From: Molly Sophia <redacted>
Date: Fri, 3 Jan 2025 12:13:18 +0000 (+0800)
Subject: common : disable KV cache shifting automatically for unsupported models (#11053)
X-Git-Tag: upstream/0.0.4488~78
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=4b0c638b9a68f577cb2066b638c9f622d91ee661;p=pkg%2Fggml%2Fsources%2Fllama.cpp

common : disable KV cache shifting automatically for unsupported models (#11053)

* Disable KV cache shifting automatically for unsupported models

instead of exiting directly

Signed-off-by: Molly Sophia <redacted>

* Update common/common.cpp

Co-authored-by: Georgi Gerganov <redacted>

---------

Signed-off-by: Molly Sophia <redacted>
Co-authored-by: Georgi Gerganov <redacted>
---

diff --git a/common/common.cpp b/common/common.cpp
index 3e37039c..4bb140ee 100644
--- a/common/common.cpp
+++ b/common/common.cpp
@@ -889,9 +889,8 @@ struct common_init_result common_init_from_params(common_params & params) {
     }
 
     if (params.ctx_shift && !llama_kv_cache_can_shift(lctx)) {
-        LOG_ERR("%s: KV cache shifting is not supported for this model (--no-context-shift to disable)'\n", __func__);
-        llama_free_model(model);
-        return iparams;
+        LOG_WRN("%s: KV cache shifting is not supported for this model, disabling KV cache shifting\n", __func__);
+        params.ctx_shift = false;
     }
 
     if (!params.control_vectors.empty()) {