Merge branch 'upstream' into concedo_experimental

# 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

View file

@ -240,6 +240,7 @@ int main(int argc, char ** argv) {
LOG_TEE("%s\n", get_system_info(params).c_str());
}
const bool add_bos = llama_should_add_bos_token(model);
GGML_ASSERT(llama_add_eos_token(model) != 1);
LOG("add_bos: %d\n", add_bos);
bool suff_rm_leading_spc = params.escape;
@ -280,10 +281,10 @@ int main(int argc, char ** argv) {
if (ctx_guidance) {
LOG("cfg_negative_prompt: \"%s\"\n", log_tostr(sparams.cfg_negative_prompt));
guidance_inp = ::llama_tokenize(ctx_guidance, sparams.cfg_negative_prompt, add_bos);
guidance_inp = ::llama_tokenize(ctx_guidance, sparams.cfg_negative_prompt, true);
LOG("guidance_inp tokenized: %s\n", LOG_TOKENS_TOSTR_PRETTY(ctx_guidance, guidance_inp).c_str());
std::vector<llama_token> original_inp = ::llama_tokenize(ctx, params.prompt, add_bos);
std::vector<llama_token> original_inp = ::llama_tokenize(ctx, params.prompt, true);
LOG("original_inp tokenized: %s\n", LOG_TOKENS_TOSTR_PRETTY(ctx, original_inp).c_str());
original_prompt_len = original_inp.size();