Skip to content

Fix qwen3 30b #8

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
JohannesGaessler merged 2 commits into from
Feb 25, 2026
Merged

Fix qwen3 30b #8

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b9ed7a9
Fix crash with Qwen-30B-A3B Q4_0
gaugarg-nv Feb 24, 2026
bbcbae2
Decide block size based on tensor quantization type
gaugarg-nv Feb 25, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions src/llama-model.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -104,8 +104,7 @@ struct ggml_backend_meta_split_state llama_meta_device_get_split_state(const str
return GGML_BACKEND_SPLIT_AXIS_MIRRORED;
};

auto get_split_granularity = [&](ggml_backend_meta_split_axis split_axis) -> int64_t {
const int64_t blck_size = split_axis == GGML_BACKEND_SPLIT_AXIS_1 && tensor->ne[1] % 256 == 0 ? 256 : 32;
auto get_split_granularity = [&](int64_t blck_size) -> int64_t {

// attention
if (std::regex_match(tensor->name, pattern_q_weight) || std::regex_match(tensor->name, pattern_q_bias) ||
Expand Down Expand Up @@ -140,7 +139,8 @@ struct ggml_backend_meta_split_state llama_meta_device_get_split_state(const str
split_state.axis = get_split_axis();
if (split_state.axis >= 0 && split_state.axis < GGML_MAX_DIMS) {
const int64_t ne_full = tensor->ne[split_state.axis];
const int64_t granularity = get_split_granularity(split_state.axis);
const int64_t blck_size = ggml_blck_size(tensor->type);
const int64_t granularity = get_split_granularity(blck_size);
GGML_ASSERT(ne_full % granularity == 0);
const float * tensor_split = ud->model->tensor_split();
std::vector<float> tensor_split_scan;
Expand Down