2023-08-11 23:17:25 +00:00
# include "ggml.h"
# include "llama.h"
2023-09-07 17:22:29 +00:00
# include "common.h"
2024-03-22 18:49:06 +00:00
# include "log.h"
2023-08-21 20:07:43 +00:00
2023-08-11 23:17:25 +00:00
# include <unordered_map>
# include <vector>
# include <cassert>
# include <climits>
# include <cstring>
# include <cstdarg>
# include <ctime>
# include <random>
# include <stdexcept>
2023-08-27 14:13:31 +00:00
# include <sstream>
2023-08-11 23:17:25 +00:00
# include <algorithm>
# include <string>
2023-08-27 14:13:31 +00:00
// GGUF keys & tensor names.
# define KV_GENERAL_ARCHITECTURE "general.architecture"
# define KV_GENERAL_NAME "general.name"
# define KV_TOKENIZER_MODEL "tokenizer.ggml.model"
# define KV_TOKENIZER_LIST "tokenizer.ggml.tokens"
# define KV_TOKENIZER_TOKEN_TYPE "tokenizer.ggml.token_type"
# define KV_TOKENIZER_SCORES "tokenizer.ggml.scores"
# define KV_TOKENIZER_BOS_ID "tokenizer.ggml.bos_token_id"
# define KV_TOKENIZER_EOS_ID "tokenizer.ggml.eos_token_id"
# define KV_TOKENIZER_UNK_ID "tokenizer.ggml.unknown_token_id"
# define KV_TOKENIZER_SEP_ID "tokenizer.ggml.seperator_token_id"
# define KV_TOKENIZER_PAD_ID "tokenizer.ggml.padding_token_id"
# define KV_TOKENIZER_HF_JSON "tokenizer.huggingface.json"
# define KV_CONTEXT_LENGTH "llama.context_length"
# define KV_EMBEDDING_LENGTH "llama.embedding_length"
# define KV_BLOCK_COUNT "llama.block_count"
# define KV_FEED_FORWARD_LENGTH "llama.feed_forward_length"
# define KV_ATTENTION_HEAD_COUNT "llama.attention.head_count"
# define KV_ATTENTION_HEAD_COUNT_KV "llama.attention.head_count_kv"
# define KV_ATTENTION_LAYERNORM_RMS_EPS "llama.attention.layer_norm_rms_epsilon"
# define KV_ROPE_DIMENSION_COUNT "llama.rope.dimension_count"
# define TN_TOKEN_EMBD "token_embd.weight"
# define TN_OUTPUT_NORM "output_norm.weight"
# define TN_OUTPUT "output.weight"
# define TN_ATTN_NORM "blk.%d.attn_norm.weight"
# define TN_ATTN_Q "blk.%d.attn_q.weight"
# define TN_ATTN_K "blk.%d.attn_k.weight"
# define TN_ATTN_V "blk.%d.attn_v.weight"
# define TN_ATTN_OUTPUT "blk.%d.attn_output.weight"
# define TN_FFN_NORM "blk.%d.ffn_norm.weight"
# define TN_FFN_GATE "blk.%d.ffn_gate.weight"
# define TN_FFN_DOWN "blk.%d.ffn_down.weight"
# define TN_FFN_UP "blk.%d.ffn_up.weight"
2023-08-11 23:17:25 +00:00
# if defined(_MSC_VER)
# pragma warning(disable: 4244 4267) // possible loss of data
# endif
2023-08-23 19:33:05 +00:00
# define LLAMA_FILE_MAGIC_GGJT 0x67676a74u // 'ggjt'
# define LLAMA_FILE_VERSION_GGJT_V3 3
2023-08-27 14:13:31 +00:00
# define TOKENIZER_NAME "llama"
# define UNKNOWN_TOKEN_ID 0
# define BOS_TOKEN_ID 1
# define EOS_TOKEN_ID 2
2023-08-11 23:17:25 +00:00
//////////////////////////////////////// llama2.c model structs and functions to load models, alloc memory etc.
typedef struct {
int dim ; // transformer dimension
int hidden_dim ; // for ffn layers
int n_layers ; // number of layers
int n_heads ; // number of query heads
int n_kv_heads ; // number of key/value heads (can be < query heads because of multiquery)
int vocab_size ; // vocabulary size, usually 256 (byte-level)
int seq_len ; // max sequence length
} Config ;
2023-09-01 09:03:49 +00:00
struct TransformerWeights {
2023-08-11 23:17:25 +00:00
// token embedding table
2024-03-22 18:49:06 +00:00
std : : vector < float > token_embedding_table ; // (vocab_size, dim)
2023-08-11 23:17:25 +00:00
// weights for rmsnorms
2024-03-22 18:49:06 +00:00
std : : vector < float > rms_att_weight ; // (layer, dim) rmsnorm weights
std : : vector < float > rms_ffn_weight ; // (layer, dim)
2023-08-11 23:17:25 +00:00
// weights for matmuls
2024-03-22 18:49:06 +00:00
std : : vector < float > wq ; // (layer, dim, dim)
std : : vector < float > wk ; // (layer, dim, dim)
std : : vector < float > wv ; // (layer, dim, dim)
std : : vector < float > wo ; // (layer, dim, dim)
2023-08-11 23:17:25 +00:00
// weights for ffn
2024-03-22 18:49:06 +00:00
std : : vector < float > w1 ; // (layer, hidden_dim, dim)
std : : vector < float > w2 ; // (layer, dim, hidden_dim)
std : : vector < float > w3 ; // (layer, hidden_dim, dim)
2023-08-11 23:17:25 +00:00
// final rmsnorm
2024-03-22 18:49:06 +00:00
std : : vector < float > rms_final_weight ; // (dim,)
2023-08-11 23:17:25 +00:00
// freq_cis for RoPE relatively positional embeddings
2024-03-22 18:49:06 +00:00
// std::vector<float> freq_cis_real; // (seq_len, dim/2)
// std::vector<float> freq_cis_imag; // (seq_len, dim/2)
2023-08-11 23:17:25 +00:00
// (optional) classifier weights for the logits, on the last layer
2024-03-22 18:49:06 +00:00
std : : vector < float > wcls ;
2023-09-01 09:03:49 +00:00
} ;
2023-08-11 23:17:25 +00:00
2024-03-22 18:49:06 +00:00
static void alloc_weights ( TransformerWeights * w , const Config * p , bool shared_weights ) {
const int n_multiqueries = p - > n_kv_heads < = 0 | | p - > n_kv_heads > = p - > n_heads ? 1 : p - > n_heads / p - > n_kv_heads ;
try {
w - > token_embedding_table . resize ( p - > vocab_size * p - > dim ) ;
LOG ( " %s: Allocating [%d] x [%d] = [%d] float space for w->token_embedding_table \n " , __func__ , p - > vocab_size , p - > dim , p - > vocab_size * p - > dim ) ;
2023-08-11 23:17:25 +00:00
2024-03-22 18:49:06 +00:00
w - > rms_att_weight . resize ( p - > n_layers * p - > dim ) ;
LOG ( " %s: Allocating [%d] x [%d] = [%d] float space for w->rms_att_weight \n " , __func__ , p - > n_layers , p - > dim , p - > n_layers * p - > dim ) ;
2023-08-11 23:17:25 +00:00
2024-03-22 18:49:06 +00:00
w - > rms_ffn_weight . resize ( p - > n_layers * p - > dim ) ;
LOG ( " %s: Allocating [%d] x [%d] = [%d] float space for w->rms_ffn_weight \n " , __func__ , p - > n_layers , p - > dim , p - > n_layers * p - > dim ) ;
2023-08-11 23:17:25 +00:00
2024-03-22 18:49:06 +00:00
w - > wq . resize ( p - > n_layers * p - > dim * p - > dim ) ;
LOG ( " %s: Allocating [%d] x [%d] x [%d] = [%d] float space for w->wq \n " , __func__ , p - > n_layers , p - > dim , p - > dim , p - > n_layers * p - > dim * p - > dim ) ;
2023-08-11 23:17:25 +00:00
2024-03-22 18:49:06 +00:00
w - > wk . resize ( p - > n_layers * p - > dim * p - > dim / n_multiqueries ) ;
LOG ( " %s: Allocating [%d] x [%d] x [%d] = [%d] float space for w->wk \n " , __func__ , p - > n_layers , p - > dim , p - > dim / n_multiqueries , p - > n_layers * p - > dim * p - > dim / n_multiqueries ) ;
2023-08-11 23:17:25 +00:00
2024-03-22 18:49:06 +00:00
w - > wv . resize ( p - > n_layers * p - > dim * p - > dim / n_multiqueries ) ;
LOG ( " %s: Allocating [%d] x [%d] x [%d] = [%d] float space for w->wv \n " , __func__ , p - > n_layers , p - > dim , p - > dim / n_multiqueries , p - > n_layers * p - > dim * p - > dim / n_multiqueries ) ;
2023-08-11 23:17:25 +00:00
2024-03-22 18:49:06 +00:00
w - > wo . resize ( p - > n_layers * p - > dim * p - > dim ) ;
LOG ( " %s: Allocating [%d] x [%d] x [%d] = [%d] float space for w->wo \n " , __func__ , p - > n_layers , p - > dim , p - > dim , p - > n_layers * p - > dim * p - > dim ) ;
2023-08-11 23:17:25 +00:00
2024-03-22 18:49:06 +00:00
w - > w1 . resize ( p - > n_layers * p - > hidden_dim * p - > dim ) ;
LOG ( " %s: Allocating [%d] x [%d] x [%d] = [%d] float space for w->w1 \n " , __func__ , p - > n_layers , p - > hidden_dim , p - > dim , p - > n_layers * p - > hidden_dim * p - > dim ) ;
2023-08-11 23:17:25 +00:00
2024-03-22 18:49:06 +00:00
w - > w2 . resize ( p - > n_layers * p - > hidden_dim * p - > dim ) ;
LOG ( " %s: Allocating [%d] x [%d] x [%d] = [%d] float space for w->w2 \n " , __func__ , p - > n_layers , p - > dim , p - > hidden_dim , p - > n_layers * p - > hidden_dim * p - > dim ) ;
2023-08-11 23:17:25 +00:00
2024-03-22 18:49:06 +00:00
w - > w3 . resize ( p - > n_layers * p - > hidden_dim * p - > dim ) ;
LOG ( " %s: Allocating [%d] x [%d] x [%d] = [%d] float space for w->w3 \n " , __func__ , p - > n_layers , p - > hidden_dim , p - > dim , p - > n_layers * p - > hidden_dim * p - > dim ) ;
2023-08-11 23:17:25 +00:00
2024-03-22 18:49:06 +00:00
w - > rms_final_weight . resize ( p - > dim ) ;
LOG ( " %s: Allocating [%d] float space for w->rms_final_weight \n " , __func__ , p - > dim ) ;
2023-08-23 19:33:05 +00:00
2024-03-22 18:49:06 +00:00
if ( shared_weights ) {
w - > wcls = { } ;
} else {
w - > wcls . resize ( p - > vocab_size * p - > dim ) ;
LOG ( " %s: Allocating [%d] x [%d] = [%d] float space for w->wcls \n " , __func__ , p - > vocab_size , p - > dim , p - > vocab_size * p - > dim ) ;
}
}
catch ( std : : length_error & ) {
die ( " Invalid configuration. Failed to allocate memory for weights " ) ;
2023-08-23 19:33:05 +00:00
}
2023-08-11 23:17:25 +00:00
}
2024-03-22 18:49:06 +00:00
static int checkpoint_init_weights ( TransformerWeights * w , const Config * p , FILE * f , bool shared_weights ) {
if ( fread ( w - > token_embedding_table . data ( ) , sizeof ( float ) , w - > token_embedding_table . size ( ) , f ) ! = w - > token_embedding_table . size ( ) ) return 1 ;
if ( fread ( w - > rms_att_weight . data ( ) , sizeof ( float ) , w - > rms_att_weight . size ( ) , f ) ! = w - > rms_att_weight . size ( ) ) return 1 ;
if ( fread ( w - > wq . data ( ) , sizeof ( float ) , w - > wq . size ( ) , f ) ! = w - > wq . size ( ) ) return 1 ;
if ( fread ( w - > wk . data ( ) , sizeof ( float ) , w - > wk . size ( ) , f ) ! = w - > wk . size ( ) ) return 1 ;
if ( fread ( w - > wv . data ( ) , sizeof ( float ) , w - > wv . size ( ) , f ) ! = w - > wv . size ( ) ) return 1 ;
if ( fread ( w - > wo . data ( ) , sizeof ( float ) , w - > wo . size ( ) , f ) ! = w - > wo . size ( ) ) return 1 ;
if ( fread ( w - > rms_ffn_weight . data ( ) , sizeof ( float ) , w - > rms_ffn_weight . size ( ) , f ) ! = w - > rms_ffn_weight . size ( ) ) return 1 ;
if ( fread ( w - > w1 . data ( ) , sizeof ( float ) , w - > w1 . size ( ) , f ) ! = w - > w1 . size ( ) ) return 1 ;
if ( fread ( w - > w2 . data ( ) , sizeof ( float ) , w - > w2 . size ( ) , f ) ! = w - > w2 . size ( ) ) return 1 ;
if ( fread ( w - > w3 . data ( ) , sizeof ( float ) , w - > w3 . size ( ) , f ) ! = w - > w3 . size ( ) ) return 1 ;
if ( fread ( w - > rms_final_weight . data ( ) , sizeof ( float ) , w - > rms_final_weight . size ( ) , f ) ! = w - > rms_final_weight . size ( ) ) return 1 ;
2023-08-23 19:33:05 +00:00
// Skip freq_cis_real & freq_cis_imag
int head_size = p - > dim / p - > n_heads ;
fseek ( f , p - > seq_len * head_size * sizeof ( float ) , SEEK_CUR ) ;
2024-03-22 18:49:06 +00:00
if ( ! shared_weights & & fread ( w - > wcls . data ( ) , sizeof ( float ) , w - > wcls . size ( ) , f ) ! = w - > wcls . size ( ) ) return 1 ;
2023-08-23 19:33:05 +00:00
// Check we didn't forget to read anything
auto curr = ftell ( f ) ;
fseek ( f , 0 , SEEK_END ) ;
auto end = ftell ( f ) ;
if ( curr ! = end ) {
2024-03-22 18:49:06 +00:00
LOG ( " %s: Error: failed to read the checkpoint file to the end (curr = %ld, end = %ld) \n " , __func__ , curr , end ) ;
2023-08-23 19:33:05 +00:00
return 1 ;
}
2023-08-11 23:17:25 +00:00
return 0 ;
}
2023-09-15 19:38:27 +00:00
static void print_sample_weights ( TransformerWeights * w ) {
2024-03-22 18:49:06 +00:00
LOG ( " ----- Quick print of first of the weight vales of all the variables \n " ) ;
LOG ( " %f \n " , w - > token_embedding_table [ 0 ] ) ;
LOG ( " %f \n " , w - > rms_att_weight [ 0 ] ) ;
LOG ( " %f \n " , w - > rms_ffn_weight [ 0 ] ) ;
LOG ( " %f \n " , w - > wq [ 0 ] ) ;
LOG ( " %f \n " , w - > wk [ 0 ] ) ;
LOG ( " %f \n " , w - > wv [ 0 ] ) ;
LOG ( " %f \n " , w - > wo [ 0 ] ) ;
LOG ( " %f \n " , w - > w1 [ 0 ] ) ;
LOG ( " %f \n " , w - > w2 [ 0 ] ) ;
LOG ( " %f \n " , w - > w3 [ 0 ] ) ;
LOG ( " %f \n " , w - > rms_att_weight [ 0 ] ) ;
if ( ! w - > wcls . empty ( ) ) LOG ( " %f \n " , w - > wcls [ 0 ] ) ;
2023-08-11 23:17:25 +00:00
}
////////////////////////////////////////////////////////////////////////////////////////////////////////////
//////////////////////////////////////// ggml structs and functions required to load models, configs and save the model.
struct llama_vocab {
using id = int32_t ;
using token = std : : string ;
2023-08-21 20:07:43 +00:00
using ttype = llama_token_type ;
2023-08-11 23:17:25 +00:00
2023-08-21 20:07:43 +00:00
struct token_data {
token text ;
2023-08-11 23:17:25 +00:00
float score ;
2023-08-21 20:07:43 +00:00
ttype type ;
2023-08-11 23:17:25 +00:00
} ;
std : : unordered_map < token , id > token_to_id ;
2023-08-21 20:07:43 +00:00
std : : vector < token_data > id_to_token ;
2023-08-11 23:17:25 +00:00
} ;
struct my_llama_hparams {
2024-03-22 18:49:06 +00:00
uint32_t n_vocab = 32000 ;
uint32_t n_ctx = 512 ; // this is provided as user input?
uint32_t n_embd = 4096 ;
uint32_t n_ff = 11008 ;
uint32_t n_mult = 4 ;
uint32_t n_head = 32 ;
uint32_t n_head_kv = 32 ;
uint32_t n_layer = 32 ;
uint32_t n_rot = 64 ;
2023-08-11 23:17:25 +00:00
bool operator ! = ( const my_llama_hparams & other ) const {
return memcmp ( this , & other , sizeof ( my_llama_hparams ) ) ;
}
} ;
struct my_llama_layer {
// normalization
struct ggml_tensor * attention_norm ;
// attention
struct ggml_tensor * wq ;
struct ggml_tensor * wk ;
struct ggml_tensor * wv ;
struct ggml_tensor * wo ;
// normalization
struct ggml_tensor * ffn_norm ;
// ff
struct ggml_tensor * w1 ;
struct ggml_tensor * w2 ;
struct ggml_tensor * w3 ;
} ;
struct my_llama_model {
struct ggml_context * ctx = NULL ;
2023-08-27 14:13:31 +00:00
std : : string name ;
2023-08-11 23:17:25 +00:00
my_llama_hparams hparams ;
struct ggml_tensor * tok_embeddings ;
struct ggml_tensor * norm ;
struct ggml_tensor * output ;
std : : vector < my_llama_layer > layers ;
uint32_t train_its = 0 ;
uint32_t train_samples = 0 ;
uint32_t train_tokens = 0 ;
} ;
struct train_params {
const char * fn_vocab_model ;
const char * fn_llama2c_model ;
const char * fn_llama2c_output_model ;
const char * fn_train_data ;
const char * fn_checkpoint_in ;
const char * fn_checkpoint_out ;
const char * fn_model_out ;
uint32_t seed ;
int n_ctx ;
int n_embd ;
int n_mult ;
int n_head ;
int n_layer ;
int n_rotmax ;
int n_threads ;
int n_batch ;
int n_examples ;
int n_predict ;
int print_info_interval ;
int print_details_interval ;
bool samples_start_after_nl ;
bool use_adam ;
bool use_flash ;
bool use_scratch ;
// only adam
int warmup ;
int cos_decay_steps ;
float cos_decay_restart ;
float cos_decay_alpha ;
int lbfgs_n_iter ;
int adam_n_iter ;
float adam_alpha ;
float adam_decay ;
int mem_model_gb ;
int mem_compute_gb ;
int mem_compute0_gb ;
int mem_compute1_gb ;
} ;
2023-09-15 19:38:27 +00:00
static void print_params ( struct my_llama_hparams * params ) {
2024-03-22 18:49:06 +00:00
LOG ( " %s: n_vocab: %u \n " , __func__ , params - > n_vocab ) ;
LOG ( " %s: n_ctx: %u \n " , __func__ , params - > n_ctx ) ;
LOG ( " %s: n_embd: %u \n " , __func__ , params - > n_embd ) ;
LOG ( " %s: n_mult: %u \n " , __func__ , params - > n_mult ) ;
LOG ( " %s: n_head: %u \n " , __func__ , params - > n_head ) ;
LOG ( " %s: n_head_kv: %u \n " , __func__ , params - > n_head_kv ) ;
LOG ( " %s: n_ff: %u \n " , __func__ , params - > n_ff ) ;
LOG ( " %s: n_layer: %u \n " , __func__ , params - > n_layer ) ;
LOG ( " %s: n_rot: %u \n " , __func__ , params - > n_rot ) ;
}
static void print_tensor_info ( const struct ggml_context * ctx ) {
for ( auto t = ggml_get_first_tensor ( ctx ) ; t ! = NULL ; t = ggml_get_next_tensor ( ctx , t ) ) {
LOG ( " %s: Allocating " , __func__ ) ;
int64_t total = 1 ;
int i = 0 ;
for ( ; i < ggml_n_dims ( t ) ; + + i ) {
if ( i > 0 ) LOG ( " x " ) ;
LOG ( " [% " PRId64 " ] " , t - > ne [ i ] ) ;
total * = t - > ne [ i ] ;
}
if ( i > 1 ) LOG ( " = [% " PRId64 " ] " , total ) ;
LOG ( " float space for %s \n " , ggml_get_name ( t ) ) ;
}
2023-08-11 23:17:25 +00:00
}
2023-09-15 19:38:27 +00:00
static void init_model ( struct my_llama_model * model ) {
2023-08-11 23:17:25 +00:00
const auto & hparams = model - > hparams ;
const uint32_t n_embd = hparams . n_embd ;
const uint32_t n_layer = hparams . n_layer ;
const uint32_t n_vocab = hparams . n_vocab ;
2024-03-22 18:49:06 +00:00
const uint32_t n_multiqueries = hparams . n_head_kv < = 0 | | hparams . n_head_kv > = hparams . n_head ? 1 : hparams . n_head / hparams . n_head_kv ;
2023-08-27 14:13:31 +00:00
const uint32_t n_ff = hparams . n_ff ;
2023-08-11 23:17:25 +00:00
struct ggml_context * ctx = model - > ctx ;
model - > train_its = 0 ;
model - > train_samples = 0 ;
model - > train_tokens = 0 ;
model - > tok_embeddings = ggml_new_tensor_2d ( ctx , GGML_TYPE_F32 , n_embd , n_vocab ) ;
model - > norm = ggml_new_tensor_1d ( ctx , GGML_TYPE_F32 , n_embd ) ;
model - > output = ggml_new_tensor_2d ( ctx , GGML_TYPE_F32 , n_embd , n_vocab ) ;
ggml_set_name ( model - > tok_embeddings , " tok_embeddings.weight " ) ;
ggml_set_name ( model - > norm , " norm.weight " ) ;
ggml_set_name ( model - > output , " output.weight " ) ;
model - > layers . resize ( n_layer ) ;
for ( uint32_t i = 0 ; i < n_layer ; + + i ) {
auto & layer = model - > layers [ i ] ;
std : : string layers_i = " layers. " + std : : to_string ( i ) ;
layer . attention_norm = ggml_new_tensor_1d ( ctx , GGML_TYPE_F32 , n_embd ) ;
layer . wq = ggml_new_tensor_2d ( ctx , GGML_TYPE_F32 , n_embd , n_embd ) ;
2024-03-22 18:49:06 +00:00
layer . wk = ggml_new_tensor_2d ( ctx , GGML_TYPE_F32 , n_embd , n_embd / n_multiqueries ) ;
layer . wv = ggml_new_tensor_2d ( ctx , GGML_TYPE_F32 , n_embd , n_embd / n_multiqueries ) ;
2023-08-11 23:17:25 +00:00
layer . wo = ggml_new_tensor_2d ( ctx , GGML_TYPE_F32 , n_embd , n_embd ) ;
layer . ffn_norm = ggml_new_tensor_1d ( ctx , GGML_TYPE_F32 , n_embd ) ;
layer . w1 = ggml_new_tensor_2d ( ctx , GGML_TYPE_F32 , n_embd , n_ff ) ;
layer . w2 = ggml_new_tensor_2d ( ctx , GGML_TYPE_F32 , n_ff , n_embd ) ;
layer . w3 = ggml_new_tensor_2d ( ctx , GGML_TYPE_F32 , n_embd , n_ff ) ;
ggml_set_name ( layer . attention_norm , ( layers_i + " .attention_norm.weight " ) . c_str ( ) ) ;
ggml_set_name ( layer . wq , ( layers_i + " .attention.wq.weight " ) . c_str ( ) ) ;
ggml_set_name ( layer . wk , ( layers_i + " .attention.wk.weight " ) . c_str ( ) ) ;
ggml_set_name ( layer . wv , ( layers_i + " .attention.wv.weight " ) . c_str ( ) ) ;
ggml_set_name ( layer . wo , ( layers_i + " .attention.wo.weight " ) . c_str ( ) ) ;
ggml_set_name ( layer . ffn_norm , ( layers_i + " .ffn_norm.weight " ) . c_str ( ) ) ;
ggml_format_name ( layer . w1 , " %s.feed_forward.w1.weight " , layers_i . c_str ( ) ) ;
ggml_format_name ( layer . w2 , " %s.feed_forward.w2.weight " , layers_i . c_str ( ) ) ;
ggml_format_name ( layer . w3 , " %s.feed_forward.w3.weight " , layers_i . c_str ( ) ) ;
}
2024-03-22 18:49:06 +00:00
print_tensor_info ( ctx ) ;
2023-08-11 23:17:25 +00:00
}
2023-09-15 19:38:27 +00:00
static float get_f32_2d ( struct ggml_tensor * tensor , int64_t i0 , int64_t i1 ) {
2023-08-11 23:17:25 +00:00
float * ptr = ( float * ) ( ( char * ) tensor - > data + i0 * tensor - > nb [ 0 ] + i1 * tensor - > nb [ 1 ] ) ;
return * ptr ;
}
2023-09-15 19:38:27 +00:00
static int32_t get_i32_2d ( struct ggml_tensor * tensor , int64_t i0 , int64_t i1 ) {
2023-08-11 23:17:25 +00:00
int32_t * ptr = ( int32_t * ) ( ( char * ) tensor - > data + i0 * tensor - > nb [ 0 ] + i1 * tensor - > nb [ 1 ] ) ;
return * ptr ;
}
2023-09-15 19:38:27 +00:00
static void print_row ( struct ggml_tensor * probs , int i ) {
2023-08-11 23:17:25 +00:00
for ( int k = 0 ; k < probs - > ne [ 0 ] ; + + k ) {
float p = get_f32_2d ( probs , k , i ) ;
2024-03-22 18:49:06 +00:00
LOG ( " %f " , p ) ;
2023-08-11 23:17:25 +00:00
}
2024-03-22 18:49:06 +00:00
LOG ( " \n " ) ;
2023-08-11 23:17:25 +00:00
}
2023-09-15 19:38:27 +00:00
static void print_matrix ( struct ggml_tensor * probs ) {
2023-12-14 15:52:08 +00:00
assert ( ggml_is_matrix ( probs ) ) ;
2023-08-11 23:17:25 +00:00
for ( int i = 0 ; i < probs - > ne [ 1 ] ; + + i ) {
for ( int k = 0 ; k < probs - > ne [ 0 ] ; + + k ) {
float p = get_f32_2d ( probs , k , i ) ;
2024-03-22 18:49:06 +00:00
LOG ( " %.2f " , p ) ;
2023-08-11 23:17:25 +00:00
}
2024-03-22 18:49:06 +00:00
LOG ( " \n " ) ;
2023-08-11 23:17:25 +00:00
}
}
struct llama_file {
// use FILE * so we don't have to re-open the file to mmap
FILE * fp ;
size_t size ;
llama_file ( const char * fname , const char * mode ) {
fp = std : : fopen ( fname , mode ) ;
if ( fp = = NULL ) {
size = 0 ;
} else {
seek ( 0 , SEEK_END ) ;
size = tell ( ) ;
seek ( 0 , SEEK_SET ) ;
}
}
size_t tell ( ) const {
# ifdef _WIN32
__int64 ret = _ftelli64 ( fp ) ;
# else
long ret = std : : ftell ( fp ) ;
# endif
GGML_ASSERT ( ret ! = - 1 ) ; // this really shouldn't fail
return ( size_t ) ret ;
}
void seek ( size_t offset , int whence ) {
# ifdef _WIN32
int ret = _fseeki64 ( fp , ( __int64 ) offset , whence ) ;
# else
int ret = std : : fseek ( fp , ( long ) offset , whence ) ;
# endif
GGML_ASSERT ( ret = = 0 ) ; // same
}
void read_raw ( void * ptr , size_t size ) {
if ( size = = 0 ) {
return ;
}
errno = 0 ;
std : : size_t ret = std : : fread ( ptr , size , 1 , fp ) ;
if ( ferror ( fp ) ) {
2023-09-07 17:22:29 +00:00
die_fmt ( " fread failed: %s " , strerror ( errno ) ) ;
2023-08-11 23:17:25 +00:00
}
if ( ret ! = 1 ) {
2023-09-07 17:22:29 +00:00
die ( " unexpectedly reached end of file " ) ;
2023-08-11 23:17:25 +00:00
}
}
std : : uint32_t read_u32 ( ) {
std : : uint32_t ret ;
read_raw ( & ret , sizeof ( ret ) ) ;
return ret ;
}
std : : float_t read_f32 ( ) {
std : : float_t ret ;
read_raw ( & ret , sizeof ( ret ) ) ;
return ret ;
}
std : : string read_string ( std : : uint32_t len ) {
std : : vector < char > chars ( len ) ;
read_raw ( chars . data ( ) , len ) ;
return std : : string ( chars . data ( ) , len ) ;
}
~ llama_file ( ) {
if ( fp ) {
std : : fclose ( fp ) ;
}
}
} ;
2023-09-15 19:38:27 +00:00
static bool is_ggml_file ( const char * filename ) {
2023-08-11 23:17:25 +00:00
llama_file file ( filename , " rb " ) ;
if ( file . size < 4 ) {
return false ;
}
2023-10-20 11:19:40 +00:00
std : : string magic = file . read_string ( 4 ) ;
2023-08-21 20:07:43 +00:00
return magic = = GGUF_MAGIC ;
2023-08-11 23:17:25 +00:00
}
2023-09-15 19:38:27 +00:00
static std : : string llama_escape_whitespaces ( const std : : string & text ) {
2023-08-27 14:13:31 +00:00
std : : ostringstream out ;
for ( char c : text ) {
if ( c = = ' ' ) out < < " \xe2 \x96 \x81 " ;
else out < < c ;
}
return out . str ( ) ;
}
2024-03-22 18:49:06 +00:00
static void load_vocab ( const char * filename , const Config * config , struct llama_vocab * vocab ) {
2023-08-27 14:13:31 +00:00
if ( is_ggml_file ( filename ) ) {
2024-03-22 18:49:06 +00:00
LOG ( " %s: Loading vocabulary from gguf file %s \n " , __func__ , filename ) ;
2023-08-27 14:13:31 +00:00
struct ggml_context * ctx_data = NULL ;
struct gguf_init_params params = {
/*.no_alloc = */ false ,
/*.ctx = */ & ctx_data ,
} ;
struct gguf_context * ctx = gguf_init_from_file ( filename , params ) ;
GGML_ASSERT ( ctx ! = NULL ) ;
const int model_idx = gguf_find_key ( ctx , KV_TOKENIZER_MODEL ) ;
GGML_ASSERT ( model_idx > = 0 ) ;
std : : string tokenizer_name = gguf_get_val_str ( ctx , model_idx ) ;
GGML_ASSERT ( tokenizer_name = = TOKENIZER_NAME ) ;
const int token_idx = gguf_find_key ( ctx , KV_TOKENIZER_LIST ) ;
GGML_ASSERT ( token_idx > = 0 ) ;
const int score_idx = gguf_find_key ( ctx , KV_TOKENIZER_SCORES ) ;
GGML_ASSERT ( score_idx > = 0 ) ;
const float * scores = ( const float * ) gguf_get_arr_data ( ctx , score_idx ) ;
const int toktype_idx = gguf_find_key ( ctx , KV_TOKENIZER_TOKEN_TYPE ) ;
GGML_ASSERT ( toktype_idx > = 0 ) ;
const int * toktypes = ( const int * ) gguf_get_arr_data ( ctx , toktype_idx ) ;
const uint32_t n_vocab = gguf_get_arr_n ( ctx , token_idx ) ;
2024-03-22 18:49:06 +00:00
if ( n_vocab ! = static_cast < uint32_t > ( config - > vocab_size ) ) {
die_fmt ( " vocab size mismatch: (gguf) %u != (llama2c) %d " , n_vocab , config - > vocab_size ) ;
}
2023-08-27 14:13:31 +00:00
vocab - > id_to_token . resize ( n_vocab ) ;
for ( uint32_t i = 0 ; i < n_vocab ; i + + ) {
std : : string word = gguf_get_arr_str ( ctx , token_idx , i ) ;
vocab - > token_to_id [ word ] = i ;
auto & token_data = vocab - > id_to_token [ i ] ;
token_data . text = std : : move ( word ) ;
token_data . score = scores [ i ] ;
token_data . type = ( llama_token_type ) toktypes [ i ] ;
}
ggml_free ( ctx_data ) ;
gguf_free ( ctx ) ;
} else {
// assume llama2.c vocabulary
2024-03-22 18:49:06 +00:00
LOG ( " %s: Assuming llama2.c vocabulary since %s is not a gguf file \n " , __func__ , filename ) ;
2023-08-11 23:17:25 +00:00
llama_file file ( filename , " rb " ) ;
2023-09-01 09:03:49 +00:00
if ( ! file . fp ) {
2023-09-07 17:22:29 +00:00
die_fmt ( " %s: %s " , strerror ( errno ) , filename ) ;
2023-09-01 09:03:49 +00:00
}
2023-08-21 20:07:43 +00:00
const int n_vocab = config - > vocab_size ;
2023-08-11 23:17:25 +00:00
/* uint32_t max_token_length = */ file . read_u32 ( ) ; // unused
vocab - > id_to_token . resize ( n_vocab ) ;
2023-08-27 14:13:31 +00:00
for ( llama_vocab : : id id = 0 ; id < n_vocab ; + + id ) {
2023-08-11 23:17:25 +00:00
float_t score = file . read_f32 ( ) ;
uint32_t len = file . read_u32 ( ) ;
2023-08-21 20:07:43 +00:00
std : : string text = file . read_string ( len ) ;
2023-08-27 14:13:31 +00:00
unsigned char byte_val ;
llama_vocab : : ttype type = LLAMA_TOKEN_TYPE_NORMAL ;
if ( id = = UNKNOWN_TOKEN_ID ) {
text = " <unk> " ;
type = LLAMA_TOKEN_TYPE_UNKNOWN ;
} else if ( id = = BOS_TOKEN_ID ) {
text = " <s> " ;
type = LLAMA_TOKEN_TYPE_CONTROL ;
} else if ( id = = EOS_TOKEN_ID ) {
text = " </s> " ;
type = LLAMA_TOKEN_TYPE_CONTROL ;
} else if ( text . empty ( ) ) {
type = LLAMA_TOKEN_TYPE_CONTROL ;
} else if ( sscanf ( text . c_str ( ) , " <0x%02hhX> " , & byte_val ) = = 1 ) {
// Text of byte tokens is already in the expected format.
type = LLAMA_TOKEN_TYPE_BYTE ;
} else {
type = LLAMA_TOKEN_TYPE_NORMAL ;
2023-08-23 19:33:05 +00:00
}
2023-08-27 14:13:31 +00:00
text = llama_escape_whitespaces ( text ) ;
vocab - > id_to_token [ id ] . text = text ;
vocab - > id_to_token [ id ] . score = score ;
vocab - > id_to_token [ id ] . type = type ;
vocab - > token_to_id . emplace ( text , id ) ;
2023-08-11 23:17:25 +00:00
}
}
}
2023-09-15 19:38:27 +00:00
static void convert_weights_ak_to_gg ( struct ggml_tensor * gg_weights , const float * karpathy_weights ) {
2024-03-22 18:49:06 +00:00
int size = 1 ;
for ( int dim = 0 ; dim < ggml_n_dims ( gg_weights ) ; + + dim ) {
size * = gg_weights - > ne [ dim ] ;
}
for ( int ct = 0 ; ct < size ; + + ct ) {
int64_t i0 = 0 ; int64_t i1 = 0 ;
int64_t i2 = 0 ; int64_t i3 = 0 ;
ggml_unravel_index ( gg_weights , ct , & i0 , & i1 , & i2 , & i3 ) ;
ggml_set_f32_nd ( gg_weights , i0 , i1 , i2 , i3 , karpathy_weights [ ct ] ) ;
2023-08-11 23:17:25 +00:00
}
}
2023-09-15 19:38:27 +00:00
static void save_as_llama_model (
struct llama_vocab * vocab , struct my_llama_model * model , TransformerWeights * w , const char * filename
) {
2023-09-01 14:00:40 +00:00
// convert AK weights into GG weights one by one.
2023-08-23 19:33:05 +00:00
// w->token_embedding_table -> model->tok_embeddings
// float* -> struct ggml_tensor
2024-03-22 18:49:06 +00:00
convert_weights_ak_to_gg ( model - > tok_embeddings , w - > token_embedding_table . data ( ) ) ;
convert_weights_ak_to_gg ( model - > output , ! w - > wcls . empty ( ) ? w - > wcls . data ( ) : w - > token_embedding_table . data ( ) ) ;
2023-08-23 19:33:05 +00:00
2024-03-22 18:49:06 +00:00
convert_weights_ak_to_gg ( model - > norm , w - > rms_final_weight . data ( ) ) ;
2023-08-23 19:33:05 +00:00
//print_row(model->norm, 0);
// for rms-att-weight
int row_length = model - > hparams . n_embd ;
2023-08-27 14:13:31 +00:00
int n_ff = model - > hparams . n_ff ;
2023-08-23 19:33:05 +00:00
2024-03-22 18:49:06 +00:00
const uint32_t n_multiqueries = model - > hparams . n_head_kv < = 0 | | model - > hparams . n_head_kv > = model - > hparams . n_head ? 1 : model - > hparams . n_head / model - > hparams . n_head_kv ;
2023-08-23 19:33:05 +00:00
for ( uint32_t i = 0 ; i < model - > hparams . n_layer ; + + i ) {
auto & layer = model - > layers [ i ] ;
// 1d
2023-09-01 14:00:40 +00:00
convert_weights_ak_to_gg ( layer . attention_norm , & w - > rms_att_weight [ i * row_length ] ) ;
convert_weights_ak_to_gg ( layer . ffn_norm , & w - > rms_ffn_weight [ i * row_length ] ) ;
2023-08-23 19:33:05 +00:00
// from 3d matrix layer x dim x dim to 2d matrix dim x dim
2023-09-01 14:00:40 +00:00
convert_weights_ak_to_gg ( layer . wq , & w - > wq [ i * row_length * row_length ] ) ;
convert_weights_ak_to_gg ( layer . wo , & w - > wo [ i * row_length * row_length ] ) ;
2024-03-22 18:49:06 +00:00
// from 3d matrix layer x dim x dim to 2d matrix dim x dim / n_multiqueries
convert_weights_ak_to_gg ( layer . wk , & w - > wk [ i * row_length * row_length / n_multiqueries ] ) ;
convert_weights_ak_to_gg ( layer . wv , & w - > wv [ i * row_length * row_length / n_multiqueries ] ) ;
2023-09-01 14:00:40 +00:00
convert_weights_ak_to_gg ( layer . w1 , & w - > w1 [ i * row_length * n_ff ] ) ;
convert_weights_ak_to_gg ( layer . w2 , & w - > w2 [ i * n_ff * row_length ] ) ;
convert_weights_ak_to_gg ( layer . w3 , & w - > w3 [ i * row_length * n_ff ] ) ;
2023-08-23 19:33:05 +00:00
}
2023-08-27 14:13:31 +00:00
struct gguf_context * ctx = gguf_init_empty ( ) ;
std : : vector < const char * > tokens ;
std : : vector < float > scores ;
std : : vector < llama_token_type > token_types ;
for ( const llama_vocab : : token_data & token_data : vocab - > id_to_token ) {
tokens . push_back ( token_data . text . c_str ( ) ) ;
scores . push_back ( token_data . score ) ;
token_types . push_back ( token_data . type ) ;
}
gguf_set_arr_str ( ctx , KV_TOKENIZER_LIST , tokens . data ( ) , tokens . size ( ) ) ;
gguf_set_arr_data ( ctx , KV_TOKENIZER_SCORES , GGUF_TYPE_FLOAT32 , scores . data ( ) , scores . size ( ) ) ;
gguf_set_arr_data ( ctx , KV_TOKENIZER_TOKEN_TYPE , GGUF_TYPE_INT32 , token_types . data ( ) , token_types . size ( ) ) ;
gguf_set_val_str ( ctx , KV_TOKENIZER_MODEL , TOKENIZER_NAME ) ;
gguf_set_val_str ( ctx , KV_GENERAL_ARCHITECTURE , " llama " ) ;
gguf_set_val_str ( ctx , KV_GENERAL_NAME , " llama " ) ;
// special tokens
gguf_set_val_u32 ( ctx , KV_TOKENIZER_UNK_ID , UNKNOWN_TOKEN_ID ) ;
gguf_set_val_u32 ( ctx , KV_TOKENIZER_BOS_ID , BOS_TOKEN_ID ) ;
gguf_set_val_u32 ( ctx , KV_TOKENIZER_EOS_ID , EOS_TOKEN_ID ) ;
gguf_set_val_u32 ( ctx , KV_TOKENIZER_SEP_ID , - 1 ) ;
gguf_set_val_u32 ( ctx , KV_TOKENIZER_PAD_ID , - 1 ) ;
gguf_set_val_u32 ( ctx , KV_CONTEXT_LENGTH , model - > hparams . n_ctx ) ;
gguf_set_val_u32 ( ctx , KV_EMBEDDING_LENGTH , model - > hparams . n_embd ) ;
gguf_set_val_u32 ( ctx , KV_FEED_FORWARD_LENGTH , model - > hparams . n_ff ) ;
gguf_set_val_u32 ( ctx , KV_ATTENTION_HEAD_COUNT , model - > hparams . n_head ) ;
2024-03-22 18:49:06 +00:00
gguf_set_val_u32 ( ctx , KV_ATTENTION_HEAD_COUNT , model - > hparams . n_head ) ;
gguf_set_val_u32 ( ctx , KV_ATTENTION_HEAD_COUNT_KV , model - > hparams . n_head_kv ) ;
2023-08-27 14:13:31 +00:00
gguf_set_val_u32 ( ctx , KV_BLOCK_COUNT , model - > hparams . n_layer ) ;
gguf_set_val_u32 ( ctx , KV_ROPE_DIMENSION_COUNT , model - > hparams . n_rot ) ;
gguf_set_val_f32 ( ctx , KV_ATTENTION_LAYERNORM_RMS_EPS , 1e-5 f ) ;
2023-08-23 19:33:05 +00:00
// write tensors
2023-08-27 14:13:31 +00:00
ggml_set_name ( model - > tok_embeddings , TN_TOKEN_EMBD ) ;
gguf_add_tensor ( ctx , model - > tok_embeddings ) ;
ggml_set_name ( model - > norm , TN_OUTPUT_NORM ) ;
gguf_add_tensor ( ctx , model - > norm ) ;
ggml_set_name ( model - > output , TN_OUTPUT ) ;
gguf_add_tensor ( ctx , model - > output ) ;
2023-08-23 19:33:05 +00:00
for ( uint32_t i = 0 ; i < model - > hparams . n_layer ; + + i ) {
auto & layer = model - > layers [ i ] ;
2023-08-27 14:13:31 +00:00
ggml_format_name ( layer . wq , TN_ATTN_Q , i ) ;
gguf_add_tensor ( ctx , layer . wq ) ;
ggml_format_name ( layer . wk , TN_ATTN_K , i ) ;
gguf_add_tensor ( ctx , layer . wk ) ;
ggml_format_name ( layer . wv , TN_ATTN_V , i ) ;
gguf_add_tensor ( ctx , layer . wv ) ;
ggml_format_name ( layer . wo , TN_ATTN_OUTPUT , i ) ;
gguf_add_tensor ( ctx , layer . wo ) ;
ggml_format_name ( layer . attention_norm , TN_ATTN_NORM , i ) ;
gguf_add_tensor ( ctx , layer . attention_norm ) ;
ggml_format_name ( layer . w1 , TN_FFN_GATE , i ) ;
gguf_add_tensor ( ctx , layer . w1 ) ;
ggml_format_name ( layer . w2 , TN_FFN_DOWN , i ) ;
gguf_add_tensor ( ctx , layer . w2 ) ;
ggml_format_name ( layer . w3 , TN_FFN_UP , i ) ;
gguf_add_tensor ( ctx , layer . w3 ) ;
ggml_format_name ( layer . ffn_norm , TN_FFN_NORM , i ) ;
gguf_add_tensor ( ctx , layer . ffn_norm ) ;
2023-08-23 19:33:05 +00:00
}
2023-08-27 14:13:31 +00:00
gguf_write_to_file ( ctx , filename , false ) ;
gguf_free ( ctx ) ;
2023-08-11 23:17:25 +00:00
}
2023-09-15 19:38:27 +00:00
static struct train_params get_default_train_params ( ) {
2023-08-11 23:17:25 +00:00
struct train_params params ;
2024-03-22 18:49:06 +00:00
params . fn_vocab_model = " models/7B/ggml-model-f16.gguf " ;
2023-08-11 23:17:25 +00:00
params . fn_llama2c_output_model = " ak_llama_model.bin " ;
2024-03-22 18:49:06 +00:00
params . fn_train_data = " shakespeare.txt " ;
params . fn_checkpoint_in = " checkpoint.bin " ;
params . fn_checkpoint_out = " checkpoint.bin " ;
params . fn_model_out = " ggml-checkpoint-f32.bin " ;
2023-08-11 23:17:25 +00:00
params . seed = - 1 ;
params . n_ctx = 128 ;
params . n_embd = 256 ;
params . n_mult = 256 ;
params . n_head = 8 ;
params . n_layer = 16 ;
params . n_rotmax = 64 ;
params . n_threads = 6 ;
params . n_batch = 8 ;
params . n_examples = 8 ;
params . n_predict = 1024 ;
params . print_info_interval = 1 ;
params . print_details_interval = 2 ;
params . samples_start_after_nl = false ;
params . use_adam = true ;
params . use_flash = true ;
params . use_scratch = true ;
// only adam
params . warmup = 100 ;
params . cos_decay_steps = 1000 ;
params . cos_decay_restart = 1.1f ;
params . cos_decay_alpha = 0.0f ;
params . lbfgs_n_iter = 16 ;
params . adam_n_iter = 16 ;
params . adam_alpha = 1e-3 f ;
params . adam_decay = 1e-3 f ;
2024-03-22 18:49:06 +00:00
params . mem_model_gb = 2 ;
params . mem_compute_gb = 24 ;
2023-08-11 23:17:25 +00:00
params . mem_compute0_gb = 8 ;
params . mem_compute1_gb = 2 ;
return params ;
}
2023-09-15 19:38:27 +00:00
static void print_usage ( int /*argc*/ , char * * argv , const struct train_params * params ) {
2023-08-11 23:17:25 +00:00
fprintf ( stderr , " usage: %s [options] \n " , argv [ 0 ] ) ;
fprintf ( stderr , " \n " ) ;
fprintf ( stderr , " options: \n " ) ;
fprintf ( stderr , " -h, --help show this help message and exit \n " ) ;
2023-08-27 14:13:31 +00:00
fprintf ( stderr , " --copy-vocab-from-model FNAME path of gguf llama model or llama2.c vocabulary from which to copy vocab (default '%s') \n " , params - > fn_vocab_model ) ;
2023-08-11 23:17:25 +00:00
fprintf ( stderr , " --llama2c-model FNAME [REQUIRED] model path from which to load Karpathy's llama2.c model \n " ) ;
fprintf ( stderr , " --llama2c-output-model FNAME model path to save the converted llama2.c model (default %s') \n " , params - > fn_llama2c_output_model ) ;
fprintf ( stderr , " \n " ) ;
}
2023-09-15 19:38:27 +00:00
static bool params_parse ( int argc , char * * argv , struct train_params * params ) {
2023-08-11 23:17:25 +00:00
bool invalid_param = false ;
bool reqd_param_found = false ;
std : : string arg ;
struct train_params default_params = get_default_train_params ( ) ;
const std : : string arg_prefix = " -- " ;
for ( int i = 1 ; i < argc ; i + + ) {
arg = argv [ i ] ;
if ( arg . compare ( 0 , arg_prefix . size ( ) , arg_prefix ) = = 0 ) {
std : : replace ( arg . begin ( ) , arg . end ( ) , ' _ ' , ' - ' ) ;
}
if ( arg = = " --copy-vocab-from-model " ) {
if ( + + i > = argc ) {
invalid_param = true ;
break ;
}
params - > fn_vocab_model = argv [ i ] ;
} else if ( arg = = " --llama2c-model " ) {
if ( + + i > = argc ) {
invalid_param = true ;
break ;
}
reqd_param_found = true ;
params - > fn_llama2c_model = argv [ i ] ;
} else if ( arg = = " --llama2c-output-model " ) {
if ( + + i > = argc ) {
invalid_param = true ;
break ;
}
params - > fn_llama2c_output_model = argv [ i ] ;
} else if ( arg = = " -h " | | arg = = " --help " ) {
print_usage ( argc , argv , & default_params ) ;
exit ( 0 ) ;
} else {
fprintf ( stderr , " error: unknown argument: %s \n " , arg . c_str ( ) ) ;
print_usage ( argc , argv , & default_params ) ;
exit ( 1 ) ;
}
}
if ( invalid_param ) {
fprintf ( stderr , " error: invalid parameter for argument: %s \n " , arg . c_str ( ) ) ;
print_usage ( argc , argv , & default_params ) ;
exit ( 1 ) ;
}
if ( ! reqd_param_found ) {
fprintf ( stderr , " error: please specify a llama2.c .bin file to be converted with argument --llama2c-model \n " ) ;
print_usage ( argc , argv , & default_params ) ;
exit ( 1 ) ;
}
return true ;
}
2023-09-15 19:38:27 +00:00
static std : : string basename ( const std : : string & path ) {
2023-09-01 09:03:49 +00:00
size_t pos = path . find_last_of ( " / \\ " ) ;
2023-08-27 14:13:31 +00:00
if ( pos = = std : : string : : npos ) {
return path ;
}
return path . substr ( pos + 1 ) ;
}
2023-08-11 23:17:25 +00:00
int main ( int argc , char * * argv ) {
struct train_params params = get_default_train_params ( ) ;
if ( ! params_parse ( argc , argv , & params ) ) {
return 1 ;
}
2024-03-22 18:49:06 +00:00
log_set_target ( stdout ) ;
2023-08-11 23:17:25 +00:00
Config config ;
2023-09-01 09:03:49 +00:00
TransformerWeights weights = { } ;
2023-08-11 23:17:25 +00:00
{
2024-03-22 18:49:06 +00:00
LOG ( " %s: Loading llama2c model from %s \n " , __func__ , params . fn_llama2c_model ) ;
2024-03-27 07:16:02 +00:00
FILE * file = fopen ( params . fn_llama2c_model , " rb " ) ;
2024-03-22 18:49:06 +00:00
if ( ! file ) {
LOG ( " %s: Unable to open the checkpoint file %s! \n " , __func__ , params . fn_llama2c_model ) ;
return 1 ;
}
2023-08-11 23:17:25 +00:00
// read in the config header
2024-03-22 18:49:06 +00:00
if ( fread ( & config , sizeof ( Config ) , 1 , file ) ! = 1 ) {
LOG ( " %s: Unable to read llama2c config from %s! \n " , __func__ , params . fn_llama2c_model ) ;
return 1 ;
}
2023-08-23 19:33:05 +00:00
auto shared_weights = config . vocab_size > 0 ;
config . vocab_size = abs ( config . vocab_size ) ;
2023-08-11 23:17:25 +00:00
// read in the Transformer weights
2024-03-22 18:49:06 +00:00
alloc_weights ( & weights , & config , shared_weights ) ;
if ( checkpoint_init_weights ( & weights , & config , file , shared_weights ) ) {
LOG ( " %s: Unable to initialize transformer weights from %s! " , __func__ , params . fn_llama2c_model ) ;
return 1 ;
}
2023-08-11 23:17:25 +00:00
fclose ( file ) ;
}
struct llama_vocab vocab ;
load_vocab ( params . fn_vocab_model , & config , & vocab ) ;
struct my_llama_model model ;
2024-03-22 18:49:06 +00:00
model . hparams . n_vocab = config . vocab_size ; //llama_n_vocab(lctx);
model . hparams . n_ctx = params . n_ctx ;
model . hparams . n_embd = config . dim ; //params.n_embd;
model . hparams . n_ff = config . hidden_dim ;
model . hparams . n_mult = 32 ; //params.n_mult;
model . hparams . n_head = config . n_heads ; //params.n_head;
model . hparams . n_head_kv = config . n_kv_heads ;
model . hparams . n_layer = config . n_layers ; //params.n_layer;
model . hparams . n_rot = std : : min ( ( uint32_t ) params . n_rotmax , model . hparams . n_embd / model . hparams . n_head ) ;
2023-08-11 23:17:25 +00:00
print_params ( & model . hparams ) ;
2024-03-22 18:49:06 +00:00
2023-08-11 23:17:25 +00:00
struct ggml_init_params lcparams ;
lcparams . mem_size = 1024ll * 1024ll * 1024ll * ( ( size_t ) params . mem_model_gb ) ;
lcparams . mem_buffer = NULL ;
lcparams . no_alloc = false ;
model . ctx = ggml_init ( lcparams ) ;
init_model ( & model ) ;
2023-08-27 14:13:31 +00:00
model . name = basename ( params . fn_llama2c_model ) ;
2023-08-11 23:17:25 +00:00
save_as_llama_model ( & vocab , & model , & weights , params . fn_llama2c_output_model ) ;
2024-03-22 18:49:06 +00:00
LOG ( " %s: Saving llama.c model file %s in ggml format at %s \n " , __func__ , params . fn_llama2c_model , params . fn_llama2c_output_model ) ;
2023-08-11 23:17:25 +00:00
ggml_free ( model . ctx ) ;
return 0 ;
}