Knowledge catalogue
Check out Chimera Arena (https://chimeraarena.com/static/showcase/index.html?lang=en)! Here’s my slightly scrappy AI-generated teaser for Chimera Arena, my card game combining AI-generated video battl
I have been running the same midrange card for a couple of years now, and every model release sends me back to the price sheets. The performance gains are real but they keep landing next to price tags
IQ3_XXS weights are just under 80GB and my slowww DDR4+7900XTX is stabilizing around 45-70/s (sometimes higher while coding depending on mtp). Looking online I'm seeing similar results for users with
Meeting notes are one of the last parts of my workflow that still depend heavily on cloud tools. I've used Fathom and lately Bluedot. Bluedot works well for me because there's no meeting bot and I get
Hey everyone, I’ve been working on a game called Anyworld. It’s a browser-based multiplayer (single player also supported) text adventure inspired by the early days of AI Dungeon, especially its brows
Hi, Im gonna go all in on this, best finetune iv got my hands on for agentic coding, period. Test it yourself, im not gonna give you any benchmarks or evaluations, use your own tests and pracices, giv
metal : add tensor API flash attention kernel for F16 KV (#29570) metal : add tensor API flash attention kernel for F16 KV cont : add tensor FA kernels for DK=DV=512 and DK=576, DV=512 cont : support
model: support nimble decision model (#29844) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/52332773 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Si
ggml-cpu : fix soft_max_back wrong output when dst aliases src1 (#27096) ggml-cpu : fix soft_max_back wrong output when dst aliases src1 GGML_OP_SOFT_MAX_BACK is listed in ggml_op_can_inplace, so the
ggml-quants : avoid invalid rounding in qkx3 scale search (#29817) ggml-quants : avoid invalid rounding in qkx3 scale search The imatrix scale search can produce an infinite, NaN, or otherwise out-of-
spec : add probabilistic sampling for simple draft and MTP (#27694) Make the drafter probabilistic and the target verify by rejection sampling Drop stale spec_draft_q before drafting Fallback to argma
CUDA: fuse shared experts into MMVQ (#29184) CUDA: fuse shared experts into MMVQ check if buffer is null move stride_col_dst to fusion args Website: https://llama.app Attestations: https://github.com/
model: add support for clef decision model (text-only) (#29831) init support for clef (text only) more static graph clean up nits nits 2 Update gguf-py/gguf/constants.py Co-authored-by: Sigbjørn Skjær
qwen4exp : halve the indexer score memory (#29825) qwen4exp : halve the indexer score memory The indexer scored all heads in one product and rectified a copy of it, so two [n_pool, n_idx_h, n_tokens]
ggml-openvino: update to 2026.4.1, optimize performance, expand ops, improve device listing. (#29852) ggml-openvino : Qwen3.5 MoE perf (#312) Squash of ravi9#312: ggml-openvino: add detailed inference
graph: gather the recurrent states once so the reserve covers every split (#29856) build_rs gathered the extra states (n_rs - n_seqs rows) with their own get_rows. The worst-case reserve has n_rs == n
ci: fix flaky ADD_ADD f16 by using the fused ADD tolerance (#29904) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/52418791 macOS/iOS: macOS Apple Silicon
chat : honor json_schema in Ling 3.0 parser (#29813) chat : honor json_schema in Ling 3.0 parser Ling 3.0 only built a grammar for tool calls and did not handle inputs.json_schema, so response_format
common : add common_is_tty() helper and fix deprecated warnings on Windows (#29860) Signed-off-by: Adrien Gallouët angt@huggingface.co Website: https://llama.app Attestations: https://github.com/ggml-
server : fix laya abort by limiting n_batch to n_ubatch (#29903) server : fix laya abort by limiting n_batch to n_ubatch Fixes #29902 Assisted-by: Claude fix(review) : rm tests, embeddings cond Websit
vendor : update cpp-httplib to 0.59.0 (#29886) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/52476117 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple S
mtmd : fix deprecated strdup warning on Windows (#29863) Signed-off-by: Adrien Gallouët angt@huggingface.co Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/
Banger paper from CMU on harness learning. (bookmark it) Also, pay attention to this important new AI engineering skill of improving agents by editing their harness code instead of their weights. Seei
Banger paper from Meta Superintelligence Labs. They find something super interesting and unexpected. (bookmark it) Base models with a light harness often solve more agentic tasks than their RL post-tr
Clef Flash (9 billion‑parameter) is a multimodal model hosted on Ollama that converts a given state and a schema of typed questions into actionable decisions. Its counterpart, Clef, is also available
.@Cloudflare's decision models are available on Ollama! Decision models allow you to classify an image, label a bug report or route a support ticket to the right team. Clef (27B): ollama pull clef Cle
I hosted my own world of warcraft private server then built a client that you can play in the browser on PC or mobile at https://jankcraft.xyz/ for free. Afterwards, I created a custom MCP and agent h
Still don't have a good cooling solution, but here are few benchmarks. I lowered the power limit (TDP) to 145 watts each. I changed the firmware on one MI50 to activate the miniDP port. Did have to us
First AutoHarness, then AutoContext, now AutoCompact. I am seeing a rising trend of work that trains models to natively support more of what the harness does. This work specifically trains agents to d
NGL, I do, but I’ve my reasons…. I do discern it’s answers but I’ve been so nomadic I’ve barely any friends my family of two aged parents won’t understand the concept of mental health and my diagnosis
So this is maybe a niche problem, but at my job I work on a huge Python codebase and every time I change some shared function I'm basically playing roulette. grep tells who mentions it in the code bas
Hi everyone, TL/DR I created NInfer 4080 to run ISTA-DASLab-Qwen-3.8-27B-GSQ at 100k context on an RTX 4080 16GB GPU using way more of the hardware capabilities (max overall: 2720 tok/s prefill, 262 t
I just made the most expressive real-time motion harness for the Reachy Mini. It was created by: > synthesizing a massive motion dataset using Astra (over 10000 samples) > fine-tuning Qwen 3.5 4B usin
I love Qwen3.8 27B so much that I decided to show my gratitude to the Alibaba ecosystem by building a dedicated RAG/search node using a cheap Mini PC from AliExpress. Turns out, my ecosystem loyalty g
'TypeSafe AI sells Jev as a frontier-class reasoner that cannot hallucinate, built by the co-inventor of ChatGPT - fast, and almost free. We ran it live on 16,379 benchmark requests, measured its late
Krea 2 Turbo — 2-Step Distillation LoRA (work in progress) Previous posts/releases - here, here and here. 🧪 A fast-preview adapter, from a project still in training. When the subject is close and fill
Language is not fundamental to human consciousness. In terms of evolution, it arrived only recently on our timeline. Consciousness evolved over millions of years through sensory awareness, emotion, an
LFM2.5-Encoder-350M is a multilingual bidirectional encoder built on the LFM2 architecture — a larger encoder for maximum downstream quality. It is a masked language model with full bidirectional atte
LWiAI Podcast #258 (recorded 26‑Sep‑2026) reviews major AI updates from the previous week. Key developments include Anthropic’s Opus 5.5 release—offering lower pricing, faster performance and robust b
Meta actually lost this, was fined 18b, shrugged and carried on The largest ever antitrust award was 5.5b (Visa & Mastercard) Any fine for slowing AI would be a fraction of this It’s not a serious arg
Full disclaimer: I've leaned heavily on Fable to develop this, but I've tested it thoroughly on my own library for a couple of months before putting it on GitHub. I live in Thailand, and it started as
Must-read paper from Meta on reliable auto-research agents. If you let coding agents run ML experiments, one silent bug like leaked eval data or a disconnected gradient can invalidate hours of trainin
Meir Orbach / CTech: Nasdaq-listed neocloud Nebius acquires Inferize, whose tech helps optimize GPU utilization and reduce AI request costs, sources say for 100M to 150M — The Israeli startup, which w
ollama pull clef open models for all. start building with @ollama and @cloudflare clef decision models today .@Cloudflare's decision models are available on Ollama! Decision models allow you to classi
one of the lessons aviation learned was that regulation is useful for gaining public trust. the AI industry refuses to learn that lesson - and then wonders why everyone hates them. I disagree with the
Sarah Perez / TechCrunch: OpenAI's DevDay 2026 announcements to turn ChatGPT into a place to discover, launch, and use software could potentially disrupt the traditional app store model — The focus of
.@OpenRouter co-founder Alex Atallah, in his first podcast since Stripe acquired the company, joins @Replit co-founder Amjad Masad and a16z's Erik Torenberg on why the future of AI is independence and
Our new extraction agents are extremely good at reasoning over tables that span multiple pages sometimes a record starts on one page and finishes on the next. extremely common for real world docs (ins
Reupload: I'm adding this again, since previous post had a totally stupid title which made people assume this app is some kind of cloud/proprietary models UI - this is totally not the case here. In ve
There are quite often posts here complaining that Krea 2 is too predictable / different seeds give almost the same output / you need to type in extremely long prompts to get good images / there is no
This is actually two models: Qwen Image 2.1 Fix v2.0 and Qwen Image 2.1 Fix Opinionated v1.0. The former surgically targets Qwen's weird noise and messy details with almost no effect on composition, a
Benchmarking an LLM here with a NVIDIA RTX 5070 12 GB VRAM here I had been working on a llama.cpp based expert streaming setup for Qwen3.8-Flash-Next 177B (UD-IQ3_XXS) on Windows. Benchmark is about 1
Reasoning from scratch, round number 6! An introduction (and implementation) of Reinforcement Learning with Verifiable Rewards (RLVR) and Group Relative Policy Optimization (GRPO). 00:00 Introduction
Disclosure: I am the creator of this project. After days of lurking and building up enough karma, I can finally post here. Every time I tried running local RAG on my own machine, I hit the exact same
Workflow & Hardware The workflow was originally based on Pixaroma’s ComfyUI workflow. I made some modifications and added Sol Attn optimization and Comfy Kitchen Sink, including the very useful Model
FieldAI Inc., a startup that develops artificial intelligence models for robots, is reportedly seeking 700 million in funding. A source told Business Insider today that the round could value the compa
I wanted to see how far I could push a fairly ordinary laptop with a huge MoE model. Turns out, Qwen3.8 Flash Next 176B can run on: RTX 3080 Laptop — 16GB VRAM 32GB system RAM SSD No 128GB/256GB RAM w
I just sent the September edition of my sponsors-only monthly newsletter. If you are a sponsor (or start a sponsorship now) you can access it here. This month: More Fable class models A pricing war 3D
Side note. The idea that current models are conscious is absurd and downright offensive. Do not confuse competent information processing and sentience, they are wholly unrelated. You can be entirely i
Follow-up to last month's post. One of the main issues people ran into was roughness or break-up on some cloned voices. 2610 is an interim update focused mostly on improving that. Reduced roughness an