vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-16 20:09:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'upstream' into concedo_experimental

Browse source 
# Conflicts:
#	.github/workflows/build.yml
#	.github/workflows/docker.yml
#	Makefile
#	README-sycl.md
#	README.md
#	ci/run.sh
#	ggml-cuda.cu
#	ggml.c
#	grammars/README.md
#	scripts/get-wikitext-2.sh
#	scripts/hf.sh
#	scripts/sync-ggml.last
#	tests/test-backend-ops.cpp
#	tests/test-grammar-integration.cpp
#	tests/test-json-schema-to-grammar.cpp
This commit is contained in:
Concedo 2024-04-14 21:18:39 +08:00
commit 9a25d77cc1
58 changed files with 6529 additions and 6121 deletions
Download patch file Download diff file Expand all files Collapse all files

12
common/common.h
View file

@ -76,6 +76,9 @@ struct gpt_params {
int32_t yarn_orig_ctx = 0; // YaRN original context length
float defrag_thold = -1.0f; // KV cache defragmentation threshold
ggml_backend_sched_eval_callback cb_eval = nullptr;
void * cb_eval_user_data = nullptr;
ggml_numa_strategy numa = GGML_NUMA_STRATEGY_DISABLED;
llama_rope_scaling_type rope_scaling_type = LLAMA_ROPE_SCALING_TYPE_UNSPECIFIED;
@ -172,6 +175,7 @@ struct gpt_params {
bool infill = false; // use infill mode
bool dump_kv_cache = false; // dump the KV cache contents for debugging purposes
bool no_kv_offload = false; // disable KV offloading
bool warmup = true; // warmup run
std::string cache_type_k = "f16"; // KV cache data type for the K
std::string cache_type_v = "f16"; // KV cache data type for the V
@ -239,14 +243,14 @@ void llama_batch_add(
std::vector<llama_token> llama_tokenize(
const struct llama_context * ctx,
const std::string & text,
bool add_bos,
bool special = false);
bool add_special,
bool parse_special = false);
std::vector<llama_token> llama_tokenize(
const struct llama_model * model,
const std::string & text,
bool add_bos,
bool special = false);
bool add_special,
bool parse_special = false);
// tokenizes a token into a piece
// should work similar to Python's `tokenizer.id_to_piece`