minus-squareAntoItaly@alien.topBtoLocalLLaMA•QuIP#: SOTA 2-bit quantization method, now implemented in text-generation-webui (experimental)linkfedilinkEnglisharrow-up1·2 years agoWow, with this quantization method, LLama 70B weighs only 17.5GB! linkfedilink
minus-squareAntoItaly@alien.topBtoLocalLLaMA•Yet another 70B Foundation Model: Aquila2-70B-ExprlinkfedilinkEnglisharrow-up1·2 years agoSource? linkfedilink
minus-squareAntoItaly@alien.topBtoLocalLLaMA•Deepseek llm 67b Chat & BaselinkfedilinkEnglisharrow-up1·2 years agoWow, this model seems very good for the Italian language! linkfedilink
minus-squareAntoItaly@alien.topBtoLocalLLaMA•Cheapest site for hosting custom LLM models?linkfedilinkarrow-up1·2 years agoReplicate $0.000575/sec for a Nvidia A40 (48GB Vram) linkfedilink
minus-squareAntoItaly@alien.topBtoLocalLLaMA•Intel's neural-chat-7b-v3-1 droppedlinkfedilinkEnglisharrow-up1·2 years agoalso they used DPO linkfedilink
minus-squareAntoItaly@alien.topBtoLocalLLaMA•Details emerge of surprise board coup that ousted CEO Sam Altman at OpenAI (Microsoft CEO Nadella "furious"; OpenAI President and three senior researchers resign)linkfedilinkarrow-up1·2 years agoOpenAIGATE linkfedilink
minus-squareAntoItaly@alien.topBtoLocalLLaMA•Sam Altman out as CEO of OpenAI. Mira Murati is the new CEO.linkfedilinkEnglisharrow-up1·2 years agoI hope GPT-3 becomes opensource with Mira Murati as CEO linkfedilink
minus-squareAntoItaly@alien.topBtoLocalLLaMA•Apple M3 Pro Chip Has 25% Less Memory Bandwidth Than M1/M2 Prolinkfedilinkarrow-up1·2 years agoFacepalm linkfedilink
Wow, with this quantization method, LLama 70B weighs only 17.5GB!