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

Merge branch 'master' into HEAD

This commit is contained in:
Georgi Gerganov 2025-12-11 14:42:56 +02:00
parent ab65b47a52 c6f6e4f96a
commit 4d10b78e23
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
29 changed files with 979 additions and 331 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
.github/workflows/build.yml vendored
View File

@ -1770,7 +1770,7 @@ jobs:
echo "Fetch llama2c model" echo "Fetch llama2c model"
wget https://huggingface.co/karpathy/tinyllamas/resolve/main/stories260K/stories260K.bin wget https://huggingface.co/karpathy/tinyllamas/resolve/main/stories260K/stories260K.bin
./bin/llama-convert-llama2c-to-ggml --copy-vocab-from-model ./tok512.bin --llama2c-model stories260K.bin --llama2c-output-model stories260K.gguf ./bin/llama-convert-llama2c-to-ggml --copy-vocab-from-model ./tok512.bin --llama2c-model stories260K.bin --llama2c-output-model stories260K.gguf
./bin/llama-cli -m stories260K.gguf -p "One day, Lily met a Shoggoth" -n 500 -c 256 ./bin/llama-completion -m stories260K.gguf -p "One day, Lily met a Shoggoth" -n 500 -c 256
ubuntu-cmake-sanitizer-riscv64-native: ubuntu-cmake-sanitizer-riscv64-native:
runs-on: RISCV64 runs-on: RISCV64

2
common/CMakeLists.txt
View File

@ -73,6 +73,8 @@ add_library(${TARGET} STATIC
ngram-cache.h ngram-cache.h
peg-parser.cpp peg-parser.cpp
peg-parser.h peg-parser.h
preset.cpp
preset.h
regex-partial.cpp regex-partial.cpp
regex-partial.h regex-partial.h
sampling.cpp sampling.cpp

82
common/arg.cpp
View File

@ -47,6 +47,7 @@
#define LLAMA_MAX_URL_LENGTH 2084 // Maximum URL Length in Chrome: 2083 #define LLAMA_MAX_URL_LENGTH 2084 // Maximum URL Length in Chrome: 2083
using json = nlohmann::ordered_json; using json = nlohmann::ordered_json;
using namespace common_arg_utils;
static std::initializer_list<enum llama_example> mmproj_examples = { static std::initializer_list<enum llama_example> mmproj_examples = {
LLAMA_EXAMPLE_MTMD, LLAMA_EXAMPLE_MTMD,
@ -64,6 +65,15 @@ static std::string read_file(const std::string & fname) {
return content; return content;
} }
static const std::vector<common_arg> & get_common_arg_defs() {
static const std::vector<common_arg> options = [] {
common_params params;
auto ctx = common_params_parser_init(params, LLAMA_EXAMPLE_SERVER, nullptr);
return ctx.options;
}();
return options;
}
common_arg & common_arg::set_examples(std::initializer_list<enum llama_example> examples) { common_arg & common_arg::set_examples(std::initializer_list<enum llama_example> examples) {
this->examples = examples; this->examples = examples;
return *this; return *this;
@ -134,7 +144,7 @@ static std::vector<std::string> break_str_into_lines(std::string input, size_t m
return result; return result;
} }
std::string common_arg::to_string() { std::string common_arg::to_string() const {
// params for printing to console // params for printing to console
const static int n_leading_spaces = 40; const static int n_leading_spaces = 40;
const static int n_char_per_line_help = 70; // TODO: detect this based on current console const static int n_char_per_line_help = 70; // TODO: detect this based on current console
@ -647,6 +657,53 @@ static void add_rpc_devices(const std::string & servers) {
} }
} }
bool common_params_parse(int argc, char ** argv, llama_example ex, std::map<common_arg, std::string> & out_map) {
common_params dummy_params;
common_params_context ctx_arg = common_params_parser_init(dummy_params, ex, nullptr);
std::unordered_map<std::string, common_arg *> arg_to_options;
for (auto & opt : ctx_arg.options) {
for (const auto & arg : opt.args) {
arg_to_options[arg] = &opt;
}
}
// TODO @ngxson : find a way to deduplicate this code
// handle command line arguments
auto check_arg = [&](int i) {
if (i+1 >= argc) {
throw std::invalid_argument("expected value for argument");
}
};
for (int i = 1; i < argc; i++) {
const std::string arg_prefix = "--";
std::string arg = argv[i];
if (arg.compare(0, arg_prefix.size(), arg_prefix) == 0) {
std::replace(arg.begin(), arg.end(), '_', '-');
}
if (arg_to_options.find(arg) == arg_to_options.end()) {
throw std::invalid_argument(string_format("error: invalid argument: %s", arg.c_str()));
}
auto opt = *arg_to_options[arg];
std::string val;
if (opt.value_hint != nullptr) {
// arg with single value
check_arg(i);
val = argv[++i];
}
if (opt.value_hint_2 != nullptr) {
// TODO: support arg with 2 values
throw std::invalid_argument("error: argument with 2 values is not yet supported\n");
}
out_map[opt] = val;
}
return true;
}
bool common_params_parse(int argc, char ** argv, common_params & params, llama_example ex, void(*print_usage)(int, char **)) { bool common_params_parse(int argc, char ** argv, common_params & params, llama_example ex, void(*print_usage)(int, char **)) {
auto ctx_arg = common_params_parser_init(params, ex, print_usage); auto ctx_arg = common_params_parser_init(params, ex, print_usage);
const common_params params_org = ctx_arg.params; // the example can modify the default params const common_params params_org = ctx_arg.params; // the example can modify the default params
@ -692,25 +749,19 @@ static std::string list_builtin_chat_templates() {
return msg.str(); return msg.str();
} }
static bool is_truthy(const std::string & value) { bool common_arg_utils::is_truthy(const std::string & value) {
return value == "on" || value == "enabled" || value == "1"; return value == "on" || value == "enabled" || value == "1";
} }
static bool is_falsey(const std::string & value) { bool common_arg_utils::is_falsey(const std::string & value) {
return value == "off" || value == "disabled" || value == "0"; return value == "off" || value == "disabled" || value == "0";
} }
static bool is_autoy(const std::string & value) { bool common_arg_utils::is_autoy(const std::string & value) {
return value == "auto" || value == "-1"; return value == "auto" || value == "-1";
} }
common_params_context common_params_parser_init(common_params & params, llama_example ex, void(*print_usage)(int, char **)) { common_params_context common_params_parser_init(common_params & params, llama_example ex, void(*print_usage)(int, char **)) {
// default values specific to example
// note: we place it here instead of inside server.cpp to allow llama-gen-docs to pick it up
if (ex == LLAMA_EXAMPLE_SERVER) {
params.use_jinja = true;
}
params.use_color = tty_can_use_colors(); params.use_color = tty_can_use_colors();
// load dynamic backends // load dynamic backends
@ -2550,6 +2601,13 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
params.models_dir = value; params.models_dir = value;
} }
).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_MODELS_DIR")); ).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_MODELS_DIR"));
add_opt(common_arg(
{"--models-preset"}, "PATH",
"path to INI file containing model presets for the router server (default: disabled)",
[](common_params & params, const std::string & value) {
params.models_preset = value;
}
).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_MODELS_PRESET"));
add_opt(common_arg( add_opt(common_arg(
{"--models-max"}, "N", {"--models-max"}, "N",
string_format("for router server, maximum number of models to load simultaneously (default: %d, 0 = unlimited)", params.models_max), string_format("for router server, maximum number of models to load simultaneously (default: %d, 0 = unlimited)", params.models_max),
@ -2566,14 +2624,14 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_NO_MODELS_AUTOLOAD")); ).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_NO_MODELS_AUTOLOAD"));
add_opt(common_arg( add_opt(common_arg(
{"--jinja"}, {"--jinja"},
string_format("use jinja template for chat (default: %s)\n", params.use_jinja ? "enabled" : "disabled"), string_format("use jinja template for chat (default: %s)", params.use_jinja ? "enabled" : "disabled"),
[](common_params & params) { [](common_params & params) {
params.use_jinja = true; params.use_jinja = true;
} }
).set_examples({LLAMA_EXAMPLE_SERVER, LLAMA_EXAMPLE_COMPLETION, LLAMA_EXAMPLE_CLI, LLAMA_EXAMPLE_MTMD}).set_env("LLAMA_ARG_JINJA")); ).set_examples({LLAMA_EXAMPLE_SERVER, LLAMA_EXAMPLE_COMPLETION, LLAMA_EXAMPLE_CLI, LLAMA_EXAMPLE_MTMD}).set_env("LLAMA_ARG_JINJA"));
add_opt(common_arg( add_opt(common_arg(
{"--no-jinja"}, {"--no-jinja"},
string_format("disable jinja template for chat (default: %s)\n", params.use_jinja ? "enabled" : "disabled"), string_format("disable jinja template for chat (default: %s)", params.use_jinja ? "disabled" : "enabled"),
[](common_params & params) { [](common_params & params) {
params.use_jinja = false; params.use_jinja = false;
} }

32
common/arg.h
View File

@ -3,8 +3,10 @@
#include "common.h" #include "common.h"
#include <set> #include <set>
#include <map>
#include <string> #include <string>
#include <vector> #include <vector>
#include <cstring>
// //
// CLI argument parsing // CLI argument parsing
@ -24,6 +26,8 @@ struct common_arg {
void (*handler_str_str)(common_params & params, const std::string &, const std::string &) = nullptr; void (*handler_str_str)(common_params & params, const std::string &, const std::string &) = nullptr;
void (*handler_int) (common_params & params, int) = nullptr; void (*handler_int) (common_params & params, int) = nullptr;
common_arg() = default;
common_arg( common_arg(
const std::initializer_list<const char *> & args, const std::initializer_list<const char *> & args,
const char * value_hint, const char * value_hint,
@ -61,9 +65,29 @@ struct common_arg {
bool is_exclude(enum llama_example ex); bool is_exclude(enum llama_example ex);
bool get_value_from_env(std::string & output) const; bool get_value_from_env(std::string & output) const;
bool has_value_from_env() const; bool has_value_from_env() const;
std::string to_string(); std::string to_string() const;
// for using as key in std::map
bool operator<(const common_arg& other) const {
if (args.empty() || other.args.empty()) {
return false;
}
return strcmp(args[0], other.args[0]) < 0;
}
bool operator==(const common_arg& other) const {
if (args.empty() || other.args.empty()) {
return false;
}
return strcmp(args[0], other.args[0]) == 0;
}
}; };
namespace common_arg_utils {
bool is_truthy(const std::string & value);
bool is_falsey(const std::string & value);
bool is_autoy(const std::string & value);
}
struct common_params_context { struct common_params_context {
enum llama_example ex = LLAMA_EXAMPLE_COMMON; enum llama_example ex = LLAMA_EXAMPLE_COMMON;
common_params & params; common_params & params;
@ -76,7 +100,11 @@ struct common_params_context {
// if one argument has invalid value, it will automatically display usage of the specific argument (and not the full usage message) // if one argument has invalid value, it will automatically display usage of the specific argument (and not the full usage message)
bool common_params_parse(int argc, char ** argv, common_params & params, llama_example ex, void(*print_usage)(int, char **) = nullptr); bool common_params_parse(int argc, char ** argv, common_params & params, llama_example ex, void(*print_usage)(int, char **) = nullptr);
// function to be used by test-arg-parser // parse input arguments from CLI into a map
// TODO: support repeated args in the future
bool common_params_parse(int argc, char ** argv, llama_example ex, std::map<common_arg, std::string> & out_map);
// initialize argument parser context - used by test-arg-parser and preset
common_params_context common_params_parser_init(common_params & params, llama_example ex, void(*print_usage)(int, char **) = nullptr); common_params_context common_params_parser_init(common_params & params, llama_example ex, void(*print_usage)(int, char **) = nullptr);
struct common_remote_params { struct common_remote_params {

9
common/common.h
View File

@ -469,7 +469,7 @@ struct common_params {
std::string public_path = ""; // NOLINT std::string public_path = ""; // NOLINT
std::string api_prefix = ""; // NOLINT std::string api_prefix = ""; // NOLINT
std::string chat_template = ""; // NOLINT std::string chat_template = ""; // NOLINT
bool use_jinja = false; // NOLINT bool use_jinja = true; // NOLINT
bool enable_chat_template = true; bool enable_chat_template = true;
common_reasoning_format reasoning_format = COMMON_REASONING_FORMAT_DEEPSEEK; common_reasoning_format reasoning_format = COMMON_REASONING_FORMAT_DEEPSEEK;
int reasoning_budget = -1; int reasoning_budget = -1;
@ -489,9 +489,10 @@ struct common_params {
bool endpoint_metrics = false; bool endpoint_metrics = false;
// router server configs // router server configs
std::string models_dir = ""; // directory containing models for the router server std::string models_dir = ""; // directory containing models for the router server
int models_max = 4; // maximum number of models to load simultaneously std::string models_preset = ""; // directory containing model presets for the router server
bool models_autoload = true; // automatically load models when requested via the router server int models_max = 4; // maximum number of models to load simultaneously
bool models_autoload = true; // automatically load models when requested via the router server
bool log_json = false; bool log_json = false;

180
common/preset.cpp Normal file
View File

@ -0,0 +1,180 @@
#include "arg.h"
#include "preset.h"
#include "peg-parser.h"
#include "log.h"
#include <fstream>
#include <sstream>
#include <filesystem>
static std::string rm_leading_dashes(const std::string & str) {
size_t pos = 0;
while (pos < str.size() && str[pos] == '-') {
++pos;
}
return str.substr(pos);
}
std::vector<std::string> common_preset::to_args() const {
std::vector<std::string> args;
for (const auto & [opt, value] : options) {
args.push_back(opt.args.back()); // use the last arg as the main arg
if (opt.value_hint == nullptr && opt.value_hint_2 == nullptr) {
// flag option, no value
if (common_arg_utils::is_falsey(value)) {
// skip the flag
args.pop_back();
}
}
if (opt.value_hint != nullptr) {
// single value
args.push_back(value);
}
if (opt.value_hint != nullptr && opt.value_hint_2 != nullptr) {
throw std::runtime_error(string_format(
"common_preset::to_args(): option '%s' has two values, which is not supported yet",
opt.args.back()
));
}
}
return args;
}
std::string common_preset::to_ini() const {
std::ostringstream ss;
ss << "[" << name << "]\n";
for (const auto & [opt, value] : options) {
auto espaced_value = value;
string_replace_all(espaced_value, "\n", "\\\n");
ss << rm_leading_dashes(opt.args.back()) << " = ";
ss << espaced_value << "\n";
}
ss << "\n";
return ss.str();
}
static std::map<std::string, std::map<std::string, std::string>> parse_ini_from_file(const std::string & path) {
std::map<std::string, std::map<std::string, std::string>> parsed;
if (!std::filesystem::exists(path)) {
throw std::runtime_error("preset file does not exist: " + path);
}
std::ifstream file(path);
if (!file.good()) {
throw std::runtime_error("failed to open server preset file: " + path);
}
std::string contents((std::istreambuf_iterator<char>(file)), std::istreambuf_iterator<char>());
static const auto parser = build_peg_parser([](auto & p) {
// newline ::= "\r\n" / "\n" / "\r"
auto newline = p.rule("newline", p.literal("\r\n") | p.literal("\n") | p.literal("\r"));
// ws ::= [ \t]*
auto ws = p.rule("ws", p.chars("[ \t]", 0, -1));
// comment ::= [;#] (!newline .)*
auto comment = p.rule("comment", p.chars("[;#]", 1, 1) + p.zero_or_more(p.negate(newline) + p.any()));
// eol ::= ws comment? (newline / EOF)
auto eol = p.rule("eol", ws + p.optional(comment) + (newline | p.end()));
// ident ::= [a-zA-Z_] [a-zA-Z0-9_.-]*
auto ident = p.rule("ident", p.chars("[a-zA-Z_]", 1, 1) + p.chars("[a-zA-Z0-9_.-]", 0, -1));
// value ::= (!eol-start .)*
auto eol_start = p.rule("eol-start", ws + (p.chars("[;#]", 1, 1) | newline | p.end()));
auto value = p.rule("value", p.zero_or_more(p.negate(eol_start) + p.any()));
// header-line ::= "[" ws ident ws "]" eol
auto header_line = p.rule("header-line", "[" + ws + p.tag("section-name", p.chars("[^]]")) + ws + "]" + eol);
// kv-line ::= ident ws "=" ws value eol
auto kv_line = p.rule("kv-line", p.tag("key", ident) + ws + "=" + ws + p.tag("value", value) + eol);
// comment-line ::= ws comment (newline / EOF)
auto comment_line = p.rule("comment-line", ws + comment + (newline | p.end()));
// blank-line ::= ws (newline / EOF)
auto blank_line = p.rule("blank-line", ws + (newline | p.end()));
// line ::= header-line / kv-line / comment-line / blank-line
auto line = p.rule("line", header_line | kv_line | comment_line | blank_line);
// ini ::= line* EOF
auto ini = p.rule("ini", p.zero_or_more(line) + p.end());
return ini;
});
common_peg_parse_context ctx(contents);
const auto result = parser.parse(ctx);
if (!result.success()) {
throw std::runtime_error("failed to parse server config file: " + path);
}
std::string current_section = COMMON_PRESET_DEFAULT_NAME;
std::string current_key;
ctx.ast.visit(result, [&](const auto & node) {
if (node.tag == "section-name") {
const std::string section = std::string(node.text);
current_section = section;
parsed[current_section] = {};
} else if (node.tag == "key") {
const std::string key = std::string(node.text);
current_key = key;
} else if (node.tag == "value" && !current_key.empty() && !current_section.empty()) {
parsed[current_section][current_key] = std::string(node.text);
current_key.clear();
}
});
return parsed;
}
static std::map<std::string, common_arg> get_map_key_opt(common_params_context & ctx_params) {
std::map<std::string, common_arg> mapping;
for (const auto & opt : ctx_params.options) {
if (opt.env != nullptr) {
mapping[opt.env] = opt;
}
for (const auto & arg : opt.args) {
mapping[rm_leading_dashes(arg)] = opt;
}
}
return mapping;
}
common_presets common_presets_load(const std::string & path, common_params_context & ctx_params) {
common_presets out;
auto key_to_opt = get_map_key_opt(ctx_params);
auto ini_data = parse_ini_from_file(path);
for (auto section : ini_data) {
common_preset preset;
if (section.first.empty()) {
preset.name = COMMON_PRESET_DEFAULT_NAME;
} else {
preset.name = section.first;
}
LOG_DBG("loading preset: %s\n", preset.name.c_str());
for (const auto & [key, value] : section.second) {
LOG_DBG("option: %s = %s\n", key.c_str(), value.c_str());
if (key_to_opt.find(key) != key_to_opt.end()) {
preset.options[key_to_opt[key]] = value;
LOG_DBG("accepted option: %s = %s\n", key.c_str(), value.c_str());
} else {
// TODO: maybe warn about unknown key?
}
}
out[preset.name] = preset;
}
return out;
}

32
common/preset.h Normal file
View File

@ -0,0 +1,32 @@
#pragma once
#include "common.h"
#include "arg.h"
#include <string>
#include <vector>
#include <map>
//
// INI preset parser and writer
//
constexpr const char * COMMON_PRESET_DEFAULT_NAME = "default";
struct common_preset {
std::string name;
// TODO: support repeated args in the future
std::map<common_arg, std::string> options;
// convert preset to CLI argument list
std::vector<std::string> to_args() const;
// convert preset to INI format string
std::string to_ini() const;
// TODO: maybe implement to_env() if needed
};
// interface for multiple presets in one file
using common_presets = std::map<std::string, common_preset>;
common_presets common_presets_load(const std::string & path, common_params_context & ctx_params);

6
docs/docker.md
View File

@ -56,7 +56,7 @@ docker run -v /path/to/models:/models ghcr.io/ggml-org/llama.cpp:light -m /model
or with a server image: or with a server image:
```bash ```bash
docker run -v /path/to/models:/models -p 8000:8000 ghcr.io/ggml-org/llama.cpp:server -m /models/7B/ggml-model-q4_0.gguf --port 8000 --host 0.0.0.0 -n 512 docker run -v /path/to/models:/models -p 8080:8080 ghcr.io/ggml-org/llama.cpp:server -m /models/7B/ggml-model-q4_0.gguf --port 8080 --host 0.0.0.0 -n 512
``` ```
## Docker With CUDA ## Docker With CUDA
@ -91,7 +91,7 @@ After building locally, Usage is similar to the non-CUDA examples, but you'll ne
```bash ```bash
docker run --gpus all -v /path/to/models:/models local/llama.cpp:full-cuda --run -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1 docker run --gpus all -v /path/to/models:/models local/llama.cpp:full-cuda --run -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1
docker run --gpus all -v /path/to/models:/models local/llama.cpp:light-cuda -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1 docker run --gpus all -v /path/to/models:/models local/llama.cpp:light-cuda -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1
docker run --gpus all -v /path/to/models:/models local/llama.cpp:server-cuda -m /models/7B/ggml-model-q4_0.gguf --port 8000 --host 0.0.0.0 -n 512 --n-gpu-layers 1 docker run --gpus all -v /path/to/models:/models local/llama.cpp:server-cuda -m /models/7B/ggml-model-q4_0.gguf --port 8080 --host 0.0.0.0 -n 512 --n-gpu-layers 1
``` ```
## Docker With MUSA ## Docker With MUSA
@ -125,5 +125,5 @@ After building locally, Usage is similar to the non-MUSA examples, but you'll ne
```bash ```bash
docker run -v /path/to/models:/models local/llama.cpp:full-musa --run -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1 docker run -v /path/to/models:/models local/llama.cpp:full-musa --run -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1
docker run -v /path/to/models:/models local/llama.cpp:light-musa -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1 docker run -v /path/to/models:/models local/llama.cpp:light-musa -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1
docker run -v /path/to/models:/models local/llama.cpp:server-musa -m /models/7B/ggml-model-q4_0.gguf --port 8000 --host 0.0.0.0 -n 512 --n-gpu-layers 1 docker run -v /path/to/models:/models local/llama.cpp:server-musa -m /models/7B/ggml-model-q4_0.gguf --port 8080 --host 0.0.0.0 -n 512 --n-gpu-layers 1
``` ```

12
ggml/include/ggml.h
View File

@ -2305,13 +2305,11 @@ extern "C" {
float stop, float stop,
float step); float step);
#define GGML_KQ_MASK_PAD 1 // q: [n_embd_k, n_batch, n_head, ne3 ]
// k: [n_embd_k, n_kv, n_head_kv, ne3 ]
// q: [n_embd_k, n_batch, n_head, ne3 ] // v: [n_embd_v, n_kv, n_head_kv, ne3 ] !! not transposed !!
// k: [n_embd_k, n_kv, n_head_kv, ne3 ] // mask: [n_kv, n_batch, ne32, ne33]
// v: [n_embd_v, n_kv, n_head_kv, ne3 ] !! not transposed !! // res: [n_embd_v, n_head, n_batch, ne3 ] !! permuted !!
// mask: [n_kv, n_batch_pad, ne32, ne33] !! n_batch_pad = GGML_PAD(n_batch, GGML_KQ_MASK_PAD) !!
// res: [n_embd_v, n_head, n_batch, ne3 ] !! permuted !!
// //
// broadcast: // broadcast:
// n_head % n_head_kv == 0 // n_head % n_head_kv == 0

73
ggml/src/ggml-cpu/ggml-cpu.c
View File

@ -187,6 +187,9 @@ typedef void * thread_ret_t;
typedef pthread_t ggml_thread_t; typedef pthread_t ggml_thread_t;
#define GGML_THREADPOOL_N_THREADS_MASK (0xffffU)
#define GGML_THREADPOOL_N_THREADS_BITS (16)
#if defined(__APPLE__) #if defined(__APPLE__)
#include <unistd.h> #include <unistd.h>
#include <mach/mach.h> #include <mach/mach.h>
@ -449,7 +452,7 @@ struct ggml_threadpool {
struct ggml_cplan * cplan; struct ggml_cplan * cplan;
// synchronization primitives // synchronization primitives
atomic_int n_graph; // incremented when there is work to be done (i.e each graph) atomic_int n_graph; // updated when there is work to be done (i.e each graph) holds graph and active thread counts.
atomic_int GGML_CACHE_ALIGN n_barrier; atomic_int GGML_CACHE_ALIGN n_barrier;
atomic_int GGML_CACHE_ALIGN n_barrier_passed; atomic_int GGML_CACHE_ALIGN n_barrier_passed;
atomic_int GGML_CACHE_ALIGN current_chunk; // currently processing chunk during Mat_Mul, shared between all the threads. atomic_int GGML_CACHE_ALIGN current_chunk; // currently processing chunk during Mat_Mul, shared between all the threads.
@ -457,12 +460,10 @@ struct ggml_threadpool {
// these are atomic as an annotation for thread-sanitizer // these are atomic as an annotation for thread-sanitizer
atomic_bool stop; // Used for stopping the threadpool altogether atomic_bool stop; // Used for stopping the threadpool altogether
atomic_bool pause; // Used for pausing the threadpool or individual threads atomic_bool pause; // Used for pausing the threadpool or individual threads
atomic_int abort; // Used for aborting processing of a graph atomic_int abort; // Used for aborting processing of a graph
struct ggml_compute_state * workers; // per thread state struct ggml_compute_state * workers; // per thread state
int n_threads_max; // number of threads in the pool int n_threads; // Number of threads in the pool
atomic_int n_threads_cur; // number of threads used in the current graph
int32_t prio; // Scheduling priority int32_t prio; // Scheduling priority
uint32_t poll; // Polling level (0 - no polling) uint32_t poll; // Polling level (0 - no polling)
@ -539,7 +540,7 @@ struct ggml_state {
static struct ggml_state g_state = {0}; static struct ggml_state g_state = {0};
void ggml_barrier(struct ggml_threadpool * tp) { void ggml_barrier(struct ggml_threadpool * tp) {
int n_threads = atomic_load_explicit(&tp->n_threads_cur, memory_order_relaxed); int n_threads = atomic_load_explicit(&tp->n_graph, memory_order_relaxed) & GGML_THREADPOOL_N_THREADS_MASK;
if (n_threads == 1) { if (n_threads == 1) {
return; return;
} }
@ -556,7 +557,7 @@ void ggml_barrier(struct ggml_threadpool * tp) {
// last thread // last thread
atomic_store_explicit(&tp->n_barrier, 0, memory_order_relaxed); atomic_store_explicit(&tp->n_barrier, 0, memory_order_relaxed);
// exit barrier (fill seq-cst fence) // exit barrier (full seq-cst fence)
atomic_fetch_add_explicit(&tp->n_barrier_passed, 1, memory_order_seq_cst); atomic_fetch_add_explicit(&tp->n_barrier_passed, 1, memory_order_seq_cst);
return; return;
} }
@ -2628,7 +2629,7 @@ static void ggml_thread_cpumask_next(const bool * global_mask, bool * local_mask
void ggml_threadpool_free(struct ggml_threadpool* threadpool) { void ggml_threadpool_free(struct ggml_threadpool* threadpool) {
if (!threadpool) return; if (!threadpool) return;
const int n_threads = threadpool->n_threads_max; const int n_threads = threadpool->n_threads;
#ifndef GGML_USE_OPENMP #ifndef GGML_USE_OPENMP
struct ggml_compute_state* workers = threadpool->workers; struct ggml_compute_state* workers = threadpool->workers;
@ -2704,7 +2705,7 @@ struct ggml_cplan ggml_graph_plan(
//GGML_PRINT_DEBUG("Threadpool is not specified. Will create a disposable threadpool : n_threads %d\n", n_threads); //GGML_PRINT_DEBUG("Threadpool is not specified. Will create a disposable threadpool : n_threads %d\n", n_threads);
} }
if (n_threads <= 0) { if (n_threads <= 0) {
n_threads = threadpool ? threadpool->n_threads_max : GGML_DEFAULT_N_THREADS; n_threads = threadpool ? threadpool->n_threads : GGML_DEFAULT_N_THREADS;
} }
#if defined(__EMSCRIPTEN__) && !defined(__EMSCRIPTEN_PTHREADS__) #if defined(__EMSCRIPTEN__) && !defined(__EMSCRIPTEN_PTHREADS__)
@ -2912,12 +2913,14 @@ static thread_ret_t ggml_graph_compute_thread(void * data) {
struct ggml_compute_params params = { struct ggml_compute_params params = {
/*.ith =*/ state->ith, /*.ith =*/ state->ith,
/*.nth =*/ atomic_load_explicit(&tp->n_threads_cur, memory_order_relaxed), /*.nth =*/ atomic_load_explicit(&tp->n_graph, memory_order_relaxed) & GGML_THREADPOOL_N_THREADS_MASK,
/*.wsize =*/ cplan->work_size, /*.wsize =*/ cplan->work_size,
/*.wdata =*/ cplan->work_data, /*.wdata =*/ cplan->work_data,
/*.threadpool=*/ tp, /*.threadpool=*/ tp,
}; };
GGML_PRINT_DEBUG("thread #%d compute-start cplan %p last-graph %d \n", state->ith, cplan, state->last_graph);
for (int node_n = 0; node_n < cgraph->n_nodes && atomic_load_explicit(&tp->abort, memory_order_relaxed) != node_n; node_n++) { for (int node_n = 0; node_n < cgraph->n_nodes && atomic_load_explicit(&tp->abort, memory_order_relaxed) != node_n; node_n++) {
struct ggml_tensor * node = cgraph->nodes[node_n]; struct ggml_tensor * node = cgraph->nodes[node_n];
@ -2939,6 +2942,8 @@ static thread_ret_t ggml_graph_compute_thread(void * data) {
} }
} }
GGML_PRINT_DEBUG("thread #%d compute-done cplan %p last-graph %d \n", state->ith, cplan, state->last_graph);
ggml_barrier(state->threadpool); ggml_barrier(state->threadpool);
return 0; return 0;
@ -2946,27 +2951,23 @@ static thread_ret_t ggml_graph_compute_thread(void * data) {
#ifndef GGML_USE_OPENMP #ifndef GGML_USE_OPENMP
// check if thread is active
static inline bool ggml_graph_compute_thread_active(struct ggml_compute_state * state) {
struct ggml_threadpool * threadpool = state->threadpool;
int n_threads = atomic_load_explicit(&threadpool->n_threads_cur, memory_order_relaxed);
return (state->ith < n_threads);
}
// check if thread is ready to proceed (exit from polling or sleeping) // check if thread is ready to proceed (exit from polling or sleeping)
// returns true if loops should exit, sets state->pending to indicate new work
static inline bool ggml_graph_compute_thread_ready(struct ggml_compute_state * state) { static inline bool ggml_graph_compute_thread_ready(struct ggml_compute_state * state) {
struct ggml_threadpool * threadpool = state->threadpool; struct ggml_threadpool * threadpool = state->threadpool;
if (state->pending || threadpool->stop || threadpool->pause) { return true; } if (state->pending || threadpool->stop || threadpool->pause) { return true; }
// check for new graph/work // check for new graph/work
int new_graph = atomic_load_explicit(&threadpool->n_graph, memory_order_relaxed); int n_graph = atomic_load_explicit(&threadpool->n_graph, memory_order_relaxed);
if (new_graph != state->last_graph) { int n_threads = n_graph & GGML_THREADPOOL_N_THREADS_MASK;
state->pending = ggml_graph_compute_thread_active(state); if (n_graph != state->last_graph) {
state->last_graph = new_graph; state->pending = (state->ith < n_threads);
state->last_graph = n_graph;
return true;
} }
return state->pending; return false;
} }
// sync thread state after polling // sync thread state after polling
@ -2983,11 +2984,6 @@ static inline void ggml_graph_compute_thread_sync(struct ggml_compute_state * st
static inline bool ggml_graph_compute_poll_for_work(struct ggml_compute_state * state) { static inline bool ggml_graph_compute_poll_for_work(struct ggml_compute_state * state) {
struct ggml_threadpool * threadpool = state->threadpool; struct ggml_threadpool * threadpool = state->threadpool;
// Skip polling for unused threads
if (!ggml_graph_compute_thread_active(state)) {
return state->pending;
}
// This seems to make 0 ... 100 a decent range for polling level across modern processors. // This seems to make 0 ... 100 a decent range for polling level across modern processors.
// Perhaps, we can adjust it dynamically based on load and things. // Perhaps, we can adjust it dynamically based on load and things.
const uint64_t n_rounds = 1024UL * 128 * threadpool->poll; const uint64_t n_rounds = 1024UL * 128 * threadpool->poll;
@ -3049,7 +3045,6 @@ static thread_ret_t ggml_graph_compute_secondary_thread(void* data) {
ggml_graph_compute_check_for_work(state); ggml_graph_compute_check_for_work(state);
if (state->pending) { if (state->pending) {
state->pending = false; state->pending = false;
ggml_graph_compute_thread(state); ggml_graph_compute_thread(state);
} }
} }
@ -3064,14 +3059,15 @@ static void ggml_graph_compute_kickoff(struct ggml_threadpool * threadpool, int
ggml_mutex_lock(&threadpool->mutex); ggml_mutex_lock(&threadpool->mutex);
GGML_PRINT_DEBUG("threadpool: n_threads_cur %d n_threads %d\n", threadpool->n_threads_cur, n_threads); // Update the number of active threads and the graph count
int n_graph = atomic_load_explicit(&threadpool->n_graph, memory_order_relaxed) >> GGML_THREADPOOL_N_THREADS_BITS;
n_graph = ((n_graph + 1) << GGML_THREADPOOL_N_THREADS_BITS) | (n_threads & GGML_THREADPOOL_N_THREADS_MASK);
// Update the number of active threads GGML_PRINT_DEBUG("compute-kickoff: n_threads %d n_graph %d\n", n_threads, n_graph);
atomic_store_explicit(&threadpool->n_threads_cur, n_threads, memory_order_relaxed);
// Indicate the graph is ready to be processed // Indicate the graph is ready to be processed
// We need the full seq-cst fence here because of the polling threads (used in thread_sync) // We need the full seq-cst fence here because of the polling threads (used in thread_sync)
atomic_fetch_add_explicit(&threadpool->n_graph, 1, memory_order_seq_cst); atomic_store_explicit(&threadpool->n_graph, n_graph, memory_order_seq_cst);
if (threadpool->pause) { if (threadpool->pause) {
// Update main thread prio and affinity to match the threadpool settings // Update main thread prio and affinity to match the threadpool settings
@ -3109,8 +3105,7 @@ static struct ggml_threadpool * ggml_threadpool_new_impl(
threadpool->pause = tpp->paused; threadpool->pause = tpp->paused;
threadpool->abort = -1; threadpool->abort = -1;
threadpool->workers = NULL; threadpool->workers = NULL;
threadpool->n_threads_max = tpp->n_threads; threadpool->n_threads = tpp->n_threads;
threadpool->n_threads_cur = tpp->n_threads;
threadpool->poll = tpp->poll; threadpool->poll = tpp->poll;
threadpool->prio = tpp->prio; threadpool->prio = tpp->prio;
threadpool->ec = GGML_STATUS_SUCCESS; threadpool->ec = GGML_STATUS_SUCCESS;
@ -3205,7 +3200,7 @@ enum ggml_status ggml_graph_compute(struct ggml_cgraph * cgraph, struct ggml_cpl
{ {
// update the number of threads from the actual number of threads that we got from OpenMP // update the number of threads from the actual number of threads that we got from OpenMP
n_threads = omp_get_num_threads(); n_threads = omp_get_num_threads();
atomic_store_explicit(&threadpool->n_threads_cur, n_threads, memory_order_relaxed); atomic_store_explicit(&threadpool->n_graph, n_threads, memory_order_relaxed);
} }
// Apply thread CPU mask and priority // Apply thread CPU mask and priority
@ -3218,13 +3213,13 @@ enum ggml_status ggml_graph_compute(struct ggml_cgraph * cgraph, struct ggml_cpl
ggml_graph_compute_thread(&threadpool->workers[ith]); ggml_graph_compute_thread(&threadpool->workers[ith]);
} }
} else { } else {
atomic_store_explicit(&threadpool->n_threads_cur, 1, memory_order_relaxed); atomic_store_explicit(&threadpool->n_graph, 1, memory_order_relaxed);
ggml_graph_compute_thread(&threadpool->workers[0]); ggml_graph_compute_thread(&threadpool->workers[0]);
} }
#else #else
if (n_threads > threadpool->n_threads_max) { if (n_threads > threadpool->n_threads) {
GGML_LOG_WARN("cplan requested more threads (%d) than available (%d)\n", n_threads, threadpool->n_threads_max); GGML_LOG_WARN("cplan requested more threads (%d) than available (%d)\n", n_threads, threadpool->n_threads);
n_threads = threadpool->n_threads_max; n_threads = threadpool->n_threads;
} }
// Kick all threads to start the new graph // Kick all threads to start the new graph

1
ggml/src/ggml-cuda/ggml-cuda.cu
View File

@ -4326,6 +4326,7 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
case GGML_UNARY_OP_EXPM1: case GGML_UNARY_OP_EXPM1:
case GGML_UNARY_OP_SOFTPLUS: case GGML_UNARY_OP_SOFTPLUS:
case GGML_UNARY_OP_ELU: case GGML_UNARY_OP_ELU:
case GGML_UNARY_OP_XIELU:
case GGML_UNARY_OP_FLOOR: case GGML_UNARY_OP_FLOOR:
case GGML_UNARY_OP_CEIL: case GGML_UNARY_OP_CEIL:
case GGML_UNARY_OP_ROUND: case GGML_UNARY_OP_ROUND:

78
ggml/src/ggml-hexagon/htp/rope-ops.c
View File

@ -73,15 +73,15 @@ static float rope_yarn_ramp(const float low, const float high, const int i0) {
return (1 - MIN(1, MAX(0, y))); return (1 - MIN(1, MAX(0, y)));
} }
static void rope_cache_init(const float theta_base, static void rope_cache_init(const float theta_base,
float freq_scale, const float freq_scale,
const float * freq_factors, const float * freq_factors,
float * corr_dims, float * corr_dims,
uint32_t ne0, const uint32_t ne0,
float ext_factor, const float ext_factor,
float mscale, const float mscale,
float * cache, float * cache,
float theta_scale) { const float theta_scale) {
// ref: https://github.com/jquesnelle/yarn/blob/master/scaled_rope/LlamaYaRNScaledRotaryEmbedding.py // ref: https://github.com/jquesnelle/yarn/blob/master/scaled_rope/LlamaYaRNScaledRotaryEmbedding.py
float theta = theta_base; float theta = theta_base;
@ -92,18 +92,19 @@ static void rope_cache_init(const float theta_base,
// Get n-d rotational scaling corrected for extrapolation // Get n-d rotational scaling corrected for extrapolation
float theta_interp = freq_scale * theta_extrap; float theta_interp = freq_scale * theta_extrap;
float theta2 = theta_interp; float theta_final = theta_interp;
float mscale_final = mscale;
if (ext_factor != 0.0f) { if (ext_factor != 0.0f) {
float ramp_mix = rope_yarn_ramp(corr_dims[0], corr_dims[1], i0) * ext_factor; float ramp_mix = rope_yarn_ramp(corr_dims[0], corr_dims[1], i0) * ext_factor;
theta2 = theta_interp * (1 - ramp_mix) + theta_extrap * ramp_mix; theta_final = theta_interp * (1 - ramp_mix) + theta_extrap * ramp_mix;
// Get n-d magnitude scaling corrected for interpolation // Get n-d magnitude scaling corrected for interpolation
mscale *= 1.0f + 0.1f * logf(1.0f / freq_scale); mscale_final *= 1.0f + 0.1f * logf(1.0f / freq_scale);
} }
cache[i0 + 0] = cosf(theta2) * mscale; cache[i0 + 0] = cosf(theta_final) * mscale_final;
cache[i0 + 1] = sinf(theta2) * mscale; cache[i0 + 1] = sinf(theta_final) * mscale_final;
theta *= theta_scale; theta *= theta_scale;
} }
@ -151,9 +152,9 @@ static void init_rope_ctx(struct rope_th_ctx * rope_ctx, struct htp_ops_context
} }
static void hvx_calc_rope_neox_f32(const float * restrict src0, static void hvx_calc_rope_neox_f32(const float * restrict src0,
float * restrict dst, float * restrict dst,
const int num_elems, const int num_elems,
const float * restrict theta_cache) { const float * restrict theta_cache) {
// for (int i = 0; i < num_elems; i += 2) { // for (int i = 0; i < num_elems; i += 2) {
//const float cos_theta = theta_cache[i + 0]; //const float cos_theta = theta_cache[i + 0];
//const float sin_theta = theta_cache[i + 1]; //const float sin_theta = theta_cache[i + 1];
@ -192,7 +193,7 @@ static void hvx_calc_rope_neox_f32(const float * restrict src0,
HVX_Vector v4 = Q6_Vqf32_vsub_Vqf32Vqf32(vx0_c, vx1_s); HVX_Vector v4 = Q6_Vqf32_vsub_Vqf32Vqf32(vx0_c, vx1_s);
HVX_Vector v5 = Q6_Vqf32_vadd_Vqf32Vqf32(vx0_s, vx1_c); HVX_Vector v5 = Q6_Vqf32_vadd_Vqf32Vqf32(vx0_s, vx1_c);
*(HVX_Vector *) dst_curr = Q6_Vsf_equals_Vqf32(v4); *(HVX_Vector *) dst_curr = Q6_Vsf_equals_Vqf32(v4);
*(HVX_Vector *) (dst_curr + half_size) = Q6_Vsf_equals_Vqf32(v5); *(HVX_Vector *) (dst_curr + half_size) = Q6_Vsf_equals_Vqf32(v5);
src0_curr += VLEN; src0_curr += VLEN;
@ -259,7 +260,7 @@ static void rope_hex_f32(struct rope_th_ctx * rope_ctx,
const uint32_t ir1, const uint32_t ir1,
int nth, int nth,
int ith, int ith,
int opt_path) { const int opt_path) {
struct htp_ops_context * octx = rope_ctx->octx; struct htp_ops_context * octx = rope_ctx->octx;
const struct htp_tensor * src0 = &octx->src0; const struct htp_tensor * src0 = &octx->src0;
@ -267,8 +268,8 @@ static void rope_hex_f32(struct rope_th_ctx * rope_ctx,
const struct htp_tensor * src2 = &octx->src2; const struct htp_tensor * src2 = &octx->src2;
struct htp_tensor * dst = &octx->dst; struct htp_tensor * dst = &octx->dst;
const int32_t mode = rope_ctx->mode; const int32_t mode = rope_ctx->mode;
const bool is_neox = mode & HTP_ROPE_TYPE_NEOX; const bool is_neox = mode & HTP_ROPE_TYPE_NEOX;
htp_rope_preamble; htp_rope_preamble;
@ -281,8 +282,9 @@ static void rope_hex_f32(struct rope_th_ctx * rope_ctx,
freq_factors = (const float *) src2->data; freq_factors = (const float *) src2->data;
} }
int ir = 0; const uint32_t i1_end = MIN(ir1, ne1);
const int32_t half_dims = rope_ctx->n_dims / 2;
const size_t remain_bytes = (ne0 - rope_ctx->n_dims) * sizeof(float);
for (uint32_t i3 = 0; i3 < ne3; i3++) { // batch for (uint32_t i3 = 0; i3 < ne3; i3++) { // batch
for (uint32_t i2 = 0; i2 < ne2; i2++) { // seq-len for (uint32_t i2 = 0; i2 < ne2; i2++) { // seq-len
const int32_t p = pos[i2]; const int32_t p = pos[i2];
@ -290,14 +292,7 @@ static void rope_hex_f32(struct rope_th_ctx * rope_ctx,
rope_cache_init(p, rope_ctx->freq_scale, freq_factors, rope_ctx->corr_dims, ne0, rope_ctx->ext_factor, rope_cache_init(p, rope_ctx->freq_scale, freq_factors, rope_ctx->corr_dims, ne0, rope_ctx->ext_factor,
rope_ctx->attn_factor, wp0, rope_ctx->theta_scale); rope_ctx->attn_factor, wp0, rope_ctx->theta_scale);
for (uint32_t i1 = 0; i1 < ne1; i1++) { // attn-heads for (uint32_t i1 = ir0; i1 < i1_end; i1++) { // attn-heads
if (ir++ < ir0) {
continue;
}
if (ir > ir1) {
break;
}
const float * src = (float *) ((char *) src0->data + i3 * nb03 + i2 * nb02 + i1 * nb01); const float * src = (float *) ((char *) src0->data + i3 * nb03 + i2 * nb02 + i1 * nb01);
float * dst_data = (float *) ((char *) dst->data + i3 * nb3 + i2 * nb2 + i1 * nb1); float * dst_data = (float *) ((char *) dst->data + i3 * nb3 + i2 * nb2 + i1 * nb1);
@ -310,6 +305,9 @@ static void rope_hex_f32(struct rope_th_ctx * rope_ctx,
} else { } else {
hvx_calc_rope_f32(src_loc, dst_data_loc, rope_ctx->n_dims, wp0); hvx_calc_rope_f32(src_loc, dst_data_loc, rope_ctx->n_dims, wp0);
} }
src_loc += rope_ctx->n_dims;
dst_data_loc += rope_ctx->n_dims;
} else { } else {
for (uint32_t i0 = 0; i0 < rope_ctx->n_dims; i0 += 2) { for (uint32_t i0 = 0; i0 < rope_ctx->n_dims; i0 += 2) {
const float cos_theta = wp0[i0 + 0]; const float cos_theta = wp0[i0 + 0];
@ -317,10 +315,10 @@ static void rope_hex_f32(struct rope_th_ctx * rope_ctx,
if (is_neox) { if (is_neox) {
const float x0 = src_loc[0]; const float x0 = src_loc[0];
const float x1 = src_loc[rope_ctx->n_dims/2]; const float x1 = src_loc[half_dims];
dst_data_loc[0] = x0 * cos_theta - x1 * sin_theta; dst_data_loc[0] = x0 * cos_theta - x1 * sin_theta;
dst_data_loc[rope_ctx->n_dims/2] = x0 * sin_theta + x1 * cos_theta; dst_data_loc[half_dims] = x0 * sin_theta + x1 * cos_theta;
src_loc += 1; src_loc += 1;
dst_data_loc += 1; dst_data_loc += 1;
@ -335,15 +333,13 @@ static void rope_hex_f32(struct rope_th_ctx * rope_ctx,
dst_data_loc += 2; dst_data_loc += 2;
} }
} }
src_loc += (is_neox ? half_dims : 0);
dst_data_loc += (is_neox ? half_dims : 0);
} }
for (uint32_t i0 = rope_ctx->n_dims; i0 < ne0; i0 += 2) { // TODO: use simd to speed up the remaining elements copy
dst_data_loc[0] = src_loc[0]; memcpy(dst_data_loc, src_loc, remain_bytes);
dst_data_loc[1] = src_loc[1];
src_loc += 2;
dst_data_loc += 2;
}
} }
} }
} }

2
ggml/src/ggml.c
View File

@ -5260,8 +5260,6 @@ struct ggml_tensor * ggml_flash_attn_ext(
if (mask) { if (mask) {
GGML_ASSERT(ggml_is_contiguous(mask)); GGML_ASSERT(ggml_is_contiguous(mask));
GGML_ASSERT(mask->ne[1] >= GGML_PAD(q->ne[1], GGML_KQ_MASK_PAD) &&
"the Flash-Attention kernel requires the mask to be padded to GGML_KQ_MASK_PAD and at least n_queries big");
//GGML_ASSERT(ggml_can_repeat_rows(mask, qk)); //GGML_ASSERT(ggml_can_repeat_rows(mask, qk));
GGML_ASSERT(q->ne[2] % mask->ne[2] == 0); GGML_ASSERT(q->ne[2] % mask->ne[2] == 0);

18
src/llama-batch.cpp
View File

@ -714,7 +714,7 @@ llama_ubatch llama_batch_allocr::ubatch_add(const std::vector<int32_t> & idxs, u
udata->seq_idx .resize(LLAMA_MAX_SEQ, -1); udata->seq_idx .resize(LLAMA_MAX_SEQ, -1);
udata->output .resize(n_tokens); udata->output .resize(n_tokens);
udata->seq_id_data.resize(n_tokens); udata->seq_id_data.reserve(n_tokens);
seq_set_t seq_set_unq; seq_set_t seq_set_unq;
@ -737,15 +737,13 @@ llama_ubatch llama_batch_allocr::ubatch_add(const std::vector<int32_t> & idxs, u
} }
udata->n_seq_id[i] = batch.n_seq_id[idxs[i]]; udata->n_seq_id[i] = batch.n_seq_id[idxs[i]];
udata->seq_id_data[i].reserve(udata->n_seq_id[i]);
for (int s = 0; s < udata->n_seq_id[i]; ++s) {
udata->seq_id_data[i].push_back(batch.seq_id[idxs[i]][s]);
}
udata->seq_id[i] = udata->seq_id_data[i].data();
udata->output[i] = batch.logits[idxs[i]]; udata->output[i] = batch.logits[idxs[i]];
for (int s = 0; s < udata->n_seq_id[i]; ++s) { for (int s = 0; s < udata->n_seq_id[i]; ++s) {
seq_set_unq.set(udata->seq_id[i][s]); const llama_seq_id seq_id = batch.seq_id[idxs[i]][s];
udata->seq_id_data.push_back(seq_id);
seq_set_unq.set(seq_id);
} }
if (udata->output[i]) { if (udata->output[i]) {
@ -753,6 +751,12 @@ llama_ubatch llama_batch_allocr::ubatch_add(const std::vector<int32_t> & idxs, u
} }
} }
llama_seq_id * seq_id_ptr = udata->seq_id_data.data();
for (size_t i = 0; i < idxs.size(); ++i) {
udata->seq_id[i] = seq_id_ptr;
seq_id_ptr += udata->n_seq_id[i];
}
for (uint32_t s = 0; s < n_seq_max; ++s) { for (uint32_t s = 0; s < n_seq_max; ++s) {
if (seq_set_unq.test(s)) { if (seq_set_unq.test(s)) {
udata->seq_idx[s] = udata->seq_id_unq.size(); udata->seq_idx[s] = udata->seq_id_unq.size();

6
src/llama-batch.h
View File

@ -56,15 +56,15 @@ struct llama_ubatch {
std::vector<float> embd; std::vector<float> embd;
std::vector<llama_pos> pos; std::vector<llama_pos> pos;
std::vector<int32_t> n_seq_id; std::vector<int32_t> n_seq_id;
std::vector<llama_seq_id *> seq_id; std::vector<llama_seq_id *> seq_id; // these point into the seq_id_data below
std::vector<llama_seq_id> seq_id_unq; std::vector<llama_seq_id> seq_id_unq;
std::vector<int32_t> seq_idx; std::vector<int32_t> seq_idx;
std::vector<int8_t> output; std::vector<int8_t> output;
std::vector<std::vector<llama_seq_id>> seq_id_data; std::vector<llama_seq_id> seq_id_data;
}; };
// the llama_ubatch pointers above point to this data if set. otherwise - points to non-owning data // the llama_ubatch pointers above point to this data if set. otherwise - point to external non-owning data
std::shared_ptr<data_t> data; std::shared_ptr<data_t> data;
}; };

8
src/llama-context.cpp
View File

@ -112,14 +112,6 @@ llama_context::llama_context(
// with causal attention, the batch size is limited by the context size // with causal attention, the batch size is limited by the context size
cparams.n_batch = cparams.causal_attn ? std::min(cparams.n_ctx, params.n_batch) : params.n_batch; cparams.n_batch = cparams.causal_attn ? std::min(cparams.n_ctx, params.n_batch) : params.n_batch;
// the batch has to be at least GGML_KQ_MASK_PAD because we will be padding the KQ_mask
// this is required by GPU kernels in order to avoid out-of-bounds accesses (e.g. ggml_flash_attn_ext)
// ref: https://github.com/ggerganov/llama.cpp/pull/5021
// TODO: this padding is not needed for the cache-less context so we should probably move it to llama_memory
if (cparams.n_batch < GGML_KQ_MASK_PAD) {
LLAMA_LOG_WARN("%s: n_batch is less than GGML_KQ_MASK_PAD - increasing to %d\n", __func__, GGML_KQ_MASK_PAD);
cparams.n_batch = GGML_KQ_MASK_PAD;
}
cparams.n_ubatch = std::min(cparams.n_batch, params.n_ubatch == 0 ? params.n_batch : params.n_ubatch); cparams.n_ubatch = std::min(cparams.n_batch, params.n_ubatch == 0 ? params.n_batch : params.n_ubatch);
cparams.op_offload = params.op_offload; cparams.op_offload = params.op_offload;

20
src/llama-graph.cpp
View File

@ -386,7 +386,7 @@ bool llm_graph_input_attn_kv::can_reuse(const llm_graph_params & params) {
//res &= self_v_idxs->ne[0] == params.ubatch.n_tokens; // TODO: need to move this to the unified cache and check there //res &= self_v_idxs->ne[0] == params.ubatch.n_tokens; // TODO: need to move this to the unified cache and check there
res &= self_kq_mask->ne[0] == mctx->get_n_kv(); res &= self_kq_mask->ne[0] == mctx->get_n_kv();
res &= self_kq_mask->ne[1] == GGML_PAD(params.ubatch.n_tokens, GGML_KQ_MASK_PAD); res &= self_kq_mask->ne[1] == params.ubatch.n_tokens;
return res; return res;
} }
@ -417,10 +417,10 @@ bool llm_graph_input_attn_kv_iswa::can_reuse(const llm_graph_params & params) {
//res &= self_v_idxs_swa->ne[0] == params.ubatch.n_tokens; // TODO: need to move this to the unified cache and check there //res &= self_v_idxs_swa->ne[0] == params.ubatch.n_tokens; // TODO: need to move this to the unified cache and check there
res &= self_kq_mask->ne[0] == mctx->get_base()->get_n_kv(); res &= self_kq_mask->ne[0] == mctx->get_base()->get_n_kv();
res &= self_kq_mask->ne[1] == GGML_PAD(params.ubatch.n_tokens, GGML_KQ_MASK_PAD); res &= self_kq_mask->ne[1] == params.ubatch.n_tokens;
res &= self_kq_mask_swa->ne[0] == mctx->get_swa()->get_n_kv(); res &= self_kq_mask_swa->ne[0] == mctx->get_swa()->get_n_kv();
res &= self_kq_mask_swa->ne[1] == GGML_PAD(params.ubatch.n_tokens, GGML_KQ_MASK_PAD); res &= self_kq_mask_swa->ne[1] == params.ubatch.n_tokens;
return res; return res;
} }
@ -453,7 +453,7 @@ void llm_graph_input_attn_cross::set_input(const llama_ubatch * ubatch) {
} }
} }
for (int i = n_tokens; i < GGML_PAD(n_tokens, GGML_KQ_MASK_PAD); ++i) { for (int i = n_tokens; i < n_tokens; ++i) {
for (int j = 0; j < n_enc; ++j) { for (int j = 0; j < n_enc; ++j) {
data[h*(n_enc*n_tokens) + i*n_enc + j] = -INFINITY; data[h*(n_enc*n_tokens) + i*n_enc + j] = -INFINITY;
} }
@ -1545,13 +1545,13 @@ llm_graph_input_attn_no_cache * llm_graph_context::build_attn_inp_no_cache() con
auto inp = std::make_unique<llm_graph_input_attn_no_cache>(hparams, cparams); auto inp = std::make_unique<llm_graph_input_attn_no_cache>(hparams, cparams);
// note: there is no KV cache, so the number of KV values is equal to the number of tokens in the batch // note: there is no KV cache, so the number of KV values is equal to the number of tokens in the batch
inp->self_kq_mask = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_tokens, GGML_PAD(n_tokens, GGML_KQ_MASK_PAD), 1, 1); inp->self_kq_mask = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_tokens, n_tokens, 1, 1);
ggml_set_input(inp->self_kq_mask); ggml_set_input(inp->self_kq_mask);
inp->self_kq_mask_cnv = cparams.flash_attn ? ggml_cast(ctx0, inp->self_kq_mask, GGML_TYPE_F16) : inp->self_kq_mask; inp->self_kq_mask_cnv = cparams.flash_attn ? ggml_cast(ctx0, inp->self_kq_mask, GGML_TYPE_F16) : inp->self_kq_mask;
if (hparams.swa_type != LLAMA_SWA_TYPE_NONE) { if (hparams.swa_type != LLAMA_SWA_TYPE_NONE) {
inp->self_kq_mask_swa = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_tokens, GGML_PAD(n_tokens, GGML_KQ_MASK_PAD), 1, 1); inp->self_kq_mask_swa = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_tokens, n_tokens, 1, 1);
ggml_set_input(inp->self_kq_mask_swa); ggml_set_input(inp->self_kq_mask_swa);
inp->self_kq_mask_swa_cnv = cparams.flash_attn ? ggml_cast(ctx0, inp->self_kq_mask_swa, GGML_TYPE_F16) : inp->self_kq_mask_swa; inp->self_kq_mask_swa_cnv = cparams.flash_attn ? ggml_cast(ctx0, inp->self_kq_mask_swa, GGML_TYPE_F16) : inp->self_kq_mask_swa;
@ -1633,7 +1633,7 @@ static std::unique_ptr<llm_graph_input_attn_kv> build_attn_inp_kv_impl(
inp->self_k_idxs = mctx_cur->build_input_k_idxs(ctx0, ubatch); inp->self_k_idxs = mctx_cur->build_input_k_idxs(ctx0, ubatch);
inp->self_v_idxs = mctx_cur->build_input_v_idxs(ctx0, ubatch); inp->self_v_idxs = mctx_cur->build_input_v_idxs(ctx0, ubatch);
inp->self_kq_mask = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_kv, GGML_PAD(n_tokens/n_stream, GGML_KQ_MASK_PAD), 1, n_stream); inp->self_kq_mask = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_kv, n_tokens/n_stream, 1, n_stream);
ggml_set_input(inp->self_kq_mask); ggml_set_input(inp->self_kq_mask);
inp->self_kq_mask_cnv = cparams.flash_attn ? ggml_cast(ctx0, inp->self_kq_mask, GGML_TYPE_F16) : inp->self_kq_mask; inp->self_kq_mask_cnv = cparams.flash_attn ? ggml_cast(ctx0, inp->self_kq_mask, GGML_TYPE_F16) : inp->self_kq_mask;
@ -1776,7 +1776,7 @@ llm_graph_input_attn_cross * llm_graph_context::build_attn_inp_cross() const {
const int32_t n_enc = !cross->v_embd.empty() ? cross->n_enc : hparams.n_ctx_train; const int32_t n_enc = !cross->v_embd.empty() ? cross->n_enc : hparams.n_ctx_train;
inp->cross_kq_mask = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_enc, GGML_PAD(n_tokens, GGML_KQ_MASK_PAD), 1, 1); inp->cross_kq_mask = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_enc, n_tokens, 1, 1);
ggml_set_input(inp->cross_kq_mask); ggml_set_input(inp->cross_kq_mask);
inp->cross_kq_mask_cnv = cparams.flash_attn ? ggml_cast(ctx0, inp->cross_kq_mask, GGML_TYPE_F16) : inp->cross_kq_mask; inp->cross_kq_mask_cnv = cparams.flash_attn ? ggml_cast(ctx0, inp->cross_kq_mask, GGML_TYPE_F16) : inp->cross_kq_mask;
@ -1842,7 +1842,7 @@ llm_graph_input_attn_kv_iswa * llm_graph_context::build_attn_inp_kv_iswa() const
inp->self_k_idxs = mctx_cur->get_base()->build_input_k_idxs(ctx0, ubatch); inp->self_k_idxs = mctx_cur->get_base()->build_input_k_idxs(ctx0, ubatch);
inp->self_v_idxs = mctx_cur->get_base()->build_input_v_idxs(ctx0, ubatch); inp->self_v_idxs = mctx_cur->get_base()->build_input_v_idxs(ctx0, ubatch);
inp->self_kq_mask = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_kv, GGML_PAD(n_tokens/n_stream, GGML_KQ_MASK_PAD), 1, n_stream); inp->self_kq_mask = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_kv, n_tokens/n_stream, 1, n_stream);
ggml_set_input(inp->self_kq_mask); ggml_set_input(inp->self_kq_mask);
ggml_set_name(inp->self_kq_mask, "self_kq_mask"); ggml_set_name(inp->self_kq_mask, "self_kq_mask");
@ -1858,7 +1858,7 @@ llm_graph_input_attn_kv_iswa * llm_graph_context::build_attn_inp_kv_iswa() const
inp->self_k_idxs_swa = mctx_cur->get_swa()->build_input_k_idxs(ctx0, ubatch); inp->self_k_idxs_swa = mctx_cur->get_swa()->build_input_k_idxs(ctx0, ubatch);
inp->self_v_idxs_swa = mctx_cur->get_swa()->build_input_v_idxs(ctx0, ubatch); inp->self_v_idxs_swa = mctx_cur->get_swa()->build_input_v_idxs(ctx0, ubatch);
inp->self_kq_mask_swa = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_kv, GGML_PAD(n_tokens/n_stream, GGML_KQ_MASK_PAD), 1, n_stream); inp->self_kq_mask_swa = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_kv, n_tokens/n_stream, 1, n_stream);
ggml_set_input(inp->self_kq_mask_swa); ggml_set_input(inp->self_kq_mask_swa);
ggml_set_name(inp->self_kq_mask_swa, "self_kq_mask_swa"); ggml_set_name(inp->self_kq_mask_swa, "self_kq_mask_swa");

5
src/llama-kv-cache.cpp
View File

@ -1232,8 +1232,7 @@ void llama_kv_cache::set_input_kq_mask(ggml_tensor * dst, const llama_ubatch * u
GGML_ASSERT(n_tokens%n_stream == 0); GGML_ASSERT(n_tokens%n_stream == 0);
// n_tps == n_tokens_per_stream // n_tps == n_tokens_per_stream
const int64_t n_tps = n_tokens/n_stream; const int64_t n_tps = n_tokens/n_stream;
const int64_t n_tps_pad = GGML_PAD(n_tps, GGML_KQ_MASK_PAD);
std::fill(data, data + ggml_nelements(dst), -INFINITY); std::fill(data, data + ggml_nelements(dst), -INFINITY);
@ -1266,7 +1265,7 @@ void llama_kv_cache::set_input_kq_mask(ggml_tensor * dst, const llama_ubatch * u
const llama_pos p1_x = is_2d ? ubatch->pos[i + ubatch->n_tokens*2] : 0; const llama_pos p1_x = is_2d ? ubatch->pos[i + ubatch->n_tokens*2] : 0;
const llama_pos p1_y = is_2d ? ubatch->pos[i + ubatch->n_tokens] : 0; const llama_pos p1_y = is_2d ? ubatch->pos[i + ubatch->n_tokens] : 0;
const uint64_t idst = n_kv*(h*n_stream*n_tps_pad + s*n_tps_pad + ii); const uint64_t idst = n_kv*(h*n_stream*n_tps + s*n_tps + ii);
for (uint32_t j = 0; j < n_kv; ++j) { for (uint32_t j = 0; j < n_kv; ++j) {
if (cells.is_empty(j)) { if (cells.is_empty(j)) {

2
tests/test-backend-ops.cpp
View File

@ -5875,7 +5875,7 @@ struct test_flash_attn_ext : public test_case {
ggml_tensor * m = nullptr; ggml_tensor * m = nullptr;
if (mask) { if (mask) {
m = ggml_new_tensor_4d(ctx, GGML_TYPE_F16, kv, GGML_PAD(nb, GGML_KQ_MASK_PAD), 1, nr23[1]); m = ggml_new_tensor_4d(ctx, GGML_TYPE_F16, kv, nb, 1, nr23[1]);
ggml_set_name(m, "m"); ggml_set_name(m, "m");
} }

170
tests/test-barrier.cpp
View File

@ -11,19 +11,7 @@
#define MAX_NARGS 2 #define MAX_NARGS 2
int main(int argc, char *argv[]) { static void test_barrier(int n_threads, int n_rounds) {
int n_threads = std::max(1, std::min(4, (int) std::thread::hardware_concurrency()));
int n_rounds = 100;
if (argc > 1) {
n_threads = std::atoi(argv[1]);
}
if (argc > 2) {
n_rounds = std::atoi(argv[2]);
}
struct ggml_init_params params = { struct ggml_init_params params = {
/* .mem_size = */ 1024*1024*1024, /* .mem_size = */ 1024*1024*1024,
/* .mem_buffer = */ NULL, /* .mem_buffer = */ NULL,
@ -56,7 +44,7 @@ int main(int argc, char *argv[]) {
exit(1); exit(1);
} }
// Create compute plan // The test runs with constant number of threads
struct ggml_cplan cplan = ggml_graph_plan(gf, n_threads, threadpool); struct ggml_cplan cplan = ggml_graph_plan(gf, n_threads, threadpool);
std::vector<uint8_t> work_data(cplan.work_size); std::vector<uint8_t> work_data(cplan.work_size);
@ -89,6 +77,160 @@ int main(int argc, char *argv[]) {
ggml_threadpool_free(threadpool); ggml_threadpool_free(threadpool);
ggml_free(ctx); ggml_free(ctx);
}
static void test_active(int n_threads, int n_rounds) {
struct ggml_init_params params = {
/* .mem_size = */ 1024*1024*1024,
/* .mem_buffer = */ NULL,
/* .no_alloc = */ false,
};
struct ggml_context * ctx = ggml_init(params);
// Create graph
struct ggml_cgraph * gf = ggml_new_graph(ctx);
// Small graph with, parallel ops with barriers
struct ggml_tensor * out = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, 64);
for (int i = 0; i < 2; i++) {
struct ggml_tensor * a = ggml_new_tensor_2d(ctx, GGML_TYPE_Q4_0, 64, 128);
out = ggml_mul_mat(ctx, a, out);
struct ggml_tensor * d = ggml_new_tensor_2d(ctx, GGML_TYPE_Q4_0, 128, 64);
out = ggml_mul_mat(ctx, d, out);
}
ggml_build_forward_expand(gf, out);
int n_nodes = ggml_graph_n_nodes(gf);
// Create threadpool
struct ggml_threadpool_params tpp = ggml_threadpool_params_default(n_threads);
struct ggml_threadpool* threadpool = ggml_threadpool_new(&tpp);
if (!threadpool) {
fprintf(stderr, "threadpool create failed : n_threads %d\n", n_threads);
exit(1);
}
std::cerr << "graph-compute with"
<< "\n n_threads: " << n_threads
<< "\n n_nodes: " << n_nodes
<< "\n n_rounds: " << n_rounds
<< "\n";
// ggml_graph_print(gf);
// In this test we keep changing the number of threads every 4th iteration
// to test for race conditions in that path
for (int i=0; i < n_rounds; i++) {
struct ggml_cplan cplan = ggml_graph_plan(gf, (i % 4) == 0 ? 1 : n_threads, threadpool);
std::vector<uint8_t> work_data(cplan.work_size);
cplan.work_data = work_data.data();
ggml_graph_compute(gf, &cplan);
}
ggml_threadpool_free(threadpool);
ggml_free(ctx);
}
static void test_multi_graph(int n_threads, int n_rounds) {
struct ggml_init_params params = {
/* .mem_size = */ 1024*1024*1024,
/* .mem_buffer = */ NULL,
/* .no_alloc = */ false,
};
struct ggml_context * ctx = ggml_init(params);
// Create graphs
struct ggml_cgraph * gf0 = ggml_new_graph(ctx);
{
// Small graph with parallel ops with barriers
struct ggml_tensor * out = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, 64);
for (int i = 0; i < 2; i++) {
struct ggml_tensor * a = ggml_new_tensor_2d(ctx, GGML_TYPE_Q4_0, 64, 128);
out = ggml_mul_mat(ctx, a, out);
struct ggml_tensor * d = ggml_new_tensor_2d(ctx, GGML_TYPE_Q4_0, 128, 64);
out = ggml_mul_mat(ctx, d, out);
}
ggml_build_forward_expand(gf0, out);
}
struct ggml_cgraph * gf1 = ggml_new_graph(ctx);
{
// Small graph with parallel ops with barriers
// Use larger tensors to make sure work_data size is larger than gf0
struct ggml_tensor * out = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, 256);
for (int i = 0; i < 4; i++) {
struct ggml_tensor * a = ggml_new_tensor_2d(ctx, GGML_TYPE_Q4_0, 256, 128);
out = ggml_mul_mat(ctx, a, out);
struct ggml_tensor * d = ggml_new_tensor_2d(ctx, GGML_TYPE_Q4_0, 128, 256);
out = ggml_mul_mat(ctx, d, out);
}
ggml_build_forward_expand(gf1, out);
}
// Create threadpool
struct ggml_threadpool_params tpp = ggml_threadpool_params_default(n_threads);
struct ggml_threadpool* threadpool = ggml_threadpool_new(&tpp);
if (!threadpool) {
fprintf(stderr, "threadpool create failed : n_threads %d\n", n_threads);
exit(1);
}
std::cerr << "graph-compute with"
<< "\n gf0 n_nodes: " << ggml_graph_n_nodes(gf0)
<< "\n gf1 n_nodes: " << ggml_graph_n_nodes(gf1)
<< "\n n_threads: " << n_threads
<< "\n n_rounds: " << n_rounds
<< "\n";
// In this test we keep changing the number of threads every 4th iteration
// and we compute two graphs back to back to test graph frequent graph switching
for (int i=0; i < n_rounds; i++) {
struct ggml_cplan cplan0 = ggml_graph_plan(gf0, (i % 4) == 0 ? 1 : n_threads, threadpool);
std::vector<uint8_t> work_data0(cplan0.work_size);
cplan0.work_data = work_data0.data();
struct ggml_cplan cplan1 = ggml_graph_plan(gf1, (i % 4) == 0 ? 1 : n_threads, threadpool);
std::vector<uint8_t> work_data1(cplan1.work_size);
cplan1.work_data = work_data1.data();
ggml_graph_compute(gf0, &cplan0);
ggml_graph_compute(gf1, &cplan1);
}
ggml_threadpool_free(threadpool);
ggml_free(ctx);
}
int main(int argc, char *argv[]) {
int n_threads = std::max(1, std::min(4, (int) std::thread::hardware_concurrency()));
int n_rounds = 100;
if (argc > 1) {
n_threads = std::atoi(argv[1]);
}
if (argc > 2) {
n_rounds = std::atoi(argv[2]);
}
test_barrier(n_threads, n_rounds);
test_active(n_threads, n_rounds * 100);
test_multi_graph(n_threads, n_rounds * 10);
return 0; return 0;
} }

4
tools/completion/completion.cpp
View File

@ -86,6 +86,10 @@ static void sigint_handler(int signo) {
int main(int argc, char ** argv) { int main(int argc, char ** argv) {
common_params params; common_params params;
g_params = &params; g_params = &params;
// disable jinja by default
params.use_jinja = false;
if (!common_params_parse(argc, argv, params, LLAMA_EXAMPLE_COMPLETION, print_usage)) { if (!common_params_parse(argc, argv, params, LLAMA_EXAMPLE_COMPLETION, print_usage)) {
return 1; return 1;
} }

158
tools/mtmd/clip.cpp
View File

@ -595,11 +595,12 @@ struct clip_graph {
cur = ggml_mul(ctx0, cur, model.mm_input_norm_w); cur = ggml_mul(ctx0, cur, model.mm_input_norm_w);
cur = ggml_add(ctx0, cur, model.mm_input_norm_b); cur = ggml_add(ctx0, cur, model.mm_input_norm_b);
cur = ggml_mul_mat(ctx0, model.mm_1_w, cur); cur = build_ffn(cur,
cur = ggml_add(ctx0, cur, model.mm_1_b); model.mm_1_w, model.mm_1_b,
cur = ggml_gelu(ctx0, cur); nullptr, nullptr,
cur = ggml_mul_mat(ctx0, model.mm_2_w, cur); model.mm_2_w, model.mm_2_b,
cur = ggml_add(ctx0, cur, model.mm_2_b); FFN_GELU,
-1);
} else if (ctx->proj_type() == PROJECTOR_TYPE_JANUS_PRO) { } else if (ctx->proj_type() == PROJECTOR_TYPE_JANUS_PRO) {
cur = build_ffn(cur, cur = build_ffn(cur,
@ -667,16 +668,12 @@ struct clip_graph {
// LlavaMultiModalProjector (always using GELU activation) // LlavaMultiModalProjector (always using GELU activation)
{ {
cur = ggml_mul_mat(ctx0, model.mm_1_w, cur); cur = build_ffn(cur,
if (model.mm_1_b) { model.mm_1_w, model.mm_1_b,
cur = ggml_add(ctx0, cur, model.mm_1_b); nullptr, nullptr,
} model.mm_2_w, model.mm_2_b,
FFN_GELU,
cur = ggml_gelu(ctx0, cur); -1);
cur = ggml_mul_mat(ctx0, model.mm_2_w, cur);
if (model.mm_2_b) {
cur = ggml_add(ctx0, cur, model.mm_2_b);
}
} }
// arrangement of the [IMG_BREAK] token // arrangement of the [IMG_BREAK] token
@ -775,10 +772,6 @@ struct clip_graph {
// if flash attn is used, we need to pad the mask and cast to f16 // if flash attn is used, we need to pad the mask and cast to f16
if (ctx->flash_attn_type == CLIP_FLASH_ATTN_TYPE_ENABLED) { if (ctx->flash_attn_type == CLIP_FLASH_ATTN_TYPE_ENABLED) {
int n_pad = GGML_PAD(window_mask->ne[1], GGML_KQ_MASK_PAD) - window_mask->ne[1];
if (n_pad > 0) {
window_mask = ggml_pad(ctx0, window_mask, 0, n_pad, 0, 0);
}
window_mask = ggml_cast(ctx0, window_mask, GGML_TYPE_F16); window_mask = ggml_cast(ctx0, window_mask, GGML_TYPE_F16);
} }
@ -791,7 +784,7 @@ struct clip_graph {
// loop over layers // loop over layers
for (int il = 0; il < n_layer; il++) { for (int il = 0; il < n_layer; il++) {
auto & layer = model.layers[il]; const auto & layer = model.layers[il];
const bool full_attn = use_window_attn ? (il + 1) % n_wa_pattern == 0 : true; const bool full_attn = use_window_attn ? (il + 1) % n_wa_pattern == 0 : true;
ggml_tensor * cur = inpL; // inpL = residual, cur = hidden_states ggml_tensor * cur = inpL; // inpL = residual, cur = hidden_states
@ -870,16 +863,12 @@ struct clip_graph {
// multimodal projection // multimodal projection
ggml_tensor * embeddings = inpL; ggml_tensor * embeddings = inpL;
embeddings = ggml_reshape_3d(ctx0, embeddings, n_embd * 4, n_pos / 4, batch_size); embeddings = ggml_reshape_3d(ctx0, embeddings, n_embd * 4, n_pos / 4, batch_size);
embeddings = build_ffn(embeddings,
embeddings = ggml_mul_mat(ctx0, model.mm_0_w, embeddings); model.mm_0_w, model.mm_0_b,
embeddings = ggml_add(ctx0, embeddings, model.mm_0_b); nullptr, nullptr,
model.mm_1_w, model.mm_1_b,
// GELU activation FFN_GELU,
embeddings = ggml_gelu(ctx0, embeddings); -1);
// Second linear layer
embeddings = ggml_mul_mat(ctx0, model.mm_1_w, embeddings);
embeddings = ggml_add(ctx0, embeddings, model.mm_1_b);
if (use_window_attn) { if (use_window_attn) {
window_idx = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, n_pos / 4); window_idx = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, n_pos / 4);
@ -1257,11 +1246,12 @@ struct clip_graph {
// projector LayerNorm uses pytorch's default eps = 1e-5 // projector LayerNorm uses pytorch's default eps = 1e-5
// ref: https://huggingface.co/OpenGVLab/InternVL3-8B-Instruct/blob/a34d3e4e129a5856abfd6aa6de79776484caa14e/modeling_internvl_chat.py#L79 // ref: https://huggingface.co/OpenGVLab/InternVL3-8B-Instruct/blob/a34d3e4e129a5856abfd6aa6de79776484caa14e/modeling_internvl_chat.py#L79
cur = build_norm(cur, model.mm_0_w, model.mm_0_b, NORM_TYPE_NORMAL, 1e-5, -1); cur = build_norm(cur, model.mm_0_w, model.mm_0_b, NORM_TYPE_NORMAL, 1e-5, -1);
cur = ggml_mul_mat(ctx0, model.mm_1_w, cur); cur = build_ffn(cur,
cur = ggml_add(ctx0, cur, model.mm_1_b); model.mm_1_w, model.mm_1_b,
cur = ggml_gelu(ctx0, cur); nullptr, nullptr,
cur = ggml_mul_mat(ctx0, model.mm_3_w, cur); model.mm_3_w, model.mm_3_b,
cur = ggml_add(ctx0, cur, model.mm_3_b); FFN_GELU,
-1);
} }
// build the graph // build the graph
@ -1412,11 +1402,12 @@ struct clip_graph {
cb(cur, "proj_inp_normed", -1); cb(cur, "proj_inp_normed", -1);
// projection mlp // projection mlp
cur = ggml_mul_mat(ctx0, model.mm_1_w, cur); cur = build_ffn(cur,
cur = ggml_add(ctx0, cur, model.mm_1_b); model.mm_1_w, model.mm_1_b,
cur = ggml_gelu(ctx0, cur); nullptr, nullptr,
cur = ggml_mul_mat(ctx0, model.mm_2_w, cur); model.mm_2_w, model.mm_2_b,
cur = ggml_add(ctx0, cur, model.mm_2_b); FFN_GELU,
-1);
cb(cur, "proj_out", -1); cb(cur, "proj_out", -1);
} }
@ -1887,9 +1878,12 @@ struct clip_graph {
} else if (ctx->proj_type() == PROJECTOR_TYPE_VOXTRAL) { } else if (ctx->proj_type() == PROJECTOR_TYPE_VOXTRAL) {
// projector // projector
cur = ggml_mul_mat(ctx0, model.mm_1_w, cur); cur = build_ffn(cur,
cur = ggml_gelu_erf(ctx0, cur); model.mm_1_w, model.mm_1_b,
cur = ggml_mul_mat(ctx0, model.mm_2_w, cur); nullptr, nullptr,
model.mm_2_w, model.mm_2_b,
FFN_GELU_ERF,
-1);
} else { } else {
GGML_ABORT("%s: unknown projector type", __func__); GGML_ABORT("%s: unknown projector type", __func__);
@ -2074,34 +2068,66 @@ private:
// self-attention // self-attention
{ {
ggml_tensor * Qcur = ggml_mul_mat(ctx0, layer.q_w, cur); ggml_tensor * Qcur = nullptr;
if (layer.q_b) { ggml_tensor * Kcur = nullptr;
Qcur = ggml_add(ctx0, Qcur, layer.q_b); ggml_tensor * Vcur = nullptr;
} if (layer.qkv_w != nullptr) {
// fused qkv
cur = ggml_mul_mat(ctx0, layer.qkv_w, cur);
if (layer.qkv_b != nullptr) {
cur = ggml_add(ctx0, cur, layer.qkv_b);
}
ggml_tensor * Kcur = ggml_mul_mat(ctx0, layer.k_w, cur); Qcur = ggml_view_3d(ctx0, cur, d_head, n_head, n_pos,
if (layer.k_b) { /* nb1 */ ggml_row_size(cur->type, d_head),
Kcur = ggml_add(ctx0, Kcur, layer.k_b); /* nb2 */ cur->nb[1],
} /* offset */ 0);
ggml_tensor * Vcur = ggml_mul_mat(ctx0, layer.v_w, cur); Kcur = ggml_view_3d(ctx0, cur, d_head, n_head, n_pos,
if (layer.v_b) { /* nb1 */ ggml_row_size(cur->type, d_head),
Vcur = ggml_add(ctx0, Vcur, layer.v_b); /* nb2 */ cur->nb[1],
} /* offset */ ggml_row_size(cur->type, n_embd));
if (layer.q_norm) { Vcur = ggml_view_3d(ctx0, cur, d_head, n_head, n_pos,
Qcur = build_norm(Qcur, layer.q_norm, NULL, norm_t, eps, il); /* nb1 */ ggml_row_size(cur->type, d_head),
cb(Qcur, "Qcur_norm", il); /* nb2 */ cur->nb[1],
} /* offset */ ggml_row_size(cur->type, 2 * n_embd));
if (layer.k_norm) { // TODO: q/k norm requires row size == n_embd, while here it's d_head
Kcur = build_norm(Kcur, layer.k_norm, NULL, norm_t, eps, il); // we can add support in the future if needed
cb(Kcur, "Kcur_norm", il); GGML_ASSERT(layer.q_norm == nullptr && layer.k_norm == nullptr);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, d_head, n_head, n_pos); } else {
Kcur = ggml_reshape_3d(ctx0, Kcur, d_head, n_head, n_pos); // separate q, k, v
Vcur = ggml_reshape_3d(ctx0, Vcur, d_head, n_head, n_pos); Qcur = ggml_mul_mat(ctx0, layer.q_w, cur);
if (layer.q_b) {
Qcur = ggml_add(ctx0, Qcur, layer.q_b);
}
Kcur = ggml_mul_mat(ctx0, layer.k_w, cur);
if (layer.k_b) {
Kcur = ggml_add(ctx0, Kcur, layer.k_b);
}
Vcur = ggml_mul_mat(ctx0, layer.v_w, cur);
if (layer.v_b) {
Vcur = ggml_add(ctx0, Vcur, layer.v_b);
}
if (layer.q_norm) {
Qcur = build_norm(Qcur, layer.q_norm, NULL, norm_t, eps, il);
cb(Qcur, "Qcur_norm", il);
}
if (layer.k_norm) {
Kcur = build_norm(Kcur, layer.k_norm, NULL, norm_t, eps, il);
cb(Kcur, "Kcur_norm", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, d_head, n_head, n_pos);
Kcur = ggml_reshape_3d(ctx0, Kcur, d_head, n_head, n_pos);
Vcur = ggml_reshape_3d(ctx0, Vcur, d_head, n_head, n_pos);
}
cb(Qcur, "Qcur", il); cb(Qcur, "Qcur", il);
cb(Kcur, "Kcur", il); cb(Kcur, "Kcur", il);

5
tools/mtmd/mtmd-cli.cpp
View File

@ -270,6 +270,7 @@ int main(int argc, char ** argv) {
ggml_time_init(); ggml_time_init();
common_params params; common_params params;
params.use_jinja = false; // disable jinja by default
params.sampling.temp = 0.2; // lower temp by default for better quality params.sampling.temp = 0.2; // lower temp by default for better quality
if (!common_params_parse(argc, argv, params, LLAMA_EXAMPLE_MTMD, show_additional_info)) { if (!common_params_parse(argc, argv, params, LLAMA_EXAMPLE_MTMD, show_additional_info)) {
@ -317,7 +318,9 @@ int main(int argc, char ** argv) {
g_is_generating = true; g_is_generating = true;
if (params.prompt.find(mtmd_default_marker()) == std::string::npos) { if (params.prompt.find(mtmd_default_marker()) == std::string::npos) {
for (size_t i = 0; i < params.image.size(); i++) { for (size_t i = 0; i < params.image.size(); i++) {
params.prompt += mtmd_default_marker(); // most models require the marker before each image
// ref: https://github.com/ggml-org/llama.cpp/pull/17616
params.prompt = mtmd_default_marker() + params.prompt;
} }
} }
common_chat_msg msg; common_chat_msg msg;

51
tools/mtmd/tests.sh
View File

@ -32,23 +32,32 @@ fi
arr_prefix=() arr_prefix=()
arr_hf=() arr_hf=()
arr_tmpl=() # chat template arr_extra_args=()
arr_file=() arr_file=()
add_test_vision() { add_test_vision() {
local hf=$1 local hf=$1
local tmpl=${2:-""} # default to empty string if not provided shift
local extra_args=""
if [ $# -gt 0 ]; then
extra_args=$(printf " %q" "$@")
fi
arr_prefix+=("[vision]") arr_prefix+=("[vision]")
arr_hf+=("$hf") arr_hf+=("$hf")
arr_tmpl+=("$tmpl") arr_extra_args+=("$extra_args")
arr_file+=("test-1.jpeg") arr_file+=("test-1.jpeg")
} }
add_test_audio() { add_test_audio() {
local hf=$1 local hf=$1
shift
local extra_args=""
if [ $# -gt 0 ]; then
extra_args=$(printf " %q" "$@")
fi
arr_prefix+=("[audio] ") arr_prefix+=("[audio] ")
arr_hf+=("$hf") arr_hf+=("$hf")
arr_tmpl+=("") # no need for chat tmpl arr_extra_args+=("$extra_args")
arr_file+=("test-2.mp3") arr_file+=("test-2.mp3")
} }
@ -56,9 +65,9 @@ add_test_vision "ggml-org/SmolVLM-500M-Instruct-GGUF:Q8_0"
add_test_vision "ggml-org/SmolVLM2-2.2B-Instruct-GGUF:Q4_K_M" add_test_vision "ggml-org/SmolVLM2-2.2B-Instruct-GGUF:Q4_K_M"
add_test_vision "ggml-org/SmolVLM2-500M-Video-Instruct-GGUF:Q8_0" add_test_vision "ggml-org/SmolVLM2-500M-Video-Instruct-GGUF:Q8_0"
add_test_vision "ggml-org/gemma-3-4b-it-GGUF:Q4_K_M" add_test_vision "ggml-org/gemma-3-4b-it-GGUF:Q4_K_M"
add_test_vision "THUDM/glm-edge-v-5b-gguf:Q4_K_M" add_test_vision "THUDM/glm-edge-v-5b-gguf:Q4_K_M" -p "name of the newspaper?<__media__>"
add_test_vision "second-state/Llava-v1.5-7B-GGUF:Q2_K" "vicuna" add_test_vision "second-state/Llava-v1.5-7B-GGUF:Q2_K" --chat-template vicuna
add_test_vision "cjpais/llava-1.6-mistral-7b-gguf:Q3_K_M" "vicuna" add_test_vision "cjpais/llava-1.6-mistral-7b-gguf:Q3_K_M" --chat-template vicuna
add_test_vision "ibm-research/granite-vision-3.2-2b-GGUF:Q4_K_M" add_test_vision "ibm-research/granite-vision-3.2-2b-GGUF:Q4_K_M"
add_test_vision "second-state/MiniCPM-Llama3-V-2_5-GGUF:Q2_K" # model from openbmb is corrupted add_test_vision "second-state/MiniCPM-Llama3-V-2_5-GGUF:Q2_K" # model from openbmb is corrupted
add_test_vision "openbmb/MiniCPM-V-2_6-gguf:Q2_K" add_test_vision "openbmb/MiniCPM-V-2_6-gguf:Q2_K"
@ -79,7 +88,7 @@ add_test_audio "ggml-org/Voxtral-Mini-3B-2507-GGUF:Q4_K_M"
# to test the big models, run: ./tests.sh big # to test the big models, run: ./tests.sh big
if [ "$RUN_BIG_TESTS" = true ]; then if [ "$RUN_BIG_TESTS" = true ]; then
add_test_vision "ggml-org/pixtral-12b-GGUF:Q4_K_M" add_test_vision "ggml-org/pixtral-12b-GGUF:Q4_K_M"
add_test_vision "ggml-org/Mistral-Small-3.1-24B-Instruct-2503-GGUF" "mistral-v7" add_test_vision "ggml-org/Mistral-Small-3.1-24B-Instruct-2503-GGUF" --chat-template mistral-v7
add_test_vision "ggml-org/Qwen2-VL-2B-Instruct-GGUF:Q4_K_M" add_test_vision "ggml-org/Qwen2-VL-2B-Instruct-GGUF:Q4_K_M"
add_test_vision "ggml-org/Qwen2-VL-7B-Instruct-GGUF:Q4_K_M" add_test_vision "ggml-org/Qwen2-VL-7B-Instruct-GGUF:Q4_K_M"
add_test_vision "ggml-org/Qwen2.5-VL-3B-Instruct-GGUF:Q4_K_M" add_test_vision "ggml-org/Qwen2.5-VL-3B-Instruct-GGUF:Q4_K_M"
@ -89,7 +98,7 @@ if [ "$RUN_BIG_TESTS" = true ]; then
add_test_vision "ggml-org/InternVL3-14B-Instruct-GGUF:Q4_K_M" add_test_vision "ggml-org/InternVL3-14B-Instruct-GGUF:Q4_K_M"
add_test_vision "ggml-org/Qwen2.5-Omni-7B-GGUF:Q4_K_M" add_test_vision "ggml-org/Qwen2.5-Omni-7B-GGUF:Q4_K_M"
# add_test_vision "ggml-org/Qwen2.5-VL-32B-Instruct-GGUF:Q4_K_M" # does not work on my mac M3 Ultra # add_test_vision "ggml-org/Qwen2.5-VL-32B-Instruct-GGUF:Q4_K_M" # does not work on my mac M3 Ultra
add_test_vision "ggml-org/Kimi-VL-A3B-Thinking-2506-GGUF:Q4_K_M" # add_test_vision "ggml-org/Kimi-VL-A3B-Thinking-2506-GGUF:Q4_K_M" # not always working
add_test_audio "ggml-org/ultravox-v0_5-llama-3_1-8b-GGUF:Q4_K_M" add_test_audio "ggml-org/ultravox-v0_5-llama-3_1-8b-GGUF:Q4_K_M"
add_test_audio "ggml-org/Qwen2.5-Omni-7B-GGUF:Q4_K_M" add_test_audio "ggml-org/Qwen2.5-Omni-7B-GGUF:Q4_K_M"
@ -122,21 +131,25 @@ for i in "${!arr_hf[@]}"; do
bin="llama-mtmd-cli" bin="llama-mtmd-cli"
prefix="${arr_prefix[$i]}" prefix="${arr_prefix[$i]}"
hf="${arr_hf[$i]}" hf="${arr_hf[$i]}"
tmpl="${arr_tmpl[$i]}" extra_args="${arr_extra_args[$i]}"
inp_file="${arr_file[$i]}" inp_file="${arr_file[$i]}"
echo "Running test with binary: $bin and HF model: $hf" echo "Running test with binary: $bin and HF model: $hf"
echo "" echo ""
echo "" echo ""
output=$(\ cmd="$(printf %q "$PROJ_ROOT/build/bin/$bin") \
"$PROJ_ROOT/build/bin/$bin" \ -hf $(printf %q "$hf") \
-hf "$hf" \ --image $(printf %q "$SCRIPT_DIR/$inp_file") \
--image $SCRIPT_DIR/$inp_file \
-p "what is the publisher name of the newspaper?" \
--temp 0 -n 128 \ --temp 0 -n 128 \
${tmpl:+--chat-template "$tmpl"} \ ${extra_args}"
2>&1 | tee /dev/tty)
# if extra_args does not contain -p, we add a default prompt
if ! [[ "$extra_args" =~ "-p" ]]; then
cmd+=" -p \"what is the publisher name of the newspaper?\""
fi
output=$(eval "$cmd" 2>&1 | tee /dev/tty)
echo "$output" > $SCRIPT_DIR/output/$bin-$(echo "$hf" | tr '/' '-').log echo "$output" > $SCRIPT_DIR/output/$bin-$(echo "$hf" | tr '/' '-').log
@ -144,9 +157,9 @@ for i in "${!arr_hf[@]}"; do
if echo "$output" | grep -iq "new york" \ if echo "$output" | grep -iq "new york" \
|| (echo "$output" | grep -iq "men" && echo "$output" | grep -iq "walk") || (echo "$output" | grep -iq "men" && echo "$output" | grep -iq "walk")
then then
result="$prefix \033[32mOK\033[0m: $bin $hf" result="$prefix \033[32mOK\033[0m: $hf"
else else
result="$prefix \033[31mFAIL\033[0m: $bin $hf" result="$prefix \033[31mFAIL\033[0m: $hf"
fi fi
echo -e "$result" echo -e "$result"
arr_res+=("$result") arr_res+=("$result")

8
tools/server/CMakeLists.txt
View File

@ -38,6 +38,14 @@ set(TARGET_SRCS
server-http.h server-http.h
server-models.cpp server-models.cpp
server-models.h server-models.h
server-task.cpp
server-task.h
server-queue.cpp
server-queue.h
server-common.cpp
server-common.h
server-context.cpp
server-context.h
) )
set(PUBLIC_ASSETS set(PUBLIC_ASSETS
index.html.gz index.html.gz

50
tools/server/README.md
View File

@ -1369,6 +1369,11 @@ llama-server
### Model sources ### Model sources
There are 3 possible sources for model files:
1. Cached models (controlled by the `LLAMA_CACHE` environment variable)
2. Custom model directory (set via the `--models-dir` argument)
3. Custom preset (set via the `--models-preset` argument)
By default, the router looks for models in the cache. You can add Hugging Face models to the cache with: By default, the router looks for models in the cache. You can add Hugging Face models to the cache with:
```sh ```sh
@ -1413,6 +1418,51 @@ llama-server -ctx 8192 -n 1024 -np 2
Note: model instances inherit both command line arguments and environment variables from the router server. Note: model instances inherit both command line arguments and environment variables from the router server.
Alternatively, you can also add GGUF based preset (see next section)
### Model presets
Model presets allow advanced users to define custom configurations using an `.ini` file:
```sh
llama-server --models-preset ./my-models.ini
```
Each section in the file defines a new preset. Keys within a section correspond to command-line arguments (without leading dashes). For example, the argument `--n-gpu-layer 123` is written as `n-gpu-layer = 123`.
Short argument forms (e.g., `c`, `ngl`) and environment variable names (e.g., `LLAMA_ARG_N_GPU_LAYERS`) are also supported as keys.
Example:
```ini
version = 1
; If the key corresponds to an existing model on the server,
; this will be used as the default config for that model
[ggml-org/MY-MODEL-GGUF:Q8_0]
; string value
chat-template = chatml
; numeric value
n-gpu-layer = 123
; flag value (for certain flags, you need to use the "no-" prefix for negation)
jinja = true
; shorthand argument (for example, context size)
c = 4096
; environment variable name
LLAMA_ARG_CACHE_RAM = 0
; file paths are relative to server's CWD
model-draft = ./my-models/draft.gguf
; but it's RECOMMENDED to use absolute path
model-draft = /Users/abc/my-models/draft.gguf
; If the key does NOT correspond to an existing model,
; you need to specify at least the model path
[custom_model]
model = /Users/abc/my-awesome-model-Q4_K_M.gguf
```
Note: some arguments are controlled by router (e.g., host, port, API key, HF repo, model alias). They will be removed or overwritten upload loading.
### Routing requests ### Routing requests
Requests are routed according to the requested model name. Requests are routed according to the requested model name.

BIN
tools/server/public/index.html.gz
View File

Binary file not shown.

267
tools/server/server-models.cpp
View File

@ -1,6 +1,7 @@
#include "server-common.h" #include "server-common.h"
#include "server-models.h" #include "server-models.h"
#include "preset.h"
#include "download.h" #include "download.h"
#include <cpp-httplib/httplib.h> // TODO: remove this once we use HTTP client from download.h #include <cpp-httplib/httplib.h> // TODO: remove this once we use HTTP client from download.h
@ -33,6 +34,10 @@
#define CMD_EXIT "exit" #define CMD_EXIT "exit"
// address for child process, this is needed because router may run on 0.0.0.0
// ref: https://github.com/ggml-org/llama.cpp/issues/17862
#define CHILD_ADDR "127.0.0.1"
static std::filesystem::path get_server_exec_path() { static std::filesystem::path get_server_exec_path() {
#if defined(_WIN32) #if defined(_WIN32)
wchar_t buf[32768] = { 0 }; // Large buffer to handle long paths wchar_t buf[32768] = { 0 }; // Large buffer to handle long paths
@ -132,6 +137,93 @@ static std::vector<local_model> list_local_models(const std::string & dir) {
return models; return models;
} }
//
// server_presets
//
server_presets::server_presets(int argc, char ** argv, common_params & base_params, const std::string & presets_path)
: ctx_params(common_params_parser_init(base_params, LLAMA_EXAMPLE_SERVER)) {
if (!presets_path.empty()) {
presets = common_presets_load(presets_path, ctx_params);
SRV_INF("Loaded %zu presets from %s\n", presets.size(), presets_path.c_str());
}
// populate reserved args (will be appended by the router)
for (auto & opt : ctx_params.options) {
if (opt.env == nullptr) {
continue;
}
std::string env = opt.env;
if (env == "LLAMA_ARG_PORT" ||
env == "LLAMA_ARG_HOST" ||
env == "LLAMA_ARG_ALIAS" ||
env == "LLAMA_ARG_API_KEY" ||
env == "LLAMA_ARG_MODELS_DIR" ||
env == "LLAMA_ARG_MODELS_MAX" ||
env == "LLAMA_ARG_MODELS_PRESET" ||
env == "LLAMA_ARG_MODEL" ||
env == "LLAMA_ARG_MMPROJ" ||
env == "LLAMA_ARG_HF_REPO" ||
env == "LLAMA_ARG_NO_MODELS_AUTOLOAD") {
control_args[env] = opt;
}
}
// read base args from router's argv
common_params_parse(argc, argv, LLAMA_EXAMPLE_SERVER, base_args);
// remove any router-controlled args from base_args
for (const auto & cargs : control_args) {
auto it = base_args.find(cargs.second);
if (it != base_args.end()) {
base_args.erase(it);
}
}
}
common_preset server_presets::get_preset(const std::string & name) {
auto it = presets.find(name);
if (it != presets.end()) {
return it->second;
}
return common_preset();
}
void server_presets::render_args(server_model_meta & meta) {
common_preset preset = meta.preset; // copy
// merging 3 kinds of args:
// 1. model-specific args (from preset)
// force removing control args if any
for (auto & cargs : control_args) {
if (preset.options.find(cargs.second) != preset.options.end()) {
SRV_WRN("Preset '%s' contains reserved arg '%s', removing it\n", preset.name.c_str(), cargs.second.args[0]);
preset.options.erase(cargs.second);
}
}
// 2. base args (from router)
// inherit from base args
for (const auto & [arg, value] : base_args) {
preset.options[arg] = value;
}
// 3. control args (from router)
// set control values
preset.options[control_args["LLAMA_ARG_HOST"]] = CHILD_ADDR;
preset.options[control_args["LLAMA_ARG_PORT"]] = std::to_string(meta.port);
preset.options[control_args["LLAMA_ARG_ALIAS"]] = meta.name;
if (meta.in_cache) {
preset.options[control_args["LLAMA_ARG_HF_REPO"]] = meta.name;
} else {
preset.options[control_args["LLAMA_ARG_MODEL"]] = meta.path;
if (!meta.path_mmproj.empty()) {
preset.options[control_args["LLAMA_ARG_MMPROJ"]] = meta.path_mmproj;
}
}
meta.args = preset.to_args();
// add back the binary path at the front
meta.args.insert(meta.args.begin(), get_server_exec_path().string());
}
// //
// server_models // server_models
// //
@ -140,7 +232,7 @@ server_models::server_models(
const common_params & params, const common_params & params,
int argc, int argc,
char ** argv, char ** argv,
char ** envp) : base_params(params) { char ** envp) : base_params(params), presets(argc, argv, base_params, params.models_preset) {
for (int i = 0; i < argc; i++) { for (int i = 0; i < argc; i++) {
base_args.push_back(std::string(argv[i])); base_args.push_back(std::string(argv[i]));
} }
@ -155,11 +247,58 @@ server_models::server_models(
LOG_WRN("failed to get server executable path: %s\n", e.what()); LOG_WRN("failed to get server executable path: %s\n", e.what());
LOG_WRN("using original argv[0] as fallback: %s\n", base_args[0].c_str()); LOG_WRN("using original argv[0] as fallback: %s\n", base_args[0].c_str());
} }
// TODO: allow refreshing cached model list load_models();
// add cached models }
void server_models::add_model(server_model_meta && meta) {
if (mapping.find(meta.name) != mapping.end()) {
throw std::runtime_error(string_format("model '%s' appears multiple times", meta.name.c_str()));
}
presets.render_args(meta); // populate meta.args
std::string name = meta.name;
mapping[name] = instance_t{
/* subproc */ std::make_shared<subprocess_s>(),
/* th */ std::thread(),
/* meta */ std::move(meta)
};
}
static std::vector<local_model> list_custom_path_models(server_presets & presets) {
// detect any custom-path models in presets
std::vector<local_model> custom_models;
for (auto & [model_name, preset] : presets.presets) {
local_model model;
model.name = model_name;
std::vector<common_arg> to_erase;
for (auto & [arg, value] : preset.options) {
std::string env(arg.env ? arg.env : "");
if (env == "LLAMA_ARG_MODEL") {
model.path = value;
to_erase.push_back(arg);
}
if (env == "LLAMA_ARG_MMPROJ") {
model.path_mmproj = value;
to_erase.push_back(arg);
}
}
for (auto & arg : to_erase) {
preset.options.erase(arg);
}
if (!model.name.empty() && !model.path.empty()) {
custom_models.push_back(model);
}
}
return custom_models;
}
// TODO: allow refreshing cached model list
void server_models::load_models() {
// loading models from 3 sources:
// 1. cached models
auto cached_models = common_list_cached_models(); auto cached_models = common_list_cached_models();
for (const auto & model : cached_models) { for (const auto & model : cached_models) {
server_model_meta meta{ server_model_meta meta{
/* preset */ presets.get_preset(model.to_string()),
/* name */ model.to_string(), /* name */ model.to_string(),
/* path */ model.manifest_path, /* path */ model.manifest_path,
/* path_mmproj */ "", // auto-detected when loading /* path_mmproj */ "", // auto-detected when loading
@ -170,21 +309,18 @@ server_models::server_models(
/* args */ std::vector<std::string>(), /* args */ std::vector<std::string>(),
/* exit_code */ 0 /* exit_code */ 0
}; };
mapping[meta.name] = instance_t{ add_model(std::move(meta));
/* subproc */ std::make_shared<subprocess_s>(),
/* th */ std::thread(),
/* meta */ meta
};
} }
// add local models specificed via --models-dir // 2. local models specificed via --models-dir
if (!params.models_dir.empty()) { if (!base_params.models_dir.empty()) {
auto local_models = list_local_models(params.models_dir); auto local_models = list_local_models(base_params.models_dir);
for (const auto & model : local_models) { for (const auto & model : local_models) {
if (mapping.find(model.name) != mapping.end()) { if (mapping.find(model.name) != mapping.end()) {
// already exists in cached models, skip // already exists in cached models, skip
continue; continue;
} }
server_model_meta meta{ server_model_meta meta{
/* preset */ presets.get_preset(model.name),
/* name */ model.name, /* name */ model.name,
/* path */ model.path, /* path */ model.path,
/* path_mmproj */ model.path_mmproj, /* path_mmproj */ model.path_mmproj,
@ -195,13 +331,31 @@ server_models::server_models(
/* args */ std::vector<std::string>(), /* args */ std::vector<std::string>(),
/* exit_code */ 0 /* exit_code */ 0
}; };
mapping[meta.name] = instance_t{ add_model(std::move(meta));
/* subproc */ std::make_shared<subprocess_s>(),
/* th */ std::thread(),
/* meta */ meta
};
} }
} }
// 3. custom-path models specified in presets
auto custom_models = list_custom_path_models(presets);
for (const auto & model : custom_models) {
server_model_meta meta{
/* preset */ presets.get_preset(model.name),
/* name */ model.name,
/* path */ model.path,
/* path_mmproj */ model.path_mmproj,
/* in_cache */ false,
/* port */ 0,
/* status */ SERVER_MODEL_STATUS_UNLOADED,
/* last_used */ 0,
/* args */ std::vector<std::string>(),
/* exit_code */ 0
};
add_model(std::move(meta));
}
// log available models
SRV_INF("Available models (%zu) (*: custom preset)\n", mapping.size());
for (const auto & [name, inst] : mapping) {
SRV_INF(" %c %s\n", inst.meta.preset.name.empty() ? ' ' : '*', name.c_str());
}
} }
void server_models::update_meta(const std::string & name, const server_model_meta & meta) { void server_models::update_meta(const std::string & name, const server_model_meta & meta) {
@ -335,19 +489,7 @@ void server_models::unload_lru() {
} }
} }
static void add_or_replace_arg(std::vector<std::string> & args, const std::string & key, const std::string & value) { void server_models::load(const std::string & name) {
for (size_t i = 0; i < args.size(); i++) {
if (args[i] == key && i + 1 < args.size()) {
args[i + 1] = value;
return;
}
}
// not found, append
args.push_back(key);
args.push_back(value);
}
void server_models::load(const std::string & name, bool auto_load) {
if (!has_model(name)) { if (!has_model(name)) {
throw std::runtime_error("model name=" + name + " is not found"); throw std::runtime_error("model name=" + name + " is not found");
} }
@ -376,26 +518,10 @@ void server_models::load(const std::string & name, bool auto_load) {
{ {
SRV_INF("spawning server instance with name=%s on port %d\n", inst.meta.name.c_str(), inst.meta.port); SRV_INF("spawning server instance with name=%s on port %d\n", inst.meta.name.c_str(), inst.meta.port);
std::vector<std::string> child_args; presets.render_args(inst.meta); // update meta.args
if (auto_load && !meta.args.empty()) {
child_args = meta.args; // copy previous args
} else {
child_args = base_args; // copy
if (inst.meta.in_cache) {
add_or_replace_arg(child_args, "-hf", inst.meta.name);
} else {
add_or_replace_arg(child_args, "-m", inst.meta.path);
if (!inst.meta.path_mmproj.empty()) {
add_or_replace_arg(child_args, "--mmproj", inst.meta.path_mmproj);
}
}
}
// set model args std::vector<std::string> child_args = inst.meta.args; // copy
add_or_replace_arg(child_args, "--port", std::to_string(inst.meta.port)); std::vector<std::string> child_env = base_env; // copy
add_or_replace_arg(child_args, "--alias", inst.meta.name);
std::vector<std::string> child_env = base_env; // copy
child_env.push_back("LLAMA_SERVER_ROUTER_PORT=" + std::to_string(base_params.port)); child_env.push_back("LLAMA_SERVER_ROUTER_PORT=" + std::to_string(base_params.port));
SRV_INF("%s", "spawning server instance with args:\n"); SRV_INF("%s", "spawning server instance with args:\n");
@ -541,7 +667,7 @@ bool server_models::ensure_model_loaded(const std::string & name) {
} }
if (meta->status == SERVER_MODEL_STATUS_UNLOADED) { if (meta->status == SERVER_MODEL_STATUS_UNLOADED) {
SRV_INF("model name=%s is not loaded, loading...\n", name.c_str()); SRV_INF("model name=%s is not loaded, loading...\n", name.c_str());
load(name, true); load(name);
} }
SRV_INF("waiting until model name=%s is fully loaded...\n", name.c_str()); SRV_INF("waiting until model name=%s is fully loaded...\n", name.c_str());
@ -571,7 +697,7 @@ server_http_res_ptr server_models::proxy_request(const server_http_req & req, co
SRV_INF("proxying request to model %s on port %d\n", name.c_str(), meta->port); SRV_INF("proxying request to model %s on port %d\n", name.c_str(), meta->port);
auto proxy = std::make_unique<server_http_proxy>( auto proxy = std::make_unique<server_http_proxy>(
method, method,
base_params.hostname, CHILD_ADDR,
meta->port, meta->port,
req.path, req.path,
req.headers, req.headers,
@ -724,38 +850,6 @@ void server_models_routes::init_routes() {
return models.proxy_request(req, method, name, true); // update last usage for POST request only return models.proxy_request(req, method, name, true); // update last usage for POST request only
}; };
this->get_router_models = [this](const server_http_req &) {
auto res = std::make_unique<server_http_res>();
json models_json = json::array();
auto all_models = models.get_all_meta();
std::time_t t = std::time(0);
for (const auto & meta : all_models) {
json status {
{"value", server_model_status_to_string(meta.status)},
{"args", meta.args},
};
if (meta.is_failed()) {
status["exit_code"] = meta.exit_code;
status["failed"] = true;
}
models_json.push_back(json {
{"id", meta.name},
{"object", "model"}, // for OAI-compat
{"owned_by", "llamacpp"}, // for OAI-compat
{"created", t}, // for OAI-compat
{"in_cache", meta.in_cache},
{"path", meta.path},
{"status", status},
// TODO: add other fields, may require reading GGUF metadata
});
}
res_ok(res, {
{"data", models_json},
{"object", "list"},
});
return res;
};
this->post_router_models_load = [this](const server_http_req & req) { this->post_router_models_load = [this](const server_http_req & req) {
auto res = std::make_unique<server_http_res>(); auto res = std::make_unique<server_http_res>();
json body = json::parse(req.body); json body = json::parse(req.body);
@ -769,7 +863,7 @@ void server_models_routes::init_routes() {
res_err(res, format_error_response("model is already loaded", ERROR_TYPE_INVALID_REQUEST)); res_err(res, format_error_response("model is already loaded", ERROR_TYPE_INVALID_REQUEST));
return res; return res;
} }
models.load(name, false); models.load(name);
res_ok(res, {{"success", true}}); res_ok(res, {{"success", true}});
return res; return res;
}; };
@ -793,9 +887,12 @@ void server_models_routes::init_routes() {
std::time_t t = std::time(0); std::time_t t = std::time(0);
for (const auto & meta : all_models) { for (const auto & meta : all_models) {
json status { json status {
{"value", server_model_status_to_string(meta.status)}, {"value", server_model_status_to_string(meta.status)},
{"args", meta.args}, {"args", meta.args},
}; };
if (!meta.preset.name.empty()) {
status["preset"] = meta.preset.to_ini();
}
if (meta.is_failed()) { if (meta.is_failed()) {
status["exit_code"] = meta.exit_code; status["exit_code"] = meta.exit_code;
status["failed"] = true; status["failed"] = true;

27
tools/server/server-models.h
View File

@ -1,6 +1,7 @@
#pragma once #pragma once
#include "common.h" #include "common.h"
#include "preset.h"
#include "server-http.h" #include "server-http.h"
#include <mutex> #include <mutex>
@ -47,6 +48,7 @@ static std::string server_model_status_to_string(server_model_status status) {
} }
struct server_model_meta { struct server_model_meta {
common_preset preset;
std::string name; std::string name;
std::string path; std::string path;
std::string path_mmproj; // only available if in_cache=false std::string path_mmproj; // only available if in_cache=false
@ -54,7 +56,7 @@ struct server_model_meta {
int port = 0; int port = 0;
server_model_status status = SERVER_MODEL_STATUS_UNLOADED; server_model_status status = SERVER_MODEL_STATUS_UNLOADED;
int64_t last_used = 0; // for LRU unloading int64_t last_used = 0; // for LRU unloading
std::vector<std::string> args; // additional args passed to the model instance (used for debugging) std::vector<std::string> args; // args passed to the model instance, will be populated by render_args()
int exit_code = 0; // exit code of the model instance process (only valid if status == FAILED) int exit_code = 0; // exit code of the model instance process (only valid if status == FAILED)
bool is_active() const { bool is_active() const {
@ -66,6 +68,19 @@ struct server_model_meta {
} }
}; };
// the server_presets struct holds the presets read from presets.ini
// as well as base args from the router server
struct server_presets {
common_presets presets;
common_params_context ctx_params;
std::map<common_arg, std::string> base_args;
std::map<std::string, common_arg> control_args; // args reserved for server control
server_presets(int argc, char ** argv, common_params & base_params, const std::string & models_dir);
common_preset get_preset(const std::string & name);
void render_args(server_model_meta & meta);
};
struct subprocess_s; struct subprocess_s;
struct server_models { struct server_models {
@ -85,14 +100,21 @@ private:
std::vector<std::string> base_args; std::vector<std::string> base_args;
std::vector<std::string> base_env; std::vector<std::string> base_env;
server_presets presets;
void update_meta(const std::string & name, const server_model_meta & meta); void update_meta(const std::string & name, const server_model_meta & meta);
// unload least recently used models if the limit is reached // unload least recently used models if the limit is reached
void unload_lru(); void unload_lru();
// not thread-safe, caller must hold mutex
void add_model(server_model_meta && meta);
public: public:
server_models(const common_params & params, int argc, char ** argv, char ** envp); server_models(const common_params & params, int argc, char ** argv, char ** envp);
void load_models();
// check if a model instance exists // check if a model instance exists
bool has_model(const std::string & name); bool has_model(const std::string & name);
@ -102,8 +124,7 @@ public:
// return a copy of all model metadata // return a copy of all model metadata
std::vector<server_model_meta> get_all_meta(); std::vector<server_model_meta> get_all_meta();
// if auto_load is true, load the model with previous args if any void load(const std::string & name);
void load(const std::string & name, bool auto_load);
void unload(const std::string & name); void unload(const std::string & name);
void unload_all(); void unload_all();