From c2a08f87b8d180115d04b8688f383d1b2761b16d Mon Sep 17 00:00:00 2001 From: anon998 <131767832+anon998@users.noreply.github.com> Date: Sun, 25 Jun 2023 08:48:36 +0000 Subject: [PATCH] fix server sampling: top k sampler first (#1977) Co-authored-by: anon --- examples/server/server.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/examples/server/server.cpp b/examples/server/server.cpp index de22d3013..79df5e847 100644 --- a/examples/server/server.cpp +++ b/examples/server/server.cpp @@ -325,10 +325,10 @@ struct llama_server_context { id = llama_sample_token_mirostat_v2(ctx, &candidates_p, mirostat_tau, mirostat_eta, &mirostat_mu); } else { // Temperature sampling + llama_sample_top_k(ctx, &candidates_p, top_k, 1); llama_sample_tail_free(ctx, &candidates_p, tfs_z, 1); llama_sample_typical(ctx, &candidates_p, typical_p, 1); llama_sample_top_p(ctx, &candidates_p, top_p, 1); - llama_sample_top_k(ctx, &candidates_p, top_k, 1); llama_sample_temperature(ctx, &candidates_p, temp); id = llama_sample_token(ctx, &candidates_p); }