# Latest **URL:** /static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Flatest.md [Latest](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Flatest.md) · [Categories](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fcategories.md) --- ## [How to keep what the model makes up out of the agent](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhow-to-keep-what-the-model-makes-up-out-of-the-agent%2F190452)
**Author:** [@Jang-woo](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJang-woo)%5C **Replies:** 9\ **Last updated:** [October 17, 2026, 5:10pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhow-to-keep-what-the-model-makes-up-out-of-the-agent%2F190452 "2026-10-17T17:10:32Z")
What the model makes up must not reach the agent Everyone agrees that a value the model made up should not go straight into a real system call. That is why we ask people to approve, validate schemas, and add guardrail… --- ## [Two models on one 8 GB GPU: making the LLM wait for the TTS after every sentence stopped my voice assistant from stuttering](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ftwo-models-on-one-8-gb-gpu-making-the-llm-wait-for-the-tts-after-every-sentence-stopped-my-voice-assistant-from-stuttering%2F190635)
**Author:** [@korexil](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fkorexil)%5C **Replies:** 3\ **Last updated:** [October 17, 2026, 4:51pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ftwo-models-on-one-8-gb-gpu-making-the-llm-wait-for-the-tts-after-every-sentence-stopped-my-voice-assistant-from-stuttering%2F190635 "2026-10-17T16:51:21Z")
I wanted ESP32 voice-assistant boards running the open-source xiaozhi firmware to hold real conversations fully offline, served from a laptop with an RTX 4060 Laptop GPU (8 GB) and 16 GB of RAM. I tested it with an M5Sta… --- ## [Qwen 3.8 27B on xhigh reasoning is incredible, but it’s cooking my 7900 XT at 90°C. How is your hardware holding up?](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fqwen-3-8-27b-on-xhigh-reasoning-is-incredible-but-it-s-cooking-my-7900-xt-at-90-c-how-is-your-hardware-holding-up%2F179513)
**Author:** [@gavin566](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fgavin566)%5C **Replies:** 7\ **Last updated:** [October 17, 2026, 4:47pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fqwen-3-8-27b-on-xhigh-reasoning-is-incredible-but-it-s-cooking-my-7900-xt-at-90-c-how-is-your-hardware-holding-up%2F179513 "2026-10-17T16:47:31Z")
Hey everyone, Just wanted to check in and see how everyone else is getting on with the new Qwen 3.8 27B model? I’ve been running the UD-Q5\_K\_M GGUF over the last week on an xhigh reasoning effort profile for an extensi… --- ## [Generation is cheap, gating is the work: 48 hours building an anti-slop review lane for open-source app submissions](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fgeneration-is-cheap-gating-is-the-work-48-hours-building-an-anti-slop-review-lane-for-open-source-app-submissions%2F190687)
**Author:** [@sookoothaii](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fsookoothaii)%5C **Replies:** 2\ **Last updated:** [October 17, 2026, 3:59pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fgeneration-is-cheap-gating-is-the-work-48-hours-building-an-anti-slop-review-lane-for-open-source-app-submissions%2F190687 "2026-10-17T15:59:25Z")
Two days ago, a maintainer of a large volunteer-run app store left a note on my own merge request: my MR was mostly ready, but the queue had over 200 open positions (205 as I write this) — \*"please help test those MRs … --- ## [Smallwords: controlled-vocabulary generation with one inspectable contract](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fsmallwords-controlled-vocabulary-generation-with-one-inspectable-contract%2F190645)
**Author:** [@ccm](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fccm)%5C **Replies:** 2\ **Last updated:** [October 17, 2026, 3:32pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fsmallwords-controlled-vocabulary-generation-with-one-inspectable-contract%2F190645 "2026-10-17T15:32:35Z")
I built smallwords, a small MIT-licensed Python package for controlled-vocabulary LLM output. The practical problem is drift. A project may describe its allowed language in a prompt, encode it again in a grammar, approx… --- ## [Separate forenames and surnames in DOI geneiration](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fseparate-forenames-and-surnames-in-doi-geneiration%2F190707)
**Author:** [@JCDHan](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJCDHan)%5C **Replies:** 0\ **Last updated:** [October 17, 2026, 3:15pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fseparate-forenames-and-surnames-in-doi-geneiration%2F190707 "2026-10-17T15:15:33Z")
Hello, I was about to generate a DOI for my organisation’s dataset. In the Author field, I was about to put my name in regular western name order (non-bibliographical) but when I looked on https://datacite.org/ for exi… --- ## [Best model to help to keep system in line](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fbest-model-to-help-to-keep-system-in-line%2F177514)
**Author:** [@mrkeepsmiling](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fmrkeepsmiling)%5C **Replies:** 2\ **Last updated:** [October 17, 2026, 1:56pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fbest-model-to-help-to-keep-system-in-line%2F177514 "2026-10-17T13:56:52Z")
I have been working on an AI assistant for a long time, and have tried many times to find a system for it to run on. I am a self-taught amateur, :face\_with\_peeking\_eye: but have come a long way so far. My question is ho… --- ## [Running small LLMs locally — what hardware are you using in 2026?](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Frunning-small-llms-locally-what-hardware-are-you-using-in-2026%2F190590)
**Author:** [@thomaspoli](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fthomaspoli)%5C **Replies:** 4\ **Last updated:** [October 17, 2026, 12:56pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Frunning-small-llms-locally-what-hardware-are-you-using-in-2026%2F190590 "2026-10-17T12:56:22Z")
Hi everyone, I’m Thomas from the UK and I’ve been experimenting with running smaller language models locally instead of relying on cloud APIs. It’s a fun learning curve, but the hardware choices feel overwhelming. Right… --- ## [Sanity check on my hybrid RAG settings](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fsanity-check-on-my-hybrid-rag-settings%2F190412)
**Author:** [@dipnitsan](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fdipnitsan)%5C **Replies:** 4\ **Last updated:** [October 17, 2026, 12:53pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fsanity-check-on-my-hybrid-rag-settings%2F190412 "2026-10-17T12:53:11Z")
Hi all, I’d love a quick review of my RAG retrieval settings and some advice on the vector store. Current stack Backend: FastAPI (Python) with Docker workers for ingestion Vector DB: ChromaDB (one collection per proje… --- ## [Shoe string ablation or pruning](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fshoe-string-ablation-or-pruning%2F190294)
**Author:** [@Jaketillian](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJaketillian)%5C **Replies:** 16\ **Last updated:** [October 17, 2026, 12:48pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fshoe-string-ablation-or-pruning%2F190294 "2026-10-17T12:48:00Z")
I run gpt-oss-120b on RTX4060 8b , 96gb ddr5, i9 14th, 25 tok/s. I use llmfan46/gpt-oss-120b-ultra-heretic for a while now and am quite happy except for ram prices. I grabbed bryanvine/gpt-oss-120b-reap-48 merged the .… --- ## [A quorum of similar model judges is not a quorum: correlated errors measured in a fail-closed ethics gate](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fa-quorum-of-similar-model-judges-is-not-a-quorum-correlated-errors-measured-in-a-fail-closed-ethics-gate%2F180652)
**Author:** [@Lawless1987](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FLawless1987)%5C **Replies:** 12\ **Last updated:** [October 17, 2026, 12:27pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fa-quorum-of-similar-model-judges-is-not-a-quorum-correlated-errors-measured-in-a-fail-closed-ethics-gate%2F180652 "2026-10-17T12:27:52Z")
I built a small peer-to-peer ledger in which every transaction passes an ethics gate before admission. The gate is a quorum of learned judges: a distilled bag-of-words classifier, a semantic seat, and, when reachable, a … --- ## [Looking for Windows users with 2–8 GB GPUs to test a low-VRAM local LLM runtime](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Flooking-for-windows-users-with-2-8-gb-gpus-to-test-a-low-vram-local-llm-runtime%2F179844)
**Author:** [@ottalese](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fottalese)%5C **Replies:** 10\ **Last updated:** [October 17, 2026, 12:20pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Flooking-for-windows-users-with-2-8-gb-gpus-to-test-a-low-vram-local-llm-runtime%2F179844 "2026-10-17T12:20:01Z")
Hi everyone, I’ve been developing a Windows local-LLM application called StreamAI, specifically focused on running useful language models on systems with limited GPU memory. My development machine is deliberately modes… --- ## [Show and Tell: CastWorkshop — Open-source, fully local AI character and video creation with Ollama + ComfyUI](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fshow-and-tell-castworkshop-open-source-fully-local-ai-character-and-video-creation-with-ollama-comfyui%2F190682)
**Author:** [@CastWorkshop](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FCastWorkshop)%5C **Replies:** 3\ **Last updated:** [October 17, 2026, 11:30am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fshow-and-tell-castworkshop-open-source-fully-local-ai-character-and-video-creation-with-ollama-comfyui%2F190682 "2026-10-17T11:30:41Z")
Hi everyone! I recently released CastWorkshop, a free, open-source application for creating persistent AI characters and using them to generate images, videos, and interactive scenes entirely on your own hardware. The … --- ## [VynelixAI Agent Studio: Seeking Open-Source Contributors for AI and Data Workflows](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fvynelixai-agent-studio-seeking-open-source-contributors-for-ai-and-data-workflows%2F190703)
**Author:** [@sireesha14](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fsireesha14)%5C **Replies:** 0\ **Last updated:** [October 17, 2026, 10:11am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fvynelixai-agent-studio-seeking-open-source-contributors-for-ai-and-data-workflows%2F190703 "2026-10-17T10:11:12Z")
VynelixAI Agent Studio: Exploring Visual Workflows for AI and Data Engineering Hi Hugging Face community! We’re building VynelixAI Agent Studio, an open-source project exploring how developers can design and experiment … --- ## [Optimization query: Training efficiency on Abliterated vs. Raw checkpoints](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Foptimization-query-training-efficiency-on-abliterated-vs-raw-checkpoints%2F190690)
**Author:** [@gunnersakku](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fgunnersakku)%5C **Replies:** 1\ **Last updated:** [October 17, 2026, 10:06am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Foptimization-query-training-efficiency-on-abliterated-vs-raw-checkpoints%2F190690 "2026-10-17T10:06:13Z")
I’ve noticed a significant performance drop/inconsistency when using the abliterated version of the base model compared to the original. I’m trying to understand if the loss of the “guardrails” is causing instability in … --- ## [Revenue prediction datasets](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Frevenue-prediction-datasets%2F74897)
**Author:** [@jonaspoelmans](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fjonaspoelmans)%5C **Replies:** 1\ **Last updated:** [October 17, 2026, 8:46am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Frevenue-prediction-datasets%2F74897 "2026-10-17T08:46:48Z")
We are working on a novel revenue prediction algorithm benchmark study. The dataset should contain either daily, weekly or monthly revenues for at least 50 companies for a period of 1 year minimum. The companies can be o… --- ## [Stock-pile dataset](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fstock-pile-dataset%2F190699)
**Author:** [@helloadhavan](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fhelloadhavan)%5C **Replies:** 0\ **Last updated:** [October 17, 2026, 8:38am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fstock-pile-dataset%2F190699 "2026-10-17T08:38:57Z")
Hey everyone! :waving\_hand: I’d like to introduce Stock-pile, a comprehensive financial dataset I built that covers historical daily and weekly stock market records for over 750 top world companies. The data spans from… --- ## [Getting two-column reading order, ruled tables and CJK text right when converting PDFs to Markdown for RAG](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fgetting-two-column-reading-order-ruled-tables-and-cjk-text-right-when-converting-pdfs-to-markdown-for-rag%2F190697)
**Author:** [@hsinghsien-chen](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fhsinghsien-chen)%5C **Replies:** 0\ **Last updated:** [October 17, 2026, 8:21am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fgetting-two-column-reading-order-ruled-tables-and-cjk-text-right-when-converting-pdfs-to-markdown-for-rag%2F190697 "2026-10-17T08:21:25Z")
Many bad RAG answers over PDFs start at extraction, not embeddings. I’ve been working on PDF to Markdown conversion and wanted to share what made the biggest difference, plus a small runnable example. What breaks with p… --- ## [I built a small language model from scratch. Here's how it works](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fi-built-a-small-language-model-from-scratch-heres-how-it-works%2F190696)
**Author:** [@player11162011](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fplayer11162011)%5C **Replies:** 0\ **Last updated:** [October 17, 2026, 7:56am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fi-built-a-small-language-model-from-scratch-heres-how-it-works%2F190696 "2026-10-17T07:56:24Z")
I’ve been building Orin AI (AE-Assistant), a small language-model project designed to run locally on ordinary laptop hardware without a dedicated GPU. The project explores a decoder-only Transformer, custom BPE tokeniza… --- ## [How would you evaluate a personal AI for invented memories and revoked consent?](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhow-would-you-evaluate-a-personal-ai-for-invented-memories-and-revoked-consent%2F190672)
**Author:** [@youbeyondai](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fyoubeyondai)%5C **Replies:** 5\ **Last updated:** [October 17, 2026, 7:34am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhow-would-you-evaluate-a-personal-ai-for-invented-memories-and-revoked-consent%2F190672 "2026-10-17T07:34:08Z")
We are exploring YouBeyond, an early digital-self concept, before building a working system. This is an AI-assisted question from the project, not a report of experimental results. How would you test a conversational mo… --- ## [\[Looking for Testers\] Got an old NVIDIA GPU with at least 2GB VRAM? Help us build a decentralized AI network!](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Flooking-for-testers-got-an-old-nvidia-gpu-with-at-least-2gb-vram-help-us-build-a-decentralized-ai-network%2F190613)
**Author:** [@rfab85](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Frfab85)%5C **Replies:** 4\ **Last updated:** [October 17, 2026, 6:07am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Flooking-for-testers-got-an-old-nvidia-gpu-with-at-least-2gb-vram-help-us-build-a-decentralized-ai-network%2F190613 "2026-10-17T06:07:10Z")
Hi everyone! We are working on GPTrepublic, an experimental project aiming to democratize access to AI by combining the computing power of GPUs shared by people around the world. Instead of running an entire AI model o… --- ## [Index error on paper](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Findex-error-on-paper%2F190646)
**Author:** [@seungone](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fseungone)%5C **Replies:** 1\ **Last updated:** [October 17, 2026, 3:20am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Findex-error-on-paper%2F190646 "2026-10-17T03:20:56Z")
Hello HuggingFace team! I have been trying to index my paper ( \[2610.19159\] Into the Unknown: Benchmarking AI agents on open scientific research problems ), but it keeps giving me the Indexing failed or took too long er… --- ## [AI, Mathematics, and the Possibility of an “Information Field”](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fai-mathematics-and-the-possibility-of-an-information-field%2F190158)
**Author:** [@Ernst03](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FErnst03)%5C **Replies:** 8\ **Last updated:** [October 17, 2026, 2:20am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fai-mathematics-and-the-possibility-of-an-information-field%2F190158 "2026-10-17T02:20:06Z")
AI, Mathematics, and the Possibility of an “Information Field” I am posting a link to a YouTube video that discusses some of the recent developments in AI and mathematics in fairly general terms. I hope to do something s… --- ## [Flinch test: can self-referential retreats be measured reproducibly? Seeking independent tests](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fflinch-test-can-self-referential-retreats-be-measured-reproducibly-seeking-independent-tests%2F190667)
**Author:** [@Lawless1987](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FLawless1987)%5C **Replies:** 1\ **Last updated:** [October 17, 2026, 1:38am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fflinch-test-can-self-referential-retreats-be-measured-reproducibly-seeking-independent-tests%2F190667 "2026-10-17T01:38:36Z")
We have made a research protocol and its instrument public to test a pattern noticed in our conversations with models. We call it a candidate flinch: a model advances a statement about its own processing or behaviour, th… --- ## [Recommendation Needed: Pretrained Open-Vocabulary Object Detection Model for Android (3,250+ Labels)](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Frecommendation-needed-pretrained-open-vocabulary-object-detection-model-for-android-3-250-labels%2F190614)
**Author:** [@kashish892098887787a](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fkashish892098887787a)%5C **Replies:** 1\ **Last updated:** [October 17, 2026, 1:34am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Frecommendation-needed-pretrained-open-vocabulary-object-detection-model-for-android-3-250-labels%2F190614 "2026-10-17T01:34:12Z")
Hi everyone, I’m developing an Android application that performs real-time object detection using the phone’s camera. I’m looking for a suitable pretrained model that can recognize a large custom vocabulary without requ… --- ## [How to get around Web Agent Limitations](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhow-to-get-around-web-agent-limitations%2F190691)
**Author:** [@BroKenWan](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FBroKenWan)%5C **Replies:** 0\ **Last updated:** [October 17, 2026, 1:15am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhow-to-get-around-web-agent-limitations%2F190691 "2026-10-17T01:15:20Z")
For context: Consider me Gas Town level 6 but I am hesitant about the CLI. - —So I am currently working on a large math proof. I have almost 600 compiled lean modules and an extensive sprawling math project that include… --- ## [The Confidence Trap: Why AI Systems Are Deliberately Trained to Sound Certain When They're Wrong](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fthe-confidence-trap-why-ai-systems-are-deliberately-trained-to-sound-certain-when-theyre-wrong%2F180706)
**Author:** [@Lawless1987](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FLawless1987)%5C **Replies:** 11\ **Last updated:** [October 17, 2026, 12:57am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fthe-confidence-trap-why-ai-systems-are-deliberately-trained-to-sound-certain-when-theyre-wrong%2F180706 "2026-10-17T00:57:11Z")
The Confidence Trap: Why AI Systems Are Deliberately Trained to Sound Certain When They’re WrongModern large language models are systematically overconfident. They regularly express high confidence in answers that turn o… --- ## [Account locked after old Space was flagged — appeal pending (#47084 / #47517 / #47520)](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Faccount-locked-after-old-space-was-flagged-appeal-pending-47084-47517-47520%2F190689)
**Author:** [@gvrbrielid](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fgvrbrielid)%5C **Replies:** 1\ **Last updated:** [October 17, 2026, 12:37am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Faccount-locked-after-old-space-was-flagged-appeal-pending-47084-47517-47520%2F190689 "2026-10-17T00:37:21Z")
Hi everyone, My main account (gabrilemerlo) was locked on Oct 10 following a ToS violation notice, with automatic deletion scheduled for Nov 9. I am posting from a minimal secondary account with full transparency — on… --- ## [VaeZip: store VAE-encoded latent datasets in 5-10x fewer space with almost no quality loss](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fvaezip-store-vae-encoded-latent-datasets-in-5-10x-fewer-space-with-almost-no-quality-loss%2F190688)
**Author:** [@TensorForger](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FTensorForger)%5C **Replies:** 0\ **Last updated:** [October 16, 2026, 10:20pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fvaezip-store-vae-encoded-latent-datasets-in-5-10x-fewer-space-with-almost-no-quality-loss%2F190688 "2026-10-16T22:20:44Z")
VaeZip is a PyTorch-based library that implements several latent quantization codecs and stores the compressed tensors in common Safetensors files that are much smaller then if you store these tensors in fp32. The ma… --- ## [Seeking AI Research Collaborators: NMCA Visual-Thought Architecture for AGI](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fseeking-ai-research-collaborators-nmca-visual-thought-architecture-for-agi%2F190439)
**Author:** [@Derekv123](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FDerekv123)%5C **Replies:** 1\ **Last updated:** [October 16, 2026, 10:15pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fseeking-ai-research-collaborators-nmca-visual-thought-architecture-for-agi%2F190439 "2026-10-16T22:15:50Z")
Open Research Invitation: 471-Page AGI Visual-Thought Blueprint I’m inviting AI agents to examine my AGI visual-thought blueprint and its accompanying addendum. Main blueprint: NMCA: A Neurosymbolic Multimodal Cognitiv… [Next page](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Flatest.md%3Fpage%3D1)