werks on top of llama.cpp commit 877b4d0c628cc70dddb5df72ed8fc14d126ca7e8
With newer commits from llama.cpp master something breaks that I don't understand. "ggml_assert: hashtable full" or something.
werks on top of llama.cpp commit 877b4d0c628cc70dddb5df72ed8fc14d126ca7e8
With newer commits from llama.cpp master something breaks that I don't understand. "ggml_assert: hashtable full" or something.