Explore
Showing results for:
"Hacker News"
1,284 results • Page 71 of 107
Matched "Hacker News"Found in Description
15 points 5 comments on Hacker News · historicmysteries.com
Matched "Hacker News"Found in Content
OpenAI and Hugging Face address security incident during model evaluation - https://news.ycombinator.com/item?id=48997548 - July 2026 (1121 comments)
Matched "Hacker News"Found in Description
346 points 597 comments on Hacker News · codeberg.org
Matched "Hacker News"Found in Description
104 points 27 comments on Hacker News · cnn.com
Matched "Hacker News"Found in Description
87 points 25 comments on Hacker News · yummymelon.com
Matched "Hacker News"Found in Description
80 points 26 comments on Hacker News · box2d.org
Matched "Hacker News"Found in Description
449 points 122 comments on Hacker News · citizendot.github.io
Matched "Hacker News"Found in Description
99 points 14 comments on Hacker News · nondescriptpointer.com
Matched "Hacker News"Found in Source
https://xcancel.com/na_announce/status/2079952538040672302 https://www.twit.community/t/john-c-dvorak-has-passed-away/2... https://www.facebook.com/aric.mackey/posts/john-c-dvorak-194...
Matched "Hacker News"Found in Source
Beginning July 27, 2026, GitHub will cut public bug bounty payouts by at least half at every severity level. Critical findings will drop from $20,000-$30,000+ to a fixed $10,000, while its permanent invite-only VIP tier will pay $30,000 or more. Reports filed before that date, including those already in GitHub's growing triage queue, will retain the previous payout terms. GitHub said the
Matched "Hacker News"Found in Source
Cybersecurity researchers have disclosed details of a new local privilege escalation (LPE) vulnerability in snap-confine that an unprivileged user can trigger to obtain root access and gain complete control of a target environment. The high-severity flaw, tracked as CVE-2026-8933 (CVSS score: 7.8), impacts default installations of Ubuntu Desktop 24.04, 25.10, and 26.04. The disclosure comes as
Matched "Hacker News"Found in Source
Hey HN, Henry & Roman here from Cactus. A small, on-device model is fast and private, but sometimes wrong, but frontier models are getting expensive pretty fast. So, we post-trained Gemma 4 E2B post-trained to know when it's wrong. Every response comes with a confidence score between 0 and 1. Developers can accept the on-device when it's high, hand off to a bigger cloud model when it's low. By routing only 15-35% of queries to Gemini 3.1 Flash-Lite, Gemma-4-E2B matches Gemini 3.1 Flash-Lite on most benchmarks. - ChartQA: 15-20% - LibriSpeech: 25-30% - MMBench, GigaSpeech, MMAU: 30-35% - MMLU-Pro: 45-55% We were always frustrated by the routing signals hybrid apps rely on: asking the model to rate itself in text (unreliable, and you're parsing prose), or token entropy heuristics (barely better than a coin flip in our tests). So we did mechanistic studies on small models, Gemma 4 particularly, and found the hidden state for different layers carry meaningful self-awareness signal for various situations. SO we extended the model with a 68k params probe layer (LayerNorm, low-rank projection, attention pooling, small MLP head) reads one intermediate layer during decoding and predicts p(wrong); confidence = 1 - p(wrong), returned as structured data, never parsed out of the answer text. Across 12 hold-out benchmarks spanning text, vision and audio, the probe averages 0.814 AUROC vs 0.549 for token entropy. The result that convinced us this is real: the probe was trained on zero audio data, yet scores 0.79-0.88 AUROC on four audio benchmarks where entropy is near-random or worse (0.32-0.52). It's reading a modality-independent correctness signal from the hidden state, not memorizing patterns from its training data. We published all weights on HuggingFace and provide copy-pase codes to run it on Transformers, MLX, Llama.cpp or Cactus. With Ollama, vLLM, SGLang etc in the works. For llama.cpp we ship a patch series you compile in once (upstreaming is planned). The code is MIT licensed; Gemma model use remains subject to the Gemma terms. GitHub: https://github.com/cactus-compute/cactus-hybrid Weights: https://huggingface.co/collections/Cactus-Compute/cactus-hyb... Some caveats: - The probe scores single-sequence decoding only, up to the first 1024 generated tokens. - Handoff works best when routing per task in a multi-step process, not per step. - Hierarchical routing is still in the works: try on-device, then DeepSeek v4 Flash, before Fable/GPT5.5/Gemini/Muse/Grok. - The technique is boutique for each model, we will share each weights as they roll out. These issues are currently being tackled at Cactus and updated weights will be shipped directly into the HuggingFace collection and GitHub repository straight up. Please let us know your thoughts, it helps us find ways to improve the design progressively. Thanks a million!
Hacker News
“GPT-5.5's API pricing is reshaping how startups build AI products”
