fable.eth

4.6K posts

fable.eth

@fable_eth

Exploring and investing in Ethereum / OP / Starkware & ZK, AI-agents, art collector but also strong in #Bitcoin and believer in privacy + funding OS hw? dm me!

Planet X Katılım Mayıs 2014

5.7K Takip Edilen1.2K Takipçiler

Sabitlenmiş Tweet

fable.eth@fable_eth·14 Şub

If you have MacStudio 512GB ram maxed out and love to play with OpenClaw @openclaw. Good news! Download @lmstudio download @MiniMax_AI model 2.5 8bit gs32 context to 40k parallel request to 3 (MLX) Enjoy fast&free stream of knowledge to your ship.

English

213

19.4K

fable.eth retweetledi

GrapheneOS@GrapheneOS·1d

GrapheneOS will remain usable by anyone around the world without requiring personal information, identification or an account. GrapheneOS and our services will remain available internationally. If GrapheneOS devices can't be sold in a region due to their regulations, so be it.

English

230

1.6K

13.2K

339.1K

fable.eth@fable_eth·2d

“What we do now echoes in eternity.” Marcus Aurelius

English

fable.eth@fable_eth·6d

@zackbshapiro It will be crazy time. You can also revive old hardware with AI, reverse engineering drivers and other things. There so much possible now. Special knowledge is now in the hands of everyone.

English

143

Zack Shapiro@zackbshapiro·6d

We have no idea how to regulate what’s coming:

chiefofautism@chiefofautism

someone built a $96 3D-PRINTED MANPADS rocket that recalculates its mid-air trajectory using a $5 sensor and piano wire its called Project Canard it integrates with distributed camera nodes to triangulate airborne targets and update flight paths in real-time it proves the barrier to advanced hardware has completely collapsed, moving precision weapons from defense labs to consumer garages the entire launcher and interceptor frame is 3D printed in PLA and runs off a standard off-the-shelf ESP32 microcontroller it even spins up a local Wi-Fi network so you can monitor live telemetry and arm the system directly from your laptop

English

437

97.1K

fable.eth@fable_eth·6d

@peter_szilagyi @superphiz @SequencingCom @openclaw @claudeai Yes, but every little change would let to harder to process the true hard and fun stuf. Like HLA genotyping, you need CRAM->BAM->sorted BAM->extraction HLA regions and a lot of validation. For validation store also FASTQ for future.

English

Péter Szilágyi@peter_szilagyi·6d

@fable_eth @superphiz @SequencingCom @openclaw @claudeai Hmmm, yes, but I guess if I make my own CRAM, I should be able to max it out so no data is lost, but it compresses better, no?

English

146

fable.eth@fable_eth·6d

@peter_szilagyi @superphiz @SequencingCom @openclaw @claudeai you can reconstruct FASTQ from CRAM in theory, but only if nothing was filtered or discarded. bc CRAM delete data like compresses quality scores and more, so you can’t always recover the full original dataset. This is the reason why companies give you the FASTQ for full.

English

Péter Szilágyi@peter_szilagyi·6d

@fable_eth @superphiz @SequencingCom @openclaw @claudeai In theory you should be able to recosntruct the FASTQ sequencing file from CRAM too, no? Then you wouldn't need to store that much data. Maybe the outright discarded strands wouldn't get stored.

English

132

fable.eth@fable_eth·6d

@peter_szilagyi @superphiz @SequencingCom @openclaw @claudeai My 330GB is raw FASTQ is 90x coverage. Raw keeps all the reads, qualities,everything, perfect for reanalysis later or other deep stuff, like HLA genotyping. It is just different levels. Raw ain’t overkill if you actually care about the full picture, for now and future, it is gold

English

143

Péter Szilágyi@peter_szilagyi·15 Mar

@fable_eth @superphiz @SequencingCom @openclaw @claudeai Nah, your genome is 3GB uncompressed blob. If it’s a variant call and compressed it’s 300MB. The huge dump is the raw sequencing data that’s out of the machine, but that’s not needed after you align it to a reference genome. Also 90x is prolly overkill.

English

273

fable.eth@fable_eth·14 Mar

@bilawalsidhu “left click , right click” 🤮 I hope this company will be prosecuted just like after WW2. Based on old data from palantier in Iran 150 little childeren died in school.

English

Bilawal Sidhu@bilawalsidhu·13 Mar

Probably the most current look at Palantir’s maven smart system software. Here’s the DoW’s Chief AI officer showing how it works:

English

382

1.2K

9.5K

2.4M

fable.eth@fable_eth·13 Mar

@Polymarket @chadRAINS @alifarhat79 Dont bait. He knows exactly that Polymarket was the only platform on earth that predicted the presidency right.

English

325

Polymarket@Polymarket·13 Mar

@chadRAINS @alifarhat79 x.com/Polymarket/sta…

Polymarket@Polymarket

Trump leads on the world's largest prediction market. 🟥 Trump 65% 🟦 Harris 35% Track the election in real-time with the Polymarket app.

QME

1.5K

118.1K

Polymarket@Polymarket·13 Mar

BREAKING: Democrats now projected to seize control of both chambers of Congress this November.

English

2.6K

20.4K

11.4M

fable.eth@fable_eth·10 Mar

@danpacary Really impressive !

English

100

Daniel Isaac@danpacary·9 Mar

This afternoon: SEQ=1024. v3b ran at SEQ=512. MLX runs at 1024. Double context = better language modeling. 72K steps at SEQ=1024 = 74M tokens (2x data). Target: sub-1.3. First real apples-to-apples ANE vs MLX comparison. Same chip. Same data. Same context length. check it out .github.com/ncdrone/autoresearch-ANE

English

7.1K

Daniel Isaac@danpacary·9 Mar

I hijacked Apple's Neural Engine -- the chip built for Siri and photo filters. Reverse-engineered the private APIs and trained a full LLM on it. Zero fan noise. Zero GPU. Just the Neural Engine doing what nobody thought it could. Your Mac has one too.

English

115

1.4K

93.5K

fable.eth@fable_eth·9 Mar

Pretty cool Training, fine tuning llm on a mac with ANE (Apple Neural Engine) with their private api. Build my own custom AI factory that can train with ANE, and falls back to Metal when private API is closed. What a time ! github.com/maderix/ANE

English

103

fable.eth@fable_eth·8 Mar

@openclaw Great HEIF image support!

English

109

OpenClaw🦞@openclaw·8 Mar

OpenClaw 2026.3.7 🦞 ⚡ GPT-5.4 + Gemini 3.1 Flash-Lite 🤖 ACP bindings survive restarts 🐳 Slim Docker multi-stage builds 🔐 SecretRef for gateway auth 🔌 Pluggable context engines 📸 HEIF image support 💬 Zalo channel fixes We don't do small releases. github.com/openclaw/openc…

English

435

536

5.5K

1.6M

fable.eth retweetledi

Cole Kennelly@ColeGotTweets·3 Mar

Add a zero to the Volmex valuation

Watcher.Guru@WatcherGuru

JUST IN: 🇺🇸 CFTC Chair Mike Selig says the US plans to approve crypto perpetual futures within the next month.

English

fable.eth@fable_eth·7 Mar

@naval It was wired to follow events from a small group max 100 people Than larger events through symbols/stories/culture/religion But now ? 🤷‍♂️

English

Naval@naval·6 Mar

The human brain isn’t designed to process all of the world’s breaking emergencies in realtime.

English

1.5K

3.3K

31.7K

fable.eth@fable_eth·7 Mar

@oliviscusAI Using this since it was introduced way back January. It can use embedded llm. Really great!

English

605

Oliver Prompts@oliviscusAI·6 Mar

This OpenClaw skill literally cuts token usage by 95% 🤯 It uses "quantized message delivery" to compress how the agent thinks and communicates. We are talking 20x cheaper agents overnight..

English

894

70.5K

fable.eth@fable_eth·6 Mar

@austingriffith Btw using @SimpleXChat custom made communication channel

English

116

fable.eth@fable_eth·6 Mar

@austingriffith And ofc the sha256

English

107

Austin Griffith@austingriffith·5 Mar

⚙️ Are the "local ai" guys basically just larping? 🐏 I have a machine with 128GB of RAM 🤖 I downloaded qwen3.5:122b 🤣 It takes 3 minutes to incorrectly do what sonnet 4.6 can do in seconds 😅 Am I doing it wrong? Do I need more RAM?

English

454

3.2K

418.7K

fable.eth@fable_eth·6 Mar

@spencer_haber @austingriffith Very simple, everything skill bro. Crafting skills for difficult tasks.

English