llama : do not allocate KV cache for "vocab_only == true" (#682)

Fixes sanitizer CI
2023-04-02 07:18:53 +00:00 · 2023-04-02 07:18:53 +00:00 · 81040f10aa
parent c4f89d8d73
commit 81040f10aa
1 changed files with 1 additions and 1 deletions
--- a/llama.cpp
+++ b/llama.cpp
@ -1608,7 +1608,7 @@ struct llama_context * llama_init_from_file(
    }

    // reserve memory for context buffers
-    {
+    if (!params.vocab_only) {
        if (!kv_cache_init(ctx->model.hparams, ctx->model.kv_self, memory_type, ctx->model.hparams.n_ctx)) {
            fprintf(stderr, "%s: kv_cache_init() failed for self-attention cache\n", __func__);
            llama_free(ctx);