Public/koboldcpp
1
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-09-20 09:38:55 +02:00
Code Issues Packages Projects Releases Wiki Activity

Merge commit '79bba02a6741de194912d370015866414faa83ad' into concedo_experimental

Browse Source 
# Conflicts:
#	.github/workflows/build-self-hosted.yml
#	.github/workflows/release.yml
#	.github/workflows/server-self-hosted.yml
#	ggml/src/ggml-opencl/ggml-opencl.cpp
#	ggml/src/ggml-opencl/kernels/flash_attn_f32_f16.cl
#	ggml/src/ggml-opencl/kernels/flash_attn_f32_q4_0.cl
#	ggml/src/ggml-opencl/kernels/flash_attn_f32_q8_0.cl
#	ggml/src/ggml-opencl/kernels/gemm_moe_mxfp4_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/mul_mv_f16_f32_l4.cl
#	ggml/src/ggml-opencl/kernels/mul_mv_q4_k_f32.cl
#	tests/test-backend-ops.cpp
#	tools/tokenize/tokenize.cpp
#	tools/ui/src/lib/components/app/chat/ChatForm/ChatFormActions/ChatFormActionAdd/ChatFormActionAddToolsSubmenu.svelte
This commit is contained in:
Concedo
2026-07-18 11:04:46 +08:00
parent 8b85ba1a33 79bba02a67
commit 4c25a3d829
180 changed files with 7455 additions and 1324 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/llama-quant.cpp
+2 -2
View File
@@ -675,7 +675,7 @@ static ggml_type llama_tensor_get_type(quantize_state_impl & qs, const llama_mod
ggml_type new_type = default_type;
// get more optimal quantization type based on the tensor shape, layer, etc.
if (!params->pure && ggml_is_quantized(default_type)) {
if (ggml_is_quantized(default_type)) {
// if the user provided tensor types - use those
bool manual = false;
if (!qs.tensor_type_patterns.empty()) {
@@ -694,7 +694,7 @@ static ggml_type llama_tensor_get_type(quantize_state_impl & qs, const llama_mod
}
// if not manual - use the standard logic for choosing the quantization type based on the selected mixture
if (!manual) {
if (!manual && !params->pure) {
new_type = llama_tensor_get_type_impl(qs, new_type, tensor, params->ftype, tm.category);
}