nikola/llama_cpp
1
0
Fork 0
Code Issues Pull Requests Actions 25 Packages Projects Releases Wiki Activity

tests : add support for qwen3 SSM archs (#24031)

Browse Source 
* tests : add support for qwen3 SSM archs

* arch : add LLM_KV_ATTENTION_RECURRENT_LAYERS

* cont : naming + TODOs
This commit is contained in:
Georgi Gerganov
2026-06-03 10:15:27 +03:00
committed by GitHub
parent d545a2a993
commit 06938ac129
25 changed files with 109 additions and 83 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/llama-model.cpp
+13 -15
View File
@@ -373,10 +373,10 @@ struct ggml_backend_meta_split_state llama_meta_device_get_split_state(const str
// count only the same type of previous layers to avoid this
auto get_il_eff = [&](const size_t il){
size_t ret = 0;
const bool il_is_recurrent = hparams.is_recurrent(il);
const bool il_is_swa = hparams.is_swa(il);
const bool il_is_recr = hparams.is_recr(il);
const bool il_is_swa = hparams.is_swa(il);
for (size_t il_prev = 0; il_prev < il; il_prev++) {
ret += hparams.is_recurrent(il_prev) == il_is_recurrent && hparams.is_swa(il_prev) == il_is_swa;
ret += hparams.is_recr(il_prev) == il_is_recr && hparams.is_swa(il_prev) == il_is_swa;
}
return ret;
};
@@ -553,7 +553,7 @@ struct ggml_backend_meta_split_state llama_meta_device_get_split_state(const str
};
auto get_split_granularity = [&](int64_t blck_size, uint32_t il, const std::vector<std::pair<int64_t, uint32_t>> & segments) -> std::vector<int64_t> {
if (hparams.is_recurrent(il)) {
if (hparams.is_recr(il)) {
// linear attention
const int64_t head_dim = hparams.ssm_d_state;
const int64_t granularity_qkv = std::lcm(blck_size, head_dim);
@@ -1076,18 +1076,16 @@ void llama_model_base::load_hparams(llama_model_loader & ml) {
std::fill(hparams.n_head_arr.begin(), hparams.n_head_arr.end(), 0);
std::fill(hparams.n_head_kv_arr.begin(), hparams.n_head_kv_arr.end(), 0);
std::fill(hparams.n_ff_arr.begin(), hparams.n_ff_arr.end(), 0);
std::fill(
hparams.recurrent_layer_arr.begin(),
hparams.recurrent_layer_arr.end(),
llm_arch_is_recurrent(ml.get_arch()));
std::fill(hparams.rope_sections.begin(), hparams.rope_sections.end(), 0);
std::fill(hparams.swa_layers.begin(), hparams.swa_layers.end(), 0);
std::fill(hparams.is_swa_impl.begin(), hparams.is_swa_impl.end(), 0);
std::fill(hparams.is_recr_impl.begin(), hparams.is_recr_impl.end(), llm_arch_is_recurrent(ml.get_arch()) ? 1 : 0);
std::fill(hparams.xielu_alpha_n.begin(), hparams.xielu_alpha_n.end(), 0.0f);
std::fill(hparams.xielu_alpha_p.begin(), hparams.xielu_alpha_p.end(), 0.0f);
std::fill(hparams.xielu_beta.begin(), hparams.xielu_beta.end(), 0.0f);
std::fill(hparams.xielu_eps.begin(), hparams.xielu_eps.end(), 0.0f);
std::fill(hparams.xielu_beta.begin(), hparams.xielu_beta.end(), 0.0f);
std::fill(hparams.xielu_eps.begin(), hparams.xielu_eps.end(), 0.0f);
std::fill(hparams.swiglu_clamp_exp.begin(), hparams.swiglu_clamp_exp.end(), 0.0f);
std::fill(hparams.swiglu_clamp_shexp.begin(), hparams.swiglu_clamp_shexp.end(), 0.0f);
@@ -2040,18 +2038,18 @@ llama_memory_i * llama_model::create_memory(const llama_memory_params & params,
filter_recr = [&](int32_t) { return true; };
} else if (arch == LLM_ARCH_NEMOTRON_H || arch == LLM_ARCH_NEMOTRON_H_MOE) {
filter_attn = [&](int32_t il) {
return !hparams.is_recurrent(il) && hparams.n_ff(il) == 0;
return !hparams.is_recr(il) && hparams.n_ff(il) == 0;
};
filter_recr = [&](int32_t il) {
return hparams.is_recurrent(il) && hparams.n_ff(il) == 0;
return hparams.is_recr(il) && hparams.n_ff(il) == 0;
};
} else if (arch == LLM_ARCH_QWEN35 || arch == LLM_ARCH_QWEN35MOE) {
const uint32_t n_main = hparams.n_layer - hparams.nextn_predict_layers;
filter_attn = [&, n_main](int32_t il) {
return (uint32_t)il < n_main && !hparams.is_recurrent(il);
return (uint32_t)il < n_main && !hparams.is_recr(il);
};
filter_recr = [&, n_main](int32_t il) {
return (uint32_t)il < n_main && hparams.is_recurrent(il);
return (uint32_t)il < n_main && hparams.is_recr(il);
};
}