Skip to content

Commit e08c44c

Browse files
authored
Remove setting hardcoded value for n_seq_max in IContextParamsExtensions
It's already set from passed argument `result.n_seq_max = @params.SeqMax;` Overwriting it with hardcoded value is removed result.n_seq_max = Math.Min(Math.Max(10U, result.n_ctx / 8U), 64U /*0x40*/); Ref: SciSharp#1353
1 parent b368e13 commit e08c44c

1 file changed

Lines changed: 1 addition & 2 deletions

File tree

LLama/Extensions/IContextParamsExtensions.cs

Lines changed: 1 addition & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -58,8 +58,7 @@ public static void ToLlamaContextParams(this IContextParams @params, out LLamaCo
5858
null => LLamaFlashAttentionType.LLAMA_FLASH_ATTENTION_TYPE_AUTO
5959
};
6060
result.kv_unified = true;
61-
result.n_seq_max = (uint)Math.Min(Math.Max(10, result.n_ctx / 8), 64);
62-
61+
6362
result.n_threads = Threads(@params.Threads);
6463
result.n_threads_batch = Threads(@params.BatchThreads);
6564

0 commit comments

Comments
 (0)