mirror of
https://github.com/ikawrakow/ik_llama.cpp.git
synced 2026-06-28 04:30:15 -05:00
11a1fea9e2
* refactor speculative decoding with companion context and draft result structures * feat: add common speculative feature handling in server context * refactor: move embedings outside server * feat: harden draft input hidden state in llama context * remove unused functions * refactor: streamline speculative feature handling and remove unused code * remove redundant code * remove more unused variables * refactor: implement speculative feature handling
1287 lines
48 KiB
C++
1287 lines
48 KiB
C++
//
|
|
// Copyright (C) 2024 Iwan Kawrakow
|
|
// Copyright (C) 2023-2024 The ggml authors
|
|
// MIT license
|
|
// SPDX-License-Identifier: MIT
|
|
//
|
|
|
|
#include "common.h"
|
|
#include "llama.h"
|
|
#include "llama-spec-features.h"
|
|
|
|
#include <cmath>
|
|
#include <cstdio>
|
|
#include <cstring>
|
|
#include <ctime>
|
|
#include <sstream>
|
|
#include <thread>
|
|
#include <mutex>
|
|
#include <vector>
|
|
#include <fstream>
|
|
#include <unordered_map>
|
|
#include <algorithm>
|
|
#include <optional>
|
|
#include <sstream>
|
|
|
|
#if defined(_MSC_VER)
|
|
#pragma warning(disable: 4244 4267) // possible loss of data
|
|
#endif
|
|
|
|
#if defined(__GNUC__) && (defined(__x86_64__) || defined(__i386__))
|
|
#include <immintrin.h>
|
|
__attribute__((target("avx2,fma")))
|
|
static inline bool add_and_check_nans_avx2(int n, const float * x, float * y, int * counts) {
|
|
int i = 0;
|
|
auto has_nans = _mm256_setzero_ps();
|
|
auto one = _mm256_set1_epi32(1);
|
|
{
|
|
__m256 vx[4], vy[4];
|
|
__m256i cy[4];
|
|
for ( ; i + 32 < n; i += 32) {
|
|
for (int k = 0; k < 4; ++k) {
|
|
vx[k] = _mm256_loadu_ps(x + i + 8*k);
|
|
vy[k] = _mm256_loadu_ps(y + i + 8*k);
|
|
cy[k] = _mm256_loadu_si256((const __m256i *)(counts + i + 8*k));
|
|
vy[k] = _mm256_fmadd_ps(vx[k], vx[k], vy[k]);
|
|
cy[k] = _mm256_add_epi32(cy[k], one);
|
|
auto mask = _mm256_cmp_ps(vx[k], vx[k], _CMP_UNORD_Q);
|
|
has_nans = _mm256_or_ps(has_nans, mask);
|
|
}
|
|
for (int k = 0; k < 4; ++k) {
|
|
_mm256_storeu_ps(y + i + 8*k, vy[k]);
|
|
_mm256_storeu_si256((__m256i *)(counts + i + 8*k), cy[k]);
|
|
}
|
|
}
|
|
}
|
|
for ( ; i + 7 < n; i += 8) {
|
|
auto vx = _mm256_loadu_ps(x + i);
|
|
auto vy = _mm256_loadu_ps(y + i);
|
|
auto cy = _mm256_loadu_si256((const __m256i *)(counts + i));
|
|
vy = _mm256_fmadd_ps(vx, vx, vy);
|
|
cy = _mm256_add_epi32(cy, one);
|
|
_mm256_storeu_ps(y + i, vy);
|
|
_mm256_storeu_si256((__m256i *)(counts + i), cy);
|
|
auto mask = _mm256_cmp_ps(vx, vx, _CMP_UNORD_Q);
|
|
has_nans = _mm256_or_ps(has_nans, mask);
|
|
}
|
|
auto has_any = _mm256_movemask_ps(has_nans);
|
|
if (has_any) {
|
|
return true;
|
|
}
|
|
for (; i < n; ++i) {
|
|
if (std::isnan(x[i])) {
|
|
return true;
|
|
}
|
|
y[i] += x[i]*x[i];
|
|
++counts[i];
|
|
}
|
|
return false;
|
|
}
|
|
#endif
|
|
static inline bool add_and_check_nans_scalar(int n, const float * x, float * y, int * counts) {
|
|
for (int i = 0; i < n; ++i) {
|
|
if (std::isnan(x[i])) {
|
|
return true;
|
|
}
|
|
y[i] += x[i]*x[i];
|
|
++counts[i];
|
|
}
|
|
return false;
|
|
}
|
|
static bool add_and_check_nans(int n, const float * x, float * y, int * counts) {
|
|
#if defined(__GNUC__) && (defined(__x86_64__) || defined(__i386__))
|
|
static const bool has_avx2 = __builtin_cpu_supports("avx2");
|
|
static const bool has_fma = __builtin_cpu_supports("fma");
|
|
if (has_avx2 && has_fma) {
|
|
return add_and_check_nans_avx2(n, x, y, counts);
|
|
}
|
|
#endif
|
|
return add_and_check_nans_scalar(n, x, y, counts);
|
|
}
|
|
|
|
void llama_set_mtp_target_context(struct llama_context * ctx, struct llama_context * target_ctx);
|
|
|
|
static llama_model * ik_load_model_from_params(const gpt_params & params, const llama_model_params & mparams) {
|
|
if (!params.hf_repo.empty() && !params.hf_file.empty()) {
|
|
return llama_load_model_from_hf(params.hf_repo.c_str(), params.hf_file.c_str(), params.model.c_str(), params.hf_token.c_str(), mparams);
|
|
}
|
|
if (!params.model_url.empty()) {
|
|
return llama_load_model_from_url(params.model_url.c_str(), params.model.c_str(), params.hf_token.c_str(), mparams);
|
|
}
|
|
|
|
return llama_model_load_from_file(params.model.c_str(), mparams);
|
|
}
|
|
|
|
static bool ik_model_has_arch(const llama_model * model, const char * expected_arch) {
|
|
char arch[64] = { 0 };
|
|
const int32_t len = llama_model_meta_val_str(model, "general.architecture", arch, sizeof(arch));
|
|
return len > 0 && std::string(arch) == expected_arch;
|
|
}
|
|
|
|
static llama_init_result ik_init_from_loaded_model(llama_model * model, gpt_params & params) {
|
|
llama_init_result iparams;
|
|
|
|
if (model == nullptr) {
|
|
return iparams;
|
|
}
|
|
|
|
auto cparams = common_context_params_to_llama(params);
|
|
|
|
llama_context * lctx = llama_init_from_model(model, cparams);
|
|
if (lctx == nullptr) {
|
|
fprintf(stderr, "%s: error: failed to create context with model '%s'\n", __func__, params.model.c_str());
|
|
llama_free_model(model);
|
|
return iparams;
|
|
}
|
|
|
|
for (auto [op, on_off] : params.offload_policy) {
|
|
llama_set_offload_policy(lctx, op, on_off);
|
|
}
|
|
|
|
if (!params.control_vectors.empty()) {
|
|
if (params.control_vector_layer_start <= 0) params.control_vector_layer_start = 1;
|
|
if (params.control_vector_layer_end <= 0) params.control_vector_layer_end = llama_n_layer(model);
|
|
|
|
const auto cvec = llama_control_vector_load(params.control_vectors);
|
|
if (cvec.n_embd == -1) {
|
|
llama_free(lctx);
|
|
llama_free_model(model);
|
|
return iparams;
|
|
}
|
|
|
|
const int err = llama_control_vector_apply(lctx,
|
|
cvec.data.data(),
|
|
cvec.data.size(),
|
|
cvec.n_embd,
|
|
params.control_vector_layer_start,
|
|
params.control_vector_layer_end);
|
|
if (err) {
|
|
llama_free(lctx);
|
|
llama_free_model(model);
|
|
return iparams;
|
|
}
|
|
}
|
|
|
|
for (auto & la : params.lora_adapters) {
|
|
llama_lora_adapter_container loaded_la;
|
|
loaded_la.path = la.path;
|
|
loaded_la.scale = la.scale;
|
|
loaded_la.adapter = llama_lora_adapter_init(model, la.path.c_str());
|
|
if (loaded_la.adapter == nullptr) {
|
|
fprintf(stderr, "%s: error: failed to apply lora adapter '%s'\n", __func__, la.path.c_str());
|
|
llama_free(lctx);
|
|
llama_free_model(model);
|
|
return iparams;
|
|
}
|
|
iparams.lora_adapters.push_back(loaded_la);
|
|
}
|
|
if (!params.lora_init_without_apply) {
|
|
llama_lora_adapters_apply(lctx, iparams.lora_adapters);
|
|
}
|
|
|
|
if (params.ignore_eos) {
|
|
params.sparams.logit_bias[llama_token_eos(model)] = -INFINITY;
|
|
}
|
|
|
|
if (params.sparams.dry_penalty_last_n == -1) {
|
|
LOG("%s: setting dry_penalty_last_n to ctx_size = %d\n", __func__, llama_n_ctx(lctx));
|
|
params.sparams.dry_penalty_last_n = llama_n_ctx(lctx);
|
|
}
|
|
|
|
if (params.warmup) {
|
|
LOG("warming up the model with an empty run\n");
|
|
|
|
std::vector<llama_token> tmp;
|
|
llama_token bos = llama_token_bos(model);
|
|
llama_token eos = llama_token_eos(model);
|
|
if (bos != -1) {
|
|
tmp.push_back(bos);
|
|
} else {
|
|
tmp.push_back(eos);
|
|
}
|
|
if (llama_model_has_encoder(model)) {
|
|
llama_encode(lctx, llama_batch_get_one(tmp.data(), tmp.size(), 0, 0));
|
|
llama_token decoder_start_token_id = llama_model_decoder_start_token(model);
|
|
if (decoder_start_token_id == LLAMA_TOKEN_NULL) {
|
|
decoder_start_token_id = bos;
|
|
}
|
|
tmp.clear();
|
|
tmp.push_back(decoder_start_token_id);
|
|
}
|
|
if (llama_model_has_decoder(model)) {
|
|
llama_decode(lctx, llama_batch_get_one(tmp.data(), std::min(tmp.size(), (size_t) params.n_batch), 0, 0));
|
|
}
|
|
llama_kv_cache_clear(lctx);
|
|
llama_synchronize(lctx);
|
|
llama_reset_timings(lctx);
|
|
}
|
|
|
|
iparams.model = model;
|
|
iparams.context = lctx;
|
|
return iparams;
|
|
}
|
|
|
|
static void print_usage(int argc, char ** argv, const gpt_params & params) {
|
|
gpt_params_print_usage(argc, argv, params);
|
|
|
|
LOG_TEE("\nexample usage:\n");
|
|
LOG_TEE("\n %s \\\n"
|
|
" -m model.gguf -f some-text.txt [-o imatrix.dat] [--process-output] [--verbosity 1] \\\n"
|
|
" [--no-ppl] [--chunk 123] [--output-frequency 10] [--save-frequency 0] \\\n"
|
|
" [--in-file imatrix-prev-0.dat --in-file imatrix-prev-1.dat ...]\n" , argv[0]);
|
|
LOG_TEE("\n");
|
|
}
|
|
|
|
struct Stats {
|
|
std::vector<float> values;
|
|
std::vector<int> counts;
|
|
int ncall = 0;
|
|
int n_as = 1;
|
|
};
|
|
|
|
class IMatrixCollector {
|
|
public:
|
|
IMatrixCollector() = default;
|
|
void set_params(gpt_params params) { m_params = std::move(params); }
|
|
bool collect_imatrix(struct ggml_tensor * t, bool ask, void * user_data);
|
|
void save_imatrix(int ncall = -1) const;
|
|
bool load_imatrix(const char * file_name);
|
|
void set_collect_lsim(bool yes_or_no) { m_collect_lsim = yes_or_no; }
|
|
void print_layer_importance();
|
|
private:
|
|
std::unordered_map<std::string, Stats> m_stats;
|
|
gpt_params m_params;
|
|
std::mutex m_mutex;
|
|
int m_last_call = 0;
|
|
int m_last_layer = 9999;
|
|
int m_last_ffn = -1;
|
|
std::vector<char> m_src1_data;
|
|
std::vector<char> m_ids; // the expert ids from ggml_mul_mat_id
|
|
std::vector<float> m_last_input;
|
|
std::vector<float> m_ffn_input;
|
|
std::vector<std::pair<double,int>> m_layer_sim;
|
|
std::vector<std::pair<double,int>> m_attn_sim;
|
|
std::vector<std::pair<double,int>> m_ffn_sim;
|
|
bool m_collect_lsim = false;
|
|
|
|
std::optional<int> layer_index(const std::string& name) const {
|
|
if (name == m_params.output_tensor_name && m_last_layer < 199) {
|
|
return m_last_layer + 1;
|
|
}
|
|
if (auto pos = name.find("blk."); pos == 0) {
|
|
pos += 4;
|
|
if (auto pos1 = name.find('.', pos); pos1 != std::string::npos) {
|
|
auto index_str = name.substr(pos, pos1 - pos);
|
|
std::istringstream str(index_str);
|
|
int index; str >> index;
|
|
if (!str.fail()) return index;
|
|
}
|
|
}
|
|
return std::nullopt;
|
|
}
|
|
|
|
static inline double cosine_similarity(int n, const float * x, const float * y) {
|
|
double sumxy = 0, sumx2 = 0, sumy2 = 0;
|
|
for (int j = 0; j < n; ++j) {
|
|
sumxy += x[j]*y[j]; sumx2 += x[j]*x[j]; sumy2 += y[j]*y[j];
|
|
}
|
|
double cos_sim = sumx2 > 0 && sumy2 > 0 ? sumxy/sqrt(sumx2*sumy2) : 0;
|
|
return cos_sim;
|
|
}
|
|
|
|
static inline void collect_cos_similarity(int nrow, int n, const float * x, const float * y, std::pair<double, int>& p) {
|
|
for (int row = 0; row < nrow; ++row) {
|
|
p.first += cosine_similarity(n, x, y);
|
|
p.second += 1;
|
|
x += n;
|
|
y += n;
|
|
}
|
|
}
|
|
|
|
static void print_layer_importance(const char * msg, const std::vector<std::pair<double, int>>& sim);
|
|
};
|
|
|
|
// remove any prefix and suffixes from the name
|
|
// CUDA0#blk.0.attn_k.weight#0 => blk.0.attn_k.weight
|
|
static std::string filter_tensor_name(const char * name) {
|
|
std::string wname;
|
|
const char * p = strchr(name, '#');
|
|
if (p != NULL) {
|
|
p = p + 1;
|
|
const char * q = strchr(p, '#');
|
|
if (q != NULL) {
|
|
wname = std::string(p, q - p);
|
|
} else {
|
|
wname = p;
|
|
}
|
|
} else {
|
|
wname = name;
|
|
}
|
|
return wname;
|
|
}
|
|
|
|
static bool is_named_imatrix_tensor(const std::string & wname, const gpt_params & params, bool collect_lsim) {
|
|
if (wname.rfind("blk.", 0) == 0) {
|
|
return true;
|
|
}
|
|
if (wname == "mtp_pre_proj.weight" || wname == "mtp_post_proj.weight") {
|
|
return true;
|
|
}
|
|
return (params.process_output || collect_lsim) && wname == params.output_tensor_name;
|
|
}
|
|
|
|
static std::string default_draft_imatrix_out_file(const std::string & target_out_file) {
|
|
if (target_out_file.empty()) {
|
|
return "imatrix-draft.dat";
|
|
}
|
|
|
|
const auto dot = target_out_file.rfind('.');
|
|
if (dot == std::string::npos || dot == 0) {
|
|
return target_out_file + "-draft";
|
|
}
|
|
|
|
return target_out_file.substr(0, dot) + "-draft" + target_out_file.substr(dot);
|
|
}
|
|
|
|
void IMatrixCollector::print_layer_importance(const char * msg, const std::vector<std::pair<double, int>>& sim) {
|
|
if (sim.empty()) return;
|
|
std::vector<std::pair<float, int>> layers;
|
|
layers.reserve(sim.size());
|
|
for (int i = 0; i < int(sim.size()); ++i) {
|
|
if (sim[i].second > 0) layers.emplace_back(float(std::abs(sim[i].first/sim[i].second)), i);
|
|
}
|
|
if (layers.empty()) return;
|
|
std::sort(layers.begin(), layers.end());
|
|
printf("%s\n", msg);
|
|
//printf("======================== sorted layer importances\n");
|
|
int j = 0;
|
|
for (auto& p : layers) {
|
|
int i = p.second;
|
|
printf("%3d: Layer %3d, <cos_sim> = %g\n", j++, i, sim[i].first/sim[i].second);
|
|
}
|
|
}
|
|
|
|
void IMatrixCollector::print_layer_importance() {
|
|
print_layer_importance("\n======================== sorted layer importances", m_layer_sim);
|
|
print_layer_importance("\n======================== sorted attention importances", m_attn_sim);
|
|
print_layer_importance("\n======================== sorted ffn importances", m_ffn_sim);
|
|
//printf("%s: have %d layers\n", __func__, int(m_layer_sim.size()));
|
|
//if (m_layer_sim.empty()) return;
|
|
//std::vector<std::pair<float, int>> layers;
|
|
//layers.reserve(m_layer_sim.size());
|
|
//for (int i = 0; i < int(m_layer_sim.size()); ++i) {
|
|
// if (m_layer_sim[i].second > 0) layers.emplace_back(float(std::abs(m_layer_sim[i].first/m_layer_sim[i].second)), i);
|
|
//}
|
|
//if (layers.empty()) return;
|
|
//std::sort(layers.begin(), layers.end());
|
|
//printf("======================== sorted layer importances\n");
|
|
//int j = 0;
|
|
//for (auto& p : layers) {
|
|
// int i = p.second;
|
|
// printf("%3d: Layer %3d, <cos_sim> = %g\n", j++, i, m_layer_sim[i].first/m_layer_sim[i].second);
|
|
//}
|
|
}
|
|
|
|
bool IMatrixCollector::collect_imatrix(struct ggml_tensor * t, bool ask, void * user_data) {
|
|
GGML_UNUSED(user_data);
|
|
|
|
const struct ggml_tensor * src0 = t->src[0];
|
|
const struct ggml_tensor * src1 = t->op == GGML_OP_FUSED_UP_GATE || t->op == GGML_OP_MOE_FUSED_UP_GATE ? t->src[2] : t->src[1];
|
|
std::string wname = filter_tensor_name(src0->name);
|
|
|
|
// when ask is true, the scheduler wants to know if we are interested in data from this tensor
|
|
// if we return true, a follow-up call will be made with ask=false in which we can do the actual collection
|
|
if (ask) {
|
|
if (t->op == GGML_OP_MUL_MAT_ID ||
|
|
t->op == GGML_OP_FUSED_UP_GATE ||
|
|
t->op == GGML_OP_MOE_FUSED_UP_GATE) return true; // collect all indirect matrix multiplications
|
|
if (t->op != GGML_OP_MUL_MAT) return false;
|
|
// why are small batches ignored (<16 tokens)?
|
|
if (src1->ne[1] < 16 || src1->type != GGML_TYPE_F32) return false;
|
|
if (!is_named_imatrix_tensor(wname, m_params, m_collect_lsim)) return false;
|
|
return true;
|
|
}
|
|
|
|
std::lock_guard<std::mutex> lock(m_mutex);
|
|
|
|
// copy the data from the GPU memory if needed
|
|
const bool is_host = ggml_backend_buffer_is_host(src1->buffer);
|
|
|
|
if (!is_host) {
|
|
auto nbytes = ggml_nbytes(src1);
|
|
m_src1_data.resize(nbytes);
|
|
ggml_backend_tensor_get(src1, m_src1_data.data(), 0, nbytes);
|
|
}
|
|
|
|
const float * data = is_host ? (const float *) src1->data : (const float *)m_src1_data.data();
|
|
|
|
if (m_collect_lsim) {
|
|
if (wname.find(".ffn_") != std::string::npos) {
|
|
if (auto index = layer_index(wname); index.has_value() && *index == m_last_layer && *index != m_last_ffn) {
|
|
int n = src1->ne[0];
|
|
int nrow = t->op == GGML_OP_MUL_MAT_ID ? src1->ne[2] : src1->ne[1];
|
|
if (t->op == GGML_OP_MUL_MAT_ID) {
|
|
GGML_ASSERT(src1->ne[1] == 1);
|
|
}
|
|
if (m_ffn_input.empty()) {
|
|
m_ffn_input.resize(nrow*n);
|
|
} else {
|
|
if ((int)m_ffn_input.size() != nrow*n) {
|
|
printf("Oops, inconsistent ffn size\n"); exit(1);
|
|
}
|
|
}
|
|
std::memcpy(m_ffn_input.data(), data, nrow*n*sizeof(float));
|
|
if (m_ffn_input.size() != m_last_input.size()) {
|
|
printf("Oops, inconsistent ffn vs last_input size\n"); exit(1);
|
|
}
|
|
if (m_attn_sim.size() < *index + 1) m_attn_sim.resize(*index + 1);
|
|
auto& p = m_attn_sim[*index];
|
|
collect_cos_similarity(nrow, n, m_ffn_input.data(), m_last_input.data(), p);
|
|
m_last_ffn = *index;
|
|
}
|
|
}
|
|
}
|
|
|
|
// this has been adapted to the new format of storing merged experts in a single 3d tensor
|
|
// ref: https://github.com/ggerganov/llama.cpp/pull/6387
|
|
if (t->op == GGML_OP_MUL_MAT_ID || t->op == GGML_OP_MOE_FUSED_UP_GATE) {
|
|
// ids -> [n_experts_used, n_tokens]
|
|
// src1 -> [cols, n_expert_used, n_tokens]
|
|
const ggml_tensor * ids = t->op == GGML_OP_MUL_MAT_ID ? t->src[2] : t->src[3];
|
|
const int n_as = src0->ne[2];
|
|
const int n_ids = ids->ne[0];
|
|
|
|
// the top-k selected expert ids are stored in the ids tensor
|
|
// for simplicity, always copy ids to host, because it is small
|
|
// take into account that ids is not contiguous!
|
|
|
|
GGML_ASSERT(ids->ne[1] == src1->ne[2]);
|
|
|
|
m_ids.resize(ggml_nbytes(ids));
|
|
ggml_backend_tensor_get(ids, m_ids.data(), 0, ggml_nbytes(ids));
|
|
|
|
auto & e = m_stats[wname];
|
|
|
|
++e.ncall;
|
|
|
|
if (e.values.empty()) {
|
|
e.values.resize(src1->ne[0]*n_as, 0);
|
|
e.counts.resize(src1->ne[0]*n_as, 0);
|
|
e.n_as = n_as;
|
|
}
|
|
else if (e.values.size() != (size_t)src1->ne[0]*n_as) {
|
|
fprintf(stderr, "Oops: inconsistent size for %s (%d vs %d)\n", wname.c_str(), (int)e.values.size(), (int)src1->ne[0]*n_as);
|
|
exit(1); //GGML_ABORT("fatal error");
|
|
}
|
|
else if (e.n_as != n_as) {
|
|
fprintf(stderr, "Oops: inconsistent n_as for %s (%d vs %d)\n", wname.c_str(), e.n_as, n_as);
|
|
}
|
|
if (m_params.verbosity > 1) {
|
|
printf("%s[%d]: %32s, %s, %5d x %5d, %d\n", __func__, m_last_call, wname.c_str(), ggml_op_name(t->op), (int)src1->ne[0], (int)src1->ne[2], (int)src1->type);
|
|
}
|
|
// loop over all possible experts, regardless if they are used or not in the batch
|
|
for (int ex = 0; ex < n_as; ++ex) {
|
|
size_t e_start = ex*src1->ne[0];
|
|
|
|
for (int idx = 0; idx < n_ids; ++idx) {
|
|
for (int row = 0; row < (int)src1->ne[2]; ++row) {
|
|
const int excur = *(const int32_t *) (m_ids.data() + row*ids->nb[1] + idx*ids->nb[0]);
|
|
|
|
GGML_ASSERT(excur >= 0 && excur < n_as); // sanity check
|
|
|
|
if (excur != ex) continue;
|
|
|
|
const int64_t i11 = idx % src1->ne[1];
|
|
const int64_t i12 = row;
|
|
const float * x = (const float *)((const char *)data + i11*src1->nb[1] + i12*src1->nb[2]);
|
|
|
|
if (add_and_check_nans(src1->ne[0], x, e.values.data() + e_start, e.counts.data() + e_start)) {
|
|
fprintf(stderr, "etected NaNs in %s\n", wname.c_str());
|
|
exit(1);
|
|
}
|
|
//for (int j = 0; j < (int)src1->ne[0]; ++j) {
|
|
// e.values[e_start + j] += x[j]*x[j];
|
|
// e.counts[e_start + j]++;
|
|
// if (!std::isfinite(e.values[e_start + j])) {
|
|
// fprintf(stderr, "%f detected in %s\n", e.values[e_start + j], wname.c_str());
|
|
// exit(1);
|
|
// }
|
|
//}
|
|
}
|
|
}
|
|
if (e.ncall > m_last_call) {
|
|
m_last_call = e.ncall;
|
|
if (m_last_call % m_params.n_out_freq == 0) {
|
|
save_imatrix();
|
|
}
|
|
if (m_params.n_save_freq > 0 && m_last_call%m_params.n_save_freq == 0) {
|
|
save_imatrix(m_last_call);
|
|
}
|
|
}
|
|
}
|
|
} else {
|
|
if (m_collect_lsim) {
|
|
// We only need to do it here and not in the MoE branch above because the first tensor in a layer
|
|
// never is a MoE tensor
|
|
if (auto index = layer_index(wname); index.has_value()) {
|
|
if (*index != m_last_layer) {
|
|
if (*index > 0) {
|
|
if (m_last_input.size() != src1->ne[0]*src1->ne[1]) {
|
|
printf("Oops: different size (%d vs %d). Tensor name was %s, m_last_layer = %d\n",
|
|
(int)(src1->ne[0]*src1->ne[1]), (int)m_last_input.size(), src0->name, m_last_layer);
|
|
exit(1);
|
|
}
|
|
if (*index > m_layer_sim.size()) m_layer_sim.resize(*index);
|
|
auto& p = m_layer_sim[*index - 1];
|
|
collect_cos_similarity(src1->ne[1], src1->ne[0], m_last_input.data(), (const float *)data, p);
|
|
if (*index == m_last_ffn + 1) {
|
|
if (*index > m_ffn_sim.size()) m_ffn_sim.resize(*index);
|
|
auto& p1 = m_ffn_sim[*index-1];
|
|
collect_cos_similarity(src1->ne[1], src1->ne[0], m_ffn_input.data(), (const float *)data, p1);
|
|
}
|
|
}
|
|
m_last_layer = *index;
|
|
if (m_last_input.empty()) {
|
|
m_last_input.resize(src1->ne[0]*src1->ne[1]);
|
|
} else {
|
|
if (m_last_input.size() != src1->ne[0]*src1->ne[1]) {
|
|
printf("Oops\n"); exit(1);
|
|
}
|
|
}
|
|
//printf("Copying src1 to m_last_input\n");
|
|
std::memcpy(m_last_input.data(), data, src1->ne[0]*src1->ne[1]*sizeof(float));
|
|
}
|
|
}
|
|
}
|
|
auto & e = m_stats[wname];
|
|
if (e.values.empty()) {
|
|
if (src0->ne[3] > 1) {
|
|
fprintf(stderr, "Unsupported 4D tensor %s\n", wname.c_str());
|
|
exit(1);
|
|
}
|
|
// If we have a 3D tensor as it is the case for the attn_k_b and attn_v_b for DeepSeek MLA models,
|
|
// than we need to compute the imatrix for each head, and not just one imatrx for all heads.
|
|
// Hence, the storage we need is src0->ne[0]*src0->ne[2].
|
|
e.values.resize(src0->ne[0]*src0->ne[2], 0);
|
|
e.counts.resize(src0->ne[0]*src0->ne[2], 0);
|
|
}
|
|
else if (e.values.size() != (size_t)(src0->ne[0]*src0->ne[2])) {
|
|
fprintf(stderr, "Oops: inconsistent size for %s (%d vs %d)\n", wname.c_str(), (int)e.values.size(), (int)src1->ne[0]);
|
|
exit(1); //GGML_ABORT("fatal error");
|
|
}
|
|
++e.ncall;
|
|
if (m_params.verbosity > 1) {
|
|
printf("%s[%d]: %32s, %s, %5d x %5d, %d\n", __func__, m_last_call, wname.c_str(), ggml_op_name(t->op), (int)src1->ne[0], (int)src1->ne[1], (int)src1->type);
|
|
}
|
|
int rk2 = src1->ne[2]/src0->ne[2];
|
|
for (int i12 = 0; i12 < (int)src1->ne[2]; ++i12) { // i.e., loop over attention heads for MLA models
|
|
int i02 = i12/rk2;
|
|
auto values = e.values.data() + i02*src0->ne[0];
|
|
auto counts = e.counts.data() + i02*src0->ne[0];
|
|
for (int i11 = 0; i11 < (int)src1->ne[1]; ++i11) {
|
|
const float * x = (const float *)((const char *)data + i11*src1->nb[1] + i12*src1->nb[2]);
|
|
if (add_and_check_nans(src1->ne[0], x, values, counts)) {
|
|
fprintf(stderr, "detected NaNs in %s\n", wname.c_str());
|
|
exit(1);
|
|
}
|
|
//for (int j = 0; j < (int)src1->ne[0]; ++j) {
|
|
// values[j] += x[j]*x[j];
|
|
// counts[j]++;
|
|
// if (!std::isfinite(values[j])) {
|
|
// fprintf(stderr, "%f detected in %s\n", values[j], wname.c_str());
|
|
// exit(1);
|
|
// }
|
|
//}
|
|
}
|
|
}
|
|
if (e.ncall > m_last_call) {
|
|
m_last_call = e.ncall;
|
|
if (m_last_call % m_params.n_out_freq == 0) {
|
|
save_imatrix();
|
|
}
|
|
if (m_params.n_save_freq > 0 && m_last_call%m_params.n_save_freq == 0) {
|
|
save_imatrix(m_last_call);
|
|
}
|
|
}
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
void IMatrixCollector::save_imatrix(int ncall) const {
|
|
auto fname = m_params.out_file;
|
|
if (fname.empty()) {
|
|
fname = "imatrix.dat";
|
|
}
|
|
|
|
if (ncall > 0) {
|
|
fname += ".at_";
|
|
fname += std::to_string(ncall);
|
|
}
|
|
|
|
// avoid writing imatrix entries that do not have full data
|
|
// this can happen with MoE models where some of the experts end up not being exercised by the provided training data
|
|
|
|
int n_entries = 0;
|
|
std::vector<std::string> to_store;
|
|
|
|
bool is_first = true; // for printing
|
|
for (const auto & kv : m_stats) {
|
|
const int n_all = kv.second.counts.size();
|
|
|
|
if (n_all == 0) {
|
|
continue;
|
|
}
|
|
|
|
int n_zeros = 0;
|
|
for (const int c : kv.second.counts) {
|
|
if (c == 0) {
|
|
n_zeros++;
|
|
}
|
|
}
|
|
|
|
if (n_zeros != 0 && is_first) {
|
|
fprintf(stderr, "\n");
|
|
is_first = false;
|
|
}
|
|
|
|
if (n_zeros == n_all) {
|
|
fprintf(stderr, "%s: entry '%40s' has no data - skipping\n", __func__, kv.first.c_str());
|
|
continue;
|
|
}
|
|
|
|
if (n_zeros > 0) {
|
|
fprintf(stderr, "%s: entry '%40s' has partial data (%.2f%%)", __func__, kv.first.c_str(), 100.0f * (n_all - n_zeros) / n_all);
|
|
bool store_it = false;
|
|
if (kv.second.n_as > 1) {
|
|
int n_per_expert = n_all / kv.second.n_as;
|
|
std::vector<int> bad_experts;
|
|
bad_experts.reserve(kv.second.n_as);
|
|
for (int i = 0; i < kv.second.n_as; ++i) {
|
|
auto counts = kv.second.counts.data() + i*n_per_expert;
|
|
int nz_i = 0;
|
|
for (int j = 0; j < n_per_expert; ++j) {
|
|
if (counts[j] == 0) ++nz_i;
|
|
}
|
|
if (nz_i > 0) bad_experts.push_back(i);
|
|
}
|
|
fprintf(stderr, " %d out of %d experts are missing data", int(bad_experts.size()), kv.second.n_as);
|
|
if (bad_experts.size() < round(kv.second.n_as * 0.05)) {
|
|
fprintf(stderr, " Storing **but be aware**\n");
|
|
store_it = true;
|
|
for (auto i : bad_experts) {
|
|
auto counts = (int *)kv.second.counts.data() + i*n_per_expert;
|
|
auto values = (float *)kv.second.values.data() + i*n_per_expert;
|
|
for (int j = 0; j < n_per_expert; ++j) {
|
|
counts[j] = 1;
|
|
values[j] = 1;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
if (!store_it) {
|
|
fprintf(stderr, " - skipping\n");
|
|
continue;
|
|
}
|
|
}
|
|
|
|
n_entries++;
|
|
to_store.push_back(kv.first);
|
|
}
|
|
|
|
if (to_store.size() < m_stats.size()) {
|
|
fprintf(stderr, "%s: warning: storing only %zu out of %zu entries\n", __func__, to_store.size(), m_stats.size());
|
|
}
|
|
|
|
std::ofstream out(fname, std::ios::binary);
|
|
out.write((const char *) &n_entries, sizeof(n_entries));
|
|
for (const auto & name : to_store) {
|
|
const auto & stat = m_stats.at(name);
|
|
int len = name.size();
|
|
out.write((const char *) &len, sizeof(len));
|
|
out.write(name.c_str(), len);
|
|
out.write((const char *) &stat.ncall, sizeof(stat.ncall));
|
|
int nval = stat.values.size();
|
|
out.write((const char *) &nval, sizeof(nval));
|
|
if (nval > 0) {
|
|
std::vector<float> tmp(nval);
|
|
for (int i = 0; i < nval; i++) {
|
|
tmp[i] = (stat.values[i] / static_cast<float>(stat.counts[i])) * static_cast<float>(stat.ncall);
|
|
}
|
|
out.write((const char*)tmp.data(), nval*sizeof(float));
|
|
}
|
|
}
|
|
|
|
// Write the number of call the matrix was computed with
|
|
out.write((const char *) &m_last_call, sizeof(m_last_call));
|
|
|
|
// Write the input filename at the end of the file to later on specify it in quantize
|
|
{
|
|
int len = m_params.prompt_file.size();
|
|
out.write((const char *) &len, sizeof(len));
|
|
out.write(m_params.prompt_file.c_str(), len);
|
|
}
|
|
|
|
if (m_params.verbosity > 0) {
|
|
fprintf(stderr, "\n%s: stored collected data after %d chunks in %s\n", __func__, m_last_call, fname.c_str());
|
|
}
|
|
}
|
|
|
|
bool IMatrixCollector::load_imatrix(const char * fname) {
|
|
std::ifstream in(fname, std::ios::binary);
|
|
if (!in) {
|
|
printf("%s: failed to open %s\n",__func__, fname);
|
|
return false;
|
|
}
|
|
int n_entries;
|
|
in.read((char*)&n_entries, sizeof(n_entries));
|
|
if (in.fail() || n_entries < 1) {
|
|
printf("%s: no data in file %s\n", __func__, fname);
|
|
return false;
|
|
}
|
|
for (int i = 0; i < n_entries; ++i) {
|
|
int len; in.read((char *)&len, sizeof(len));
|
|
std::vector<char> name_as_vec(len+1);
|
|
in.read((char *)name_as_vec.data(), len);
|
|
if (in.fail()) {
|
|
printf("%s: failed reading name for entry %d from %s\n",__func__,i+1, fname);
|
|
return false;
|
|
}
|
|
name_as_vec[len] = 0;
|
|
std::string name{name_as_vec.data()};
|
|
auto & e = m_stats[std::move(name)];
|
|
int ncall;
|
|
in.read((char*)&ncall, sizeof(ncall));
|
|
int nval;
|
|
in.read((char *)&nval, sizeof(nval));
|
|
if (in.fail() || nval < 1) {
|
|
printf("%s: failed reading number of values for entry %d\n",__func__,i);
|
|
m_stats = {};
|
|
return false;
|
|
}
|
|
|
|
if (e.values.empty()) {
|
|
e.values.resize(nval, 0);
|
|
e.counts.resize(nval, 0);
|
|
}
|
|
|
|
std::vector<float> tmp(nval);
|
|
in.read((char*)tmp.data(), nval*sizeof(float));
|
|
if (in.fail()) {
|
|
printf("%s: failed reading data for entry %d\n",__func__,i);
|
|
m_stats = {};
|
|
return false;
|
|
}
|
|
|
|
// Recreate the state as expected by save_imatrix(), and corerct for weighted sum.
|
|
for (int i = 0; i < nval; i++) {
|
|
e.values[i] += tmp[i];
|
|
e.counts[i] += ncall;
|
|
}
|
|
e.ncall += ncall;
|
|
|
|
}
|
|
return true;
|
|
}
|
|
|
|
static IMatrixCollector g_target_collector;
|
|
static IMatrixCollector g_draft_collector;
|
|
|
|
static IMatrixCollector * ik_get_imatrix_collector(void * user_data) {
|
|
return user_data != nullptr ? static_cast<IMatrixCollector *>(user_data) : &g_target_collector;
|
|
}
|
|
|
|
static bool ik_collect_imatrix(struct ggml_tensor * t, bool ask, void * user_data) {
|
|
return ik_get_imatrix_collector(user_data)->collect_imatrix(t, ask, user_data);
|
|
}
|
|
|
|
|
|
struct results_log_softmax {
|
|
double log_softmax;
|
|
float logit;
|
|
float prob;
|
|
};
|
|
|
|
static std::vector<float> softmax(const std::vector<float> & logits) {
|
|
std::vector<float> probs(logits.size());
|
|
float max_logit = logits[0];
|
|
for (float v : logits) {
|
|
max_logit = std::max(max_logit, v);
|
|
}
|
|
double sum_exp = 0.0;
|
|
for (size_t i = 0; i < logits.size(); i++) {
|
|
// Subtract the maximum logit value from the current logit value for numerical stability
|
|
const float logit = logits[i] - max_logit;
|
|
const float exp_logit = expf(logit);
|
|
sum_exp += exp_logit;
|
|
probs[i] = exp_logit;
|
|
}
|
|
for (size_t i = 0; i < probs.size(); i++) {
|
|
probs[i] /= sum_exp;
|
|
}
|
|
return probs;
|
|
}
|
|
|
|
static results_log_softmax log_softmax(int n_vocab, const float * logits, int tok) {
|
|
float max_logit = logits[0];
|
|
for (int i = 1; i < n_vocab; ++i) {
|
|
max_logit = std::max(max_logit, logits[i]);
|
|
}
|
|
double sum_exp = 0.0;
|
|
for (int i = 0; i < n_vocab; ++i) {
|
|
sum_exp += expf(logits[i] - max_logit);
|
|
}
|
|
return {logits[tok] - max_logit - log(sum_exp), logits[tok], expf(logits[tok] - max_logit) / (float) sum_exp};
|
|
}
|
|
|
|
static void process_logits(
|
|
int n_vocab, const float * logits, const int * tokens, int n_token, std::vector<std::thread> & workers,
|
|
double & nll, double & nll2, float * logit_history, float * prob_history) {
|
|
std::mutex mutex;
|
|
int counter = 0;
|
|
auto compute = [&mutex, &counter, &nll, &nll2, logit_history, prob_history, n_vocab, logits, tokens, n_token] () {
|
|
double local_nll = 0;
|
|
double local_nll2 = 0;
|
|
while (true) {
|
|
std::unique_lock<std::mutex> lock(mutex);
|
|
int i = counter++;
|
|
if (i >= n_token) {
|
|
nll += local_nll; nll2 += local_nll2;
|
|
break;
|
|
}
|
|
lock.unlock();
|
|
const results_log_softmax results = log_softmax(n_vocab, logits + i*n_vocab, tokens[i+1]);
|
|
const double v = -results.log_softmax;
|
|
local_nll += v;
|
|
local_nll2 += v*v;
|
|
|
|
logit_history[i] = results.logit;
|
|
prob_history[i] = results.prob;
|
|
}
|
|
};
|
|
for (auto & w : workers) {
|
|
w = std::thread(compute);
|
|
}
|
|
compute();
|
|
for (auto & w : workers) {
|
|
w.join();
|
|
}
|
|
}
|
|
|
|
static gpt_params build_draft_imatrix_params(const gpt_params & params) {
|
|
gpt_params draft_params = params;
|
|
|
|
draft_params.model = params.speculative.model;
|
|
draft_params.model_url.clear();
|
|
draft_params.hf_repo.clear();
|
|
draft_params.hf_file.clear();
|
|
draft_params.n_ctx = params.speculative.n_ctx > 0 ? params.speculative.n_ctx : params.n_ctx;
|
|
draft_params.n_gpu_layers = params.speculative.n_gpu_layers >= 0 ? params.speculative.n_gpu_layers : params.n_gpu_layers;
|
|
draft_params.has_mtp = true;
|
|
draft_params.warmup = false;
|
|
draft_params.out_file = params.out_file_draft.empty() ? default_draft_imatrix_out_file(params.out_file) : params.out_file_draft;
|
|
draft_params.out_file_draft.clear();
|
|
draft_params.cb_eval = ik_collect_imatrix;
|
|
draft_params.cb_eval_user_data = nullptr;
|
|
|
|
if (params.speculative.n_threads > 0) {
|
|
draft_params.n_threads = params.speculative.n_threads;
|
|
}
|
|
if (params.speculative.n_threads_batch > 0) {
|
|
draft_params.n_threads_batch = params.speculative.n_threads_batch;
|
|
}
|
|
if (!params.speculative.devices.empty()) {
|
|
draft_params.devices = params.speculative.devices;
|
|
}
|
|
if (!params.speculative.cache_type_k.empty()) {
|
|
draft_params.cache_type_k = params.speculative.cache_type_k;
|
|
}
|
|
if (!params.speculative.cache_type_v.empty()) {
|
|
draft_params.cache_type_v = params.speculative.cache_type_v;
|
|
}
|
|
|
|
return draft_params;
|
|
}
|
|
|
|
static bool compute_draft_imatrix_batch(
|
|
llama_context * ctx_tgt,
|
|
llama_context * ctx_dft,
|
|
llama_token * draft_tokens,
|
|
int batch_start,
|
|
int batch_size,
|
|
int batch_pos) {
|
|
const float * hidden = llama_get_embeddings(ctx_tgt);
|
|
const int n_embd_tgt = llama_mtp_state_n_embd(ctx_tgt);
|
|
const int n_embd_dft = llama_mtp_state_n_embd(ctx_dft);
|
|
|
|
if (hidden == nullptr || n_embd_tgt <= 0 || n_embd_dft <= 0) {
|
|
fprintf(stderr, "%s: missing target hidden state for paired draft calibration\n", __func__);
|
|
return false;
|
|
}
|
|
|
|
if (n_embd_tgt != n_embd_dft) {
|
|
fprintf(stderr, "%s: hidden width mismatch between target (%d) and draft (%d)\n",
|
|
__func__, n_embd_tgt, n_embd_dft);
|
|
return false;
|
|
}
|
|
|
|
llama_set_mtp_op_type(ctx_dft, MTP_OP_DRAFT_GEN);
|
|
if (!llama_set_draft_input_hidden_state_copy(ctx_dft, hidden, (size_t) batch_size * n_embd_dft)) {
|
|
llama_set_mtp_op_type(ctx_dft, MTP_OP_NONE);
|
|
fprintf(stderr, "%s: failed to stage paired draft hidden snapshot\n", __func__);
|
|
return false;
|
|
}
|
|
const int ret = llama_decode(ctx_dft, llama_batch_get_one(draft_tokens + batch_start, batch_size, batch_pos, 0));
|
|
llama_set_mtp_op_type(ctx_dft, MTP_OP_NONE);
|
|
|
|
if (ret != 0) {
|
|
fprintf(stderr, "%s: paired draft eval failed\n", __func__);
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
static bool compute_imatrix(llama_context * ctx, const gpt_params & params, llama_context * ctx_dft = nullptr) {
|
|
const bool add_bos = llama_should_add_bos_token(llama_get_model(ctx));
|
|
GGML_ASSERT(llama_add_eos_token(llama_get_model(ctx)) != 1);
|
|
const int n_ctx = llama_n_ctx(ctx);
|
|
|
|
auto tim1 = std::chrono::high_resolution_clock::now();
|
|
fprintf(stderr, "%s: tokenizing the input ..\n", __func__);
|
|
|
|
std::vector<llama_token> tokens = ::common_tokenize(ctx, params.prompt, true);
|
|
|
|
auto tim2 = std::chrono::high_resolution_clock::now();
|
|
fprintf(stderr, "%s: tokenization took %g ms\n",__func__,1e-3*std::chrono::duration_cast<std::chrono::microseconds>(tim2-tim1).count());
|
|
|
|
if (params.i_chunk > 0) {
|
|
if (size_t((params.i_chunk + 2)*n_ctx) >= tokens.size()) {
|
|
fprintf(stderr, "%s: there will be not enough tokens left after removing %d chunks\n", __func__, params.i_chunk);
|
|
return false;
|
|
}
|
|
fprintf(stderr, "%s: removing initial %d chunks (%d tokens)\n", __func__, params.i_chunk, params.i_chunk*n_ctx);
|
|
tokens.erase(tokens.begin(), tokens.begin() + params.i_chunk*n_ctx);
|
|
}
|
|
|
|
if (int(tokens.size()) < 2*n_ctx) {
|
|
fprintf(stderr, "%s: you need at least %d tokens for a context of %d tokens\n",__func__,2*n_ctx,
|
|
n_ctx);
|
|
fprintf(stderr, "%s: the data file you provided tokenizes to only %zu tokens\n",__func__,tokens.size());
|
|
return false;
|
|
}
|
|
|
|
std::vector<float> logit_history;
|
|
std::vector<float> prob_history;
|
|
|
|
if (params.compute_ppl) {
|
|
logit_history.resize(tokens.size());
|
|
prob_history.resize(tokens.size());
|
|
}
|
|
|
|
const int n_chunk_max = tokens.size() / n_ctx;
|
|
|
|
const int n_chunk = params.n_chunks < 0 ? n_chunk_max : std::min(params.n_chunks, n_chunk_max);
|
|
const int n_vocab = llama_n_vocab(llama_get_model(ctx));
|
|
const int n_batch = params.n_batch;
|
|
|
|
int count = 0;
|
|
double nll = 0.0;
|
|
double nll2 = 0.0;
|
|
|
|
fprintf(stderr, "%s: computing over %d chunks with batch_size %d\n", __func__, n_chunk, n_batch);
|
|
|
|
std::vector<std::thread> workers(std::thread::hardware_concurrency() - 1);
|
|
|
|
const int num_batches = (n_ctx + n_batch - 1) / n_batch;
|
|
|
|
std::vector<float> logits;
|
|
if (params.compute_ppl && num_batches > 1) {
|
|
logits.reserve((size_t)n_ctx * n_vocab);
|
|
}
|
|
|
|
for (int i = 0; i < n_chunk; ++i) {
|
|
const int start = i * n_ctx;
|
|
const int end = start + n_ctx;
|
|
|
|
std::vector<float> logits;
|
|
|
|
const auto t_start = std::chrono::high_resolution_clock::now();
|
|
|
|
// clear the KV cache
|
|
llama_kv_cache_clear(ctx);
|
|
if (ctx_dft != nullptr) {
|
|
llama_kv_cache_clear(ctx_dft);
|
|
}
|
|
|
|
for (int j = 0; j < num_batches; ++j) {
|
|
const int batch_start = start + j * n_batch;
|
|
const int batch_size = std::min(end - batch_start, n_batch);
|
|
|
|
// save original token and restore it after eval
|
|
const auto token_org = tokens[batch_start];
|
|
|
|
// add BOS token for the first batch of each chunk
|
|
if (add_bos && j == 0) {
|
|
tokens[batch_start] = llama_token_bos(llama_get_model(ctx));
|
|
}
|
|
|
|
// TODO: use batch.logits to save computations instead of relying on logits_all == true
|
|
if (llama_decode(ctx, llama_batch_get_one(tokens.data() + batch_start, batch_size, j * n_batch, 0))) {
|
|
fprintf(stderr, "%s : failed to eval\n", __func__);
|
|
return false;
|
|
}
|
|
|
|
if (ctx_dft != nullptr && !compute_draft_imatrix_batch(ctx, ctx_dft, tokens.data(), batch_start, batch_size, j * n_batch)) {
|
|
return false;
|
|
}
|
|
|
|
// restore the original token in case it was set to BOS
|
|
tokens[batch_start] = token_org;
|
|
|
|
if (params.compute_ppl && num_batches > 1) {
|
|
const auto * batch_logits = llama_get_logits(ctx);
|
|
logits.insert(logits.end(), batch_logits, batch_logits + batch_size * n_vocab);
|
|
}
|
|
}
|
|
|
|
const auto t_end = std::chrono::high_resolution_clock::now();
|
|
|
|
if (i == 0) {
|
|
const float t_total = std::chrono::duration<float>(t_end - t_start).count();
|
|
fprintf(stderr, "%s: %.2f seconds per pass - ETA ", __func__, t_total);
|
|
int total_seconds = (int)(t_total * n_chunk);
|
|
if (total_seconds >= 60*60) {
|
|
fprintf(stderr, "%d hours ", total_seconds / (60*60));
|
|
total_seconds = total_seconds % (60*60);
|
|
}
|
|
fprintf(stderr, "%.2f minutes\n", total_seconds / 60.0);
|
|
}
|
|
|
|
if (params.compute_ppl) {
|
|
const int first = n_ctx/2;
|
|
const auto all_logits = num_batches > 1 ? logits.data() : llama_get_logits(ctx);
|
|
process_logits(n_vocab, all_logits + first*n_vocab, tokens.data() + start + first, n_ctx - 1 - first,
|
|
workers, nll, nll2, logit_history.data() + start + first, prob_history.data() + start + first);
|
|
count += n_ctx - first - 1;
|
|
|
|
printf("[%d]%.4lf,", i + 1, std::exp(nll / count));
|
|
fflush(stdout);
|
|
|
|
logits.clear();
|
|
}
|
|
}
|
|
printf("\n");
|
|
|
|
if (params.compute_ppl) {
|
|
nll2 /= count;
|
|
nll /= count;
|
|
const double ppl = exp(nll);
|
|
nll2 -= nll * nll;
|
|
if (nll2 > 0) {
|
|
nll2 = sqrt(nll2/(count-1));
|
|
printf("Final estimate: PPL = %.4lf +/- %.5lf\n", ppl, nll2*ppl);
|
|
} else {
|
|
printf("Unexpected negative standard deviation of log(prob)\n");
|
|
}
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
int main(int argc, char ** argv) {
|
|
gpt_params params;
|
|
|
|
params.n_ctx = 512;
|
|
params.logits_all = true;
|
|
params.verbosity = 1;
|
|
|
|
bool lsim = false;
|
|
//
|
|
// Do not pollute common with totally imatrix specific arguments as it was done in mainline.
|
|
// Instead, parse imatrix specific args here, push unknown args into a new array of args,
|
|
// and pass that to gpt_params_parse().
|
|
//
|
|
std::vector<char*> args;
|
|
args.reserve(argc);
|
|
args.push_back(argv[0]);
|
|
for (int i = 1; i < argc; ++i) {
|
|
std::string arg{argv[i]};
|
|
if (arg == "-lsim" || arg == "--layer-similarity") {
|
|
lsim = true;
|
|
} else {
|
|
args.push_back(argv[i]);
|
|
}
|
|
}
|
|
|
|
if (!gpt_params_parse(args.size(), args.data(), params)) {
|
|
print_usage(argc, argv, params);
|
|
return 1;
|
|
}
|
|
|
|
params.n_batch = std::min(params.n_batch, params.n_ctx);
|
|
|
|
g_target_collector.set_params(params);
|
|
g_target_collector.set_collect_lsim(lsim);
|
|
|
|
for (const auto & in_file : params.in_files) {
|
|
printf("%s : loading imatrix from '%s'\n", __func__, in_file.c_str());
|
|
if (!g_target_collector.load_imatrix(in_file.c_str())) {
|
|
fprintf(stderr, "%s : failed to load %s\n", __func__, in_file.c_str());
|
|
return 1;
|
|
}
|
|
}
|
|
|
|
if (params.in_files.size() > 1) {
|
|
printf("%s : saving combined imatrix to '%s'\n", __func__, params.out_file.c_str());
|
|
g_target_collector.save_imatrix();
|
|
}
|
|
|
|
llama_backend_init();
|
|
llama_numa_init(params.numa);
|
|
|
|
gpt_params target_params = params;
|
|
target_params.warmup = false;
|
|
|
|
const bool has_draft_model = !params.speculative.model.empty();
|
|
if (!has_draft_model) {
|
|
// pass the callback to the backend scheduler
|
|
// it will be executed for each node during the graph computation
|
|
target_params.cb_eval = ik_collect_imatrix;
|
|
target_params.cb_eval_user_data = &g_target_collector;
|
|
}
|
|
|
|
llama_init_result llama_init;
|
|
llama_model * model_dft = nullptr;
|
|
llama_context * ctx_dft = nullptr;
|
|
bool use_paired_gemma4_mtp = false;
|
|
|
|
if (has_draft_model) {
|
|
gpt_params draft_params = build_draft_imatrix_params(params);
|
|
g_draft_collector.set_params(draft_params);
|
|
g_draft_collector.set_collect_lsim(lsim);
|
|
draft_params.cb_eval_user_data = &g_draft_collector;
|
|
auto mparams_dft = common_model_params_to_llama(draft_params);
|
|
|
|
model_dft = ik_load_model_from_params(draft_params, mparams_dft);
|
|
if (model_dft == nullptr) {
|
|
fprintf(stderr, "%s : failed to load draft model '%s'\n", __func__, draft_params.model.c_str());
|
|
llama_backend_free();
|
|
return 1;
|
|
}
|
|
|
|
if (!llama_model_is_gemma4_mtp_assistant(model_dft)) {
|
|
fprintf(stderr, "%s : paired imatrix mode currently supports Gemma 4 assistant draft models only\n", __func__);
|
|
llama_free_model(model_dft);
|
|
llama_backend_free();
|
|
return 1;
|
|
}
|
|
|
|
target_params.has_mtp = true;
|
|
target_params.cb_eval = ik_collect_imatrix;
|
|
target_params.cb_eval_user_data = &g_target_collector;
|
|
|
|
fprintf(stderr, "%s : paired imatrix outputs: target='%s', draft='%s'\n",
|
|
__func__, target_params.out_file.c_str(), draft_params.out_file.c_str());
|
|
|
|
auto mparams_tgt = common_model_params_to_llama(target_params);
|
|
llama_model * model_tgt = ik_load_model_from_params(target_params, mparams_tgt);
|
|
if (model_tgt == nullptr) {
|
|
fprintf(stderr, "%s : failed to load target model '%s'\n", __func__, target_params.model.c_str());
|
|
llama_free_model(model_dft);
|
|
llama_backend_free();
|
|
return 1;
|
|
}
|
|
|
|
if (!ik_model_has_arch(model_tgt, "gemma4")) {
|
|
fprintf(stderr, "%s : paired imatrix mode currently supports Gemma 4 target models only\n", __func__);
|
|
llama_free_model(model_tgt);
|
|
llama_free_model(model_dft);
|
|
llama_backend_free();
|
|
return 1;
|
|
}
|
|
|
|
llama_init = ik_init_from_loaded_model(model_tgt, target_params);
|
|
if (llama_init.model == nullptr || llama_init.context == nullptr) {
|
|
llama_free_model(model_dft);
|
|
llama_backend_free();
|
|
return 1;
|
|
}
|
|
|
|
auto draft_init = ik_init_from_loaded_model(model_dft, draft_params);
|
|
model_dft = draft_init.model;
|
|
ctx_dft = draft_init.context;
|
|
if (model_dft == nullptr || ctx_dft == nullptr) {
|
|
llama_free(llama_init.context);
|
|
llama_free_model(llama_init.model);
|
|
llama_backend_free();
|
|
return 1;
|
|
}
|
|
|
|
llama_set_mtp_target_context(ctx_dft, llama_init.context);
|
|
use_paired_gemma4_mtp = true;
|
|
} else {
|
|
llama_init = llama_init_from_gpt_params(target_params);
|
|
}
|
|
|
|
llama_model * model = llama_init.model;
|
|
llama_context * ctx = llama_init.context;
|
|
if (model == nullptr || ctx == nullptr) {
|
|
fprintf(stderr, "%s : failed to init\n", __func__);
|
|
return 1;
|
|
}
|
|
|
|
if (!use_paired_gemma4_mtp && llama_model_is_gemma4_mtp_assistant(model) && !params.process_output) {
|
|
fprintf(stderr, "%s: warning: standalone Gemma 4 assistant imatrix does not exercise the assistant layers. Use '-m <target> -md <assistant> -mtp' for meaningful calibration.\n", __func__);
|
|
}
|
|
|
|
const int n_ctx_train = llama_n_ctx_train(model);
|
|
if (params.n_ctx > n_ctx_train) {
|
|
fprintf(stderr, "%s: warning: model was trained on only %d context tokens (%d specified)\n",
|
|
__func__, n_ctx_train, params.n_ctx);
|
|
}
|
|
|
|
// print system information
|
|
{
|
|
fprintf(stderr, "\n");
|
|
fprintf(stderr, "%s\n", gpt_params_get_system_info(params).c_str());
|
|
}
|
|
|
|
if (!compute_imatrix(ctx, params, ctx_dft)) {
|
|
if (ctx_dft != nullptr) {
|
|
llama_free(ctx_dft);
|
|
}
|
|
if (model_dft != nullptr) {
|
|
llama_free_model(model_dft);
|
|
}
|
|
llama_free(ctx);
|
|
llama_free_model(model);
|
|
llama_backend_free();
|
|
return 1;
|
|
}
|
|
|
|
g_target_collector.save_imatrix();
|
|
g_target_collector.print_layer_importance();
|
|
|
|
if (ctx_dft != nullptr) {
|
|
g_draft_collector.save_imatrix();
|
|
g_draft_collector.print_layer_importance();
|
|
}
|
|
|
|
llama_print_timings(ctx);
|
|
|
|
if (ctx_dft != nullptr) {
|
|
llama_free(ctx_dft);
|
|
}
|
|
if (model_dft != nullptr) {
|
|
llama_free_model(model_dft);
|
|
}
|
|
|
|
llama_free(ctx);
|
|
llama_free_model(model);
|
|
|
|
llama_backend_free();
|
|
|
|
return 0;
|
|
}
|