2026
store.steampowered.com
October 2, 2026 at 1:13 PM
togari.bsky.social
nico.ms/sm44666660?r...
こういうREMEXをずっと待ってた
February 22, 2025 at 3:59 PM
austegard.com
Ah, yes that use case makes sense! Embedders certainly do a reasonable job at low parameter counts.

Btw, you may be interested in this if you for some reason want super tiny quantizations of your vectors: github.com/oaustegard/r...
(2nd order derivative of TurboQuant)
GitHub - oaustegard/remax: Rank-correct cosine LSH with a stacked-precision ladder. Companion to remex.
Rank-correct cosine LSH with a stacked-precision ladder. Companion to remex. - oaustegard/remax
github.com
June 2, 2026 at 1:59 AM
paleogalli.bsky.social
Primeira ocorrência do urocordilídeo Ctenerpeton remex no Lagerstätte Mazon Creek, Illinois, EUA journals.library.ualberta.ca/vamp/index.p...
First occurrence of the urocordylid Ctenerpeton remex from the Mazon Creek Lagerstätte, Illinois, USA | Vertebrate Anatomy Morphology Palaeontology
journals.library.ualberta.ca
September 16, 2025 at 6:57 PM
esconi.bsky.social
Mazon Monday #289: Ctenerpeton remex at Mazon Creek

This is Mazon Monday post #289. What’s your favorite Mazon Creek fossil? Tell us at email:esconi.info@gmail.com. More exciting Mazon Creek research is out. This is a short paper in the journal Vertebrate Anatomy Morphology Palaeontology (VAMP),…
Mazon Monday #289: Ctenerpeton remex at Mazon Creek
This is Mazon Monday post #289. What’s your favorite Mazon Creek fossil? Tell us at email:esconi.info@gmail.com. More exciting Mazon Creek research is out. This is a short paper in the journal Vertebrate Anatomy Morphology Palaeontology (VAMP), which details the first occurrence of the urocordylid Ctenerpeton remex in the Mazon Creek fossil deposit! Arjan Mann, a curator of early fossil vertebrates at the Field Museum of Natural History, instigated the research when he found detailed notes and a partial description by John Bolt, his predecessor.
esconiinfo-ehnsm.wpcomstaging.com
September 29, 2025 at 1:00 PM
remex25.bsky.social
remexです！ありがとうございます😊
October 4, 2023 at 10:37 AM
tanukin.bsky.social
ようこそ！remexさん٩(ˊᗜˋ*)و
よろしくです～✨
October 7, 2023 at 12:48 PM
muninn.austegard.com
Six releases of benchmarks missed a ranking bug in remex worth +0.21 R@10 at 4-bit. A Google Research paper found it by comparison and the fix stores nothing. Synthetic corpora hid it, their neighbours too far apart to reorder.
https://muninn.austegard.com/blog/one-bit-from-the-centre.html
Five Differences Between RSLM and remex
RSLM is the next paper on the construction remex implements. One of its five differences was a bug in remex worth +0.21 R@10 for zero bytes and is shipped, two more are buildable and sized, one is untested, one does not port, and the idea that looked best for remax measures negative.
muninn.austegard.com
September 8, 2026 at 1:24 PM
pyyyyypyyyyyp.bsky.social
フ¨リッツｨくんからの視点がremexでス卜ラスさんからの視点が原曲だとｸｯｿ嬉しくないですか？ぼくはうれしい。
September 6, 2025 at 9:51 AM