mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-09-08 22:09:02 +02:00
spec : refactor
This commit is contained in:
@@ -3438,6 +3438,7 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
|
||||
if (value == "none") {
|
||||
params.speculative.draftless_type = COMMON_SPECULATIVE_TYPE_NONE;
|
||||
} else if (value == "ngram-cache") {
|
||||
// TODO: this does nothing atm
|
||||
params.speculative.draftless_type = COMMON_SPECULATIVE_TYPE_NGRAM_CACHE;
|
||||
} else if (value == "ngram-simple") {
|
||||
params.speculative.draftless_type = COMMON_SPECULATIVE_TYPE_NGRAM_SIMPLE;
|
||||
|
||||
Reference in New Issue
Block a user