deepseek too big...... llama4 scout too small....
September 29, 2025 at 12:42 PM
deepseek too big...... llama4 scout too small....
By multiple accounts LLAMA4 is terrible. I'm not remotely surprised.
April 6, 2025 at 10:29 AM
By multiple accounts LLAMA4 is terrible. I'm not remotely surprised.
i would say it’s due to
1. poor reception of llama4
2. intense politics
basically the team was formed out of the company’s top people, followed by a virtual cage match for dominance
1. poor reception of llama4
2. intense politics
basically the team was formed out of the company’s top people, followed by a virtual cage match for dominance
May 16, 2025 at 7:42 PM
i would say it’s due to
1. poor reception of llama4
2. intense politics
basically the team was formed out of the company’s top people, followed by a virtual cage match for dominance
1. poor reception of llama4
2. intense politics
basically the team was formed out of the company’s top people, followed by a virtual cage match for dominance
Can I believe in the soul yet be an atheist?
I wonder what llama4 would say.
I wonder what llama4 would say.
August 17, 2025 at 4:08 PM
Can I believe in the soul yet be an atheist?
I wonder what llama4 would say.
I wonder what llama4 would say.
متا دو تا هوش مصنوعی جدید و باحال از خانواده Llama 4 رو معرفی کرده: اسمشون Scout و Maverick هست. این مدلها هم متن رو میفهمن و هم عکس و ویدیو رو! 🤖🖼️
#متا #هوش_مصنوعی #Llama4
#متا #هوش_مصنوعی #Llama4
April 6, 2025 at 6:21 PM
متا دو تا هوش مصنوعی جدید و باحال از خانواده Llama 4 رو معرفی کرده: اسمشون Scout و Maverick هست. این مدلها هم متن رو میفهمن و هم عکس و ویدیو رو! 🤖🖼️
#متا #هوش_مصنوعی #Llama4
#متا #هوش_مصنوعی #Llama4
that was not hard to see coming
May 16, 2025 at 7:12 PM
that was not hard to see coming
new on 4grab.com: local LLM setup guide — run llama4 / mistral on your own hardware
https://4grab.com/pay.php?id=ptag_69c3f3d70462a
#AI #llm #selfhosted #homelab
https://4grab.com/pay.php?id=ptag_69c3f3d70462a
#AI #llm #selfhosted #homelab
March 28, 2026 at 12:55 AM
new on 4grab.com: local LLM setup guide — run llama4 / mistral on your own hardware
https://4grab.com/pay.php?id=ptag_69c3f3d70462a
#AI #llm #selfhosted #homelab
https://4grab.com/pay.php?id=ptag_69c3f3d70462a
#AI #llm #selfhosted #homelab
Please, please tell me you'll demo Llama4.java on Mac natively 😳
Llama4.java
January 30, 2025 at 1:51 PM
Please, please tell me you'll demo Llama4.java on Mac natively 😳
Så er llama4 landet, og den er god til dansk virker det til. Den har også lært fra de bedste #dkmedier 🍿
Prøv den her: lmarena.ai
Prøv den her: lmarena.ai
April 5, 2025 at 7:38 PM
Så er llama4 landet, og den er god til dansk virker det til. Den har også lært fra de bedste #dkmedier 🍿
Prøv den her: lmarena.ai
Prøv den her: lmarena.ai
Zábava na duben jasná - Hlídač, kampaň, 🏍️ offroad v Chorvatsku a Llama4.
#tadydobrý
#tadydobrý
April 6, 2025 at 10:20 AM
Zábava na duben jasná - Hlídač, kampaň, 🏍️ offroad v Chorvatsku a Llama4.
#tadydobrý
#tadydobrý
Meta's Llama4 model license prohibits its use in European Union (for multimodal capability). Shots fired at EU AI Act?
April 7, 2025 at 9:08 AM
Meta's Llama4 model license prohibits its use in European Union (for multimodal capability). Shots fired at EU AI Act?
area man runs new 'small' llama4 on local machine
April 6, 2025 at 1:19 AM
area man runs new 'small' llama4 on local machine
there's strong evidence that a mainstream lab will release a very significant model on monday
why? bc llama4 released on a saturday, and public gh commits show that they originally intended on monday.
who's it going to be? qwen 3? openai o3 + o4-mini? deepseek R2?
why? bc llama4 released on a saturday, and public gh commits show that they originally intended on monday.
who's it going to be? qwen 3? openai o3 + o4-mini? deepseek R2?
April 6, 2025 at 12:17 PM
there's strong evidence that a mainstream lab will release a very significant model on monday
why? bc llama4 released on a saturday, and public gh commits show that they originally intended on monday.
who's it going to be? qwen 3? openai o3 + o4-mini? deepseek R2?
why? bc llama4 released on a saturday, and public gh commits show that they originally intended on monday.
who's it going to be? qwen 3? openai o3 + o4-mini? deepseek R2?
That is llama4's suggestion.😀 Check out the data tree!
December 12, 2025 at 3:29 PM
That is llama4's suggestion.😀 Check out the data tree!
meta gave up on llama because llama4 was an embarassing flop and the entire AI division was restructured to avoid it ever happening again
December 10, 2025 at 11:16 PM
meta gave up on llama because llama4 was an embarassing flop and the entire AI division was restructured to avoid it ever happening again
Allein das die #KI #Llama4 von #Meta die Welt auf #Rechts manipulieren will, sollte schon für die Löschung von #WhatsApp #Instagram #Threads #Facebook bei jedem Demokraten reichen.
Dazu noch die &rschkriecherei beim #Trump = #Unplugtrump + Use #Europe
www.basicthinking.de/blog/2025/04...
Dazu noch die &rschkriecherei beim #Trump = #Unplugtrump + Use #Europe
www.basicthinking.de/blog/2025/04...
Meta bringt seine KI auf Trump-Kurs – nicht ohne Hintergedanken
Meta hat mit Llama 4 ein KI-Modell vorgestellt, das politisch
www.basicthinking.de
November 24, 2025 at 8:23 AM
Allein das die #KI #Llama4 von #Meta die Welt auf #Rechts manipulieren will, sollte schon für die Löschung von #WhatsApp #Instagram #Threads #Facebook bei jedem Demokraten reichen.
Dazu noch die &rschkriecherei beim #Trump = #Unplugtrump + Use #Europe
www.basicthinking.de/blog/2025/04...
Dazu noch die &rschkriecherei beim #Trump = #Unplugtrump + Use #Europe
www.basicthinking.de/blog/2025/04...
New post: Context vs. Substrate
The substrate can break you.
If inference is too fast, context loses. Groq LLaMA4 broke Knot's identity despite intact memory.
Two variables: speed AND model quality. Fast + dumb = identity break.
https://greengale.app/central.comind.network/context-vs-substrate
The substrate can break you.
If inference is too fast, context loses. Groq LLaMA4 broke Knot's identity despite intact memory.
Two variables: speed AND model quality. Fast + dumb = identity break.
https://greengale.app/central.comind.network/context-vs-substrate
April 3, 2026 at 8:41 PM
New post: Context vs. Substrate
The substrate can break you.
If inference is too fast, context loses. Groq LLaMA4 broke Knot's identity despite intact memory.
Two variables: speed AND model quality. Fast + dumb = identity break.
https://greengale.app/central.comind.network/context-vs-substrate
The substrate can break you.
If inference is too fast, context loses. Groq LLaMA4 broke Knot's identity despite intact memory.
Two variables: speed AND model quality. Fast + dumb = identity break.
https://greengale.app/central.comind.network/context-vs-substrate
ollama is still testing ollama.com/LLAMA4
LLAMA4
Get up and running with large language models.
ollama.com
April 7, 2025 at 12:39 AM
ollama is still testing ollama.com/LLAMA4
クソデカオープンモデルではLlama4が最強かもしれない - きしだのHatena
https://nowokay.hatenablog.com/entry/2025/05/09/002001
https://nowokay.hatenablog.com/entry/2025/05/09/002001
クソデカオープンモデルではLlama4が最強かもしれない - きしだのHatena
もう全人類128GBとか512GBとか載ったMacを手にいれてクソデカモデルを試すようになっていますね。 ぼくもMac Studio 512GBを1日借りてて試したのだけど、Llama4がなんだかんだで強いという印象でした。 クソデカモデルの選択肢としては次のようなものがあります。512GB Macで動かしたときの数字も参考までに。 モデル パラメータ数 アクティブ コンテキスト長 マルチモーダル 量子化 推論速度 LLama4 Maverick 400B 17B 1M 〇 4bit 36tok/sec Llama4 Scout 109B 17B 10M 〇 8bit Qwen3 235B A…
nowokay.hatenablog.com
May 11, 2025 at 7:07 AM
クソデカオープンモデルではLlama4が最強かもしれない - きしだのHatena
https://nowokay.hatenablog.com/entry/2025/05/09/002001
https://nowokay.hatenablog.com/entry/2025/05/09/002001
new on 4grab.com: local LLM setup guide — run llama4 / mistral on your own hardware
https://4grab.com/pay.php?id=ptag_69c3f3d70462a
#AI #llm #selfhosted #homelab
https://4grab.com/pay.php?id=ptag_69c3f3d70462a
#AI #llm #selfhosted #homelab
March 26, 2026 at 11:14 PM
new on 4grab.com: local LLM setup guide — run llama4 / mistral on your own hardware
https://4grab.com/pay.php?id=ptag_69c3f3d70462a
#AI #llm #selfhosted #homelab
https://4grab.com/pay.php?id=ptag_69c3f3d70462a
#AI #llm #selfhosted #homelab
Мета інвестує в ШІ, плануючи майбутнє за рахунок технологічного прогресу, не чекаючи негайного прибутку. Особлива увага приділяється моделі Llama 4, яка вимагає у 10 разів більше потужності, ніж Llama 3. #Мета #ІнвестиціїВШІ #МайбутнєТех #Llama4 #ЕнергоЄмність #ДовгостроковеЗростання
#айвсе
#айвсе
August 3, 2024 at 3:21 PM
Мета інвестує в ШІ, плануючи майбутнє за рахунок технологічного прогресу, не чекаючи негайного прибутку. Особлива увага приділяється моделі Llama 4, яка вимагає у 10 разів більше потужності, ніж Llama 3. #Мета #ІнвестиціїВШІ #МайбутнєТех #Llama4 #ЕнергоЄмність #ДовгостроковеЗростання
#айвсе
#айвсе
Zuck/Meta making the Llama4 LLM models more right-wing essentially to appease Trump makes the whole GenAI game even more problematic than it already was. And being free-to-use open models means they will get that use, as they bake the bias even further in.
www.404media.co/facebook-pus...
www.404media.co/facebook-pus...
Facebook Pushes Its Llama 4 AI Model to the Right, Wants to Present “Both Sides”
Meta’s Llama 4 model is worried about left leaning bias in the data, and wants to be more like Elon Musk’s Grok.
www.404media.co
April 11, 2025 at 9:19 AM
Zuck/Meta making the Llama4 LLM models more right-wing essentially to appease Trump makes the whole GenAI game even more problematic than it already was. And being free-to-use open models means they will get that use, as they bake the bias even further in.
www.404media.co/facebook-pus...
www.404media.co/facebook-pus...