春雨为观,香月为邻 | 喜欢玩游戏时看风景 | 想成为优秀的程序员 | 看上去是女装大佬实际并不是( | 上海老阿姨 | qwqwqwq

Shanghai
Joined July 2008
Qwen-Image-2.1 running on Intel B580! Thanks to @nihui for qwenimage-ncnn-vulkan
1
11
989
nihui retweeted
Visualizing to learn model internals in the agentic era can be surprisingly low-friction. I had Codex one-shot architecture notes earlier for several recent OSS models—DeepSeek V4/V4.1, GLM 5.3, and Kimi K3—specifically from a kernel-development perspective: tensor contracts, exact shapes/layouts, attention variants, KV-cache representation, precision choices, dataflow, and operator boundaries. Sharing in case they’re useful to others: lei.chat/models/
5
19
1
116
10,472
好耶,libwaifu支持Qwen-Image-2.1成功🎉 github.com/ling0322/libwaifu
1
1
24
2,527
I wanted to add extra GPU options to the benchmark suite, sooo I added @nihui's vkpeak to cover a range of options. Only the @RadxaComputer X4L on sbc.compare/vkpeak so far, but as with others, new boards will be tested with it, and others backfilled later! For full data, the table at the bottom of the comparison pages such as sbc.compare/radxa-x4l is where it will live.
2
1
6
763
Vulkan `vkpeak` benchmark for the new Arduino Ventuno Q reaches 400+ FP32 GFLOPS. ./vkpeak 0 device = Adreno623 driver = turnip Mesa driver / Mesa 25.2.8-0ubuntu0.24.04.2 fp32-scalar = 417.44 GFLOPS fp32-vec4 = 438.42 GFLOPS fp16-scalar = 594.99 GFLOPS fp16-vec4 = 853.58 GFLOPS int32-scalar = 109.97 GIOPS int32-vec4 = 111.09 GIOPS int16-scalar = 283.50 GIOPS int16-vec4 = 294.14 GIOPS int64-scalar = 11.31 GIOPS int64-vec4 = 6.33 GIOPS int8-dotprod = 621.29 GIOPS copy-h2h = 10.11 GBPS copy-h2d = 11.41 GBPS copy-d2h = 9.12 GBPS copy-d2d = 13.66 GBPS @arduino
1
1
4
793
nihui retweeted
Tibo
@thsottiauxSep 7
你最牛了!
472
212
120
3,970
1,386,715
nihui retweeted
YuzukiNeko - A Pico Shape Cute CatBoard, With Allwinner F101 SoC Hardware OpenSource under CC0: github.com/YuzukiHD/YuzukiNe… Linux Kernel: github.com/YuzukiHD/linux-ma… Buildroot: github.com/YuzukiHD/buildroo… Zephyr: github.com/YuzukiHD/zephyr
8
42
5
437
18,217
From dense models to 122B and 295B MoE, local inference should work with the hardware you have—not require a datacenter. Palm-Infra is open source. Built by Tencent's #PalmAI team, it brings Metal, CPU SIMD and SSD expert offload to Apple Silicon.
3
4
3
15
20,560
11k views in 1 day for a technical 1h 40min video! My hope for humanity has been restored. There's still people out there watching long videos :) youtube.com/watch?v=aQv9pUl9…
23
36
819
21,136
nihui retweeted
太好了,上周做的 DSV4 flash 的适配可以直接用在 DeepSeek V4 Flash 0731 的版本上 5070ti 跑 DeepSeek V4 flash )
3
1
10
3,269
Two failed attempts. One working solution. Building a real-time forklift proximity alert system on a Raspberry Pi 5 taught me the biggest bottleneck wasn't the model, it was the deployment stack. What ONNX Runtime, INT8 quantization and NCNN taught me: medium.com/p/two-failed-atte…
3
1
4
1,441
nihui retweeted
先让 ncnn 能运行起 GPT-oss-20b 然后再一步步完善补充 ncnn MoE Runtime 的伟业!Make ncnn Great Again!
1
5
1,768
第一次参与线下的公开演讲,总之就是非常激动!🤩
65
4
116
8,655
轨道交通享受者
139
58
3
2,651
77,457
Didn't expect the PoC prototype worked! I just built a zero runtime dependencies CNN inference server using #golang by wrapping #ncnn (Nihui Convolutional Neural Network) into #wazero WebAssembly runtime, and it cross compile without needing CGO support 🔥
1
1
3
1,624
nihui retweeted
codex for open source! just granted about another huge batch including some that you might recognize! tensorflow/tensorflow n8n-io/n8n twbs/bootstrap github/gitignore ytdl-org/youtube-dl vercel/next.js 30-seconds/30-seconds-of-code kubernetes/kubernetes papers-we-love/papers-we-love angular/angular neovim/neovim microsoft/web-dev-for-beginners florinpop17/app-ideas bitcoin/bitcoin gin-gonic/gin microsoft/playwright laravel/laravel gothinkster/realworld spring-projects/spring-boot tensorflow/models apple/swift unclecode/crawl4ai tldr-pages/tldr snowpackjs/astro embedchain/embedchain vim/vim pingcap/tidb jonnyburger/remotion aspnet/aspnetcore seleniumhq/selenium jqlang/jq immutable-js/immutable-js anncwb/vue-vben-admin pynecone-io/pynecone martinvonz/jj serverless-stack/serverless-stack manojvivek/responsively-app trekhleb/homemade-machine-learning sipeed/picoclaw spicetify/spicetify-cli vueuse/vueuse guidance-ai/guidance nautechsystems/nautilus_trader hshoff/vx preservim/nerdtree officedev/office-ui-fabric-react carlospolop/peass-ng reduxjs/reselect adonisjs/adonis-framework rizinorg/cutter facebookresearch/llama-recipes stackexchange/dapper resendlabs/react-email tomav/docker-mailserver lichess-org/lila google/libphonenumber apache/incubator-brpc googlechrome/chrome-app-samples hwchase17/langchainjs fanux/sealos argoproj/argo argoproj/argo-workflows rjsf-team/react-jsonschema-form secureauthcorp/impacket scylladb/scylla uuidjs/uuid cayleygraph/cayley cesiumgs/cesium eclipse-vertx/vert.x pyodide/pyodide jetstack/cert-manager rileytestut/altstore sunnyyoung/wechattweak-macos pydanny/cookiecutter-django pandas-profiling/pandas-profiling espanso/espanso ansible-semaphore/semaphore k9mail/k-9 nock/nock dotnet/aspnetcore.docs selectize/selectize.js mozilla-mobile/firefox-ios wanghongenpin/network_proxy_flutter webpack-contrib/webpack-bundle-analyzer alicevision/meshroom actions/virtual-environments jxnl/instructor theramu/fay svprogresshud/svprogresshud lexikos/autohotkey_l lipis/flag-icon-css redpanda-data/redpanda vega/vega mrjbq7/ta-lib uber/ludwig keplergl/kepler.gl devicons/devicon crossplane/crossplane openaccess-ai-collective/axolotl go-shiori/shiori audiokit/audiokit pyroscope-io/pyroscope px4/px4-autopilot quickwit-oss/quickwit vuecomponent/ant-design-vue-pro divanteltd/vue-storefront k2-fsa/sherpa-onnx jantimon/html-webpack-plugin mockery/mockery automattic/node-canvas divio/django-cms containers/skopeo kubernetes/kompose lucia-auth/lucia microsoft/fluentui-system-icons triton-inference-server/server pressly/goose altair-viz/altair pwndbg/pwndbg maplibre/maplibre-gl-js webtorrent/webtorrent-desktop hackmdio/codimd
31
8
5
327
22,161
Replying to @AgnerFog_
@AgnerFog_ updated the microarchitecture.pdf with the #AMD #Zen5 "1or2 clk" latency issue: agner.org/optimize/
4
17
4,637
これを超える人は知らない。
論文がペンネームで良い事案、歴史的には統計学のスチューデントがあり、最近の物性の数理物理的にはHal Tasakiがとりあえず例がある。他にもペンネーム利用の例はどのくらいあるのだろうか。
19
1,683
126
14,441
1,174,549
困晕
5
2
13
2,612