Post
1,065
We've released smaller GGUF builds of Clef, Clef-Flash, Perplexity's Decider and Kev for stock llama.cpp, plus Gemma 4 E4B and E2B. Each file is checked against its original, with the full comparisons on its card.
One example: Clef-Flash goes from 18.2 GB to 4.9 GB and gives the original's answer on 95% of 2,000 test decisions on CPU. turintech/Clef-Flash-Artemis-GGUF
Which model would you like smaller next, and on what hardware would you run it? Reply here or in our feedback thread: turintech/README#1
Decision models: https://huggingface.co/collections/turintech/artemis-edge-decision-models-6accea74eb7016a886f25fdb
Compressed LLMs: https://huggingface.co/collections/turintech/artemis-edge-compressed-llms-6acd251457ba2b9f33ad9d8a
One example: Clef-Flash goes from 18.2 GB to 4.9 GB and gives the original's answer on 95% of 2,000 test decisions on CPU. turintech/Clef-Flash-Artemis-GGUF
Which model would you like smaller next, and on what hardware would you run it? Reply here or in our feedback thread: turintech/README#1
Decision models: https://huggingface.co/collections/turintech/artemis-edge-decision-models-6accea74eb7016a886f25fdb
Compressed LLMs: https://huggingface.co/collections/turintech/artemis-edge-compressed-llms-6acd251457ba2b9f33ad9d8a