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

* tests : add support for qwen3 SSM archs * arch : add LLM_KV_ATTENTION_RECURRENT_LAYERS * cont : naming + TODOs
2026-06-03 10:15:27 +03:00
parent d545a2a993
commit 06938ac129
25 changed files with 109 additions and 83 deletions
@@ -15,7 +15,8 @@ void llama_model_llama4::load_arch_hparams(llama_model_loader & ml) {
        hparams.n_attn_temp_floor_scale = 8192;
        hparams.f_attn_temp_scale       = 0.1f;
        hparams.f_attn_temp_offset      = 1.0f;
-        uint32_t swa_period             = 4; // pattern: 3 chunked - 1 full
+
+        uint32_t swa_period = 4; // pattern: 3 chunked - 1 full
        ml.get_key_or_arr(LLM_KV_ATTENTION_SLIDING_WINDOW_PATTERN, swa_period, false);
        hparams.set_swa_pattern(swa_period);