From 09d83f04013f9e8551c3ff54449cf28e1ca00784 Mon Sep 17 00:00:00 2001 From: Adam Treat Date: Thu, 5 Oct 2023 10:52:04 -0400 Subject: [PATCH] Delete TODO now that we have q8_0. --- llama.cpp | 1 - 1 file changed, 1 deletion(-) diff --git a/llama.cpp b/llama.cpp index 858494244..f5e0eac81 100644 --- a/llama.cpp +++ b/llama.cpp @@ -6510,7 +6510,6 @@ struct llama_context * llama_new_context_with_model( #undef LLAMA_METAL_CHECK_BUF } #elif defined(GGML_USE_KOMPUTE) - // TODO(cebtenzzre): we need to check the type of each tensor because Q8_0 is not currently supported if (ggml_vk_has_device() && params.n_gpu_layers > 0 && (model->arch == LLM_ARCH_LLAMA || model->arch == LLM_ARCH_FALCON) && (model->ftype == LLAMA_FTYPE_ALL_F32