@taller2006 Como aplica un ciudadano para esas ayudas de alquiler, transporte y comida?
O es solo para los hijos de políticos?
Español
N G Dreams
1.6K posts
El hijo de Noroña no solamente recibe anualmente más de $1,400,000 pesos tabién le pagamos la renta, el transporte y hasta la despensa.
Manuel Lopez San Martin@MLopezSanMartin
Noroña defiende sueldo de su hijo Kin Yael Villafaña Morán en la CFE Es una “campaña canalla”, dice
Español
@soldni At sonnet price this would be 5k USD per day minimum, giving a $1.3 million dollars per year for this agent.
That is pretty low, an efficient engineer should put around $10 millions in tokens per year.
English
@akhusharmuta Se ve con madres la oposición @AccionNacional
Por desgracia, @PartidoMorenaMx va a durar mucho si siguen con esas pendejadas
Español
Wey está gente que pedo?
Lo deberían de meter a la cárcel por traición a la patria alv
Emilio Vallejo Rangel-Larios@EmilioVallejoRL
Esto es una teoría de conspiración, pero a México realmente le convendría concesionar la península de Baja California al movimiento sionista, a cambio de un 15% del PIB nominal durante 99 años. Nuestros hermanos judíos solamente traen riqueza, prosperidad y desarrollo allá donde se instalan. Máxime al no estar rodeados de fundamentalistas envidiosos de su éxito. Am Yisrael Chai 🇮🇱🤝🏻🇲🇽
Español
@brockpierson Right now, many non Americans are paying an extra increase in gas prices due to American politics, therefore, people whose life is being affected by American politics feel the need to speak about it.
English
Why is it that on 𝕏 so many non Americans primarily speak about American politics?
English
@Chimp80984636 @thetrueshelby It was intentional. Here is the original AI Image
x.com/i/status/20191…
English
@thetrueshelby That's a meaningless AI slop artifact from an old man who doesn't know how to prompt engineer image-gens. Reading into the glitches is pointless, the figure also has more than 2 arms. That's clearly not intentional. What is intentional is the subject of the image...
English
@gabriberton Yeah, I just realized you meant that, sorry for the confusion. 😅
English
@NGDreaming It's table 2 here.
I mean 4 digits like 12.34, not like 0.1234, but that's pretty much the same
arxiv.org/abs/2506.05176
English
Looks like Qwen3 embedding authors don't follow me, the tables in the paper all use 4 digits
@Alibaba_Qwen in case you guys ever write another paper :)
Gabriele Berton@gabriberton
Qwen3 embeddings (Jun 2025) This paper fine tunes Qwen3 text-only models for retrieval and reranking The retrieval model takes the text, appends an EOS token (End Of Sentence) whose output is used as embedding. Dimensions very from 1024 for 0.6B LLM to 4096 for 8B. [1/N]
English
@gabriberton Oo were you refering as 4 digits including non-decimals? I always liked the 2 decimals as a standard, only avoidable when dealing with large numbers (>1000). It looks more structured, even in their table the 20.1 looks out of place.
Only one decimal is weird, maybe just natural
English
@NGDreaming They're using 4 digits, the 4th digit is absolutely meaningless
x.com/gabriberton/st…
Gabriele Berton@gabriberton
Please people stop cluttering your tables with 4 or more digits on your results Outperforming a competitor by 0.01% is meaningless
English
@gabriberton What is the issue? They are using only 2 decimals which is standard. Or is it in another table they use the 4 decimals? 🤔
English
@elonmusk The red and green path in the UI gives me anxiety, like if it was braking and accelerating 😂
I liked more the blue.
English
@AdrianoAvrntura @MameShitPst No jugó.. solo en el barcelona B y C.
Tiene 0 apariciones en el Barcelona
Español
Ese cabrón se cree leyenda jajajajajaja nadie se acuerda que jugó ahí
Español
N G Dreams retweetet
N G Dreams retweetet
Introducing Muse Spark, the first in the Muse family of models developed by Meta Superintelligence Labs.
Muse Spark is a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration.
Muse Spark is available today at meta.ai and the Meta AI app. We’re also making it available in private preview via API to select partners, and we hope to open-source future versions of the model.
Learn more: go.meta.me/43ea00
English
Iran Consulate - Hyderabad@IraninHyderabad
Iran's 10-point conditions that the US has accepted as "workable": The US is fundamentally committed to: 🔹 Non-aggression 🔹 Continuation of Iran's control over the Strait of Hormuz 🔹 Acceptance of enrichment 🔹 Lifting all primary sanctions 🔹 Lifting all secondary sanctions 🔹 Termination of all UN Security Council resolutions 🔹 Termination of all IAEA Board of Governors resolutions 🔹 Payment of compensation to Iran 🔹 Withdrawal of US combat forces from the region 🔹 Cessation of war on all fronts, including against the heroic Islamic Resistance of Lebanon. #WarOnIran #TrumpsWar @PressTV
English
@Yuchenj_UW I thought AGI meant Artificial GENERAL Intelligent, or is it Artificial CODING Intelligence?
It is the best coding model but let's stop calling AGI to everything.
English
Anthropic is truly unstoppable.
Mythos is crushing Claude Opus 4.6 across every serious agentic coding benchmark.
It has found vulnerabilities in the Linux kernel, a 27-year-old vulnerability in OpenBSD, and a 16-year-old vulnerability in FFmpeg.
No wonder folks at big labs keep telling me AGI is already here.
English
@Jose26745772 @_BonFCB De hecho, esos mismos rivales le hubieran tocado a Portugal (lugar de EEUU) ya que estaban en la misma llave.
Lo único que tenían que haber hecho era ganarle a EEUU en fase de grupos o no haber sido goleado por Alemania (cosa que Ghana y EEUU si lograron).
Español
@willdepue The small ones (E4B and E2B) are really bad on GPQA compared with Qwen3.5-4B and 9B (76% and 81% respectively)
English
x.com/demishassabis/…
looks like a sick model though! 84% gpqa is wild
Demis Hassabis@demishassabis
Gemma 4 has incredible capabilities for such small models. Follow @googlegemma for the latest information.
English
can we not post lm arena elo score. it’s literally worth 200 points if you just crank Markdown and response length. remember Llama 4 lmsys? this is not a good eval
Demis Hassabis@demishassabis
Gemma 4 outperforms models over 10x their size! (note the x-axis is log scale!)
English
@PeteHegseth @POTUS Two weeks ago Trump said they only had 48 hours, then it was 5 days more and then it was another 5 days?
What did I miss? Why Iran is not listening and why the US is doing nothing about it?
English
@scaling01 The smallest variants are nowhere close to Qwen 3.5 models in benchmarks, but they include audio and let's hope they shine in real world usage.
English
