Show HN: PicoLM v1.0-rc2 ("Yura Kana"). Run an LLM on Digital Unix
PicoLM - But It's Actually Useful What's New v1.0-rc2 ("Yura Kana") RoPE scaling: linear, YaRN, and LongRoPE methods. Noble effort, but it was missing basically everything you need to actually use an LLM. So I went in and added hundreds of commits across quantization acceleration, model support, HTTP server, GPU backends, and cross-platform fixes.
- ▪PicoLM - But It's Actually Useful What's New v1.0-rc2 ("Yura Kana") RoPE scaling: linear, YaRN, and LongRoPE methods.
- ▪Noble effort, but it was missing basically everything you need to actually use an LLM.
- ▪So I went in and added hundreds of commits across quantization acceleration, model support, HTTP server, GPU backends, and cross-platform fixes.
Hacker News (AI / LLM) files mainly under ai. We currently carry 6,153 of its stories.
Story provenance
Source · retrieval · rights · ranking — open for full record
inspect →
Story provenance
Attribution is not the same as permission. This drawer separates discovery metadata, excerpts, WeSearch-generated summaries, reuse status, and whether the publisher receives the visit. Nothing here claims a legal grant the publisher has not made.
Record
| Original publisher | GitHub |
| Canonical URL | https://github.com/whoreson/picolm/ |
| Publication time | Wed, 23 Sep 2026 16:12:19 +0000 |
| Retrieval time | 2026-09-23T16:29:30.850Z |
| Last seen | 2026-09-23T16:29:30.850Z |
| Headline source | Publisher (no WeSearch rewrite) |
| Excerpt source | publisher body |
| Excerpt method | First ~120 words (~800 chars) of extracted publisher body, fair-use limited. |
| Summary | WeSearch · cerebras-chat (WeSearch summarizer) |
| Summary source text | contentText |
| Citation coverage | Summary is a WeSearch-generated derivative; primary citation is the original publisher URL. |
| Cluster | 3_0EdU4Pu-Oj · 1 stories |
| Cluster logic | Grouped by semantic title/content similarity across sources within a rolling window. Same-publisher template collisions are excluded from coverage comparison. |
| Ranking reason | Story pages are not engagement-ranked. Hub feeds use recency, with optional source-diversified chronological ordering (cap consecutive stories per source). No personalized ranking. |
| Publisher visit | Yes — open original |
| Substitutes article? | No — link-out required for full text |
Rights status (four layers)
WeSearch handling by dimension
| Indexing | May the item be indexed (stored, ranked, made findable)? | Allowed |
| Snippet | May a short excerpt of the publisher's text be shown? | Allowed |
| AI summary | May WeSearch generate its own short summary of the article? | Limited |
| Retrieval / RAG | May the content be exposed for third-party retrieval-augmented generation? | Not asserted |
| Model training | May the content be used to train AI models? | Not asserted |
| Commercial reuse | May the content be reused commercially? | Not permitted |
Basis: Derived from the published RSS/Atom feed. Contact: [email protected]. Reviewed: 2026-07-24.
Opening excerpt (first ~120 words) tap to expand
PicoLM - But It's Actually Useful What's New v1.0-rc2 ("Yura Kana") RoPE scaling: linear, YaRN, and LongRoPE methods. Pre-computed cos/sin tables modified per GGUF rope.scaling metadata (freq_scale, ext_factor, attn_factor, beta_fast/slow), enabling extended context beyond model defaults Sliding window attention (SWA), model-independent with unit tests Tokenizer rewritten: O(n^2) -> O(n log n) SPM via priority queue + hash table, UTF-8 fallback always splits on failed lookup IQ4_NL (GGUF type 20) full support: LUT-based non-linear 4-bit quantization with AVX512 VPERMD, AVX2 VPSHUFB, NEON vqtbl1q_u8 dequant, I8MM/AVX2/NEON int8 MAC matmul, Lloyd-Max 4-bit quantize Quantized tiled GEMM engine (C port of llama.cpp's TinyBLAS Q0): Q8_0/Q4_0/Q5_0 x Q8_0, AVX2 sign-trick+maddubs/dpbusd, AVX-512…
Excerpt limited to ~120 words for fair-use compliance. The full article is at GitHub.