████████████████████████████████████████████
█████████
Joined October 2019
- Tweets14.2K
- Following1.9K
- Followers162
- Likes72.1K
T retweeted
🚨 ALERT: Bloomberg Intelligence strategist Mike McGlone warns Bitcoin could fall toward $10,000, particularly if the US stock market suffers a 20% drawdown.
T retweeted
• اگر نام یهوه را از آخر به اول و طور دیگری بچینیم،می شود حوا. در مطالب قبل گفتیم که ذات اولیه خالق این دنیا مؤنث است. یهوه همان حوا بود که سیب (𝚽) را به آدم تعارف کرد. در گنوسیسم نیز سوفیا و یلداباوث (مادر/فرزند) یکی هستند، زیرا یلداباوث سایه سوفیا بود. یعنی یک خدای دوجنسه.
T retweeted
Liveball
Gingabet
CricfyTV
Sportzfy
Playfly
Starball Tv
Epic sports
Vipbox
Footybite
FCTV33
Youcine
Camel1live
Fawanews
RBTV
YasenTV
Koralive
808sports
Sportzspeed 24
T retweeted
The pyramid diagram explores π, geometric proportions, circles and angular relationships.
In trading, these concepts can inspire research into time cycles and potential turning points—but only historical backtesting can establish whether they offer an edge.
Geometry creates the hypothesis. Price data delivers the verdict.
T retweeted
A jailbroken PS5 can now run:
- PS2 : github.com/Swordpdf/PS5SX2
- Xbox : github.com/ZiZc3/XPSemu/rele…
- Xbox 360 : github.com/BrinooTk/PS5X360
- Wii U + 3DS : github.com/premohq/PS5CEMU-H…
- SNES : github.com/MisterTemaki/snes…
- Retroarch: github.com/mihawk-99/PS5_Ret…
- Switch: github.com/blackbearreloaded…
- GameCube + Wii : github.com/elripalda/Porpois…
Native ports for PS2, Switch, Xbox 360, Wii U, 3DS, GameCube, and more are already available.
PS5 Games 'Shadow of the Colossus' is now fully playable on any Windows PC.
No emulator, original game code runs directly on Windows & Vulkan translation.
Unlocked resolution, 60 FPS option, and active developer updates, It looks incredible.
- github.com/AjncSystem/Shadow…
T retweeted
GitHub 上全是免费的东西，但几乎没人在用。
这个月最火的 10 个仓库，从最简单的到最硬核的：
谁都能用的
1. VoiceStudio：克隆声音、给视频配音、转录 646 种语言，免费，全程不用离开你的电脑（57K ⭐）
github.com/debpalash/VoiceSt…
2. ppt-master：丢给它一个 PDF 或 Word，它给你返回一个真正的 PowerPoint，可编辑，带动画（59K ⭐）
github.com/hugohe3/ppt-maste…
3. @bilawalsidhu 的 gods-eye-view：浏览器里的间谍卫星模拟器，3D 地球上叠加真实的飞机、船只和摄像头（50K ⭐）
github.com/bilawalsidhu/gods…
4. REA：你让 AI 看另一个 app 的某个功能，它就能给你讲清楚这功能是怎么做的，不需要拿到代码（65K ⭐）
github.com/morluto/rea
5. @cathrynlavery 的 diagram-design：用文字描述一个图表，它按你的品牌风格画出来，44 种类型（49K ⭐）
github.com/cathrynlavery/dia…
6. paperclip：相当于拥有一家自主智能体公司为你干活，这个开源 app 让你像管理员工一样指挥它们（99K ⭐）
github.com/paperclipai/paper…
让 Claude Code 干得更好
7. @mattpocockuk 的 skills：一个真正的工程师每天在用的那些，小巧，容易改（284K ⭐）
github.com/mattpocock/skills
8. ECC：skills、记忆和安全打包在一起，支持 Claude Code、Codex、Cursor 和 OpenCode（276K ⭐）
github.com/affaan-m/ECC
9. andrej-karpathy-skills：一个文件，4 条规则，来自 Karpathy 说的模型写代码时犯的错（218K ⭐）
github.com/multica-ai/andrej…
10. claude-mem：存下你的智能体做过的事，总结好，传给下一次会话，让它不用从零开始（99K ⭐）
github.com/thedotmack/claude…
觉得有意思就分享 ♻️
丢了之前先存下来 🔖
T retweeted
🧲 Community torrents · 15 Oct 2026
716 model torrents added:
• orcarouter/Qwen3.8-Flash-Next-Uncensored-GGUF
• ggml-org/Qwen3-1.7B-GGUF
• K-intelligence/Midm-2.0-Base-Instruct
• IFM/K2-Horizon-7B
• EldanRing/Winnow-12B
• ggml-org/Clef-Flash-GGUF
• lukey03/Qwen3.5-9B-abliterated-GGUF
• christopherthompson81/VibeVoice-ASR-Streaming-1.5B-GGUF
• Qwen/Qwen-Image-2512
• QuantaPlanta/MiMo-V2.6-Flash-MOPD-MXFP4-GGUF
• amd/GLM-5.2-MXFP4
• allenai/olmOCR-7B-0825-FP8
• bartowski/THUDM_GLM-4-32B-0414-GGUF
• peculiar-ragdoll/Cyber-Tiel-Coder-35B-A3B-GGUF
• AtomicChat/Ornith-1.5-9B-GGUF
• bartowski/Qwen2.5-Coder-7B-Instruct-abliterated-GGUF
• Hob-forge/Kolibri-1-GGUF
• Lucebox/DeepSeek-V4.1-Flash-ROCMFP23-GGUF
• mradermacher/Huihui-Qwen3.5-4B-abliterated-GGUF
• mradermacher/Huihui-Qwen3-Coder-30B-A3B-Instruct-abliterated-i1-GGUF
• realrebelai/SCAIL-2_GGUF
• laion/CLIP-ViT-H-14-frozen-xlm-roberta-large-laion5B-s13B-b90k
• onnx-community/Florence-2-base-ft
• jartine/phi-2-llamafile
• Hikari07jp/MiMo-V2.6-Distill-Qwen-9B-Ablitrated
• alexrzem/flux-loras
• zai-org/GLM-4.5V
• autotrust/GLM-5.3-Flash-GGUF-DGX-Spark
• bartowski/Cloudflare_clef-flash-GGUF
• IFM/K2-Horizon-MoVA-36B-A4B-GGUF
• dealignai/Ornith-1.5-35B-A3B-UNCENSORED-GGUF
• unsloth/GLM-4.7-Flash-REAP-23B-A3B-GGUF
• bartowski/Cloudflare_clef-GGUF
• pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF
• SC117/Ornith-1.0-9B-heretic-MTP-GGUF
• zerofata/MS3.2-PaintedFantasy-v4.1-24B-GGUF
• moonshotai/Moonlight-16B-A3B-Instruct
• bartowski/zai-org_GLM-4.7-Flash-GGUF
• toeniant/Qwen-SEA-LION-v4.5-27B-IT-FP8-Dynamic
• lmstudio-community/GLM-4.7-Flash-GGUF
• lmstudio-community/GLM-4.6V-Flash-GGUF
• nvidia/DeepSeek-V4-Flash-0731-NVFP4
• DavidAU/GLM-4.7-Flash-Uncensored-Heretic-NEO-CODE-Imatrix-MAX-GGUF
• AtomicChat/Ornith-1.5-35B-A3B-GGUF
• zai-org/GLM-5.1-FP8
• westlake-repl/Evolla-10B-hf
• brandonmusic/GLM-5.2-EXL3-TR3v4-3.5bpw-MTP78
• nvidia/DeepSeek-V4.1-Flash-NVFP4
• suno/bark
• microsoft/Phi-4-reasoning
• + 666 more
Browse new magnets ↓
T retweeted
CHINA ACABA DE LIBERAR EL CÓDIGO DE UN OCR QUE SE TRAGA PDF's DE 100 PÁGINAS DE GOLPE
se llama Unlimited OCR y está en GitHub, gratis.
solo 3b de parámetros. corre en local.
las demás herramientas de OCR parten tu documento en páginas y se pierden por el camino. esta lee todo de corrido, sin cortes.
→ contexto de 32k de un tiro
→ multilingüe de fábrica
→ 93% en el benchmark de análisis
→ error bajo 0,11 tras 40 páginas
→ corre en tu propio hardware
→ transformers, vllm, sglang, docker
el OBR en la nube (textract, google vision, azure) cobra de $1,50 a $15 por cada 1.000 páginas.
este corre en tu máquina. gratis. para siempre.
Baidu lo hizo para llevar el ocr de deepseek un paso más allá. 26,2k estrellas en GitHub y aún se habla poco de él.
100% open source. os dejo el repo abajo.
T retweeted
You no longer need an expensive GPU to run powerful AI models locally
Colibri is a free GitHub project with nearly 42K stars that runs large AI models on laptops with as little as 8GB of RAM, using smart memory management and supporting Claude Code.
Your laptop could become your own private AI server, without a monthly subscription.
Repo: github.com/JustVugg/colibri
T retweeted
THE SQUARE OF THE MOON
The Moon counts in nines, and every line comes to 369.
In 1533 Heinrich Cornelius Agrippa printed the complete Three Books of Occult Philosophy at Cologne. Book II, chapter 22, gives seven magic squares, one for each planet: Saturn 3×3, Jupiter 4×4, Mars 5×5, the Sun 6×6, Venus 7×7, Mercury 8×8, the Moon 9×9. He had first sent a draft of the work to the abbot Johannes Trithemius in 1510.
The Moon's square holds every number from 1 to 81 once. Every row, column and diagonal adds to 369. The whole square totals 3,321.
Agrippa gives each of these numbers a Hebrew name. For the 81 cells, the divine name אלים, Elim: 1 + 30 + 10 + 40 = 81. For each line of 369, the spirit Hasmodai, חשמודאי: 8 + 300 + 40 + 6 + 4 + 1 + 10 = 369.
The squares weren't his invention. The 3×3 Lo Shu comes from Chinese tradition. The Brethren of Purity in 10th-century Basra wrote about them in Arabic. In 1514 Albrecht Dürer engraved a 4×4 square, Jupiter's size, into Melencolia I, with the year 1514 in its bottom row. In 1801 Francis Barrett copied Agrippa's tables into The Magus.
Tonight's Moon is a waxing crescent in Sagittarius, three days from its first quarter. In alchemy the Moon is silver, the whitening, the light that grows by reflection.
Long before anyone made "3, 6, 9" a meme, the Moon's square was counting in it. What else is hiding in plain arithmetic?
As above, so below. ⛈️
T retweeted
RUN UNCENSORED AI MODELS ON LOCALLY
how to set it up
1. install Ollama:
ollama.com/download
2. choose a model for your hardware
low VRAM
ollama.com/s1gnature/gemma4-…
8–12GB VRAM
huggingface.co/igorls/gemma-…
16GB+ VRAM
ollama.com/search?q=gemma-4-…
3. copy the run command from the model page and launch it in your terminal
visual interface: lmstudio.ai/download
T retweeted
i still can't believe this fits
uncensored⚠️ 27 billion parameter model on a 12gb rtx 3060, the whole model in a 6gb file, answering at 50 tok/s with the mtp head i grafted on. five years ago a 27b on a gaming card wasn't even a conversation, that class of model lived on servers with datacenter gpus, now it runs next to your steam library
ternary weights are how, every weight is -1, 0 or +1, about 1.6 bits each, and it still does real work, on a 16gb card it ran four agents building four games at the same time
8gb cards aren't left out either, the lean version of that head keeps a 64k window on an 8gb card at 46 tok/s
and the models keep getting smaller and smarter, something new drops almost every week, and every drop raises the floor of what a normal gaming pc can run
what card is sitting in your rig right now?
i'm running a 27b uncensored🚨model at 50 tok/s on a 12gb rtx 3060 and it does not say no to anything ⚠️
model just finished full harmbench red-team suite against it, 200 behaviors, and it refused zero. not a few. zero. aligned models sit in the 90s.
the model is ternary bonsai 2 with the heretic uncensoring pass, the weights are ternary, just -1, 0 and +1, so the whole 27b is a file small enough that a gaming card from 2020 holds it with room to spare, and it's a reasoning model. it thinks before it answers, and it still keeps the pace
the speed is the head i grafted on, a draft model that predicts the next token before the big model checks it. speculative decoding. the card fires a guess, the 27b confirms, and you get 50 tok/s instead of 39
the whole thing, weights, head and a 128k context window, sits in 10.7 of the 12 gb. kv cache at q4. nothing exotic, one card, one power plug
why uncensored. because it's mine. no api keys, no terms of service, no model deciding what an adult is allowed to ask. the refusals left in it are ones it chose, and there are almost none
that's the whole point of running local. the policy lives in your house, not someone else's
T retweeted
Best movies to brainwash yourself with:
Limitless
Catch Me If You Can
Ocean’s Twelve
Casino Royale
The Prestige
Michael
Inception
Entourage
Scarface
The Founder
Watch what serves you.
Your mind is always absorbing identities, beliefs, and patterns.
Choose them consciously.
T retweeted
AI is now being used to help decompile games you own.
Decompwtf is an open-source project that lets AI coding agents work through a game’s compiled code and reconstruct it as C/C++.
>The reconstructed code is compiled and compared against the original game binary producing the exact same machine code.
>It supports projects targeting systems including N64, PS1, PS2, GameCube, Wii, GBA and DS.
>The project can also continue existing community decompilation projects, potentially making large-scale decompilation work much faster.
>It runs locally, with no account or game data uploads required. github.com/i12bp8/decomp.wtf
