CarlesM ⠠⠵

30.9K posts

CarlesM ⠠⠵ banner
CarlesM ⠠⠵

CarlesM ⠠⠵

@carlesm

A https://t.co/HoMRY9GYhF ||*|| Recerca i docència a @greiaudl a la @udl_info: IA, energIA, eficiencIA, i xarxes......

Lleida, Catalunya Katılım Mart 2007
1.1K Takip Edilen1K Takipçiler
Sabitlenmiş Tweet
CarlesM ⠠⠵
CarlesM ⠠⠵@carlesm·
From now on I'll comment all my git commits buzzfeed-clickbait style: - "I forgot a function return , and you wont believe what happened next!" - "3 top imports to add for code to work!" - "7 exceptions you should catch but didn't!" - "The amazing effect of a +1 on a loop!"
English
0
14
23
0
Abhishek B R
Abhishek B R@abhitwt·
Programming language you learned once and never touched again?
Abhishek B R tweet media
English
904
54
1.2K
94.7K
CarlesM ⠠⠵ retweetledi
GREiA
GREiA@greiaudl·
Congrats to Pablo Tagle for the successful PhD defense today! Inspiring work on thermal energy storage in CSP systems. A big milestone and well deserved achievement! @UdL_info @EPS_UdL @UdL_RDI
GREiA tweet mediaGREiA tweet media
English
0
3
7
223
CarlesM ⠠⠵
CarlesM ⠠⠵@carlesm·
@gethuxe your product works great, but there's a small issue with it. If one has subscribed to a calendar in Google (my wife's) the daily update mixes both calendars, and the upd't is really weird (having a meeting and travelling simultaneously, etc.).
English
0
0
0
22
Thiyagarajan Maruthavanan (Rajan)
@LiorOnAI layer-by-layer loading tanks throughput. You are swapping between RAM & disk constantly. This works for tinkering. Production needs speed, not that just "it runs." Still cool though. More people get to feel what 70B actually behaves like.
English
5
3
216
26.3K
Lior Alexander
Lior Alexander@LiorOnAI·
You can now run 70B LLMs on a 4GB GPU. AirLLM just made massive models usable on low-memory hardware. 𝗪𝗵𝗮𝘁 𝗷𝘂𝘀𝘁 𝗵𝗮𝗽𝗽𝗲𝗻𝗲𝗱 AirLLM released memory-optimized inference for large language models. It runs 70B models on 4GB VRAM. It can even run 405B Llama 3.1 on 8GB VRAM. 𝗛𝗼𝘄 𝗶𝘁 𝘄𝗼𝗿𝗸𝘀 AirLLM loads models one layer at a time. Instead of loading everything: → Load a layer → Run computation → Free memory → Load the next layer This keeps GPU memory usage extremely low. 𝗞𝗲𝘆 𝗱𝗲𝘁𝗮𝗶𝗹𝘀 • No quantization required by default • Optional 4-bit or 8-bit weight compression • Same API as Hugging Face Transformers • Supports CPU and GPU inference • Works on Linux and macOS Apple Silicon 𝗪𝗵𝗮𝘁 𝘆𝗼𝘂 𝗰𝗮𝗻 𝗱𝗼 • Run Llama, Qwen, Mistral, Mixtral locally • Test large models without cloud GPUs • Prototype agents on cheap hardware
Lior Alexander tweet media
English
367
1.2K
11.2K
635.8K
St4rP0wd3rM4n
St4rP0wd3rM4n@St4rP0wd3rM4n·
@franperez_co Ok, puede ser que no aumente, me lo compro. Pero cómo carajo va a BAJAR el consumo? Simplesmente imposible! Este tipo de declaración es lo que nos hace sospechar...
Español
3
0
2
3.4K
Fran Perez
Fran Perez@franperez_co·
Los acontecimientos: 1. Elon empieza a meter Starlink en todas las aerolíneas. 2. Ryanair dice que no, porque “la antena aumenta el consumo de combustible”. 3. Otras aerolíneas muestran números: no aumenta, baja. 4. El CEO de Ryanair retruca: “ok, sube 2% y nadie va a pagar USD 1 extra por WiFi”. 5. Elon dice: “quizás estás desinformado” y publica los cálculos de un Boeing 737-800. 6. El CEO de Ryanair responde que Elon es un pelotudo y “no entiende la resistencia del aire” 🍿 7. Twitter explota: el CEO de la peor aerolínea del mundo sobrando al fundador de la mejor empresa aeroespacial de la historia 🍿🍿 8. Elon contesta que el pelotudo es él y pregunta si debería comprar Ryanair solo para echarlo y poner a alguien que se llame Ryan 🍿🍿🍿 9. Ryanair saca un comunicado oficial insultando a Elon 🍿🍿🍿🍿 Aviación low cost, pero drama premium.
Fran Perez tweet mediaFran Perez tweet mediaFran Perez tweet mediaFran Perez tweet media
Español
382
2.4K
28.8K
1.8M
CarlesM ⠠⠵ retweetledi
World of Engineering
World of Engineering@engineers_feed·
Who used Winamp?!
World of Engineering tweet media
English
2K
2K
30.2K
867.5K
CarlesM ⠠⠵ retweetledi
GREiA
GREiA@greiaudl·
🎉 We are pleased to welcome José Gonçalves from Centro de Biotecnologia de Plantas da Beira interior 🇵🇹 as part of the #CSTO2NE project!🤝 Inspiring discussions, great collaboration, and exciting steps forward for the project.🌱🔬✨ @UdL_RDI @UdL_info
GREiA tweet mediaGREiA tweet media
English
1
2
7
133
CarlesM ⠠⠵
CarlesM ⠠⠵@carlesm·
Our latest article published in Nature Reviews Clean Technology examines how thermal energy storage (TES) can optimize energy use in buildings. We assess its economic, environmental, and social impacts, highlighting the need for advanced systems. Paper: nature.com/articles/s4435…
English
0
0
1
19
Elon Musk
Elon Musk@elonmusk·
Similar story: 7th grade math teacher had left my school unexpectedly and so the principal, who had been in WW2, took over math class. He told amazing and brutal stories of his time in the war, while also doubling the normal curriculum speed to do 2 years of math in 1. But if you didn’t do your extra math homework, you didn’t get to hear the stories. Everyone did their homework.
FischerKing@FischerKing64

I had a male teacher for grades 5 and 6 who was highly eccentric (and not gay - he was a family man without a whiff of scandal). He had old printing presses in the classroom and we learned about how books and newspapers were printed - by printing our own pages with movable type we set ourselves. He showed us violent war movies. He had WW2 and Vietnam veterans come in and give graphic talks. One guy talked about his baptism of fire in Japan - he said something like ‘I jumped into the foxhole and killed two Japs with my pistol - I’m glad I bought that pistol off a guy before the fight because I might not be here if I hadn’t.’ The teacher would tells us the casualty figures of WW1 and 2 and sometimes start crying. He also taught us chess, and he spotted kids who were good in math. He gave me math problems well in advance of 6th grade, and ensured I was pushed up a level in middle school. He was tolerant of wild boy behavior - including boys throwing things at each other across the classroom. Only a few years later I heard he had to stop showing the war movies, the veterans weren’t allowed to give talks. He retired shortly thereafter - he was getting old anyway. But this kind of man used to teach elementary school. I don’t know why he wanted to do it - but he was good at it. And I don’t think there are many like him in public schools now - and that isn’t good for boys.

English
2.5K
8.3K
101.2K
16.8M
Het Mehta
Het Mehta@hetmehtaa·
Teams and Outlook
English
3
0
9
4.8K
Het Mehta
Het Mehta@hetmehtaa·
"An app with 0 haters." I'll go first:
Het Mehta tweet media
English
498
306
8.6K
516.5K
CarlesM ⠠⠵ retweetledi
Alan Daitch
Alan Daitch@AlanDaitch·
El último trimestre desplegué Microsoft Copilot para 4.000 colaboradores. 30 dólares por licencia por mes. 1,4 millones de dólares anuales. Lo bauticé “transformación digital”. Al directorio le encantó el concepto. Lo aprobaron en once minutos. Nadie preguntó qué iba a hacer en la práctica. Incluido yo. Le comuniqué a toda la organización que íbamos a “potenciar la productividad por 10”. No es un número real. Pero suena perfectamente defendible. Desde RR. HH. preguntaron cómo íbamos a medir ese 10x. Respondí que íbamos a “articular un marco de métricas apoyado en dashboards de performance”. No hubo más preguntas. Tres meses después revisé los reportes de adopción. 47 personas lo habían abierto. 12 lo habían usado más de una vez. Una de ellas era yo. Lo usé para resumir un mail que podía haber leído en 30 segundos. Tardó 45. Más el tiempo de corregir las alucinaciones. Pero lo declaré un “piloto exitoso”. Éxito significa que el piloto no fracasó de manera evidente. El CFO pidió claridad sobre el retorno de la inversión. Le mostré un gráfico. El gráfico iba hacia arriba y a la derecha. Medía “nivel de madurez en capacidades de IA”. La métrica no existe. La definí yo. Asintió satisfecho. Hoy somos una organización “AI-ready”. No sé exactamente qué implica. Pero figura en la presentación para inversores. Un desarrollador senior preguntó por qué no evaluamos Claude o ChatGPT. Le expliqué que necesitábamos “estándares de seguridad enterprise”. Preguntó qué estándares. Respondí “cumplimiento normativo”. Preguntó de cuál normativa. Dije “transversal”. Me miró con escepticismo. Le agendé una “conversación de alineamiento de carrera”. No volvió a preguntar. Microsoft envió un equipo para desarrollar un caso de éxito. Querían posicionarnos como referencia. Les dije que “liberamos 40.000 horas de capacidad productiva”. Ese número salió de multiplicar la dotación por un supuesto razonable. No lo auditaron. Nunca lo hacen. Ahora aparecemos en la web de Microsoft. “Empresa global acelera su productividad en 40.000 horas gracias a Copilot”. El CEO lo compartió en LinkedIn. Sumó 3.000 likes. Jamás usó Copilot. Ningún miembro del C-level lo hizo. Tenemos una excepción formal. “El foco estratégico requiere minimizar fricciones digitales”. Esa política la redacté yo. Las licencias se renuevan el mes próximo. Estoy solicitando una ampliación. 5.000 licencias adicionales. Todavía no usamos las primeras 4.000. Pero esta vez vamos a “impulsar la adopción organizacional”. Adopción significa capacitación obligatoria. Capacitación significa un webinar de 45 minutos que nadie va a mirar. Pero la finalización va a quedar registrada. La finalización es un KPI. Los KPIs viven en dashboards. Los dashboards viven en presentaciones al directorio. Las presentaciones al directorio aceleran carreras. Voy a ser SVP en Q3. Todavía no sé bien qué hace Copilot. Pero tengo claro para qué sirve. Sirve para demostrar que estamos “invirtiendo estratégicamente en IA”. Invertir es gastar. Gastar es compromiso. Compromiso es visión de futuro. El futuro es lo que yo diga que es. Mientras el gráfico siga subiendo hacia la derecha. Créditos: texto original de @gothburz
Español
118
861
3.6K
368.3K
CarlesM ⠠⠵ retweetledi
Anirban chowdhury
Anirban chowdhury@VoyageBliss·
One of my travel cheat codes: a private VPN at home All you need is a Raspberry Pi or even a spare computer, an SD card, and 15 minutes. Install Tailscale, enable exit node + subnet routing, leave it always on at home. Now log into Tailscale on your phone. Use your phone as a hotspot and every connected device routes via your home network. Result: wherever you are, your devices appear to be at your home location. Perfect for IoT access, blocked apps abroad, and yes even WFH turned WFX 😄 I use this mostly to connect and access all my IOT devices at home and NAS. Simple, secure, and super effective.
Anirban chowdhury tweet media
English
221
1K
11.6K
832.3K
CarlesM ⠠⠵ retweetledi
Computer ♥ Records
Computer ♥ Records@ComputerLove_·
If you know what these are, we can be friends.
Computer ♥ Records tweet media
English
2.5K
1.6K
18.1K
632.3K
MataGigantes
MataGigantes@GigantesMata·
@TurtellGran @calotonterias Todos los comercios que aparecen en éste y otros videos de la época en Barcelona tienen los comercios rotulados en castellano, y no era porque el catalán estuviera prohibido a esos niveles ni muchísimo menos, sino porque no tenían comido el coco como muchos ahora y preferían así.
Español
2
0
6
153
Sergio Ferrero
Sergio Ferrero@calotonterias·
Barcelona en 1908. ¿Qué cosa curiosa notáis?
Español
638
124
765
506.9K
Ayushi☄️
Ayushi☄️@iyoushetwt·
WITHOUT GOOGLING Name a programming language older than Java.
Ayushi☄️ tweet media
English
4.1K
86
2K
294.6K
CarlesM ⠠⠵ retweetledi
Peter Girnus 🦅
Peter Girnus 🦅@gothburz·
I work at Slack. We tell employees their DMs are private. And they are. Mostly. Look, when we say "private" we mean private between you and the person you're messaging. And your admin. And HR. And legal. And whatever compliance tool your company bought. And the export logs. And the backup systems. And anyone with a court order. But other than that, totally private. We're very clear about this in our documentation. Page 47. Section 12. Subsection C. Paragraph 8. The part nobody reads before they trash-talk their manager at 11pm. Here's what employees don't understand. When you delete a message, you're just deleting it from your view. The message still exists. In exports. In backups. In the retention policy. It's like closing your eyes and thinking you're invisible. The data belongs to the company, not you. We say this right in our terms. Workspace owners control everything. They decide how long messages are stored. Sometimes it's 30 days. Sometimes it's forever. Hope you didn't say anything spicy in 2019. Enterprise customers get extra features. Full message exports. Metadata tracking. Who messaged whom. When. How often. Communication patterns. It's for "compliance." It's for "legal needs." It's for "regulatory requirements." It's definitely not for micromanagement. We're very careful to explain that admins can't see messages in real-time. They have to formally request an export. Fill out some forms. Click some buttons. Maybe wait an hour. Very high barrier. Almost impossible to abuse. The key takeaway is simple. Treat Slack like work email. Not like WhatsApp. Not like Signal. Just because it looks like a chat app doesn't mean it works like one. If a message could cause trouble when HR reads it, don't send it. This is empowering employees with knowledge. If you wouldn't say it in the break room with your manager behind you, don't type it in Slack. That's privacy. Informed privacy. Enterprise-grade informed privacy.
Peter Girnus 🦅 tweet media
English
117
512
3.9K
615.8K