I believe that If you create a mixture of experts solution, it is worth to hear this inspiring Interview.
https://open.spotify.com/episode/3w1yB5fCkxkT2VNnHBJQHs?si=Ammq5Qe3TPaHr4QmXDe1UA&utm_source=copy-link
#agenticai #mixtureofexperts
https://open.spotify.com/episode/3w1yB5fCkxkT2VNnHBJQHs?si=Ammq5Qe3TPaHr4QmXDe1UA&utm_source=copy-link
#agenticai #mixtureofexperts
09 | HELLO, I’M HERE
I'M SORRY, DAVE · Episode
open.spotify.com
November 5, 2026 at 7:58 AM
I believe that If you create a mixture of experts solution, it is worth to hear this inspiring Interview.
https://open.spotify.com/episode/3w1yB5fCkxkT2VNnHBJQHs?si=Ammq5Qe3TPaHr4QmXDe1UA&utm_source=copy-link
#agenticai #mixtureofexperts
https://open.spotify.com/episode/3w1yB5fCkxkT2VNnHBJQHs?si=Ammq5Qe3TPaHr4QmXDe1UA&utm_source=copy-link
#agenticai #mixtureofexperts
2.4 trillion is not what running a model costs. It's what owning it costs!
The name holds two numbers: 2.4T stored, 95B active per token. Everyone quotes the disk number. Repo 4.89TB. Smallest heroic quant 397GB. The Apache 2.0 27B: 18GB, on a machine you own. #MixtureOfExperts #LocalLLM
The name holds two numbers: 2.4T stored, 95B active per token. Everyone quotes the disk number. Repo 4.89TB. Smallest heroic quant 397GB. The Apache 2.0 27B: 18GB, on a machine you own. #MixtureOfExperts #LocalLLM
October 29, 2026 at 3:59 AM
2.4 trillion is not what running a model costs. It's what owning it costs!
The name holds two numbers: 2.4T stored, 95B active per token. Everyone quotes the disk number. Repo 4.89TB. Smallest heroic quant 397GB. The Apache 2.0 27B: 18GB, on a machine you own. #MixtureOfExperts #LocalLLM
The name holds two numbers: 2.4T stored, 95B active per token. Everyone quotes the disk number. Repo 4.89TB. Smallest heroic quant 397GB. The Apache 2.0 27B: 18GB, on a machine you own. #MixtureOfExperts #LocalLLM
2.4 trillion is not what running a model costs. It's what owning it costs!
The name holds two numbers: 2.4T stored, 95B active per token. Everyone quotes the disk number. Repo 4.89TB. Smallest heroic quant 397GB. The Apache 2.0 27B: 18GB, on a machine you own. #MixtureOfExperts #LocalLLM
The name holds two numbers: 2.4T stored, 95B active per token. Everyone quotes the disk number. Repo 4.89TB. Smallest heroic quant 397GB. The Apache 2.0 27B: 18GB, on a machine you own. #MixtureOfExperts #LocalLLM
October 29, 2026 at 3:59 AM
2.4 trillion is not what running a model costs. It's what owning it costs!
The name holds two numbers: 2.4T stored, 95B active per token. Everyone quotes the disk number. Repo 4.89TB. Smallest heroic quant 397GB. The Apache 2.0 27B: 18GB, on a machine you own. #MixtureOfExperts #LocalLLM
The name holds two numbers: 2.4T stored, 95B active per token. Everyone quotes the disk number. Repo 4.89TB. Smallest heroic quant 397GB. The Apache 2.0 27B: 18GB, on a machine you own. #MixtureOfExperts #LocalLLM
CoRover just dropped three BharatGPT models on the government‑backed AIKosh platform – think speech‑to‑text, Mixture‑of‑Experts and a new Indic language powerhouse. Curious? Dive into the details! #BharatGPT #AIKosh #MixtureOfExperts
🔗 aidailypost.com/news/corover...
🔗 aidailypost.com/news/corover...
October 27, 2026 at 11:18 PM
CoRover just dropped three BharatGPT models on the government‑backed AIKosh platform – think speech‑to‑text, Mixture‑of‑Experts and a new Indic language powerhouse. Curious? Dive into the details! #BharatGPT #AIKosh #MixtureOfExperts
🔗 aidailypost.com/news/corover...
🔗 aidailypost.com/news/corover...
Adolescente entrenó modelo de IA de 780B parámetros
Un chico de 16 años entrenó solo un modelo de IA de 780 mil millones de parámetros. Así lo armó, con qué infraestructura y qué falta confirmar.
#ganymede11 #mixtureofexperts #afonsofaro #merkium #modelosdelenguaje
Un chico de 16 años entrenó solo un modelo de IA de 780 mil millones de parámetros. Así lo armó, con qué infraestructura y qué falta confirmar.
#ganymede11 #mixtureofexperts #afonsofaro #merkium #modelosdelenguaje
Adolescente entrenó modelo de IA de 780B parámetros
Afonso Faro armó Ganymede 1.1 solo, con arquitectura MoE y 450+ GPUs H200 en tres países. Te contamos qué está confirmado y qué no.
blog.donweb.com
October 22, 2026 at 4:17 AM
Adolescente entrenó modelo de IA de 780B parámetros
Un chico de 16 años entrenó solo un modelo de IA de 780 mil millones de parámetros. Así lo armó, con qué infraestructura y qué falta confirmar.
#ganymede11 #mixtureofexperts #afonsofaro #merkium #modelosdelenguaje
Un chico de 16 años entrenó solo un modelo de IA de 780 mil millones de parámetros. Así lo armó, con qué infraestructura y qué falta confirmar.
#ganymede11 #mixtureofexperts #afonsofaro #merkium #modelosdelenguaje
#callforreading #openaccess #AI We propose EXP4-LoRA, a contextual bandit framework that treats LoRA expert selection as an online decision problem. Please access the full article via www.sciltp.com/journals/tai...
#contextualbandit #LoRA #mixtureofexperts #onlinelearning #LLMrouting
#contextualbandit #LoRA #mixtureofexperts #onlinelearning #LLMrouting
October 20, 2026 at 5:36 AM
#callforreading #openaccess #AI We propose EXP4-LoRA, a contextual bandit framework that treats LoRA expert selection as an online decision problem. Please access the full article via www.sciltp.com/journals/tai...
#contextualbandit #LoRA #mixtureofexperts #onlinelearning #LLMrouting
#contextualbandit #LoRA #mixtureofexperts #onlinelearning #LLMrouting
MiniMax just dropped the M3.1 Flash at $0.15/M—428B params, Space Bunny tricks, huge context window, and a crazy price split. Kimi K3’s match table blows up the cost chart. Dive into the benchmarks and see why it matters. #MiniMaxM3_1 #MixtureOfExperts #LLMBenchmarks
🔗
🔗
October 19, 2026 at 3:39 PM
MiniMax just dropped the M3.1 Flash at $0.15/M—428B params, Space Bunny tricks, huge context window, and a crazy price split. Kimi K3’s match table blows up the cost chart. Dive into the benchmarks and see why it matters. #MiniMaxM3_1 #MixtureOfExperts #LLMBenchmarks
🔗
🔗
A 35B AI model just ran on an iPhone using only 3 GB of memory.
MiniCPM V4.7 uses mixture of experts and look-ahead loading to stay fast on a phone.
The catch: that 3 GB is memory, not storage. 📱
#AItools #MiniCPM #OpenSource #OnDeviceAI #LocalAI #MixtureOfExperts
MiniCPM V4.7 uses mixture of experts and look-ahead loading to stay fast on a phone.
The catch: that 3 GB is memory, not storage. 📱
#AItools #MiniCPM #OpenSource #OnDeviceAI #LocalAI #MixtureOfExperts
October 18, 2026 at 6:49 AM
A 35B AI model just ran on an iPhone using only 3 GB of memory.
MiniCPM V4.7 uses mixture of experts and look-ahead loading to stay fast on a phone.
The catch: that 3 GB is memory, not storage. 📱
#AItools #MiniCPM #OpenSource #OnDeviceAI #LocalAI #MixtureOfExperts
MiniCPM V4.7 uses mixture of experts and look-ahead loading to stay fast on a phone.
The catch: that 3 GB is memory, not storage. 📱
#AItools #MiniCPM #OpenSource #OnDeviceAI #LocalAI #MixtureOfExperts
www.ibm.com/think/podcas...
Ik vind dit de Marketing Giller van het jaar.
Ik kan Microsoft niet koppelen aan 📲 Android via Phone Link; dus Telefoon nummer niet op mijn Account.
Whatsapp doet het ook niet meer....
Ik vind dit de Marketing Giller van het jaar.
Ik kan Microsoft niet koppelen aan 📲 Android via Phone Link; dus Telefoon nummer niet op mijn Account.
Whatsapp doet het ook niet meer....
Listen to Mixture of Experts
Watch episode 129 of Mixture of Experts: Decision models take off & Anthropic bans cruelty to Claude. Mixture of Experts–an IBM Think podcast. #MoEPodcast #MixtureOfExperts #IBM
www.ibm.com
October 18, 2026 at 1:06 PM
www.ibm.com/think/podcas...
Ik vind dit de Marketing Giller van het jaar.
Ik kan Microsoft niet koppelen aan 📲 Android via Phone Link; dus Telefoon nummer niet op mijn Account.
Whatsapp doet het ook niet meer....
Ik vind dit de Marketing Giller van het jaar.
Ik kan Microsoft niet koppelen aan 📲 Android via Phone Link; dus Telefoon nummer niet op mijn Account.
Whatsapp doet het ook niet meer....
🚀 Reflection's 501B Beam is now free on Ollama Cloud! Dive into the 501‑billion‑parameter mixture‑of‑experts model—open‑weight, zero data retention, perfect for software‑engineering and agentic tasks. Grab it while it lasts! #ReflectionBeam #MixtureOfExperts #OllamaCloud
🔗
🔗
October 16, 2026 at 2:54 PM
🚀 Reflection's 501B Beam is now free on Ollama Cloud! Dive into the 501‑billion‑parameter mixture‑of‑experts model—open‑weight, zero data retention, perfect for software‑engineering and agentic tasks. Grab it while it lasts! #ReflectionBeam #MixtureOfExperts #OllamaCloud
🔗
🔗
🚀 Nvidia's new Nemotron 3.5 Super VL just crushed Gemini Ultra in agent benchmarks—thanks to its Mixture‑of‑Experts multimodal magic. Open model vibes are shaking up enterprise AI agents. Curious? Dive in! #Nemotron3_5SuperVL #MixtureOfExperts #EnterpriseAIAgents
🔗 aidailypost.com/news/nvidia-...
🔗 aidailypost.com/news/nvidia-...
October 15, 2026 at 4:52 PM
🚀 Nvidia's new Nemotron 3.5 Super VL just crushed Gemini Ultra in agent benchmarks—thanks to its Mixture‑of‑Experts multimodal magic. Open model vibes are shaking up enterprise AI agents. Curious? Dive in! #Nemotron3_5SuperVL #MixtureOfExperts #EnterpriseAIAgents
🔗 aidailypost.com/news/nvidia-...
🔗 aidailypost.com/news/nvidia-...
New arXiv work proposes the Superposed Specialisation Hypothesis for Mixture of Experts models, arguing experts specialise in fine-grained features rather than single domains, and introduces RouterInterp for…
#AI #MachineLearning #MixtureOfExperts #Interpretability
https://arxiv.org/abs/2610.11775
#AI #MachineLearning #MixtureOfExperts #Interpretability
https://arxiv.org/abs/2610.11775
October 10, 2026 at 10:01 PM
New arXiv work proposes the Superposed Specialisation Hypothesis for Mixture of Experts models, arguing experts specialise in fine-grained features rather than single domains, and introduces RouterInterp for…
#AI #MachineLearning #MixtureOfExperts #Interpretability
https://arxiv.org/abs/2610.11775
#AI #MachineLearning #MixtureOfExperts #Interpretability
https://arxiv.org/abs/2610.11775
DivMoE introduces a fine-grained MoE upcycling framework that addresses routing collapse when experts come from a single source model, outperforming naive fine-grained approaches on Qwen3-1.7B.
#AIresearch #MixtureOfExperts #LLMs #ModelUpcycling
https://arxiv.org/abs/2610.11317
#AIresearch #MixtureOfExperts #LLMs #ModelUpcycling
https://arxiv.org/abs/2610.11317
October 10, 2026 at 10:00 AM
DivMoE introduces a fine-grained MoE upcycling framework that addresses routing collapse when experts come from a single source model, outperforming naive fine-grained approaches on Qwen3-1.7B.
#AIresearch #MixtureOfExperts #LLMs #ModelUpcycling
https://arxiv.org/abs/2610.11317
#AIresearch #MixtureOfExperts #LLMs #ModelUpcycling
https://arxiv.org/abs/2610.11317
Технология Mixture of Experts: новый стандарт ИИ для финансов
Архитектура Mixture of Experts (MoE) позволяет финансовым организациям использовать мощь нейросетей с высокой скоростью и точностью за счет разделения…
Читать полностью
#ИскусственныйИнтеллект #ФинансовыеТехнологии #MixtureOfExperts
Архитектура Mixture of Experts (MoE) позволяет финансовым организациям использовать мощь нейросетей с высокой скоростью и точностью за счет разделения…
Читать полностью
#ИскусственныйИнтеллект #ФинансовыеТехнологии #MixtureOfExperts
October 9, 2026 at 8:43 AM
Технология Mixture of Experts: новый стандарт ИИ для финансов
Архитектура Mixture of Experts (MoE) позволяет финансовым организациям использовать мощь нейросетей с высокой скоростью и точностью за счет разделения…
Читать полностью
#ИскусственныйИнтеллект #ФинансовыеТехнологии #MixtureOfExperts
Архитектура Mixture of Experts (MoE) позволяет финансовым организациям использовать мощь нейросетей с высокой скоростью и точностью за счет разделения…
Читать полностью
#ИскусственныйИнтеллект #ФинансовыеТехнологии #MixtureOfExperts
Un MoE de 188M en una GPU gratis: qué probó y qué no
¿Se puede armar un modelo MoE desde cero con una GPU gratis? Un dev diseñó 188M parámetros con 51M activos. Mirá qué midió y qué falta confirmar
#moe #deepseekmoe #llm #colab #mixtureofexperts
¿Se puede armar un modelo MoE desde cero con una GPU gratis? Un dev diseñó 188M parámetros con 51M activos. Mirá qué midió y qué falta confirmar
#moe #deepseekmoe #llm #colab #mixtureofexperts
Un MoE de 188M en una GPU gratis: qué probó y qué no
188M parámetros totales, 51M activos por token y 64 micro-expertos por capa. Qué midió Shivam Kumar con su modelo MoE desde cero y qué falta confirmar.
blog.donweb.com
October 7, 2026 at 7:12 AM
Un MoE de 188M en una GPU gratis: qué probó y qué no
¿Se puede armar un modelo MoE desde cero con una GPU gratis? Un dev diseñó 188M parámetros con 51M activos. Mirá qué midió y qué falta confirmar
#moe #deepseekmoe #llm #colab #mixtureofexperts
¿Se puede armar un modelo MoE desde cero con una GPU gratis? Un dev diseñó 188M parámetros con 51M activos. Mirá qué midió y qué falta confirmar
#moe #deepseekmoe #llm #colab #mixtureofexperts
🚀 Mistral AI just dropped “Le Chonk” – a 1.05 trillion‑parameter Mixture‑of‑Experts model with vision encoder, 1 M token context, and ready for Nvidia Grace Blackwell. Open‑weight and massive. Dive in to see what this means for AI! #MistralAI #TrillionParams #MixtureOfExperts
🔗
🔗
October 6, 2026 at 6:13 PM
🚀 Mistral AI just dropped “Le Chonk” – a 1.05 trillion‑parameter Mixture‑of‑Experts model with vision encoder, 1 M token context, and ready for Nvidia Grace Blackwell. Open‑weight and massive. Dive in to see what this means for AI! #MistralAI #TrillionParams #MixtureOfExperts
🔗
🔗
Reflection AI just dropped its Beam model—an open‑weight, mixture‑of‑experts beast that hits reasoning benchmarks on par with Chinese rivals while using far less inference compute. Curious how they did it? Dive in. #ReflectionAI #BeamModel #MixtureOfExperts
🔗 aidailypost.com/news/reflect...
🔗 aidailypost.com/news/reflect...
October 5, 2026 at 8:38 PM
Reflection AI just dropped its Beam model—an open‑weight, mixture‑of‑experts beast that hits reasoning benchmarks on par with Chinese rivals while using far less inference compute. Curious how they did it? Dive in. #ReflectionAI #BeamModel #MixtureOfExperts
🔗 aidailypost.com/news/reflect...
🔗 aidailypost.com/news/reflect...
New arXiv work shows that a small set of routed experts in sparse MoE LLMs can be targeted to weaken refusal behavior without retraining, and that router-gradient sensitivity outperforms activation frequency for…
#AISafety #LLM #MixtureOfExperts #CyberSecurity
https://arxiv.org/abs/2610.02910
#AISafety #LLM #MixtureOfExperts #CyberSecurity
https://arxiv.org/abs/2610.02910
October 5, 2026 at 4:01 PM
New arXiv work shows that a small set of routed experts in sparse MoE LLMs can be targeted to weaken refusal behavior without retraining, and that router-gradient sensitivity outperforms activation frequency for…
#AISafety #LLM #MixtureOfExperts #CyberSecurity
https://arxiv.org/abs/2610.02910
#AISafety #LLM #MixtureOfExperts #CyberSecurity
https://arxiv.org/abs/2610.02910
ألمانيا دخلت سباق النماذج مفتوحة الأوزان… بنموذج حجمه 78 مليار، بس بيشتغل بـ 3.46 مليار بس في كل كلمة! 🇩🇪
في 3 أكتوبر (يوم الوحدة الألمانية)، شركة Aleph Alpha نزّلت Kolibri: نموذج استدلال
#Kolibri #AlephAlpha #MixtureOfExperts #OpenWeights #ذكاء_اصطناعي_عربي #حسام_الدين_حسن #خبير_اونلاين
في 3 أكتوبر (يوم الوحدة الألمانية)، شركة Aleph Alpha نزّلت Kolibri: نموذج استدلال
#Kolibri #AlephAlpha #MixtureOfExperts #OpenWeights #ذكاء_اصطناعي_عربي #حسام_الدين_حسن #خبير_اونلاين
October 4, 2026 at 1:19 PM
ألمانيا دخلت سباق النماذج مفتوحة الأوزان… بنموذج حجمه 78 مليار، بس بيشتغل بـ 3.46 مليار بس في كل كلمة! 🇩🇪
في 3 أكتوبر (يوم الوحدة الألمانية)، شركة Aleph Alpha نزّلت Kolibri: نموذج استدلال
#Kolibri #AlephAlpha #MixtureOfExperts #OpenWeights #ذكاء_اصطناعي_عربي #حسام_الدين_حسن #خبير_اونلاين
في 3 أكتوبر (يوم الوحدة الألمانية)، شركة Aleph Alpha نزّلت Kolibri: نموذج استدلال
#Kolibri #AlephAlpha #MixtureOfExperts #OpenWeights #ذكاء_اصطناعي_عربي #حسام_الدين_حسن #خبير_اونلاين
🐦 #AlephAlpha released #Kolibri, an open-weight #LLM for German and English (Apache 2.0). #MixtureOfExperts: 78.1B parameters, only 3.46B active per token. Trained from scratch in Germany and Finland #SovereignAI #opensource #AI
🧵👇 🇩🇪
🧵👇 🇩🇪
October 3, 2026 at 3:26 PM
🐦 #AlephAlpha released #Kolibri, an open-weight #LLM for German and English (Apache 2.0). #MixtureOfExperts: 78.1B parameters, only 3.46B active per token. Trained from scratch in Germany and Finland #SovereignAI #opensource #AI
🧵👇 🇩🇪
🧵👇 🇩🇪
🚀 H Company just smashed the OSWorld benchmark—Holo4 27B scored 85.2% at just $0.08 per task! See how its Mixture‑of‑Experts architecture powers GUI and tool‑calling agents. Curious? Dive into the details. #Holo4 #OSWorld #MixtureOfExperts
🔗 aidailypost.com/news/h-compa...
🔗 aidailypost.com/news/h-compa...
September 29, 2026 at 8:03 AM
🚀 H Company just smashed the OSWorld benchmark—Holo4 27B scored 85.2% at just $0.08 per task! See how its Mixture‑of‑Experts architecture powers GUI and tool‑calling agents. Curious? Dive into the details. #Holo4 #OSWorld #MixtureOfExperts
🔗 aidailypost.com/news/h-compa...
🔗 aidailypost.com/news/h-compa...
Mixture of Experts buff.ly/WnKpoKs
Mixture of Experts —IBM podcast
Listen to the latest episode of Mixture of Experts–an IBM podcast. Your weekly AI news podcast! #MoEPodcast #MixtureOfExperts #IBM
buff.ly
September 26, 2026 at 8:02 PM
Mixture of Experts buff.ly/WnKpoKs
Cracking the JAX MoE training puzzle: variable expert token counts, routing tricks, and the NVIDIA Transformer Engine boost. See how DeepSeek‑V3 and GB200 handle conditional computation. Dive in for the nitty‑gritty! #JAX #MixtureOfExperts #ConditionalComputation
🔗 aidailypost.com/news/jax-moe...
🔗 aidailypost.com/news/jax-moe...
September 14, 2026 at 4:54 PM
Cracking the JAX MoE training puzzle: variable expert token counts, routing tricks, and the NVIDIA Transformer Engine boost. See how DeepSeek‑V3 and GB200 handle conditional computation. Dive in for the nitty‑gritty! #JAX #MixtureOfExperts #ConditionalComputation
🔗 aidailypost.com/news/jax-moe...
🔗 aidailypost.com/news/jax-moe...
#Term: #MixtureOfBlockAttention (#Moba) – #ArtificialIntelligence - https://with.ga/y4ww8
"Mixture of Block Attention (MoBA) is an efficient, #SparseAttention mechanism for #Transformer models that applies the routing logic of #MixtureOfExperts (#Moe) to sequence blocks instead of...
"Mixture of Block Attention (MoBA) is an efficient, #SparseAttention mechanism for #Transformer models that applies the routing logic of #MixtureOfExperts (#Moe) to sequence blocks instead of...
September 11, 2026 at 9:36 AM
#Term: #MixtureOfBlockAttention (#Moba) – #ArtificialIntelligence - https://with.ga/y4ww8
"Mixture of Block Attention (MoBA) is an efficient, #SparseAttention mechanism for #Transformer models that applies the routing logic of #MixtureOfExperts (#Moe) to sequence blocks instead of...
"Mixture of Block Attention (MoBA) is an efficient, #SparseAttention mechanism for #Transformer models that applies the routing logic of #MixtureOfExperts (#Moe) to sequence blocks instead of...
#Term: #MixtureOfBlockAttention (#Moba) – #ArtificialIntelligence - https://with.ga/y4ww8
"Mixture of Block Attention (MoBA) is an efficient, #SparseAttention mechanism for #Transformer models that applies the routing logic of #MixtureOfExperts (#Moe) to sequence blocks instead of...
"Mixture of Block Attention (MoBA) is an efficient, #SparseAttention mechanism for #Transformer models that applies the routing logic of #MixtureOfExperts (#Moe) to sequence blocks instead of...
September 11, 2026 at 9:30 AM
#Term: #MixtureOfBlockAttention (#Moba) – #ArtificialIntelligence - https://with.ga/y4ww8
"Mixture of Block Attention (MoBA) is an efficient, #SparseAttention mechanism for #Transformer models that applies the routing logic of #MixtureOfExperts (#Moe) to sequence blocks instead of...
"Mixture of Block Attention (MoBA) is an efficient, #SparseAttention mechanism for #Transformer models that applies the routing logic of #MixtureOfExperts (#Moe) to sequence blocks instead of...