context : fix SWA-related warning for multiple sequences (#14045)

ggerganov · web-flow · commit 487a5e040142 · 2025-06-06T13:29:18.000+03:00
diff --git a/src/llama-context.cpp b/src/llama-context.cpp
@@ -123,7 +123,7 @@ llama_context::llama_context(
                 __func__, n_ctx_per_seq, hparams.n_ctx_train);
     }
 
-    if (!params.swa_full && cparams.n_seq_max > 1) {
+    if (!params.swa_full && cparams.n_seq_max > 1 && hparams.is_swa_any()) {
         LLAMA_LOG_WARN("%s: requested n_seq_max (%u) > 1, but swa_full is not enabled -- performance may be degraded: %s\n",
                 __func__, cparams.n_seq_max, "https://github.com/ggml-org/llama.cpp/pull/13845#issuecomment-2924800573");
     }

Original file line number	Diff line number	Diff line change
`@@ -123,7 +123,7 @@ llama_context::llama_context(`
`123`	`123`	`__func__, n_ctx_per_seq, hparams.n_ctx_train);`
`124`	`124`	`}`
`125`	`125`
`126`		`- if (!params.swa_full && cparams.n_seq_max > 1) {`
	`126`	`+ if (!params.swa_full && cparams.n_seq_max > 1 && hparams.is_swa_any()) {`
`127`	`127`	`LLAMA_LOG_WARN("%s: requested n_seq_max (%u) > 1, but swa_full is not enabled -- performance may be degraded: %s\n",`
`128`	`128`	`__func__, cparams.n_seq_max, "https://github.com/ggml-org/llama.cpp/pull/13845#issuecomment-2924800573");`
`129`	`129`	`}`