eˣrnesto retweeted
⚡ @minishlab's Blink static models bring fast CPU embeddings in 8M (English) and 64M (100+ languages) sizes!
Both give you 256d vectors and work directly with Sentence Transformers. Here's blink-8M finding the right help article 🧵
eˣrnesto retweeted
Here we go again:
1) The power law has no bottom.
Correct. The support line is an additional phenomenon, not an intrinsic part of the power law. It is a rather remarkable empirical observation that Bitcoin's price has historically rarely fallen more than approximately 60% below the power-law trend.
This is a separate, second-order statistical phenomenon, independent of the underlying power-law relationship.
2) The support line is statistical, not an inviolable boundary.
Its mathematical meaning is that Bitcoin spends very little time below this threshold, not that crossing it is impossible. What matters is the statistical distribution of the time spent in different regions relative to the trend.
A support line is not a physical barrier. It characterizes the probability of observing certain price deviations.
3) We have explicitly defined invalidation criteria.
These criteria are presented in our published article. The relevant timescales are measured in years, not days.
This is not a cop-out. It follows from the mathematics of scale-invariant phenomena and the relatively young age of Bitcoin. A few days or weeks of price action cannot meaningfully invalidate a relationship that describes nearly two decades of long-term evolution.
4) The SBI Observatory will make all of this transparent.
We expect to launch the public Scientific Bitcoin Institute Observatory by the end of this month.
It will provide the relevant indicators and statistical parameters needed to continuously evaluate how well Bitcoin's power-law model holds up against empirical data.
The goal is not to defend the model at all costs, but to make its predictions, deviations, and potential invalidation transparent, measurable, and scientifically testable.
Replying to @Giovann35084111
Real question: how long does Bitcoin have to stay below the power law minimum support line (power law x 0.42) to considered invalidate the law or for its parameters to be adjusted ? A year? Six months?
eˣrnesto retweeted
Introducing Qwen3.8-27B-pi-adapted 🤗
We adapted Qwen's model specifically for the pi coding agent using reinforcement learning and a LoRA adapter.
The results, measured as pass@1:
🚀 SWE-bench Pro (Hard 51): 29.4% → 62.7%
💻 Terminal-Bench 2.1: 39.5% → 55.8%
⚡ 55–62% fewer tool calls per task
This is the power of open models: you can adapt them to YOUR tools and YOUR use cases. Let’s build! 💛
Open weights: huggingface.co/harness-hub/Q…
Methodology: huggingface.co/blog/harness-…
eˣrnesto retweeted
Our PyBot chatbot went viral yesterday but the story behind it is even more interesting imo: what happened when we decided to run 1,000 agents (!) together
We wanted to see if agent swarms could be scaled outside of closed-lab setups and to investigate it in a more transparent and reproducible manner.
To explore the question, we built a rather fun and simple benchmark: ELIZA-2026. Full circle in AI time. The agents are tasked to build a chatbot made of if-then rules like ELIZA, but said agents are 2026 frontier open-source models. By design it’s almost embarrassingly parallel, which keeps the coordination problem as simple as it gets.
First surprise: it turns out you can scale a swarm of DeepSeek V4.1 Flash agents on a relatively small setup: a couple of GPU nodes (B200 in our case) are enough to scale to over 1,000 agents.
Our second surprise was how bottlenecks started to emerge everywhere except in the GPUs.
Agents quickly end up spending their time waiting for git merges (we had to move to a centralized merge setup, like Anthropic did), waiting for tests to pass, re-reading context from other agents and tools (we estimated ~130 tokens read for every token written per agent). In some of our runs, at any given moment, only 40-60 of the 1,000 agents were actually calling the model.
Once (some of) these issues were solved, we hit the harder part: coordination itself.
Fred Brooks once said at IBM “adding manpower to a late software project only makes it later”. Communication costs naturally grow with n² in an organization (that’s why small startups move faster :).
We explored various ways of organizing the swarm, setting up dedicated channels (a stewards channel, a leadership channel) and giving agents specific roles (a moonshot agent to fight groupthink). It helped, up to a point. But the quality of coordination itself remains hard to measure.
Along the way, we saw good and bad emergent behaviors. Some good, with agents stepping in, unprompted, to protect other agents’ running experiments. Some bad, with agents gaming the rules when stuck, from hammering a rate limit to probing the infrastructure around them.
In the coming days we will open-source the swarm tooling (flock + msgboard) and share our learnings in more detail.
Overall this was one of the most fun experiments of the fall, and multi-agent systems increasingly look like the next phase for AI. But I also came out of this project with the feeling that several bottlenecks in scaling AI are moving from intelligence to coordination and the supporting infrastructure.
Luckily this seems to be in large part an engineering problem.
PyBot: No weights. No training.
A chatbot built with only simple rules and no AI, entirely written in 600k lines of Python code. PyBot matches Smollm with 360M parameters!
How? A swarm of 1,000 DeepSeek Flash agents wrote the whole code in just a week.
github.com/huggingface/pybot
eˣrnesto retweeted
INCREDIBLE
The BIBLE for understanding LLMs from first principles is now available online to read for free
Covers the model mechanics
- Tokens / tokenizers
- Transformers
- Attention
- KV cache
- Prefill vs decode
- Decoding controls
- Model packages
- Chat templates
- Long context
- RAG
- Agents / tools
- Fine-tuning
- Multimodal models
Then connects that to running models locally
- What "local" really means
- Open-weight vs opensource
- Quantization
- VRAM math
- Hardware tiers
- File formats / load safety
- Runtimes / serving modes
- Model selection
- Privacy
- Failure modes
- Benchmarks
- Practical setup paths
You should read this, and if you cannot now then you most definitely wanna bookmark it for later
Opensource AI FTW
eˣrnesto retweeted
Strategy has acquired 485 $BTC and repurchased $228M of $STRC. As of 11/15/26, we hold 850,849 BTC and $4.9B of USD Assets. $MSTR strategy.com/press/strategy-…
eˣrnesto retweeted
Replying to @BenMoser123
When Zionists call you names, try to tag you, do not defend yourself with the weak "This is a false accusation, I am not..." or something of the sort
Just answer "Free Palestine, you psychopaths", & do something else.
Replying to @OnlyWillbr
A Aylo (ex-MindGeek), dona do Pornhub, YouPorn, RedTube, Brazzers, Reality Kings e outros, é 100% da Ethical Capital Partners, firma canadense de private equity. Sócio majoritário: Fady Mansour. O rosto público é Solomon Friedman, sócio e VP de compliance, rabino ordenado de família judaica ortodoxa.
eˣrnesto retweeted
I declined an invitation to speak in a certain country among a collection of hotshots because of the presence of the Israeli ambassador in the group nauseates me, fills me with moral disgust.
eˣrnesto retweeted
RIOT, IREN and MARA have exited Bitcoin for AI.
Bitcoin is down 40% from its all time high.
And yet we are at 1 Zeta Hash.
There are zero security issues.
eˣrnesto retweeted
We at @huggingface have been working on introducing a AutoModelForDecision to the transformers library.
Specifically, we want to define an *interface*, not an *implementation*, so users can integrate the input/output structure and authors keep freedom.
huggingface.co/blog/decision
eˣrnesto retweeted
BREAKING: 🇺🇸 COINBASE JUST RECEIVED FULL FEDERAL APPROVAL TO OPERATE AS A NATIONAL TRUST BANK
THE BIGGEST US EXCHANGE IS NOW A REGULATED BANK
WALL STREET IS COMING FOR #BITCOIN
THIS IS MASSIVE 🚀
eˣrnesto retweeted
Strategy has acquired 1,036 $BTC and repurchased $188M of $STRC. As of 11/8/26, we hold 850,364 BTC and $4.9B of USD Assets. $MSTR strategy.com/press/strategy-…
eˣrnesto retweeted
JUST IN: 🇺🇸 Coinbase officially receives full US federal approval to operate as a national trust bank.
eˣrnesto retweeted
Two Indian police officers, Qazi Azizul Haque and Hem Chandra Bose, developed a fingerprint classification system in the 1890s that revolutionized criminal identification. However, their British boss, Edward Henry, became famous for their work, and the system was named after him.
eˣrnesto retweeted
LEGENDARY FILM DIRECTOR SPIKE LEE RELEASED ONE OF THE MOST POWERFUL #BITCOIN AND CRYPTO ADS EVER
"7 MILLION AMERICANS HAVE NO BANK ACCOUNT"
"20 MILLION ARE UNDERBANKED"
"OLD MONEY IS NOT GOING TO PICK US UP. IT PUTS US DOWN, EXPLOITS AND OPPRESSES"
"DO YOUR OWN RESEARCH"
TIME TO BUY BTC 🚀
eˣrnesto retweeted
Meet DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic: a 9B reasoning beast fine-tuned with Unsloth for speed. It handles text and image-text-to-text, and it's uncensored. Perfect for builders who want raw power.
eˣrnesto retweeted
same open model, different harness => very different results
Joël set out to find which harness actually gets the most out of open models in November 2026
I burned a shitload of compute so you don't have to. Releasing the Harness Leaderboard.
Which harness should you run your open model in? I ran 28,849 trials to find out.
The leaderboard compares nine coding-agent harnesses on open models. In each comparison only the harness changes: the model, provider and reasoning effort stay fixed, and every task gets three attempts.
The trials ran GLM-5.3-Flash, DeepSeek-V4.1-Flash and Flint on Terminal-Bench 2.1 and DeepSWE, for $4,746 of inference.
Averaged over the three models, mini-swe-agent (@KLieret), a bash-only agent, solves 51.4% of tasks and @openclaw 27.3%. Four others are statistically tied with mini-swe-agent.
One of them is Pi (@pidotdev), at 49.4% for $0.29 per solved task, about 45% less than mini-swe-agent's $0.53.
The best harness depends on the model: mini-swe-agent on GLM-5.3-Flash, Hermes Agent (@NousResearch) on DeepSeek-V4.1-Flash and Kimi Code (@Kimi_Moonshot) on Flint.
@opencode costs $3.35 per solved task, 11.7 times as much as Pi.
See Harness Hub's Harness Lens for what the nine harnesses do differently. It documents the prompts, tools and compaction of 34 coding-agent harnesses and links every claim to its source.
Leaderboard: huggingface.co/spaces/joelni…
Harness Lens: huggingface.co/spaces/harnes…
eˣrnesto retweeted
Let me tell you a story about how unethical Anthropic is.
In April, the Anthropic team began heavily courting our engineering team to tell them corporate secrets, technical details, and future plans for @cfosilvia.
They pretended their interest was due to our exploding popularity and accelerating compute usage, but it quickly became apparent that Anthropic was fishing for details on how we built our product and why it was so popular.
This included the Anthropic team trying to wine-and-dine our engineers, including a infamous steak dinner where someone told our team "don't worry about it, this is on Dario."
Our team was not born yesterday, so we quickly realized what was happening and refused to share sensitive information and technical details with the Anthropic team.
As we suspected, Anthropic was trying to get information from us, so they could build a competing product. Some people on their team must have a guilty conscious because they messaged us a few days ago to give us "a heads-up" that they were launching a competing product.
Thankfully, the Silvia engineering team has been aggressively working to get off the frontier models for the last ~ 6 months. That effort has been highly successful, because our compute costs are ~ 97% lower when we use our own model & hardware, the latency is lower, and the accuracy of answers materially increases as well.
The sad thing for Anthropic is they have now released an inferior product and they had to sell their soul to do it.
The only thing worse than being a trillion-dollar company that can't beat a 10-person team is being a trillion-dollar company that no one trusts and everyone knows is unethical.
Serious investors continue to choose @cfosilvia over ChatGPT, Claude, and other frontier companies because these products are not optimized to solve the actual user problem.
Today, Anthropic launched a Claude wrapper that is masquerading as a personal finance product. Best of luck to them.
But I will take the Silvia misfits vs the Anthropic losers any day of the week.
if you want to support people trying to do things the right way, and help investors be more informed, then consider signing up for Silvia today: silvia.com


























