AntoItaly

joined 1 year ago
sorted by: new top controversial old
QuIP#: SOTA 2-bit quantization method, now implemented in text-generation-webui (experimental) in c/localllama@poweruser.forum
[–] AntoItaly@alien.top 1 points 11 months ago (1 children)

Wow, with this quantization method, LLama 70B weighs only 17.5GB!

Yet another 70B Foundation Model: Aquila2-70B-Expr in c/localllama@poweruser.forum
Deepseek llm 67b Chat & Base in c/localllama@poweruser.forum
[–] AntoItaly@alien.top 1 points 11 months ago

Wow, this model seems very good for the Italian language!

Cheapest site for hosting custom LLM models? in c/localllama@poweruser.forum
[–] AntoItaly@alien.top 1 points 11 months ago (2 children)

Replicate $0.000575/sec for a Nvidia A40 (48GB Vram)

Intel's neural-chat-7b-v3-1 dropped in c/localllama@poweruser.forum
[–] AntoItaly@alien.top 1 points 11 months ago

also they used DPO

Details emerge of surprise board coup that ousted CEO Sam Altman at OpenAI (Microsoft CEO Nadella "furious"; OpenAI President and three senior researchers resign) in c/localllama@poweruser.forum
[–] AntoItaly@alien.top 1 points 11 months ago

OpenAIGATE

Sam Altman out as CEO of OpenAI. Mira Murati is the new CEO. in c/localllama@poweruser.forum
[–] AntoItaly@alien.top 1 points 11 months ago (1 children)

I hope GPT-3 becomes opensource with Mira Murati as CEO

Apple M3 Pro Chip Has 25% Less Memory Bandwidth Than M1/M2 Pro in c/localllama@poweruser.forum
[–] AntoItaly@alien.top 1 points 1 year ago

Facepalm