Skip to content

spec : save the dynamic ngram cache file #22055

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
petersid2022 wants to merge 1 commit into
base: master
Choose a base branch
from
+21 −21
Open

spec : save the dynamic ngram cache file #22055

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
42 changes: 21 additions & 21 deletions common/speculative.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,7 @@
#include <cassert>
#include <cmath>
#include <cstring>
#include <fstream>
#include <iomanip>
#include <map>
#include <cinttypes>
Expand Down Expand Up @@ -2080,9 +2081,6 @@ struct common_speculative_impl_ngram_cache : public common_speculative_impl {

uint16_t n_draft;

bool save_dynamic;
bool save_static;

struct seq_info {
size_t cache_size = 0; // number of tokens in n-gram cache

Expand All @@ -2098,14 +2096,10 @@ struct common_speculative_impl_ngram_cache : public common_speculative_impl {
uint32_t n_seq,
uint16_t n_draft,
const std::string & path_static,
const std::string & path_dynamic,
bool save_dynamic,
bool save_static)
const std::string & path_dynamic)
: common_speculative_impl(COMMON_SPECULATIVE_TYPE_NGRAM_CACHE, n_seq, n_draft)
, params(params.ngram_cache)
, n_draft(n_draft)
, save_dynamic(save_dynamic)
, save_static(save_static)
{
SPC_TRC("%s", "adding speculative implementation 'ngram-cache'\n");
SPC_TRC("- n_draft=%d, cache_static=%s, cache_dynamic=%s\n",
Expand All @@ -2122,7 +2116,7 @@ struct common_speculative_impl_ngram_cache : public common_speculative_impl {
for (auto & sinfo : sinfos) {
sinfo.ngram_cache_static = ngram_cache_static;
}
} catch (...) {
} catch (std::ifstream::failure const &) {
SPC_ERR("failed to open static lookup cache: %s", path_static.c_str());
GGML_ABORT("Couldn't read static lookup cache");
}
Expand All @@ -2135,13 +2129,25 @@ struct common_speculative_impl_ngram_cache : public common_speculative_impl {
for (auto & sinfo : sinfos) {
sinfo.ngram_cache_dynamic = ngram_cache_dynamic;
}
} catch (...) {
SPC_ERR("failed to open dynamic lookup cache: %s", path_dynamic.c_str());
GGML_ABORT("Couldn't read dynamic lookup cache");
} catch (std::ifstream::failure const &) {
// missing file is created on destruction
}
}
}

~common_speculative_impl_ngram_cache() override {
if (params.lookup_cache_dynamic.empty() || sinfos.empty()) {
return;
}

auto & dynamic_cache = sinfos.front().ngram_cache_dynamic;
for (auto & sinfo : sinfos) {
common_ngram_cache_merge(dynamic_cache, sinfo.ngram_cache_context);
}

common_ngram_cache_save(dynamic_cache, params.lookup_cache_dynamic);
}

void begin(llama_seq_id /*seq_id*/, const llama_tokens & /*prompt*/) override {
// noop
}
Expand Down Expand Up @@ -2237,20 +2243,14 @@ static common_ngram_map get_common_ngram_map(
return common_ngram_map(size_key, size_value, key_only, min_hits);
}

static common_speculative_impl_ngram_cache create_state_ngram_cache(
static std::unique_ptr<common_speculative_impl_ngram_cache> create_state_ngram_cache(
const common_speculative_config & config,
uint32_t n_seq,
const std::string & path_static,
const std::string & path_dynamic) {
uint16_t n_draft = 8; // TODO get from config?

// TODO bool param in common/common.h to set save_static/save_dynamic?
bool save_static = false;
bool save_dynamic = false;

common_speculative_impl_ngram_cache state(config.params, n_seq, n_draft, path_static, path_dynamic, save_static, save_dynamic);

return state;
return std::make_unique<common_speculative_impl_ngram_cache>(config.params, n_seq, n_draft, path_static, path_dynamic);
}

std::string common_speculative_type_name_str(const std::vector<common_speculative_type> & types) {
Expand Down Expand Up @@ -2736,7 +2736,7 @@ common_speculative * common_speculative_init(common_params_speculative & params,
config, n_seq,
params.ngram_cache.lookup_cache_static,
params.ngram_cache.lookup_cache_dynamic);
impls.push_back(std::make_unique<common_speculative_impl_ngram_cache>(state));
impls.push_back(std::move(state));
break;
}
default:
Expand Down