#llama70b — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #llama70b, aggregated by home.social.
-
🚀🤡 In what sounds like a fever dream concocted by a caffeinated coder, somebody managed to finagle a Llama 70B model onto an RTX 3090 using an NVMe-to-GPU #magic #trick. Meanwhile, #GitHub buzzwords are flying like confetti at a tech bro's birthday party. But hey, at least we can always count on these geniuses to overcomplicate the simple. 🧐💻
https://github.com/xaskasdf/ntransformer #Llama70B #RTX3090 #NVMe #techhumor #HackerNews #ngated -
Ah yes, the age-old dilemma: how to make a Llama smaller and smarter at the same time. 🤓 Because everyone knows that what the world truly needs is a 224x compressed Llama-70B—just imagine the savings on imaginary Llama feed and storage! 🙄 #ProblemSolved #TotallyEssentialResearch
https://zenodo.org/records/17873275 #LlamaCompression #Llama70B #SmartLlama #TechHumor #AIResearch #HackerNews #ngated -
Post-transformer inference: 224× compression of Llama-70B with improved accuracy
https://zenodo.org/records/17873275
#HackerNews #PostTransformer #Inference #Llama70B #Compression #ImprovedAccuracy