happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

common : fix split model migration (#21019) 

Sadly the manifest does not list all required files, i honestly thought
it was the case

Without the files listed we don't have the sha256, so if the first file
is valid, and all others have the correct size, then we can assume we
are good and do the migration...

Here my test:

    $ find /home/angt/.cache/llama.cpp
    /home/angt/.cache/llama.cpp
    /home/angt/.cache/llama.cpp/angt_test-split-model-stories260K_stories260K-f32-00002-of-00002.gguf
    /home/angt/.cache/llama.cpp/angt_test-split-model-stories260K_stories260K-f32-00001-of-00002.gguf
    /home/angt/.cache/llama.cpp/angt_test-split-model-stories260K_stories260K-f32-00001-of-00002.gguf.etag
    /home/angt/.cache/llama.cpp/angt_test-split-model-stories260K_stories260K-f32-00002-of-00002.gguf.etag
    /home/angt/.cache/llama.cpp/manifest=angt=test-split-model-stories260K=latest.json

    $ build/bin/llama-server
    ================================================================================
    WARNING: Migrating cache to HuggingFace cache directory
      Old cache: /home/angt/.cache/llama.cpp/
      New cache: /home/angt/.cache/huggingface/hub
    This one-time migration moves models previously downloaded with -hf
    from the legacy llama.cpp cache to the standard HuggingFace cache.
    Models downloaded with --model-url are not affected.
    ================================================================================
    migrate_file: migrated angt_test-split-model-stories260K_stories260K-f32-00001-of-00002.gguf -> /home/angt/.cache/huggingface/hub/models--angt--test-split-model-stories260K/snapshots/68c3ea2061e8c7688455fab07597dde0f4d7f0db/stories260K-f32-00001-of-00002.gguf
    migrate_file: migrated angt_test-split-model-stories260K_stories260K-f32-00002-of-00002.gguf -> /home/angt/.cache/huggingface/hub/models--angt--test-split-model-stories260K/snapshots/68c3ea2061e8c7688455fab07597dde0f4d7f0db/stories260K-f32-00002-of-00002.gguf
    migrate_old_cache_to_hf_cache: migration complete, deleting manifest: /home/angt/.cache/llama.cpp/manifest=angt=test-split-model-stories260K=latest.json

    $ find /home/angt/.cache/llama.cpp /home/angt/.cache/huggingface
    /home/angt/.cache/llama.cpp
    /home/angt/.cache/huggingface
    /home/angt/.cache/huggingface/hub
    /home/angt/.cache/huggingface/hub/models--angt--test-split-model-stories260K
    /home/angt/.cache/huggingface/hub/models--angt--test-split-model-stories260K/blobs
    /home/angt/.cache/huggingface/hub/models--angt--test-split-model-stories260K/blobs/50d019817c2626eb9e8a41f361ff5bfa538757e6f708a3076cd3356354a75694
    /home/angt/.cache/huggingface/hub/models--angt--test-split-model-stories260K/blobs/7b273e1dbfab11dc67dce479deb5923fef27c39cbf56a20b3a928a47b77dab3c
    /home/angt/.cache/huggingface/hub/models--angt--test-split-model-stories260K/refs
    /home/angt/.cache/huggingface/hub/models--angt--test-split-model-stories260K/refs/main
    /home/angt/.cache/huggingface/hub/models--angt--test-split-model-stories260K/snapshots
    /home/angt/.cache/huggingface/hub/models--angt--test-split-model-stories260K/snapshots/68c3ea2061e8c7688455fab07597dde0f4d7f0db
    /home/angt/.cache/huggingface/hub/models--angt--test-split-model-stories260K/snapshots/68c3ea2061e8c7688455fab07597dde0f4d7f0db/stories260K-f32-00002-of-00002.gguf
    /home/angt/.cache/huggingface/hub/models--angt--test-split-model-stories260K/snapshots/68c3ea2061e8c7688455fab07597dde0f4d7f0db/stories260K-f32-00001-of-00002.gguf

Signed-off-by: Adrien Gallouët <angt@huggingface.co>
This commit is contained in:
Adrien Gallouët 2026-03-26 12:04:37 +01:00 committed by GitHub
parent 112c78159f
commit 3cba8bba18
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 163 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

197
common/hf-cache.cpp
View File

@ -325,9 +325,15 @@ hf_files get_repo_files(const std::string & repo_id,
if (item["lfs"].contains("oid") && item["lfs"]["oid"].is_string()) {
file.oid = item["lfs"]["oid"].get<std::string>();
}
if (item["lfs"].contains("size") && item["lfs"]["size"].is_number()) {
file.size = item["lfs"]["size"].get<size_t>();
}
} else if (item.contains("oid") && item["oid"].is_string()) {
file.oid = item["oid"].get<std::string>();
}
if (file.size == 0 && item.contains("size") && item["size"].is_number()) {
file.size = item["size"].get<size_t>();
}
if (!file.oid.empty() && !is_valid_oid(file.oid)) {
LOG_WRN("%s: skip invalid oid: %s\n", __func__, file.oid.c_str());
@ -487,6 +493,34 @@ std::string finalize_file(const hf_file & file) {
// delete everything after this line, one day
// copied from download.cpp without the tag part
struct gguf_split_info {
std::string prefix; // tag included
int index;
int count;
};
static gguf_split_info get_gguf_split_info(const std::string & path) {
static const std::regex re_split("^(.+)-([0-9]{5})-of-([0-9]{5})$", std::regex::icase);
std::smatch m;
std::string prefix = path;
if (!string_remove_suffix(prefix, ".gguf")) {
return {};
}
int index = 1;
int count = 1;
if (std::regex_match(prefix, m, re_split)) {
index = std::stoi(m[2].str());
count = std::stoi(m[3].str());
prefix = m[1].str();
}
return {std::move(prefix), index, count};
}
static std::pair<std::string, std::string> parse_manifest_name(std::string & filename) {
static const std::regex re(R"(^manifest=([^=]+)=([^=]+)=.*\.json$)");
std::smatch match;
@ -504,25 +538,30 @@ static std::string make_old_cache_filename(const std::string & owner,
return result;
}
static void migrate_single_file(const fs::path & old_cache,
const std::string & owner,
const std::string & repo,
const nl::json & node,
const hf_files & files) {
struct migrate_file {
std::string path;
std::string sha256;
size_t size;
fs::path old_path;
fs::path etag_path;
const hf_file * file;
};
if (!node.contains("rfilename") ||
!node.contains("lfs") ||
!node["lfs"].contains("sha256")) {
return;
}
using migrate_files = std::vector<migrate_file>;
std::string path = node["rfilename"];
std::string sha256 = node["lfs"]["sha256"];
static bool collect_file(const fs::path & old_cache,
const std::string & owner,
const std::string & repo,
const std::string & path,
const std::string & sha256,
const hf_files & files,
migrate_files & to_migrate) {
const hf_file * file = nullptr;
const hf_file * file_info = nullptr;
for (const auto & f : files) {
if (f.path == path) {
file_info = &f;
file = &f;
break;
}
}
@ -532,41 +571,105 @@ static void migrate_single_file(const fs::path & old_cache,
fs::path etag_path = old_path.string() + ".etag";
if (!fs::exists(old_path)) {
if (fs::exists(etag_path)) {
LOG_WRN("%s: %s is orphan, deleting...\n", __func__, etag_path.string().c_str());
fs::remove(etag_path);
if (file && fs::exists(file->final_path)) {
return true;
}
return;
LOG_WRN("%s: %s not found in old cache or HF cache\n", __func__, old_filename.c_str());
return false;
}
if (!file_info) {
LOG_WRN("%s: %s not found in current repo, ignoring...\n", __func__, old_filename.c_str());
return;
} else if (!sha256.empty() && !file_info->oid.empty() && sha256 != file_info->oid) {
LOG_WRN("%s: %s is not up to date (sha256 mismatch), ignoring...\n", __func__, old_filename.c_str());
return;
if (!file) {
LOG_WRN("%s: %s not found in current repo\n", __func__, old_filename.c_str());
return false;
}
if (!sha256.empty() && !file->oid.empty() && sha256 != file->oid) {
LOG_WRN("%s: %s is not up to date (sha256 mismatch)\n", __func__, old_filename.c_str());
return false;
}
if (file->size > 0) {
size_t size = fs::file_size(old_path);
if (size != file->size) {
LOG_WRN("%s: %s has wrong size %zu (expected %zu)\n", __func__, old_filename.c_str(), size, file->size);
return false;
}
}
to_migrate.push_back({path, sha256, file->size, old_path, etag_path, file});
return true;
}
static bool collect_files(const fs::path & old_cache,
const std::string & owner,
const std::string & repo,
const nl::json & node,
const hf_files & files,
migrate_files & to_migrate) {
if (!node.contains("rfilename") ||
!node.contains("lfs") ||
!node["lfs"].contains("sha256")) {
return true;
}
std::string path = node["rfilename"];
std::string sha256 = node["lfs"]["sha256"];
auto split = get_gguf_split_info(path);
if (split.count <= 1) {
return collect_file(old_cache, owner, repo, path, sha256, files, to_migrate);
}
std::vector<std::pair<std::string, std::string>> splits;
for (const auto & f : files) {
auto split_f = get_gguf_split_info(f.path);
if (split_f.count == split.count && split_f.prefix == split.prefix) {
// sadly the manifest only provides the sha256 of the first file (index == 1)
// the rest will be verified using the size...
std::string f_sha256 = (split_f.index == 1) ? sha256 : "";
splits.emplace_back(f.path, f_sha256);
}
}
if ((int)splits.size() != split.count) {
LOG_WRN("%s: expected %d split files but found %d in repo\n", __func__, split.count, (int)splits.size());
return false;
}
for (const auto & [f_path, f_sha256] : splits) {
if (!collect_file(old_cache, owner, repo, f_path, f_sha256, files, to_migrate)) {
return false;
}
}
return true;
}
static bool migrate_file(const migrate_file & file) {
std::error_code ec;
fs::path new_path(file_info->local_path);
fs::path new_path(file.file->local_path);
fs::create_directories(new_path.parent_path(), ec);
if (!fs::exists(new_path, ec)) {
fs::rename(old_path, new_path, ec);
fs::rename(file.old_path, new_path, ec);
if (ec) {
fs::copy_file(old_path, new_path, ec);
fs::copy_file(file.old_path, new_path, ec);
if (ec) {
LOG_WRN("%s: failed to move/copy %s: %s\n", __func__, old_path.string().c_str(), ec.message().c_str());
return;
LOG_ERR("%s: failed to move/copy %s: %s\n", __func__, file.old_path.string().c_str(), ec.message().c_str());
return false;
}
}
fs::remove(old_path, ec);
fs::remove(file.old_path, ec);
}
fs::remove(etag_path, ec);
fs::remove(file.etag_path, ec);
std::string filename = finalize_file(*file_info);
LOG_INF("%s: migrated %s -> %s\n", __func__, old_filename.c_str(), filename.c_str());
std::string filename = finalize_file(*file.file);
LOG_INF("%s: migrated %s -> %s\n", __func__, file.old_path.filename().string().c_str(), filename.c_str());
return true;
}
void migrate_old_cache_to_hf_cache(const std::string & token, bool offline) {
@ -614,19 +717,43 @@ void migrate_old_cache_to_hf_cache(const std::string & token, bool offline) {
continue;
}
migrate_files to_migrate;
bool ok = true;
try {
std::ifstream manifest(entry.path());
auto json = nl::json::parse(manifest);
for (const char * key : {"ggufFile", "mmprojFile"}) {
if (json.contains(key)) {
migrate_single_file(old_cache, owner, repo, json[key], files);
if (!collect_files(old_cache, owner, repo, json[key], files, to_migrate)) {
ok = false;
break;
}
}
}
} catch (const std::exception & e) {
LOG_WRN("%s: failed to parse manifest %s: %s\n", __func__, filename.c_str(), e.what());
continue;
}
if (!ok) {
LOG_WRN("%s: migration skipped: one or more files failed validation\n", __func__);
continue;
}
for (const auto & file : to_migrate) {
if (!migrate_file(file)) {
ok = false;
break;
}
}
if (!ok) {
LOG_WRN("%s: migration failed: could not migrate all files\n", __func__);
continue;
}
LOG_INF("%s: migration complete, deleting manifest: %s\n", __func__, entry.path().string().c_str());
fs::remove(entry.path());
}
}

1
common/hf-cache.h
View File

@ -14,6 +14,7 @@ struct hf_file {
std::string final_path;
std::string oid;
std::string repo_id;
size_t size = 0; // only for the migration
};
using hf_files = std::vector<hf_file>;