2024-07-23 13:10:17 +03:00
#include "llama-sampling.h"
2024-09-07 15:16:19 +03:00
#include "llama-vocab.h"
#include "llama-grammar.h"
2024-07-23 13:10:17 +03:00
#include <algorithm>
2024-09-24 09:03:17 +03:00
#include <cassert>
2024-07-23 13:10:17 +03:00
#include <cfloat>
2024-09-10 18:04:25 +02:00
#include <chrono>
2024-09-09 17:10:46 +02:00
#include <cmath>
2024-09-24 09:03:17 +03:00
#include <cstdlib>
#include <cstring>
#include <ctime>
2024-07-23 13:10:17 +03:00
#include <numeric>
2024-09-07 15:16:19 +03:00
#include <random>
2024-07-23 13:10:17 +03:00
#include <unordered_map>
2024-09-09 17:10:46 +02:00
static int llama_sample_dist ( llama_token_data_array * cur_p , std :: mt19937 & rng ) {
// iterator for the probabilities
#ifdef __GNUC__
2024-09-08 15:52:07 +02:00
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wunused-local-typedefs"
2024-09-09 17:10:46 +02:00
#endif
2024-09-08 15:52:07 +02:00
struct probs_iterator {
typedef std :: input_iterator_tag iterator_category ;
typedef float value_type ;
typedef float * pointer ;
typedef float & reference ;
2024-09-09 17:10:46 +02:00
typedef ptrdiff_t difference_type ;
2024-09-08 15:52:07 +02:00
2024-09-09 17:10:46 +02:00
const llama_token_data * data ;
2024-09-08 15:52:07 +02:00
2024-09-09 17:10:46 +02:00
bool operator == ( const probs_iterator & other ) const { return data == other . data ; }
bool operator != ( const probs_iterator & other ) const { return data != other . data ; }
const float & operator * () const { return data -> p ; }
probs_iterator & operator ++ () { ++ data ; return * this ; }
probs_iterator operator ++ ( int ) { probs_iterator tmp = * this ; ++ data ; return tmp ; }
2024-09-08 15:52:07 +02:00
};
2024-09-09 17:10:46 +02:00
#ifdef __GNUC__
2024-09-08 15:52:07 +02:00
#pragma GCC diagnostic pop
#endif
2024-09-07 15:16:19 +03:00
2024-09-09 17:10:46 +02:00
std :: discrete_distribution < int > dist ( probs_iterator { cur_p -> data }, probs_iterator { cur_p -> data + cur_p -> size });
2024-09-07 15:16:19 +03:00
return dist ( rng );
}
2024-09-09 17:10:46 +02:00
/*
2024-07-23 13:10:17 +03:00
static void llama_log_softmax(float * array, size_t size) {
float max_l = *std::max_element(array, array + size);
float sum = 0.f;
for (size_t i = 0; i < size; ++i) {
float p = expf(array[i] - max_l);
sum += p;
array[i] = p;
}
for (size_t i = 0; i < size; ++i) {
array[i] = logf(array[i] / sum);
}
}
2024-09-09 17:10:46 +02:00
*/
2024-07-23 13:10:17 +03:00
2024-09-07 15:16:19 +03:00
static void llama_sampler_softmax_impl ( llama_token_data_array * cur_p ) {
GGML_ASSERT ( cur_p -> size > 0 );
2024-07-23 13:10:17 +03:00
// Sort the logits in descending order
2024-09-07 15:16:19 +03:00
if ( ! cur_p -> sorted ) {
std :: sort ( cur_p -> data , cur_p -> data + cur_p -> size , []( const llama_token_data & a , const llama_token_data & b ) {
2024-07-23 13:10:17 +03:00
return a . logit > b . logit ;
});
2024-09-07 15:16:19 +03:00
cur_p -> sorted = true ;
2024-07-23 13:10:17 +03:00
}
2024-09-07 15:16:19 +03:00
float max_l = cur_p -> data [ 0 ]. logit ;
2024-07-23 13:10:17 +03:00
float cum_sum = 0.0f ;
2024-09-07 15:16:19 +03:00
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
float p = expf ( cur_p -> data [ i ]. logit - max_l );
cur_p -> data [ i ]. p = p ;
2024-07-23 13:10:17 +03:00
cum_sum += p ;
}
2024-09-07 15:16:19 +03:00
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
cur_p -> data [ i ]. p /= cum_sum ;
2024-07-23 13:10:17 +03:00
}
}
2024-09-07 15:16:19 +03:00
static void llama_sampler_top_k_impl ( llama_token_data_array * cur_p , int32_t k ) {
2024-07-23 13:10:17 +03:00
// TODO: move bucket sort to separate function so that top_p/tail_free/typical/softmax first is equally fast
2024-09-07 15:16:19 +03:00
// if (k >= (int32_t)cur_p->size) {
2024-07-23 13:10:17 +03:00
// return;
// }
if ( k <= 0 ) {
2024-09-07 15:16:19 +03:00
k = cur_p -> size ;
2024-07-23 13:10:17 +03:00
}
2024-09-07 15:16:19 +03:00
k = std :: min ( k , ( int ) cur_p -> size );
2024-07-23 13:10:17 +03:00
// Sort scores in descending order
2024-09-07 15:16:19 +03:00
if ( ! cur_p -> sorted ) {
2024-07-23 13:10:17 +03:00
auto comp = []( const llama_token_data & a , const llama_token_data & b ) {
return a . logit > b . logit ;
};
if ( k <= 128 ) {
2024-09-07 15:16:19 +03:00
std :: partial_sort ( cur_p -> data , cur_p -> data + k , cur_p -> data + cur_p -> size , comp );
2024-07-23 13:10:17 +03:00
} else {
constexpr int nbuckets = 128 ;
constexpr float bucket_low = - 10.0f ;
constexpr float bucket_high = 10.0f ;
constexpr float bucket_scale = nbuckets / ( bucket_high - bucket_low );
2024-08-12 17:46:03 +08:00
constexpr float bucket_inter = - bucket_low * bucket_scale ;
2024-07-23 13:10:17 +03:00
2024-09-07 15:16:19 +03:00
std :: vector < int > bucket_idx ( cur_p -> size );
2024-07-23 13:10:17 +03:00
std :: vector < int > histo ( nbuckets , 0 );
2024-09-07 15:16:19 +03:00
for ( int i = 0 ; i < ( int ) cur_p -> size ; ++ i ) {
const float val = cur_p -> data [ i ]. logit ;
2024-08-12 17:46:03 +08:00
int ib = int ( bucket_scale * val + bucket_inter ); //nbuckets * (val - bucket_low) / (bucket_high - bucket_low);
2024-07-23 13:10:17 +03:00
ib = std :: max ( 0 , std :: min ( nbuckets - 1 , ib ));
bucket_idx [ i ] = ib ;
++ histo [ ib ];
}
int nhave = 0 ;
int ib = nbuckets - 1 ;
for ( ; ib >= 0 ; -- ib ) {
nhave += histo [ ib ];
2024-09-07 15:16:19 +03:00
if ( nhave >= k ) {
break ;
}
2024-07-23 13:10:17 +03:00
}
std :: vector < llama_token_data > tmp_tokens ( nhave );
2024-09-07 15:16:19 +03:00
auto * ptr = tmp_tokens . data ();
2024-07-23 13:10:17 +03:00
std :: vector < llama_token_data *> bucket_ptrs ;
bucket_ptrs . reserve ( nbuckets - ib );
for ( int j = nbuckets - 1 ; j >= ib ; -- j ) {
bucket_ptrs . push_back ( ptr );
ptr += histo [ j ];
}
2024-09-07 15:16:19 +03:00
for ( int i = 0 ; i < ( int ) cur_p -> size ; ++ i ) {
2024-07-23 13:10:17 +03:00
int j = bucket_idx [ i ];
if ( j >= ib ) {
2024-09-07 15:16:19 +03:00
* bucket_ptrs [ nbuckets - 1 - j ] ++ = cur_p -> data [ i ];
2024-07-23 13:10:17 +03:00
}
}
ptr = tmp_tokens . data ();
int ndone = 0 ;
for ( int j = nbuckets - 1 ; j > ib ; -- j ) {
std :: sort ( ptr , ptr + histo [ j ], comp );
ptr += histo [ j ];
ndone += histo [ j ];
}
std :: partial_sort ( ptr , ptr + k - ndone , ptr + histo [ ib ], comp );
2024-09-07 15:16:19 +03:00
std :: memcpy ( cur_p -> data , tmp_tokens . data (), k * sizeof ( llama_token_data ));
2024-07-23 13:10:17 +03:00
}
2024-09-07 15:16:19 +03:00
cur_p -> sorted = true ;
2024-07-23 13:10:17 +03:00
}
2024-09-07 15:16:19 +03:00
cur_p -> size = k ;
2024-07-23 13:10:17 +03:00
}
2024-09-10 18:04:25 +02:00
static uint32_t get_rng_seed ( uint32_t seed ) {
if ( seed == LLAMA_DEFAULT_SEED ) {
// use system clock if std::random_device is not a true RNG
static bool is_rd_prng = std :: random_device (). entropy () == 0 ;
if ( is_rd_prng ) {
return ( uint32_t ) std :: chrono :: system_clock :: now (). time_since_epoch (). count ();
}
std :: random_device rd ;
return rd ();
}
return seed ;
}
2024-09-07 15:16:19 +03:00
// llama_sampler API
const char * llama_sampler_name ( const struct llama_sampler * smpl ) {
if ( ! smpl -> iface ) {
return "(null)" ;
}
return smpl -> iface -> name ( smpl );
}
void llama_sampler_accept ( struct llama_sampler * smpl , llama_token token ) {
if ( smpl -> iface -> accept ) {
smpl -> iface -> accept ( smpl , token );
2024-07-23 13:10:17 +03:00
}
}
2024-09-07 15:16:19 +03:00
void llama_sampler_apply ( struct llama_sampler * smpl , struct llama_token_data_array * cur_p ) {
GGML_ASSERT ( smpl -> iface -> apply );
smpl -> iface -> apply ( smpl , cur_p );
2024-07-23 13:10:17 +03:00
}
2024-09-07 15:16:19 +03:00
void llama_sampler_reset ( struct llama_sampler * smpl ) {
if ( smpl -> iface -> reset ) {
smpl -> iface -> reset ( smpl );
2024-07-23 13:10:17 +03:00
}
}
2024-09-07 15:16:19 +03:00
struct llama_sampler * llama_sampler_clone ( const struct llama_sampler * smpl ) {
if ( smpl -> iface -> clone ) {
return smpl -> iface -> clone ( smpl );
2024-07-23 13:10:17 +03:00
}
2024-09-07 15:16:19 +03:00
if ( smpl -> ctx == nullptr ) {
return new llama_sampler {
/* .iface = */ smpl -> iface ,
/* .ctx = */ nullptr ,
};
2024-07-23 13:10:17 +03:00
}
2024-09-07 15:16:19 +03:00
GGML_ABORT ( "the sampler does not support cloning" );
2024-07-23 13:10:17 +03:00
}
2024-09-07 15:16:19 +03:00
void llama_sampler_free ( struct llama_sampler * smpl ) {
if ( smpl == nullptr ) {
return ;
2024-07-23 13:10:17 +03:00
}
2024-09-07 15:16:19 +03:00
if ( smpl -> iface -> free ) {
smpl -> iface -> free ( smpl );
}
delete smpl ;
2024-07-23 13:10:17 +03:00
}
2024-09-07 15:16:19 +03:00
llama_token llama_sampler_sample ( struct llama_sampler * smpl , struct llama_context * ctx , int32_t idx ) {
const auto * logits = llama_get_logits_ith ( ctx , idx );
2024-07-23 13:10:17 +03:00
2024-09-07 15:16:19 +03:00
const int n_vocab = llama_n_vocab ( llama_get_model ( ctx ));
// TODO: do not allocate each time
2024-09-18 13:42:36 +02:00
std :: vector < llama_token_data > cur ;
cur . reserve ( n_vocab );
2024-09-07 15:16:19 +03:00
for ( llama_token token_id = 0 ; token_id < n_vocab ; token_id ++ ) {
2024-09-18 13:42:36 +02:00
cur . emplace_back ( llama_token_data { token_id , logits [ token_id ], 0.0f });
2024-09-07 15:16:19 +03:00
}
2024-09-09 17:10:46 +02:00
llama_token_data_array cur_p = {
/* .data = */ cur . data (),
/* .size = */ cur . size (),
/* .selected = */ - 1 ,
/* .sorted = */ false ,
};
2024-09-07 15:16:19 +03:00
llama_sampler_apply ( smpl , & cur_p );
2024-09-09 17:10:46 +02:00
GGML_ASSERT ( cur_p . selected >= 0 && cur_p . selected < ( int32_t ) cur_p . size );
auto token = cur_p . data [ cur_p . selected ]. id ;
llama_sampler_accept ( smpl , token );
return token ;
2024-09-07 15:16:19 +03:00
}
// sampler chain
2024-09-09 17:10:46 +02:00
static const char * llama_sampler_chain_name ( const struct llama_sampler * /*smpl*/ ) {
return "chain" ;
}
static void llama_sampler_chain_accept ( struct llama_sampler * smpl , llama_token token ) {
auto * chain = ( llama_sampler_chain * ) smpl -> ctx ;
time_meas tm ( chain -> t_sample_us , chain -> params . no_perf );
for ( auto * smpl : chain -> samplers ) {
llama_sampler_accept ( smpl , token );
}
chain -> n_sample ++ ;
}
static void llama_sampler_chain_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
auto * chain = ( llama_sampler_chain * ) smpl -> ctx ;
time_meas tm ( chain -> t_sample_us , chain -> params . no_perf );
for ( auto * smpl : chain -> samplers ) {
llama_sampler_apply ( smpl , cur_p );
}
}
static void llama_sampler_chain_reset ( struct llama_sampler * smpl ) {
auto * chain = ( llama_sampler_chain * ) smpl -> ctx ;
for ( auto * smpl : chain -> samplers ) {
llama_sampler_reset ( smpl );
}
chain -> t_sample_us = 0 ;
chain -> n_sample = 0 ;
}
static struct llama_sampler * llama_sampler_chain_clone ( const struct llama_sampler * smpl ) {
const auto * chain_src = ( const llama_sampler_chain * ) smpl -> ctx ;
auto * result = llama_sampler_chain_init ( chain_src -> params );
for ( auto * smpl : chain_src -> samplers ) {
llama_sampler_chain_add ( result , llama_sampler_clone ( smpl ));
}
return result ;
}
static void llama_sampler_chain_free ( struct llama_sampler * smpl ) {
auto * chain = ( llama_sampler_chain * ) smpl -> ctx ;
for ( auto * smpl : chain -> samplers ) {
llama_sampler_free ( smpl );
}
delete chain ;
}
2024-09-07 15:16:19 +03:00
static struct llama_sampler_i llama_sampler_chain_i = {
2024-09-09 17:10:46 +02:00
/* .name = */ llama_sampler_chain_name ,
/* .accept = */ llama_sampler_chain_accept ,
/* .apply = */ llama_sampler_chain_apply ,
/* .reset = */ llama_sampler_chain_reset ,
/* .clone = */ llama_sampler_chain_clone ,
/* .free = */ llama_sampler_chain_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_chain_init ( struct llama_sampler_chain_params params ) {
return new llama_sampler {
/* .iface = */ & llama_sampler_chain_i ,
/* .ctx = */ new llama_sampler_chain {
/* .params = */ params ,
/* .samplers = */ {},
/* .t_sample_us = */ 0 ,
/* .n_sample = */ 0 ,
},
};
}
void llama_sampler_chain_add ( struct llama_sampler * chain , struct llama_sampler * smpl ) {
auto * p = ( llama_sampler_chain * ) chain -> ctx ;
p -> samplers . push_back ( smpl );
}
struct llama_sampler * llama_sampler_chain_get ( const struct llama_sampler * chain , int32_t i ) {
const auto * p = ( const llama_sampler_chain * ) chain -> ctx ;
2024-09-13 04:54:49 +03:00
if ( i < 0 || ( size_t ) i >= p -> samplers . size ()) {
2024-09-07 15:16:19 +03:00
return nullptr ;
}
return p -> samplers [ i ];
}
2024-09-13 04:54:49 +03:00
struct llama_sampler * llama_sampler_chain_remove ( struct llama_sampler * chain , int32_t i ) {
auto * p = ( llama_sampler_chain * ) chain -> ctx ;
if ( i < 0 || ( size_t ) i >= p -> samplers . size ()) {
return nullptr ;
}
auto * result = p -> samplers [ i ];
p -> samplers . erase ( p -> samplers . begin () + i );
return result ;
}
2024-09-07 15:16:19 +03:00
int llama_sampler_chain_n ( const struct llama_sampler * chain ) {
const auto * p = ( const llama_sampler_chain * ) chain -> ctx ;
return p -> samplers . size ();
}
//
// samplers
//
// greedy
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_greedy_name ( const struct llama_sampler * /*smpl*/ ) {
return "greedy" ;
}
static void llama_sampler_greedy_apply ( struct llama_sampler * /*smpl*/ , llama_token_data_array * cur_p ) {
cur_p -> selected = 0 ;
for ( size_t i = 1 ; i < cur_p -> size ; ++ i ) {
if ( cur_p -> data [ i ]. logit > cur_p -> data [ cur_p -> selected ]. logit ) {
cur_p -> selected = i ;
2024-09-07 15:16:19 +03:00
}
2024-09-08 15:52:07 +02:00
}
}
static struct llama_sampler_i llama_sampler_greedy_i = {
/* .name = */ llama_sampler_greedy_name ,
/* .accept = */ nullptr ,
/* .apply = */ llama_sampler_greedy_apply ,
2024-09-07 15:16:19 +03:00
/* .reset = */ nullptr ,
/* .clone = */ nullptr ,
/* .free = */ nullptr ,
};
struct llama_sampler * llama_sampler_init_greedy () {
return new llama_sampler {
/* .iface = */ & llama_sampler_greedy_i ,
/* .ctx = */ nullptr ,
};
}
// dist
struct llama_sampler_dist {
const uint32_t seed ;
2024-09-10 18:04:25 +02:00
uint32_t seed_cur ;
2024-09-07 15:16:19 +03:00
std :: mt19937 rng ;
};
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_dist_name ( const struct llama_sampler * /*smpl*/ ) {
return "dist" ;
}
static void llama_sampler_dist_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
auto * ctx = ( llama_sampler_dist * ) smpl -> ctx ;
2024-09-09 17:10:46 +02:00
cur_p -> selected = llama_sample_dist ( cur_p , ctx -> rng );
2024-09-08 15:52:07 +02:00
}
static struct llama_sampler * llama_sampler_dist_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_dist * ) smpl -> ctx ;
auto * result = llama_sampler_init_dist ( ctx -> seed );
// copy the state
{
auto * result_ctx = ( llama_sampler_dist * ) result -> ctx ;
result_ctx -> rng = ctx -> rng ;
}
return result ;
}
static void llama_sampler_dist_reset ( struct llama_sampler * smpl ) {
auto * ctx = ( llama_sampler_dist * ) smpl -> ctx ;
2024-09-10 18:04:25 +02:00
ctx -> seed_cur = get_rng_seed ( ctx -> seed );
ctx -> rng . seed ( ctx -> seed_cur );
2024-09-08 15:52:07 +02:00
}
static void llama_sampler_dist_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_dist * ) smpl -> ctx ;
}
2024-09-07 15:16:19 +03:00
static struct llama_sampler_i llama_sampler_dist_i = {
2024-09-08 15:52:07 +02:00
/* .name = */ llama_sampler_dist_name ,
2024-09-07 15:16:19 +03:00
/* .accept = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .apply = */ llama_sampler_dist_apply ,
/* .reset = */ llama_sampler_dist_reset ,
/* .clone = */ llama_sampler_dist_clone ,
/* .free = */ llama_sampler_dist_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_init_dist ( uint32_t seed ) {
2024-09-10 18:04:25 +02:00
auto seed_cur = get_rng_seed ( seed );
2024-09-07 15:16:19 +03:00
return new llama_sampler {
/* .iface = */ & llama_sampler_dist_i ,
/* .ctx = */ new llama_sampler_dist {
2024-09-10 18:04:25 +02:00
/* .seed = */ seed ,
/* .seed_cur = */ seed_cur ,
/* .rng = */ std :: mt19937 ( seed_cur ),
2024-09-07 15:16:19 +03:00
},
};
}
// softmax
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_softmax_name ( const struct llama_sampler * /*smpl*/ ) {
return "softmax" ;
}
static void llama_sampler_softmax_apply ( struct llama_sampler * /*smpl*/ , llama_token_data_array * cur_p ) {
llama_sampler_softmax_impl ( cur_p );
}
2024-09-07 15:16:19 +03:00
static struct llama_sampler_i llama_sampler_softmax_i = {
2024-09-08 15:52:07 +02:00
/* .name = */ llama_sampler_softmax_name ,
2024-09-07 15:16:19 +03:00
/* .accept = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .apply = */ llama_sampler_softmax_apply ,
2024-09-07 15:16:19 +03:00
/* .reset = */ nullptr ,
/* .clone = */ nullptr ,
/* .free = */ nullptr ,
};
struct llama_sampler * llama_sampler_init_softmax () {
return new llama_sampler {
/* .iface = */ & llama_sampler_softmax_i ,
/* .ctx = */ nullptr ,
};
}
// top-k
struct llama_sampler_top_k {
const int32_t k ;
};
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_top_k_name ( const struct llama_sampler * /*smpl*/ ) {
return "top-k" ;
}
static void llama_sampler_top_k_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
const auto * ctx = ( llama_sampler_top_k * ) smpl -> ctx ;
llama_sampler_top_k_impl ( cur_p , ctx -> k );
}
static struct llama_sampler * llama_sampler_top_k_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_top_k * ) smpl -> ctx ;
return llama_sampler_init_top_k ( ctx -> k );
}
static void llama_sampler_top_k_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_top_k * ) smpl -> ctx ;
}
2024-09-07 15:16:19 +03:00
static struct llama_sampler_i llama_sampler_top_k_i = {
2024-09-08 15:52:07 +02:00
/* .name = */ llama_sampler_top_k_name ,
2024-09-07 15:16:19 +03:00
/* .accept = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .apply = */ llama_sampler_top_k_apply ,
2024-09-07 15:16:19 +03:00
/* .reset = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .clone = */ llama_sampler_top_k_clone ,
/* .free = */ llama_sampler_top_k_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_init_top_k ( int32_t k ) {
return new llama_sampler {
/* .iface = */ & llama_sampler_top_k_i ,
/* .ctx = */ new llama_sampler_top_k {
/* .k = */ k ,
},
};
}
// top-p
struct llama_sampler_top_p {
const float p ;
const size_t min_keep ;
};
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_top_p_name ( const struct llama_sampler * /*smpl*/ ) {
return "top-p" ;
}
static void llama_sampler_top_p_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
const auto * ctx = ( llama_sampler_top_p * ) smpl -> ctx ;
if ( ctx -> p >= 1.0f ) {
return ;
}
llama_sampler_softmax_impl ( cur_p );
// Compute the cumulative probabilities
float cum_sum = 0.0f ;
size_t last_idx = cur_p -> size ;
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
cum_sum += cur_p -> data [ i ]. p ;
// Check if the running sum is at least p or if we have kept at least min_keep tokens
// we set the last index to i+1 to indicate that the current iterate should be included in the set
if ( cum_sum >= ctx -> p && i + 1 >= ctx -> min_keep ) {
last_idx = i + 1 ;
break ;
}
}
// Resize the output vector to keep only the top-p tokens
cur_p -> size = last_idx ;
}
static struct llama_sampler * llama_sampler_top_p_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_top_p * ) smpl -> ctx ;
return llama_sampler_init_top_p ( ctx -> p , ctx -> min_keep );
}
static void llama_sampler_top_p_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_top_p * ) smpl -> ctx ;
}
2024-09-07 15:16:19 +03:00
static struct llama_sampler_i llama_sampler_top_p_i = {
2024-09-08 15:52:07 +02:00
/* .name = */ llama_sampler_top_p_name ,
2024-09-07 15:16:19 +03:00
/* .accept = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .apply = */ llama_sampler_top_p_apply ,
2024-09-07 15:16:19 +03:00
/* .reset = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .clone = */ llama_sampler_top_p_clone ,
/* .free = */ llama_sampler_top_p_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_init_top_p ( float p , size_t min_keep ) {
return new llama_sampler {
/* .iface = */ & llama_sampler_top_p_i ,
/* .ctx = */ new llama_sampler_top_p {
/* .p = */ p ,
/* .min_keep = */ min_keep ,
},
};
}
// min-p
struct llama_sampler_min_p {
const float p ;
const size_t min_keep ;
};
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_min_p_name ( const struct llama_sampler * /*smpl*/ ) {
return "min-p" ;
}
static void llama_sampler_min_p_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
const auto * ctx = ( llama_sampler_min_p * ) smpl -> ctx ;
if ( ctx -> p <= 0.0f || ! cur_p -> size ) {
return ;
}
bool min_p_applied = false ;
// if the cur_p aren't sorted, try the unsorted implementation first
if ( ! cur_p -> sorted ) {
std :: vector < llama_token_data > filtered_tokens ;
float max_logit = - FLT_MAX ;
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
max_logit = std :: max ( max_logit , cur_p -> data [ i ]. logit );
}
const float min_logit = max_logit + logf ( ctx -> p ); // min logit for p_i >= p * p_max
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
if ( cur_p -> data [ i ]. logit >= min_logit ) {
filtered_tokens . push_back ( cur_p -> data [ i ]);
}
}
// if we have enough values the operation was a success
if ( filtered_tokens . size () >= ctx -> min_keep ) {
memcpy ( cur_p -> data , filtered_tokens . data (), filtered_tokens . size () * sizeof ( llama_token_data ));
cur_p -> size = filtered_tokens . size ();
min_p_applied = true ;
}
}
// if the cur_p are sorted or the unsorted implementation failed, use this implementation
if ( ! min_p_applied ) {
// Sort the logits in descending order
if ( ! cur_p -> sorted ) {
std :: sort ( cur_p -> data , cur_p -> data + cur_p -> size , []( const llama_token_data & a , const llama_token_data & b ) {
return a . logit > b . logit ;
});
cur_p -> sorted = true ;
}
const float min_logit = cur_p -> data [ 0 ]. logit + logf ( ctx -> p ); // min logit for p_i >= p * p_max
size_t i = 1 ; // first token always matches
for (; i < cur_p -> size ; ++ i ) {
if ( cur_p -> data [ i ]. logit < min_logit && i >= ctx -> min_keep ) {
break ; // prob too small
}
}
// Resize the output vector to keep only the matching tokens
cur_p -> size = i ;
}
}
static struct llama_sampler * llama_sampler_min_p_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_min_p * ) smpl -> ctx ;
return llama_sampler_init_min_p ( ctx -> p , ctx -> min_keep );
}
static void llama_sampler_min_p_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_min_p * ) smpl -> ctx ;
}
2024-09-07 15:16:19 +03:00
static struct llama_sampler_i llama_sampler_min_p_i = {
2024-09-08 15:52:07 +02:00
/* .name = */ llama_sampler_min_p_name ,
2024-09-07 15:16:19 +03:00
/* .accept = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .apply = */ llama_sampler_min_p_apply ,
2024-09-07 15:16:19 +03:00
/* .reset = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .clone = */ llama_sampler_min_p_clone ,
/* .free = */ llama_sampler_min_p_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_init_min_p ( float p , size_t min_keep ) {
return new llama_sampler {
/* .iface = */ & llama_sampler_min_p_i ,
/* .ctx = */ new llama_sampler_min_p {
/* .p = */ p ,
/* .min_keep = */ min_keep ,
},
};
}
// tail-free
struct llama_sampler_tail_free {
const float z ;
const size_t min_keep ;
};
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_tail_free_name ( const struct llama_sampler * /*smpl*/ ) {
return "tail-free" ;
}
static void llama_sampler_tail_free_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
const auto * ctx = ( llama_sampler_tail_free * ) smpl -> ctx ;
if ( ctx -> z >= 1.0f || cur_p -> size <= 2 ) {
return ;
}
llama_sampler_softmax_impl ( cur_p );
// Compute the first and second derivatives
std :: vector < float > first_derivatives ( cur_p -> size - 1 );
std :: vector < float > second_derivatives ( cur_p -> size - 2 );
for ( size_t i = 0 ; i < first_derivatives . size (); ++ i ) {
first_derivatives [ i ] = cur_p -> data [ i ]. p - cur_p -> data [ i + 1 ]. p ;
}
for ( size_t i = 0 ; i < second_derivatives . size (); ++ i ) {
second_derivatives [ i ] = first_derivatives [ i ] - first_derivatives [ i + 1 ];
}
// Calculate absolute value of second derivatives
for ( size_t i = 0 ; i < second_derivatives . size (); ++ i ) {
second_derivatives [ i ] = std :: abs ( second_derivatives [ i ]);
}
// Normalize the second derivatives
{
const float second_derivatives_sum = std :: accumulate ( second_derivatives . begin (), second_derivatives . end (), 0.0f );
if ( second_derivatives_sum > 1e-6 f ) {
for ( float & value : second_derivatives ) {
value /= second_derivatives_sum ;
}
} else {
for ( float & value : second_derivatives ) {
value = 1.0f / second_derivatives . size ();
}
}
}
float cum_sum = 0.0f ;
size_t last_idx = cur_p -> size ;
for ( size_t i = 0 ; i < second_derivatives . size (); ++ i ) {
cum_sum += second_derivatives [ i ];
// Check if the running sum is greater than z or if we have kept at least min_keep tokens
if ( cum_sum > ctx -> z && i >= ctx -> min_keep ) {
last_idx = i ;
break ;
}
}
// Resize the output vector to keep only the tokens above the tail location
cur_p -> size = last_idx ;
}
static struct llama_sampler * llama_sampler_tail_free_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_tail_free * ) smpl -> ctx ;
return llama_sampler_init_tail_free ( ctx -> z , ctx -> min_keep );
}
static void llama_sampler_tail_free_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_tail_free * ) smpl -> ctx ;
}
2024-09-07 15:16:19 +03:00
static struct llama_sampler_i llama_sampler_tail_free_i = {
2024-09-08 15:52:07 +02:00
/* .name = */ llama_sampler_tail_free_name ,
2024-09-07 15:16:19 +03:00
/* .accept = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .apply = */ llama_sampler_tail_free_apply ,
2024-09-07 15:16:19 +03:00
/* .reset = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .clone = */ llama_sampler_tail_free_clone ,
/* .free = */ llama_sampler_tail_free_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_init_tail_free ( float z , size_t min_keep ) {
return new llama_sampler {
/* .iface = */ & llama_sampler_tail_free_i ,
/* .ctx = */ new llama_sampler_tail_free {
/* .z = */ z ,
/*. min_keep = */ min_keep ,
},
};
}
// typical
struct llama_sampler_typical {
const float p ;
const size_t min_keep ;
};
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_typical_name ( const struct llama_sampler * /*smpl*/ ) {
return "typical" ;
}
static void llama_sampler_typical_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
const auto * ctx = ( llama_sampler_typical * ) smpl -> ctx ;
// Reference implementation:
// https://github.com/huggingface/transformers/compare/main...cimeister:typical-sampling:typical-pr
if ( ctx -> p >= 1.0f ) {
return ;
}
// Compute the softmax of logits and calculate entropy
llama_sampler_softmax_impl ( cur_p );
float entropy = 0.0f ;
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
entropy += - cur_p -> data [ i ]. p * logf ( cur_p -> data [ i ]. p );
}
// Compute the absolute difference between negative log probability and entropy for each candidate
std :: vector < float > shifted_scores ;
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
float shifted_score = fabsf ( - logf ( cur_p -> data [ i ]. p ) - entropy );
shifted_scores . push_back ( shifted_score );
}
// Sort tokens based on the shifted_scores and their corresponding indices
std :: vector < size_t > indices ( cur_p -> size );
std :: iota ( indices . begin (), indices . end (), 0 );
std :: sort ( indices . begin (), indices . end (), [ & ]( size_t a , size_t b ) {
return shifted_scores [ a ] < shifted_scores [ b ];
});
// Compute the cumulative probabilities
float cum_sum = 0.0f ;
size_t last_idx = indices . size ();
for ( size_t i = 0 ; i < indices . size (); ++ i ) {
size_t idx = indices [ i ];
cum_sum += cur_p -> data [ idx ]. p ;
// Check if the running sum is greater than typical or if we have kept at least min_keep tokens
if ( cum_sum > ctx -> p && i >= ctx -> min_keep - 1 ) {
last_idx = i + 1 ;
break ;
}
}
// Resize the output vector to keep only the locally typical tokens
std :: vector < llama_token_data > cur_p_new ;
for ( size_t i = 0 ; i < last_idx ; ++ i ) {
size_t idx = indices [ i ];
cur_p_new . push_back ( cur_p -> data [ idx ]);
}
// Replace the data in cur_p with the cur_p_new data
std :: copy ( cur_p_new . begin (), cur_p_new . end (), cur_p -> data );
cur_p -> size = cur_p_new . size ();
cur_p -> sorted = false ;
}
static struct llama_sampler * llama_sampler_typical_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_typical * ) smpl -> ctx ;
return llama_sampler_init_typical ( ctx -> p , ctx -> min_keep );
}
static void llama_sampler_typical_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_typical * ) smpl -> ctx ;
}
2024-09-07 15:16:19 +03:00
static struct llama_sampler_i llama_sampler_typical_i = {
2024-09-08 15:52:07 +02:00
/* .name = */ llama_sampler_typical_name ,
2024-09-07 15:16:19 +03:00
/* .accept = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .apply = */ llama_sampler_typical_apply ,
2024-09-07 15:16:19 +03:00
/* .reset = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .clone = */ llama_sampler_typical_clone ,
/* .free = */ llama_sampler_typical_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_init_typical ( float p , size_t min_keep ) {
return new llama_sampler {
/* .iface = */ & llama_sampler_typical_i ,
/* .ctx = */ new llama_sampler_typical {
/* .p = */ p ,
/* .min_keep = */ min_keep ,
},
};
}
// temp
struct llama_sampler_temp {
const float temp ;
};
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_temp_name ( const struct llama_sampler * /*smpl*/ ) {
return "temp" ;
}
static void llama_sampler_temp_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
const auto * ctx = ( llama_sampler_temp * ) smpl -> ctx ;
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
cur_p -> data [ i ]. logit /= ctx -> temp ;
}
}
static struct llama_sampler * llama_sampler_temp_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_temp * ) smpl -> ctx ;
return llama_sampler_init_temp ( ctx -> temp );
}
static void llama_sampler_temp_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_temp * ) smpl -> ctx ;
}
2024-09-07 15:16:19 +03:00
static struct llama_sampler_i llama_sampler_temp_i = {
2024-09-08 15:52:07 +02:00
/* .name = */ llama_sampler_temp_name ,
2024-09-07 15:16:19 +03:00
/* .accept = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .apply = */ llama_sampler_temp_apply ,
2024-09-07 15:16:19 +03:00
/* .reset = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .clone = */ llama_sampler_temp_clone ,
/* .free = */ llama_sampler_temp_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_init_temp ( float temp ) {
return new llama_sampler {
/* .iface = */ & llama_sampler_temp_i ,
/* .ctx = */ new llama_sampler_temp {
/*.temp = */ temp ,
},
};
}
// temp-ext
struct llama_sampler_temp_ext {
const float temp ;
const float delta ;
const float exponent ;
};
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_temp_ext_name ( const struct llama_sampler * /*smpl*/ ) {
return "temp-ext" ;
}
2024-09-07 15:16:19 +03:00
2024-09-08 15:52:07 +02:00
static void llama_sampler_temp_ext_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
const auto * ctx = ( llama_sampler_temp_ext * ) smpl -> ctx ;
if ( ctx -> delta > 0 ) {
const float min_temp = std :: max ( 0.0f , ctx -> temp - ctx -> delta );
const float max_temp = ctx -> temp + ctx -> delta ;
float exponent_val = ctx -> exponent ;
// no need to do anything if there is only one (or zero) candidates
if ( cur_p -> size <= 1 ) {
return ;
2024-09-07 15:16:19 +03:00
}
2024-09-08 15:52:07 +02:00
// Calculate maximum possible entropy
float max_entropy = - logf ( 1.0f / cur_p -> size );
llama_sampler_softmax_impl ( cur_p );
// Calculate entropy of the softmax probabilities
float entropy = 0.0f ;
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
float prob = cur_p -> data [ i ]. p ;
if ( prob > 0.0f ) { // Ensure no log(0)
entropy -= prob * logf ( prob );
}
}
// Normalize the entropy (max_entropy cannot be 0 here because we checked cur_p->size != 1 above)
float normalized_entropy = entropy / max_entropy ;
// Map the normalized entropy to the desired temperature range using the power function
float dyn_temp = min_temp + ( max_temp - min_temp ) * powf ( normalized_entropy , exponent_val );
#ifdef DEBUG
LLAMA_LOG_INFO ( "Your text maxtemp value is: %f \n " , max_temp );
LLAMA_LOG_INFO ( "Entropy: %f \n " , entropy );
LLAMA_LOG_INFO ( "Max Possible Entropy: %f \n " , max_entropy );
LLAMA_LOG_INFO ( "Normalized Entropy: %f \n " , normalized_entropy );
LLAMA_LOG_INFO ( "Exponent: %f \n " , exponent_val );
LLAMA_LOG_INFO ( "Dynamic Temperature (dyn_temp): %f \n " , dyn_temp );
#endif
// Apply the dynamically calculated temperature scaling
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
cur_p -> data [ i ]. logit /= dyn_temp ;
}
// Re-compute softmax probabilities after scaling logits with dynamic temperature
const double max_l_double = cur_p -> data [ 0 ]. logit ;
double cum_sum_double = 0.0 ;
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
double p = exp ( cur_p -> data [ i ]. logit - max_l_double );
cur_p -> data [ i ]. p = p ; // Store the scaled probability
cum_sum_double += p ;
}
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
cur_p -> data [ i ]. p /= cum_sum_double ; // Re-normalize the probabilities
}
#ifdef DEBUG
// Print the updated top 25 probabilities after temperature scaling
LLAMA_LOG_INFO ( " \n Updated Top 25 Probabilities After Dynamic Temperature Scaling (in percentages): \n " );
for ( size_t i = 0 ; i < 25 && i < cur_p -> size ; ++ i ) {
LLAMA_LOG_INFO ( "Token %zu: %f%% \n " , i + 1 , cur_p -> data [ i ]. p * 100.0f );
}
#endif
} else {
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
cur_p -> data [ i ]. logit /= ctx -> temp ;
}
}
}
static struct llama_sampler * llama_sampler_temp_ext_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_temp_ext * ) smpl -> ctx ;
return llama_sampler_init_temp_ext ( ctx -> temp , ctx -> delta , ctx -> exponent );
}
static void llama_sampler_temp_ext_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_temp_ext * ) smpl -> ctx ;
}
static struct llama_sampler_i llama_sampler_temp_ext_i = {
/* .name = */ llama_sampler_temp_ext_name ,
/* .accept = */ nullptr ,
/* .apply = */ llama_sampler_temp_ext_apply ,
2024-09-07 15:16:19 +03:00
/* .reset = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .clone = */ llama_sampler_temp_ext_clone ,
/* .free = */ llama_sampler_temp_ext_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_init_temp_ext ( float temp , float delta , float exponent ) {
return new llama_sampler {
/* .iface = */ & llama_sampler_temp_ext_i ,
/* .ctx = */ new llama_sampler_temp_ext {
/* .temp = */ temp ,
/* .delta = */ delta ,
/* .exponent = */ exponent ,
},
};
}
2024-10-15 15:54:55 +05:00
// xtc
struct llama_sampler_xtc {
const float probability ;
const float threshold ;
const size_t min_keep ;
const uint32_t seed ;
uint32_t seed_cur ;
std :: mt19937 rng ;
};
static const char * llama_sampler_xtc_name ( const struct llama_sampler * /*smpl*/ ) {
return "xtc" ;
}
static void llama_sample_xtc_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
auto * ctx = ( llama_sampler_xtc * ) smpl -> ctx ;
if ( ctx -> probability <= 0.0f
|| ctx -> threshold > 0.5f
|| cur_p -> size < 2 ) {
return ;
}
std :: uniform_real_distribution < float > distribution ( 0.0f , 1.0f );
float chance = distribution ( ctx -> rng );
if ( chance > ctx -> probability ) return ;
// in case it's not sorted/recalculated yet
llama_sampler_softmax_impl ( cur_p );
int pos_last = 0 ;
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
if ( cur_p -> data [ i ]. p >= ctx -> threshold ) {
pos_last = i ;
} else break ;
}
if ( cur_p -> size - pos_last >= ctx -> min_keep && pos_last > 0 ) {
cur_p -> data += pos_last ;
cur_p -> size -= pos_last ;
}
}
static struct llama_sampler * llama_sampler_xtc_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_xtc * ) smpl -> ctx ;
auto * result = llama_sampler_init_xtc ( ctx -> probability , ctx -> threshold , ctx -> min_keep , ctx -> seed );
// copy the state
{
auto * result_ctx = ( llama_sampler_xtc * ) result -> ctx ;
result_ctx -> rng = ctx -> rng ;
}
return result ;
}
static void llama_sampler_xtc_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_xtc * ) smpl -> ctx ;
}
static void llama_sampler_xtc_reset ( struct llama_sampler * smpl ) {
auto * ctx = ( llama_sampler_xtc * ) smpl -> ctx ;
ctx -> seed_cur = get_rng_seed ( ctx -> seed );
ctx -> rng . seed ( ctx -> seed_cur );
}
static struct llama_sampler_i llama_sampler_xtc_i = {
/* .name = */ llama_sampler_xtc_name ,
/* .accept = */ nullptr ,
/* .apply = */ llama_sample_xtc_apply ,
/* .reset = */ llama_sampler_xtc_reset ,
/* .clone = */ llama_sampler_xtc_clone ,
/* .free = */ llama_sampler_xtc_free ,
};
struct llama_sampler * llama_sampler_init_xtc ( float p , float t , size_t min_keep , uint32_t seed ) {
auto seed_cur = get_rng_seed ( seed );
return new llama_sampler {
/* .iface = */ & llama_sampler_xtc_i ,
/* .ctx = */ new llama_sampler_xtc {
/* .probability = */ p ,
/* .threshold = */ t ,
/* .min_keep = */ min_keep ,
/* .seed = */ seed ,
/* .seed_cur = */ seed_cur ,
/* .rng = */ std :: mt19937 ( seed_cur ),
},
};
}
2024-09-07 15:16:19 +03:00
// mirostat
struct llama_sampler_mirostat {
const int32_t n_vocab ;
const uint32_t seed ;
2024-09-10 18:04:25 +02:00
uint32_t seed_cur ;
2024-09-07 15:16:19 +03:00
const float tau ;
const float eta ;
const int32_t m ;
float mu ;
std :: mt19937 rng ;
};
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_mirostat_name ( const struct llama_sampler * /*smpl*/ ) {
return "mirostat" ;
}
static void llama_sampler_mirostat_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
auto * ctx = ( llama_sampler_mirostat * ) smpl -> ctx ;
llama_sampler_softmax_impl ( cur_p );
// Estimate s_hat using the most probable m tokens
float s_hat = 0.0 ;
float sum_ti_bi = 0.0 ;
float sum_ti_sq = 0.0 ;
for ( size_t i = 0 ; i < size_t ( ctx -> m - 1 ) && i < cur_p -> size - 1 ; ++ i ) {
float t_i = logf ( float ( i + 2 ) / float ( i + 1 ));
float b_i = logf ( cur_p -> data [ i ]. p / cur_p -> data [ i + 1 ]. p );
sum_ti_bi += t_i * b_i ;
sum_ti_sq += t_i * t_i ;
}
s_hat = sum_ti_bi / sum_ti_sq ;
// Compute k from the estimated s_hat and target surprise value
float epsilon_hat = s_hat - 1 ;
float k = powf (( epsilon_hat * powf ( 2 , ctx -> mu )) / ( 1 - powf ( ctx -> n_vocab , - epsilon_hat )), 1 / s_hat );
llama_sampler_top_k_impl ( cur_p , std :: max ( int ( k ), 1 ));
llama_sampler_softmax_impl ( cur_p );
2024-09-09 17:10:46 +02:00
const int idx = llama_sample_dist ( cur_p , ctx -> rng );
2024-09-08 15:52:07 +02:00
cur_p -> selected = idx ;
float observed_surprise = - log2f ( cur_p -> data [ idx ]. p );
float e = observed_surprise - ctx -> tau ;
// Update mu using the learning rate and error
ctx -> mu = ctx -> mu - ctx -> eta * e ;
}
static struct llama_sampler * llama_sampler_mirostat_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_mirostat * ) smpl -> ctx ;
auto * result = llama_sampler_init_mirostat ( ctx -> n_vocab , ctx -> seed , ctx -> tau , ctx -> eta , ctx -> m );
// copy the state
{
auto * result_ctx = ( llama_sampler_mirostat * ) smpl -> ctx ;
result_ctx -> mu = ctx -> mu ;
result_ctx -> rng = ctx -> rng ;
}
return result ;
}
static void llama_sampler_mirostat_reset ( struct llama_sampler * smpl ) {
auto * ctx = ( llama_sampler_mirostat * ) smpl -> ctx ;
ctx -> mu = 2.0f * ctx -> tau ;
2024-09-10 18:04:25 +02:00
ctx -> seed_cur = get_rng_seed ( ctx -> seed );
ctx -> rng . seed ( ctx -> seed_cur );
2024-09-08 15:52:07 +02:00
}
static void llama_sampler_mirostat_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_mirostat * ) smpl -> ctx ;
}
2024-09-07 15:16:19 +03:00
static struct llama_sampler_i llama_sampler_mirostat_i = {
2024-09-08 15:52:07 +02:00
/* .name = */ llama_sampler_mirostat_name ,
2024-09-07 15:16:19 +03:00
/* .accept = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .apply = */ llama_sampler_mirostat_apply ,
/* .reset = */ llama_sampler_mirostat_reset ,
/* .clone = */ llama_sampler_mirostat_clone ,
/* .free = */ llama_sampler_mirostat_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_init_mirostat ( int32_t n_vocab , uint32_t seed , float tau , float eta , int32_t m ) {
2024-09-10 18:04:25 +02:00
auto seed_cur = get_rng_seed ( seed );
2024-09-07 15:16:19 +03:00
return new llama_sampler {
/* .iface = */ & llama_sampler_mirostat_i ,
/* .ctx = */ new llama_sampler_mirostat {
2024-09-10 18:04:25 +02:00
/* .n_vocab = */ n_vocab ,
/* .seed = */ seed ,
/* .seed_cur = */ seed_cur ,
/* .tau = */ tau ,
/* .eta = */ eta ,
/* .m = */ m ,
/* .mu = */ 2.0f * tau ,
/* .rng = */ std :: mt19937 ( seed_cur ),
2024-09-07 15:16:19 +03:00
},
};
}
// mirostat v2
struct llama_sampler_mirostat_v2 {
const uint32_t seed ;
2024-09-10 18:04:25 +02:00
uint32_t seed_cur ;
2024-09-07 15:16:19 +03:00
const float tau ;
const float eta ;
float mu ;
std :: mt19937 rng ;
};
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_mirostat_v2_name ( const struct llama_sampler * /*smpl*/ ) {
return "mirostat-v2" ;
}
static void llama_sampler_mirostat_v2_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
auto * ctx = ( llama_sampler_mirostat_v2 * ) smpl -> ctx ;
llama_sampler_softmax_impl ( cur_p );
// Truncate the words with surprise values greater than mu
cur_p -> size = std :: distance ( cur_p -> data , std :: find_if ( cur_p -> data , cur_p -> data + cur_p -> size , [ & ]( const llama_token_data & candidate ) {
return - log2f ( candidate . p ) > ctx -> mu ;
}));
if ( cur_p -> size == 0 ) {
cur_p -> size = 1 ;
}
// Normalize the probabilities of the remaining words
llama_sampler_softmax_impl ( cur_p );
2024-09-09 17:10:46 +02:00
const int idx = llama_sample_dist ( cur_p , ctx -> rng );
2024-09-08 15:52:07 +02:00
cur_p -> selected = idx ;
float observed_surprise = - log2f ( cur_p -> data [ idx ]. p );
float e = observed_surprise - ctx -> tau ;
// Update mu using the learning rate and error
ctx -> mu = ctx -> mu - ctx -> eta * e ;
}
static void llama_sampler_mirostat_v2_reset ( struct llama_sampler * smpl ) {
auto * ctx = ( llama_sampler_mirostat_v2 * ) smpl -> ctx ;
ctx -> mu = 2.0f * ctx -> tau ;
2024-09-10 18:04:25 +02:00
ctx -> seed_cur = get_rng_seed ( ctx -> seed );
ctx -> rng . seed ( ctx -> seed_cur );
2024-09-08 15:52:07 +02:00
}
static struct llama_sampler * llama_sampler_mirostat_v2_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_mirostat_v2 * ) smpl -> ctx ;
auto * result = llama_sampler_init_mirostat_v2 ( ctx -> seed , ctx -> tau , ctx -> eta );
// copy the state
{
auto * result_ctx = ( llama_sampler_mirostat_v2 * ) result -> ctx ;
result_ctx -> mu = ctx -> mu ;
result_ctx -> rng = ctx -> rng ;
}
return result ;
}
static void llama_sampler_mirostat_v2_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_mirostat_v2 * ) smpl -> ctx ;
}
2024-09-07 15:16:19 +03:00
static struct llama_sampler_i llama_sampler_mirostat_v2_i = {
2024-09-08 15:52:07 +02:00
/* .name = */ llama_sampler_mirostat_v2_name ,
2024-09-07 15:16:19 +03:00
/* .accept = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .apply = */ llama_sampler_mirostat_v2_apply ,
/* .reset = */ llama_sampler_mirostat_v2_reset ,
/* .clone = */ llama_sampler_mirostat_v2_clone ,
/* .free = */ llama_sampler_mirostat_v2_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_init_mirostat_v2 ( uint32_t seed , float tau , float eta ) {
2024-09-10 18:04:25 +02:00
auto seed_cur = get_rng_seed ( seed );
2024-09-07 15:16:19 +03:00
return new llama_sampler {
/* .iface = */ & llama_sampler_mirostat_v2_i ,
/* .ctx = */ new llama_sampler_mirostat_v2 {
2024-09-10 18:04:25 +02:00
/* .seed = */ seed ,
/* .seed_cur = */ seed_cur ,
/* .tau = */ tau ,
/* .eta = */ eta ,
/* .mu = */ 2.0f * tau ,
/* .rng = */ std :: mt19937 ( seed_cur ),
2024-09-07 15:16:19 +03:00
},
};
}
// grammar
struct llama_sampler_grammar {
const struct llama_vocab * vocab ;
std :: string grammar_str ;
std :: string grammar_root ;
struct llama_grammar * grammar ;
};
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_grammar_name ( const struct llama_sampler * /*smpl*/ ) {
return "grammar" ;
}
2024-09-07 15:16:19 +03:00
2024-09-08 15:52:07 +02:00
static void llama_sampler_grammar_accept_impl ( struct llama_sampler * smpl , llama_token token ) {
auto * ctx = ( llama_sampler_grammar * ) smpl -> ctx ;
if ( ctx -> grammar ) {
llama_grammar_accept_impl ( * ctx -> grammar , token );
}
}
2024-09-07 15:16:19 +03:00
2024-09-08 15:52:07 +02:00
static void llama_sampler_grammar_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
auto * ctx = ( llama_sampler_grammar * ) smpl -> ctx ;
if ( ctx -> grammar ) {
llama_grammar_apply_impl ( * ctx -> grammar , cur_p );
}
}
static void llama_sampler_grammar_reset ( struct llama_sampler * smpl ) {
auto * ctx = ( llama_sampler_grammar * ) smpl -> ctx ;
if ( ! ctx -> grammar ) {
return ;
}
auto * grammar_new = llama_grammar_init_impl ( ctx -> grammar -> vocab , ctx -> grammar_str . c_str (), ctx -> grammar_root . c_str ());
llama_grammar_free_impl ( ctx -> grammar );
ctx -> grammar = grammar_new ;
}
static struct llama_sampler * llama_sampler_grammar_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_grammar * ) smpl -> ctx ;
auto * result = llama_sampler_init_grammar_impl ( * ctx -> vocab , nullptr , nullptr );
// copy the state
{
auto * result_ctx = ( llama_sampler_grammar * ) result -> ctx ;
if ( ctx -> grammar ) {
result_ctx -> grammar_str = ctx -> grammar_str ;
result_ctx -> grammar_root = ctx -> grammar_root ;
result_ctx -> grammar = llama_grammar_clone_impl ( * ctx -> grammar );
}
}
return result ;
}
static void llama_sampler_grammar_free ( struct llama_sampler * smpl ) {
const auto * ctx = ( llama_sampler_grammar * ) smpl -> ctx ;
if ( ctx -> grammar ) {
2024-09-07 15:16:19 +03:00
llama_grammar_free_impl ( ctx -> grammar );
2024-09-08 15:52:07 +02:00
}
2024-09-07 15:16:19 +03:00
2024-09-08 15:52:07 +02:00
delete ctx ;
}
2024-09-07 15:16:19 +03:00
2024-09-08 15:52:07 +02:00
static struct llama_sampler_i llama_sampler_grammar_i = {
/* .name = */ llama_sampler_grammar_name ,
/* .accept = */ llama_sampler_grammar_accept_impl ,
/* .apply = */ llama_sampler_grammar_apply ,
/* .reset = */ llama_sampler_grammar_reset ,
/* .clone = */ llama_sampler_grammar_clone ,
/* .free = */ llama_sampler_grammar_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_init_grammar_impl ( const struct llama_vocab & vocab , const char * grammar_str , const char * grammar_root ) {
auto * ctx = new llama_sampler_grammar ;
if ( grammar_str != nullptr && grammar_str [ 0 ] != '\0' ) {
* ctx = {
/* .vocab = */ & vocab ,
/* .grammar_str = */ grammar_str ,
/* .grammar_root = */ grammar_root ,
/* .grammar = */ llama_grammar_init_impl ( & vocab , grammar_str , grammar_root ),
};
} else {
* ctx = {
/* .vocab = */ & vocab ,
/* .grammar_str = */ {},
/* .grammar_root = */ {},
/* .grammar = */ nullptr ,
};
2024-07-23 13:10:17 +03:00
}
2024-09-07 15:16:19 +03:00
return new llama_sampler {
/* .iface = */ & llama_sampler_grammar_i ,
/* .ctx = */ ctx ,
};
2024-07-23 13:10:17 +03:00
}
2024-09-07 15:16:19 +03:00
// penalties
struct llama_sampler_penalties {
const int32_t n_vocab ;
const llama_token special_eos_id ;
const llama_token linefeed_id ;
const int32_t penalty_last_n ;
const float penalty_repeat ;
const float penalty_freq ;
const float penalty_present ;
const bool penalize_nl ;
const bool ignore_eos ;
ring_buffer < llama_token > prev ;
};
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_penalties_name ( const struct llama_sampler * /*smpl*/ ) {
return "penalties" ;
}
static void llama_sampler_penalties_accept ( struct llama_sampler * smpl , llama_token token ) {
auto * ctx = ( llama_sampler_penalties * ) smpl -> ctx ;
if ( ctx -> penalty_last_n == 0 ) {
return ;
}
ctx -> prev . push_back ( token );
}
static void llama_sampler_penalties_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
auto * ctx = ( llama_sampler_penalties * ) smpl -> ctx ;
if ( ctx -> ignore_eos ) {
assert ( ctx -> special_eos_id >= 0 );
// optimistically check if the candidates are not yet sorted/shuffled/truncated
if ( cur_p -> size > ( size_t ) ctx -> special_eos_id && cur_p -> data [ ctx -> special_eos_id ]. id == ctx -> special_eos_id ) {
cur_p -> data [ ctx -> special_eos_id ]. logit = - INFINITY ;
} else {
// else, search for the special EOS token
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
if ( cur_p -> data [ i ]. id == ctx -> special_eos_id ) {
cur_p -> data [ i ]. logit = - INFINITY ;
break ;
}
}
}
}
if (( ctx -> penalty_last_n == 0 ) ||
( ctx -> penalty_repeat == 1.0f && ctx -> penalty_freq == 0.0f && ctx -> penalty_present == 0.0f )) {
return ;
}
bool nl_found = false ;
size_t nl_idx = 0 ;
float nl_logit = - INFINITY ;
if ( ! ctx -> penalize_nl ) {
assert ( ctx -> linefeed_id >= 0 );
// optimistically check if the candidates are not yet sorted/shuffled/truncated
if ( cur_p -> size > ( size_t ) ctx -> linefeed_id && cur_p -> data [ ctx -> linefeed_id ]. id == ctx -> linefeed_id ) {
nl_found = true ;
nl_idx = ctx -> linefeed_id ;
nl_logit = cur_p -> data [ ctx -> linefeed_id ]. logit ;
} else {
// else, search for the linefeed token
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
if ( cur_p -> data [ i ]. id == ctx -> linefeed_id ) {
nl_found = true ;
nl_idx = i ;
nl_logit = cur_p -> data [ i ]. logit ;
break ;
}
}
}
}
// Create a frequency map to count occurrences of each token in last_tokens
// TODO: optimize this by maintaining the token count in the sampler context
using llama_token_cnt = std :: unordered_map < llama_token , int > ;
llama_token_cnt token_count ;
for ( int i = 0 ; i < std :: min < int > ( ctx -> penalty_last_n , ctx -> prev . size ()); ++ i ) {
token_count [ ctx -> prev . rat ( i )] ++ ;
}
// Apply frequency and presence penalties to the cur_p
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
const auto token_iter = token_count . find ( cur_p -> data [ i ]. id );
if ( token_iter == token_count . end ()) {
continue ;
}
const int count = token_iter -> second ;
// The academic publication that described this technique actually just only divided, but that would cause tokens with negative logits to become more likely, which is obviously wrong.
// This is common fix for this problem, which is to multiply by the penalty instead of dividing.
if ( cur_p -> data [ i ]. logit <= 0 ) {
cur_p -> data [ i ]. logit *= ctx -> penalty_repeat ;
} else {
cur_p -> data [ i ]. logit /= ctx -> penalty_repeat ;
}
cur_p -> data [ i ]. logit -= float ( count ) * ctx -> penalty_freq + float ( count > 0 ) * ctx -> penalty_present ;
}
cur_p -> sorted = false ;
if ( ! ctx -> penalize_nl && nl_found ) {
// restore the logit of the newline token if it was penalized
cur_p -> data [ nl_idx ]. logit = nl_logit ;
}
}
static void llama_sampler_penalties_reset ( struct llama_sampler * smpl ) {
auto * ctx = ( llama_sampler_penalties * ) smpl -> ctx ;
ctx -> prev . clear ();
}
static struct llama_sampler * llama_sampler_penalties_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_penalties * ) smpl -> ctx ;
auto * result = llama_sampler_init_penalties (
ctx -> n_vocab ,
ctx -> special_eos_id ,
ctx -> linefeed_id ,
ctx -> penalty_last_n ,
ctx -> penalty_repeat ,
ctx -> penalty_freq ,
ctx -> penalty_present ,
ctx -> penalize_nl ,
ctx -> ignore_eos );
// copy the state
{
auto * result_ctx = ( llama_sampler_penalties * ) result -> ctx ;
result_ctx -> prev = ctx -> prev ;
}
return result ;
}
static void llama_sampler_penalties_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_penalties * ) smpl -> ctx ;
}
2024-09-07 15:16:19 +03:00
static struct llama_sampler_i llama_sampler_penalties_i = {
2024-09-08 15:52:07 +02:00
/* .name = */ llama_sampler_penalties_name ,
/* .accept = */ llama_sampler_penalties_accept ,
/* .apply = */ llama_sampler_penalties_apply ,
/* .reset = */ llama_sampler_penalties_reset ,
/* .clone = */ llama_sampler_penalties_clone ,
/* .free = */ llama_sampler_penalties_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_init_penalties (
int32_t n_vocab ,
llama_token special_eos_id ,
llama_token linefeed_id ,
int32_t penalty_last_n ,
float penalty_repeat ,
float penalty_freq ,
float penalty_present ,
bool penalize_nl ,
bool ignore_eos ) {
if ( linefeed_id == LLAMA_TOKEN_NULL ) {
2024-09-08 15:52:07 +02:00
penalize_nl = true ;
2024-09-07 15:16:19 +03:00
}
if ( special_eos_id == LLAMA_TOKEN_NULL ) {
2024-09-08 15:52:07 +02:00
ignore_eos = false ;
2024-09-07 15:16:19 +03:00
}
2024-09-10 18:04:25 +02:00
penalty_last_n = std :: max ( penalty_last_n , 0 );
2024-09-07 15:16:19 +03:00
return new llama_sampler {
/* .iface = */ & llama_sampler_penalties_i ,
/* .ctx = */ new llama_sampler_penalties {
/* .n_vocab = */ n_vocab ,
/* .special_eos_id = */ special_eos_id ,
/* .linefeed_id = */ linefeed_id ,
/* .penalty_last_n = */ penalty_last_n ,
/* .penalty_repeat = */ penalty_repeat ,
/* .penalty_freq = */ penalty_freq ,
/* .penalty_present = */ penalty_present ,
/* .penalize_nl = */ penalize_nl ,
/* .ignore_eos = */ ignore_eos ,
/* .prev = */ ring_buffer < llama_token > ( penalty_last_n ),
},
};
}
// logit-bias
struct llama_sampler_logit_bias {
const int32_t n_vocab ;
const std :: vector < llama_logit_bias > logit_bias ;
std :: vector < llama_logit_bias > to_search ;
};
2024-09-08 15:52:07 +02:00
static const char * llama_sampler_logit_bias_name ( const struct llama_sampler * /*smpl*/ ) {
return "logit-bias" ;
}
static void llama_sampler_logit_bias_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
auto * ctx = ( llama_sampler_logit_bias * ) smpl -> ctx ;
2024-09-09 17:10:46 +02:00
if ( ctx -> logit_bias . empty ()) {
return ;
}
2024-09-08 15:52:07 +02:00
ctx -> to_search . clear ();
// update the candidates that have not been shuffled in the vocabulary (i.e. idx == id)
for ( const auto & lb : ctx -> logit_bias ) {
if ( lb . token >= 0 && cur_p -> size > ( size_t ) lb . token && cur_p -> data [ lb . token ]. id == lb . token ) {
cur_p -> data [ lb . token ]. logit += lb . bias ;
} else {
ctx -> to_search . push_back ( lb );
}
}
2024-09-09 17:10:46 +02:00
if ( ctx -> to_search . empty ()) {
return ;
}
2024-09-08 15:52:07 +02:00
// search for the remaining candidates that were not found in the previous step
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
for ( const auto & lb : ctx -> to_search ) {
if ( cur_p -> data [ i ]. id == lb . token ) {
cur_p -> data [ i ]. logit += lb . bias ;
break ;
}
}
}
}
2024-09-10 18:04:25 +02:00
2024-09-08 15:52:07 +02:00
static struct llama_sampler * llama_sampler_logit_bias_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_logit_bias * ) smpl -> ctx ;
return llama_sampler_init_logit_bias ( ctx -> n_vocab , ctx -> logit_bias . size (), ctx -> logit_bias . data ());
}
static void llama_sampler_logit_bias_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_logit_bias * ) smpl -> ctx ;
}
2024-09-07 15:16:19 +03:00
static struct llama_sampler_i llama_sampler_logit_bias_i = {
2024-09-08 15:52:07 +02:00
/* .name = */ llama_sampler_logit_bias_name ,
2024-09-07 15:16:19 +03:00
/* .accept = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .apply = */ llama_sampler_logit_bias_apply ,
2024-09-07 15:16:19 +03:00
/* .reset = */ nullptr ,
2024-09-08 15:52:07 +02:00
/* .clone = */ llama_sampler_logit_bias_clone ,
/* .free = */ llama_sampler_logit_bias_free ,
2024-09-07 15:16:19 +03:00
};
struct llama_sampler * llama_sampler_init_logit_bias (
int32_t n_vocab ,
int32_t n_logit_bias ,
const llama_logit_bias * logit_bias ) {
return new llama_sampler {
/* .iface = */ & llama_sampler_logit_bias_i ,
/* .ctx = */ new llama_sampler_logit_bias {
/* .n_vocab = */ n_vocab ,
/* .logit_bias = */ std :: vector < llama_logit_bias > ( logit_bias , logit_bias + n_logit_bias ),
/* .to_search = */ {},
},
};
2024-07-23 13:10:17 +03:00
}
2024-09-10 18:04:25 +02:00
2024-10-15 16:35:33 +03:00
// infill
//#define GGML_DEBUG_SAMPLER_INFILL
struct llama_sampler_infill {
const struct llama_vocab * vocab ;
2024-10-17 22:32:47 +03:00
std :: vector < char > buf0 ;
std :: vector < char > buf1 ;
2024-10-15 16:35:33 +03:00
};
static const char * llama_sampler_infill_name ( const struct llama_sampler * /*smpl*/ ) {
return "infill" ;
}
static void llama_sampler_infill_apply ( struct llama_sampler * smpl , llama_token_data_array * cur_p ) {
auto * ctx = ( llama_sampler_infill * ) smpl -> ctx ;
llama_sampler_softmax_impl ( cur_p );
#if defined(GGML_DEBUG_SAMPLER_INFILL)
#define LOG_DBG_CUR LLAMA_LOG_DEBUG
#else
#define LOG_DBG_CUR(...)
#endif
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
LOG_DBG_CUR ( "%s: cur_p[%3zu] = { id: %6d, p: %.6f, logit: %6.3f } \n " , __func__ , i , cur_p -> data [ i ]. id , cur_p -> data [ i ]. p , cur_p -> data [ i ]. logit );
}
float p_txt_sum = 0.0f ;
float p_eog_sum = 0.0f ;
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
if ( llama_token_is_eog_impl ( * ctx -> vocab , cur_p -> data [ i ]. id )) {
p_eog_sum += cur_p -> data [ i ]. p ;
} else {
p_txt_sum += cur_p -> data [ i ]. p ;
}
}
const float rat = p_eog_sum == 0.0 ? INFINITY : p_txt_sum / p_eog_sum ; GGML_UNUSED ( rat );
LOG_DBG_CUR ( "%s: p_txt_sum = %.2f, p_eog_sum = %.2f, rat = %.2f, n = %zu \n " , __func__ , p_txt_sum , p_eog_sum , rat , cur_p -> size );
if ( 3 * p_eog_sum * cur_p -> size > p_txt_sum ) {
LOG_DBG_CUR ( "%s: the ratio p_txt/p_eog = %.2f is too low -> sampling EOG \n " , __func__ , p_txt_sum / p_eog_sum );
// keep just the EOG tokens
const auto size_org = cur_p -> size ;
cur_p -> size = 0 ;
float p_sum = 0.0f ;
for ( size_t i = 0 ; i < size_org ; ++ i ) {
if ( llama_token_is_eog_impl ( * ctx -> vocab , cur_p -> data [ i ]. id )) {
p_sum += cur_p -> data [ i ]. p ;
cur_p -> data [ cur_p -> size ++ ] = cur_p -> data [ i ];
}
}
// normalize probs
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
cur_p -> data [ i ]. p /= p_sum ;
}
return ;
}
size_t n_combined = 0 ; GGML_UNUSED ( n_combined );
// combine tokens with common prefix
2024-10-17 22:32:47 +03:00
for ( size_t i0 = 0 ; i0 < cur_p -> size ; ++ i0 ) {
for ( size_t i1 = 0 ; i1 < cur_p -> size ; ++ i1 ) {
if ( cur_p -> data [ i0 ]. logit == - INFINITY ) {
2024-10-15 16:35:33 +03:00
break ;
}
2024-10-17 22:32:47 +03:00
if ( i0 == i1 || cur_p -> data [ i1 ]. logit == - INFINITY ) {
2024-10-15 16:35:33 +03:00
continue ;
}
2024-10-17 22:32:47 +03:00
int len0 = llama_token_to_piece_impl ( * ctx -> vocab , cur_p -> data [ i0 ]. id , ctx -> buf0 . data (), ctx -> buf0 . size (), 0 , false );
if ( len0 < 0 ) {
ctx -> buf0 . resize ( len0 );
len0 = llama_token_to_piece_impl ( * ctx -> vocab , cur_p -> data [ i0 ]. id , ctx -> buf0 . data (), ctx -> buf0 . size (), 0 , false );
assert ( len0 > 0 );
}
int len1 = llama_token_to_piece_impl ( * ctx -> vocab , cur_p -> data [ i1 ]. id , ctx -> buf1 . data (), ctx -> buf1 . size (), 0 , false );
if ( len1 < 0 ) {
ctx -> buf1 . resize ( len1 );
len1 = llama_token_to_piece_impl ( * ctx -> vocab , cur_p -> data [ i1 ]. id , ctx -> buf1 . data (), ctx -> buf1 . size (), 0 , false );
assert ( len1 > 0 );
}
// token i0 is a prefix of token i1
if ( len0 > 0 && len0 <= len1 && memcmp ( ctx -> buf0 . data (), ctx -> buf1 . data (), len0 ) == 0 ) {
int dst = i0 ;
int src = i1 ;
// merge into the token with higher probability
if ( cur_p -> data [ i1 ]. p > cur_p -> data [ i0 ]. p ) {
std :: swap ( dst , src );
2024-10-15 16:35:33 +03:00
}
2024-10-17 22:32:47 +03:00
cur_p -> data [ dst ]. p += cur_p -> data [ src ]. p ;
cur_p -> data [ src ]. logit = - INFINITY ;
cur_p -> data [ src ]. p = 0.0f ;
2024-10-15 16:35:33 +03:00
n_combined ++ ;
}
}
}
size_t n_non_eog = 0 ;
size_t size_org = cur_p -> size ;
float p_sum = 0.0f ;
float thold = 0.2f ;
cur_p -> size = 0 ;
LOG_DBG_CUR ( "%s: n_combined = %zu, applying thold = %.3f \n " , __func__ , n_combined , thold );
for ( size_t i = 0 ; i < size_org ; ++ i ) {
const bool is_eog = llama_token_is_eog_impl ( * ctx -> vocab , cur_p -> data [ i ]. id );
if ( cur_p -> data [ i ]. p < thold && ! is_eog ) {
continue ;
}
if ( ! is_eog ) {
++ n_non_eog ;
}
p_sum += cur_p -> data [ i ]. p ;
// keep this token
cur_p -> data [ cur_p -> size ++ ] = cur_p -> data [ i ];
}
LOG_DBG_CUR ( "%s: n_non_eog = %zu \n " , __func__ , n_non_eog );
// if no non-EOG tokens are left -> reduce cur_p to single EOT token
if ( n_non_eog == 0 ) {
cur_p -> size = 1 ;
cur_p -> data [ 0 ]. id = llama_token_eot_impl ( * ctx -> vocab );
cur_p -> data [ 0 ]. logit = 1.0f ;
return ;
}
// normalize probs
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
cur_p -> data [ i ]. p /= p_sum ;
LOG_DBG_CUR ( "%s: cur_p[%3zu] = { id: %6d, p: %.6f, logit: %6.3f } \n " , __func__ , i , cur_p -> data [ i ]. id , cur_p -> data [ i ]. p , cur_p -> data [ i ]. logit );
}
size_org = cur_p -> size ;
p_sum = 0.0f ;
thold = 1.0 / ( n_non_eog + 1 );
cur_p -> size = 0 ;
LOG_DBG_CUR ( "%s: applying thold = %.3f \n " , __func__ , thold );
for ( size_t i = 0 ; i < size_org ; ++ i ) {
const bool is_eog = llama_token_is_eog_impl ( * ctx -> vocab , cur_p -> data [ i ]. id );
if ( cur_p -> data [ i ]. p < thold && ! is_eog ) {
continue ;
}
p_sum += cur_p -> data [ i ]. p ;
cur_p -> data [ cur_p -> size ++ ] = cur_p -> data [ i ];
}
// normalize probs
for ( size_t i = 0 ; i < cur_p -> size ; ++ i ) {
cur_p -> data [ i ]. p /= p_sum ;
LOG_DBG_CUR ( "%s: cur_p[%3zu] = { id: %6d, p: %.6f, logit: %6.3f } \n " , __func__ , i , cur_p -> data [ i ]. id , cur_p -> data [ i ]. p , cur_p -> data [ i ]. logit );
}
#undef LOG_DBG_CUR
}
static struct llama_sampler * llama_sampler_infill_clone ( const struct llama_sampler * smpl ) {
const auto * ctx = ( const llama_sampler_infill * ) smpl -> ctx ;
return llama_sampler_init_infill_impl ( * ctx -> vocab );
}
static void llama_sampler_infill_free ( struct llama_sampler * smpl ) {
delete ( llama_sampler_infill * ) smpl -> ctx ;
}
static struct llama_sampler_i llama_sampler_infill_i = {
/* .name = */ llama_sampler_infill_name ,
/* .accept = */ nullptr ,
/* .apply = */ llama_sampler_infill_apply ,
/* .reset = */ nullptr ,
/* .clone = */ llama_sampler_infill_clone ,
/* .free = */ llama_sampler_infill_free ,
};
struct llama_sampler * llama_sampler_init_infill_impl (
const struct llama_vocab & vocab ) {
return new llama_sampler {
/* .iface = */ & llama_sampler_infill_i ,
/* .ctx = */ new llama_sampler_infill {
/* .vocab = */ & vocab ,
2024-10-17 22:32:47 +03:00
/* .buf0 = */ std :: vector < char > ( 512 ),
/* .buf1 = */ std :: vector < char > ( 512 ),
2024-10-15 16:35:33 +03:00
},
};
}
2024-09-10 18:04:25 +02:00
// utils
uint32_t llama_sampler_get_seed ( const struct llama_sampler * smpl ) {
if ( smpl -> iface == & llama_sampler_dist_i ) {
return (( const llama_sampler_dist * ) smpl -> ctx ) -> seed_cur ;
}
if ( smpl -> iface == & llama_sampler_mirostat_i ) {
return (( const llama_sampler_mirostat * ) smpl -> ctx ) -> seed_cur ;
}
if ( smpl -> iface == & llama_sampler_mirostat_v2_i ) {
return (( const llama_sampler_mirostat_v2 * ) smpl -> ctx ) -> seed_cur ;
}
if ( smpl -> iface == & llama_sampler_chain_i ) {
const auto * ctx = ( const llama_sampler_chain * ) smpl -> ctx ;
for ( auto it = ctx -> samplers . rbegin (); it != ctx -> samplers . rend (); ++ it ) {
const uint32_t seed = llama_sampler_get_seed ( * it );
if ( seed != LLAMA_DEFAULT_SEED ) {
return seed ;
}
}
}
return LLAMA_DEFAULT_SEED ;
}
2024-09-13 09:53:38 +03:00
// perf
struct llama_perf_sampler_data llama_perf_sampler ( const struct llama_sampler * chain ) {
struct llama_perf_sampler_data data = {};
if ( chain == nullptr || chain -> iface != & llama_sampler_chain_i ) {
GGML_ABORT ( "%s: invalid sampler passed - requires a sampler created with llama_sampler_chain_init() \n " , __func__ );
}
const auto * ctx = ( const struct llama_sampler_chain * ) chain -> ctx ;
data . t_sample_ms = 1e-3 * ctx -> t_sample_us ;
data . n_sample = std :: max ( 0 , ctx -> n_sample );
return data ;
}
void llama_perf_sampler_print ( const struct llama_sampler * chain ) {
const auto data = llama_perf_sampler ( chain );
LLAMA_LOG_INFO ( "%s: sampling time = %10.2f ms / %5d runs (%8.2f ms per token, %8.2f tokens per second) \n " ,
__func__ , data . t_sample_ms , data . n_sample , data . t_sample_ms / data . n_sample , 1e3 / data . t_sample_ms * data . n_sample );
}
void llama_perf_sampler_reset ( struct llama_sampler * chain ) {
if ( chain == nullptr || chain -> iface != & llama_sampler_chain_i ) {
GGML_ABORT ( "%s: invalid sampler passed - requires a sampler created with llama_sampler_chain_init() \n " , __func__ );
}
auto * ctx = ( struct llama_sampler_chain * ) chain -> ctx ;
ctx -> t_sample_us = ctx -> n_sample = 0 ;
}