nodes/city96-ComfyUI-GGUF
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Show the correct type when using fallback quantization

Browse Source
This commit is contained in:
blepping
2024-08-23 12:37:31 -06:00
parent cea1b47ab3
commit 489aaecf7a
1 changed files with 6 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/lcpp.patch
+6 -6
View File
@@ -1,5 +1,5 @@
diff --git a/src/llama.cpp b/src/llama.cpp
index 5ab65ea9..ef63e28f 100644
index 5ab65ea9..6d684fef 100644
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -212,6 +212,9 @@ enum llm_arch {
@@ -156,15 +156,15 @@ index 5ab65ea9..ef63e28f 100644
if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L || ftype == LLAMA_FTYPE_MOSTLY_IQ3_M) {
new_type = GGML_TYPE_Q4_K;
}
@@ -16107,6 +16140,8 @@ static ggml_type llama_tensor_get_type(quantize_state_internal & qs, ggml_type n
@@ -16105,6 +16138,8 @@ static ggml_type llama_tensor_get_type(quantize_state_internal & qs, ggml_type n
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));
++qs.n_fallback;
+ // Force FP16 fallback - needed due to Conv2D
+ new_type = GGML_TYPE_F16;
LLAMA_LOG_WARN(" - using fallback quantization %s\n", ggml_type_name(new_type));
++qs.n_fallback;
}
return new_type;
@@ -17432,6 +17467,9 @@ enum llama_rope_type llama_rope_type(const struct llama_model * model) {
case LLM_ARCH_T5:
case LLM_ARCH_T5ENCODER: