frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Glq a port trellis quantization of large language models as a vLLM plugin

https://github.com/cnygaard/glq
1•acd•48m ago

Comments

acd•48m ago
glq is a re implementation gpl port of QTIP Trellis encoding of large language models as vLLM plugin. glq is efficient between 2-4 bits per word where it has an edge over scalar quantization. Fit larger LLMs on consumer hardware GPUs. glq extents QTIP up to 8 bits per word by running the original algorithm two times with a residual.

1.Incoherence processing — two-sided randomized Hadamard transform (via FWHT) makes weights ~i.i.d. Gaussian 2. BlockLDLQ rounding — sequential sweep with Hessian error feedback 3. Viterbi encoding — bitshift trellis + computed Gaussian code 4. Decode — stateless bit-shift arithmetic, fused into matmul https://github.com/cnygaard/glq/blob/main/docs/trellis-3inst...

glq comes with a user friendly installer which installs glq, vLLM and gradio chat interface and optionally a local coding agent. If you choose glq-chat glq sizes which model fits in VRAM downloads the model and starts a chat, on a 24GB Nvidia card Gemma4 is choosen. glq also ships with picode and then Qwen 3.8 27B is used.

glq also works on pure CPU. Because the LLM weights are compressed and are decompressed on the fly less DRAM/VRAM bandwidth is used.

The target group of glq is PC gamers with Linux and a Nvidia gaming card. If you happen to have an AMD Epyc 9005 server CPU the cpu encoding/decoding should be usable.

Original QTIP paper not affiliated but grateful to the original authors, https://arxiv.org/pdf/2406.11235

AI doesn't take over. We give it the keys

https://mauriciogomes.com/ai-doesnt-take-over/
1•mauricio•25s ago•0 comments

Show HN: We served 5k dynamic websites from a 2-vCPU, 4 GB VPS

https://github.com/Kooboo/Kooboo/blob/main/Docs/5000-sites-benchmark.md
2•zguoqi•1m ago•0 comments

Update from Buckmaster: Sébastien's statement is wildly false

https://mastodon.social/@tristanbuckmaster/117247207157737650
1•pred_•1m ago•0 comments

I Brought Pokemon into Rust

https://www.youtube.com/watch?v=-oZ67G7AC3I
1•blizow•2m ago•0 comments

The Inner Ring

https://www.lewissociety.org/innerring/
1•sir_unix•2m ago•0 comments

Show HN: Give Agents your design system to build artifacts you're proud of

https://www.valet.dev/blog/work-that-looks-like-yours
1•miradu•3m ago•1 comments

Bending Spoons to buy Miro for $1.36B, 90% less than its 2022 valuation

https://techcrunch.com/2026/09/10/bending-spoons-to-buy-collaboration-tools-maker-miro-for-1-36b-...
1•jaredwiener•3m ago•0 comments

The Absurd, Unending Quest to Floodproof the New York City Subway

https://www.nytimes.com/2026/09/08/magazine/new-york-city-subway-mta-flooding.html
3•bookofjoe•4m ago•1 comments

Launch deserves more than a tweet

https://pressdrop.io
1•isjackback•5m ago•1 comments

Churchill and Lincoln have had it with subscription software [video]

https://www.youtube.com/watch?v=NMY6FiiJQUM
1•phughes1980•7m ago•0 comments

Operation Smart Kettle – Börzels Blog

https://blog.boerzel.de/de/blog/operation-smart-kettle
1•marbartolome•7m ago•0 comments

Killdeer: Independent VM hosting for developers, managed through your terminal

https://killdeer.digital/
1•thunderbong•8m ago•0 comments

Show HN: Xfinlab – a financial intelligence API with an MCP server for AI agents

https://github.com/lnanology/Xfinlab
1•lnanology•8m ago•0 comments

Shopify is moving from React Native to Native

https://twitter.com/mustafa01ali/status/2098047492772249730
1•tosh•9m ago•0 comments

Moduloku

https://www.fractalkitty.com/moduloku/
1•wonger_•10m ago•0 comments

MailPace now has a Domains API

https://mailpace.com/blog/changelog/domains-api
1•albertgoeswoof•11m ago•0 comments

Angelarium – The Encyclopedia of Angels

https://www.angelarium.net
1•bramadityaw•12m ago•0 comments

The Public Suffix List Is a Disaster

https://ethlimo.substack.com/p/the-public-suffix-list-is-a-disaster
3•woodenspoongang•12m ago•0 comments

Models are worse at reviewing their own code

https://www.greptile.com/blog/model-inversion
1•wslh•13m ago•0 comments

Google Doubles Down on WebMCP

https://nekuda.substack.com/p/google-doubles-down-on-webmcp-at
1•amai•13m ago•0 comments

You (Still) Can't Vibe-Code a Product

https://pawelbrodzinski.substack.com/p/you-still-cant-vibe-code-a-product
2•flail•13m ago•0 comments

Guardrail Your Agents with Lean

https://github.com/DebarghaG/LeanGuard
4•barthelomew•14m ago•0 comments

Autonomous Vehicles as the Next General-Purpose Technology

https://humanprogress.org/autonomous-vehicles-as-the-next-general-purpose-technology/
1•surprisetalk•14m ago•0 comments

The Four-Color Theorem Gets a Rare New Proof

https://www.quantamagazine.org/the-four-color-theorem-gets-a-rare-new-proof-20260910/
1•pavel_lishin•15m ago•0 comments

Why some experts increasingly fear AI will take over

https://www.bbc.co.uk/news/articles/c74edv9887eo
1•edward•15m ago•1 comments

Show HN: Model pricing board for DeepSeek Harness: 7k models, cheapest route

https://github.com/vitas/dsh-model-pricing
1•vitass•15m ago•0 comments

Show HN: Skillctl – audit context cost and conflicts across your agent skills

https://github.com/zongwu233/skillctl
1•zongwu233•17m ago•0 comments

Show HN: See Sounds on Your Webcam

https://soundmap.darebuild.com/
1•jimhi•17m ago•0 comments

OpenVPN 2.7.7 Released

https://forums.openvpn.net/d/46625-openvpn-277-released
1•m_montazeri•18m ago•0 comments

AI models don't kill people – people kill people

https://www.theregister.com/ai-and-ml/2026/09/09/ai-models-dont-kill-people-people-kill-people/52...
1•joebuckwilliams•19m ago•1 comments