From f6ea7a093c6b7eb348071d158290902ceb7459e8 Mon Sep 17 00:00:00 2001 From: Georgi Gerganov Date: Mon, 15 Jul 2024 10:27:07 +0300 Subject: [PATCH] llama : change fallback type IQ4_NL -> Q4_0 ggml-ci --- src/llama.cpp | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/src/llama.cpp b/src/llama.cpp index 4a9903cc3..9312e24d5 100644 --- a/src/llama.cpp +++ b/src/llama.cpp @@ -18090,10 +18090,10 @@ static ggml_type llama_tensor_get_type(quantize_state_internal & qs, ggml_type n case GGML_TYPE_IQ1_M: case GGML_TYPE_Q2_K: case GGML_TYPE_Q3_K: - case GGML_TYPE_IQ4_XS: new_type = GGML_TYPE_IQ4_NL; break; - case GGML_TYPE_Q4_K: new_type = GGML_TYPE_Q5_0; break; - case GGML_TYPE_Q5_K: new_type = GGML_TYPE_Q5_1; break; - case GGML_TYPE_Q6_K: new_type = GGML_TYPE_Q8_0; break; + case GGML_TYPE_IQ4_XS: new_type = GGML_TYPE_Q4_0; break; + case GGML_TYPE_Q4_K: new_type = GGML_TYPE_Q5_0; break; + case GGML_TYPE_Q5_K: new_type = GGML_TYPE_Q5_1; break; + case GGML_TYPE_Q6_K: new_type = GGML_TYPE_Q8_0; break; default: throw std::runtime_error("\nUnsupported tensor size encountered\n"); } LLAMA_LOG_WARN(" - using fallback quantization %s\n", ggml_type_name(new_type));