vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-08-21 22:35:48 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fixed dflash

Browse source
This commit is contained in:
Concedo 2026-08-14 22:38:55 +08:00
parent 22806c84d3
commit 531cac9cd3
1 changed files with 15 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

15
gpttype_adapter.cpp
View file

@ -279,6 +279,18 @@ static bool speculative_draft_type_needs_preprocess_kv_rollback(common_speculati
|| type == COMMON_SPECULATIVE_TYPE_DRAFT_DSPARK;
}
static void speculative_apply_block_draft_output_limits(llama_context_params & ctx_params, common_speculative_type type)
{
if(type != COMMON_SPECULATIVE_TYPE_DRAFT_DFLASH && type != COMMON_SPECULATIVE_TYPE_DRAFT_DSPARK)
{
return;
}
const uint32_t per_seq = std::max<uint32_t>(1, speculative_chunk_amt + 1);
ctx_params.n_outputs_max = std::max<uint32_t>(ctx_params.n_outputs_max, per_seq);
ctx_params.n_outputs_max_per_seq = std::max<uint32_t>(ctx_params.n_outputs_max_per_seq, per_seq);
}
static inline void string_trim_whitespace(std::string & s) {
auto nul = std::find(s.begin(), s.end(), '\0'); //remove everything after the first NUL
if (nul != s.end()) {
@ -658,6 +670,7 @@ static size_t estimate_draft_autofit_tax_mb(
draft_ctx_params.n_outputs_max = std::max<uint32_t>(1, base_ctx_params.n_seq_max); //match the real MTP draft context so the autofit tax doesn't over-reserve the draft compute buffer at n_batch*n_vocab (~2GB on large-vocab models like Gemma)
measure_model_bytes = has_draft_model;
}
speculative_apply_block_draft_output_limits(draft_ctx_params, draft_spec_type_estimate);
std::vector<ggml_backend_dev_t> devs;
uint32_t hp_ngl = 0;
@ -1020,11 +1033,13 @@ static void speculative_decoding_setup(std::string spec_model_filename, llama_co
{
printf("Detected DSpark draft model, using llama.cpp DSpark speculative decoding.\n");
draft_ctx_params.ctx_other = main_ctx;
speculative_apply_block_draft_output_limits(draft_ctx_params, draft_spec_type);
}
else if(draft_spec_type == COMMON_SPECULATIVE_TYPE_DRAFT_DFLASH)
{
printf("Detected DFlash draft model, using llama.cpp DFlash speculative decoding.\n");
draft_ctx_params.ctx_other = main_ctx;
speculative_apply_block_draft_output_limits(draft_ctx_params, draft_spec_type);
}
draft_ctx = llama_init_from_model(draftmodel, draft_ctx_params);
if(draft_ctx == NULL)