ci: server: fix server tests after ggml-org#6638

phymbert · phymbert · commit 658451be8bf7 · 2024-04-26T21:44:18.000+02:00
diff --git a/examples/server/server.cpp b/examples/server/server.cpp
@@ -1208,7 +1208,7 @@ struct server_context {
         }
 
         auto n_ctx_train = llama_n_ctx_train(model);
-        if (slot.params.n_predict < 1 && slot.ga_n == 1
+        if (slot.params.n_predict < 1 && slot.n_predict < 1 && slot.ga_n == 1
                     && slot.n_prompt_tokens + slot.n_decoded >= n_ctx_train) {
             LOG_WARNING("n_predict is not set and self-context extend is disabled."
                         " Limiting generated tokens to n_ctx_train to avoid EOS-less generation infinite loop", {

Original file line number	Diff line number	Diff line change
`@@ -1208,7 +1208,7 @@ struct server_context {`
`1208`	`1208`	`}`
`1209`	`1209`
`1210`	`1210`	`auto n_ctx_train = llama_n_ctx_train(model);`
`1211`		`- if (slot.params.n_predict < 1 && slot.ga_n == 1`
	`1211`	`+ if (slot.params.n_predict < 1 && slot.n_predict < 1 && slot.ga_n == 1`
`1212`	`1212`	`&& slot.n_prompt_tokens + slot.n_decoded >= n_ctx_train) {`
`1213`	`1213`	`LOG_WARNING("n_predict is not set and self-context extend is disabled."`
`1214`	`1214`	`" Limiting generated tokens to n_ctx_train to avoid EOS-less generation infinite loop", {`