Archives
/
gpt4all
mirror of https://github.com/nomic-ai/gpt4all
2
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

fix: helly changes

Browse Source
pull/520/head
Zach Nussbaum 1 year ago committed by Adam Treat
parent d30be81506
commit 61e2aabadb
1 changed files with 162 additions and 250 deletions
Show Stats Download Patch File Download Diff File

412
llmodel/mpt.cpp
Unescape Escape View File

@ -19,37 +19,31 @@
static const size_t MB = 1024*1024;
// default hparams (MPT 7B)
struct mpt_hparams {
int32_t n_vocab = 50432;
int32_t n_ctx = 2048;
int32_t n_embd = 4096;
int32_t n_head = 32;
int32_t n_layer = 32;
// this isn't used should we remove?
int32_t n_rot = 0;
int32_t f16 = 1;
int32_t n_vocab = 50432;
int32_t n_ctx = 2048;
int32_t n_embd = 4096;
int32_t n_head = 32;
int32_t n_layer = 32;
float alibi_bias_max = 8;
float clip_qkv = 0;
int32_t expand = 4;
int32_t f16 = 1;
};
struct mpt_layer {
// normalization
struct ggml_tensor * norm_1_g;
struct ggml_tensor * norm_1_b;
struct ggml_tensor * norm_2_g;
struct ggml_tensor * norm_2_b;
struct ggml_tensor * norm_1_w;
struct ggml_tensor * norm_2_w;
// attention
struct ggml_tensor * c_attn_q_proj_w;
struct ggml_tensor * c_attn_k_proj_w;
struct ggml_tensor * c_attn_v_proj_w;
struct ggml_tensor * c_attn_proj_w;
struct ggml_tensor * attn_Wqkv_w;
struct ggml_tensor * attn_out_proj_w;
// ff
struct ggml_tensor * up_proj_w;
struct ggml_tensor * up_proj_b;
struct ggml_tensor * down_proj_w;
struct ggml_tensor * down_proj_b;
struct ggml_tensor * ffn_up_proj_w;
struct ggml_tensor * ffn_down_proj_w;
};
struct mpt_buffer {
@ -90,8 +84,7 @@ struct mpt_model {
mpt_hparams hparams;
// normalization
struct ggml_tensor * ln_f_g;
struct ggml_tensor * ln_f_b;
struct ggml_tensor * norm_f_w;
struct ggml_tensor * wte; // position embedding
@ -159,10 +152,11 @@ struct mpt_vocab {
bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & model, mpt_vocab & vocab) {
printf("%s: loading model from '%s' - please wait ...\n", __func__, fname.c_str());
// verify magic
{
uint32_t magic;
fin.read((char *) &magic, sizeof(magic));
if (magic != 0x67676d6d) {
if (magic != 0x67676d6d) {
fprintf(stderr, "%s: invalid model file '%s' (bad magic)\n", __func__, fname.c_str());
return false;
}
@ -174,23 +168,26 @@ bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & mod
fin.read((char *) &hparams.n_vocab, sizeof(hparams.n_vocab));
fin.read((char *) &hparams.n_ctx, sizeof(hparams.n_ctx));
fin.read((char *) &hparams.n_embd, sizeof(hparams.n_embd));
fin.read((char *) &hparams.n_head, sizeof(hparams.n_head));
fin.read((char *) &hparams.n_layer, sizeof(hparams.n_layer));
fin.read((char *) &hparams.n_rot, sizeof(hparams.n_rot));
fin.read((char *) &hparams.f16, sizeof(hparams.f16));
printf("%s: n_vocab = %d\n", __func__, hparams.n_vocab);
printf("%s: n_ctx = %d\n", __func__, hparams.n_ctx);
printf("%s: n_embd = %d\n", __func__, hparams.n_embd);
printf("%s: n_head = %d\n", __func__, hparams.n_head);
printf("%s: n_layer = %d\n", __func__, hparams.n_layer);
printf("%s: n_rot = %d\n", __func__, hparams.n_rot);
printf("%s: f16 = %d\n", __func__, hparams.f16);
fin.read((char *) &hparams.n_head, sizeof(hparams.n_head));
fin.read((char *) &hparams.n_embd, sizeof(hparams.n_embd));
fin.read((char *) &hparams.alibi_bias_max, sizeof(hparams.alibi_bias_max));
fin.read((char *) &hparams.clip_qkv, sizeof(hparams.clip_qkv));
fin.read((char *) &hparams.f16, sizeof(hparams.f16));
printf("%s: n_vocab = %d\n", __func__, hparams.n_vocab);
printf("%s: n_ctx = %d\n", __func__, hparams.n_ctx);
printf("%s: n_embd = %d\n", __func__, hparams.n_embd);
printf("%s: n_head = %d\n", __func__, hparams.n_head);
printf("%s: n_layer = %d\n", __func__, hparams.n_layer);
printf("%s: alibi_bias_max = %f\n", __func__, hparams.alibi_bias_max);
printf("%s: clip_qkv = %f\n", __func__, hparams.clip_qkv);
printf("%s: ftype = %d\n", __func__, hparams.f16);
}
// load vocab
{
int32_t n_vocab = 0;
int32_t n_vocab = model.hparams.n_vocab;
fin.read((char *) &n_vocab, sizeof(n_vocab));
if (n_vocab != model.hparams.n_vocab) {
@ -203,14 +200,24 @@ bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & mod
for (int i = 0; i < n_vocab; i++) {
uint32_t len;
fin.read((char *) &len, sizeof(len));
bool special = false;
if (len & (1<<31)) {
len = len &~ (1<<31);
special = true;
}
word.resize(len);
fin.read((char *) word.data(), len);
if (len > 0) {
word.resize(len);
fin.read((char *) word.data(), len);
vocab.token_to_id[word] = i;
vocab.id_to_token[i] = word;
}
vocab.token_to_id[word] = i;
vocab.id_to_token[i] = word;
// TODO: this only kind-of works, the gpt_tokenize can still incorrectly
// tokenize special tokens
}
}
// for the big tensors, we have the option to store the data in 16-bit floats or quantized
// in order to save memory and also to speed up the computation
ggml_type wtype = GGML_TYPE_COUNT;
@ -227,8 +234,6 @@ bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & mod
}
}
const ggml_type wtype2 = GGML_TYPE_F32;
auto & ctx = model.ctx;
size_t ctx_size = 0;
@ -240,38 +245,26 @@ bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & mod
const int n_layer = hparams.n_layer;
const int n_ctx = hparams.n_ctx;
const int n_vocab = hparams.n_vocab;
const int expand = hparams.expand;
ctx_size += n_embd*ggml_type_sizef(GGML_TYPE_F32); // ln_f_g
ctx_size += n_embd*ggml_type_sizef(GGML_TYPE_F32); // ln_f_b
ctx_size += n_embd*n_vocab*ggml_type_sizef(wtype); // wte
// with weight tying i don't think we need this
// ctx_size += n_embd*n_vocab*ggml_type_sizef(wtype); // lmh_g
// ctx_size += n_vocab*ggml_type_sizef(GGML_TYPE_F32); // lmh_b
ctx_size += n_layer*(n_embd*ggml_type_sizef(GGML_TYPE_F32)); // ln_1_g
ctx_size += n_layer*(n_embd*ggml_type_sizef(GGML_TYPE_F32)); // ln_1_b
ctx_size += n_layer*(n_embd*n_embd*ggml_type_sizef(wtype)); // c_attn_q_proj_w
ctx_size += n_layer*(n_embd*n_embd*ggml_type_sizef(wtype)); // c_attn_k_proj_w
ctx_size += n_layer*(n_embd*n_embd*ggml_type_sizef(wtype)); // c_attn_v_proj_w
ctx_size += n_embd*ggml_type_sizef(GGML_TYPE_F32); // ln_f_w
ctx_size += n_layer*(n_embd*n_embd*ggml_type_sizef(wtype)); // c_attn_proj_w
ctx_size += n_embd*n_vocab*ggml_type_sizef(GGML_TYPE_F32); // wte
ctx_size += n_layer*(4*n_embd*n_embd*ggml_type_sizef(wtype)); // c_mlp_fc_w
ctx_size += n_layer*( 4*n_embd*ggml_type_sizef(GGML_TYPE_F32)); // c_mlp_fc_b
ctx_size += n_layer*(n_embd*ggml_type_sizef(GGML_TYPE_F32)); // norm_1_w
ctx_size += n_layer*(n_embd*ggml_type_sizef(GGML_TYPE_F32)); // norm_2_w
ctx_size += n_layer*(4*n_embd*n_embd*ggml_type_sizef(wtype)); // c_mlp_proj_w
ctx_size += n_layer*( n_embd*ggml_type_sizef(GGML_TYPE_F32)); // c_mlp_proj_b
ctx_size += n_layer*(3*n_embd*n_embd*ggml_type_sizef(wtype)); // attn_Wqkv_w
ctx_size += n_layer*(n_embd*n_embd*ggml_type_sizef(wtype)); // attn_out_proj_w
ctx_size += n_ctx*n_layer*n_embd*ggml_type_sizef(GGML_TYPE_F32); // memory_k
ctx_size += n_ctx*n_layer*n_embd*ggml_type_sizef(GGML_TYPE_F32); // memory_v
ctx_size += n_layer*(expand*n_embd*n_embd*ggml_type_sizef(wtype)); // ffn_up_proj_w
ctx_size += n_layer*(expand*n_embd*n_embd*ggml_type_sizef(wtype)); // ffn_down_proj_w
ctx_size += n_ctx*n_layer*n_embd*ggml_type_sizef(GGML_TYPE_F16); // memory_k
ctx_size += n_ctx*n_layer*n_embd*ggml_type_sizef(GGML_TYPE_F16); // memory_v
ctx_size += n_layer*(n_embd*ggml_type_sizef(GGML_TYPE_F32)); // ln_2_g
ctx_size += n_layer*(n_embd*ggml_type_sizef(GGML_TYPE_F32)); // ln_2_b
// TODO: what is this??
// TODO probably less now?
ctx_size += (5 + 10*n_layer)*256; // object overhead
printf("%s: ggml ctx size = %6.2f MB\n", __func__, ctx_size/(1024.0*1024.0));
@ -282,6 +275,7 @@ bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & mod
struct ggml_init_params params = {
.mem_size = ctx_size,
.mem_buffer = NULL,
.no_alloc = false,
};
model.ctx = ggml_init(params);
@ -290,7 +284,8 @@ bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & mod
return false;
}
}
// prepare memory for the weights
// prepare memory for the weights
{
const auto & hparams = model.hparams;
@ -298,66 +293,40 @@ bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & mod
const int n_layer = hparams.n_layer;
const int n_ctx = hparams.n_ctx;
const int n_vocab = hparams.n_vocab;
const int expand = hparams.expand;
model.layers.resize(n_layer);
model.wte = ggml_new_tensor_2d(ctx, wtype, n_embd, n_vocab);
model.ln_f_g = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, n_embd);
model.ln_f_b = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, n_embd);
// we don't need this because of weight tying
// model.lmh_g = ggml_new_tensor_2d(ctx, wtype, n_embd, n_vocab);
// model.lmh_b = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, n_vocab);
model.wte = ggml_new_tensor_2d(ctx, GGML_TYPE_F32, n_embd, n_vocab);
model.norm_f_w = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, n_embd);
// map by name
model.tensors["transformer.wte.weight"] = model.wte;
model.tensors["transformer.ln_f.weight"] = model.ln_f_g;
model.tensors["transformer.ln_f.bias"] = model.ln_f_b;
model.tensors["transformer.norm_f.weight"] = model.norm_f_w;
for (int i = 0; i < n_layer; ++i) {
auto & layer = model.layers[i];
layer.norm_1_g = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, n_embd);
layer.norm_1_b = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, n_embd);
layer.norm_2_g = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, n_embd);
layer.norm_2_b = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, n_embd);
layer.c_attn_q_proj_w = ggml_new_tensor_2d(ctx, wtype, n_embd, n_embd);
layer.c_attn_k_proj_w = ggml_new_tensor_2d(ctx, wtype, n_embd, n_embd);
layer.c_attn_v_proj_w = ggml_new_tensor_2d(ctx, wtype, n_embd, n_embd);
layer.c_attn_proj_w = ggml_new_tensor_2d(ctx, wtype, n_embd, n_embd);
layer.norm_1_w = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, n_embd);
layer.norm_2_w = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, n_embd);
layer.up_proj_w = ggml_new_tensor_2d(ctx, wtype, n_embd, 4*n_embd);
layer.up_proj_b = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, 4*n_embd);
layer.down_proj_w = ggml_new_tensor_2d(ctx, wtype, 4*n_embd, n_embd);
layer.down_proj_b = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, n_embd);
layer.attn_Wqkv_w = ggml_new_tensor_2d(ctx, wtype, n_embd, n_embd * 3);
layer.attn_out_proj_w = ggml_new_tensor_2d(ctx, wtype, n_embd, n_embd);
layer.ffn_up_proj_w = ggml_new_tensor_2d(ctx, wtype, n_embd, expand*n_embd);
layer.ffn_down_proj_w = ggml_new_tensor_2d(ctx, wtype, expand*n_embd, n_embd);
// map by name
model.tensors["transformer.blocks." + std::to_string(i) + ".norm_1.weight"] = layer.norm_1_g;
model.tensors["transformer.blocks." + std::to_string(i) + ".norm_1.bias"] = layer.norm_1_b;
model.tensors["transformer.blocks." + std::to_string(i) + ".attn.q_proj.weight"] = layer.c_attn_q_proj_w;
model.tensors["transformer.blocks." + std::to_string(i) + ".attn.k_proj.weight"] = layer.c_attn_k_proj_w;
model.tensors["transformer.blocks." + std::to_string(i) + ".attn.v_proj.weight"] = layer.c_attn_v_proj_w;
model.tensors["transformer.blocks." + std::to_string(i) + ".attn.out_proj.weight"] = layer.c_attn_proj_w;
model.tensors["transformer.blocks." + std::to_string(i) + ".ffn.up_proj.weight"] = layer.up_proj_w;
model.tensors["transformer.blocks." + std::to_string(i) + ".ffn.up_proj.bias"] = layer.up_proj_b;
model.tensors["transformer.blocks." + std::to_string(i) + ".norm_1.weight"] = layer.norm_1_w;
model.tensors["transformer.blocks." + std::to_string(i) + ".norm_2.weight"] = layer.norm_2_w;
model.tensors["transformer.blocks." + std::to_string(i) + ".attn.Wqkv.weight"] = layer.attn_Wqkv_w;
model.tensors["transformer.blocks." + std::to_string(i) + ".attn.out_proj.weight"] = layer.attn_out_proj_w;
model.tensors["transformer.blocks." + std::to_string(i) + ".ffn.down_proj.weight"] = layer.down_proj_w;
model.tensors["transformer.blocks." + std::to_string(i) + ".ffn.down_proj.bias"] = layer.down_proj_b;
model.tensors["transformer.blocks." + std::to_string(i) + ".norm_2.weight"] = layer.norm_2_g;
model.tensors["transformer.blocks." + std::to_string(i) + ".norm_2.bias"] = layer.norm_2_b;
model.tensors["transformer.blocks." + std::to_string(i) + ".ffn.up_proj.weight"] = layer.ffn_up_proj_w;
model.tensors["transformer.blocks." + std::to_string(i) + ".ffn.down_proj.weight"] = layer.ffn_down_proj_w;
}
}
// key + value memory
// key + value memory
{
const auto & hparams = model.hparams;
@ -377,7 +346,8 @@ bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & mod
const size_t memory_size = ggml_nbytes(model.kv_self.k) + ggml_nbytes(model.kv_self.v);
printf("%s: kv self size = %7.2f MB\n", __func__, memory_size / 1024.0 / 1024.0);
}
// load weights
// load weights
{
int n_tensors = 0;
size_t total_size = 0;
@ -387,13 +357,11 @@ bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & mod
while (true) {
int32_t n_dims;
int32_t length;
int32_t ftype;
int32_t ttype;
fin.read(reinterpret_cast<char *>(&n_dims), sizeof(n_dims));
fin.read(reinterpret_cast<char *>(&length), sizeof(length));
fin.read(reinterpret_cast<char *>(&ftype), sizeof(ftype));
fprintf(stderr, "%s: n_dims = %d, length = %d, ftype = %d\n", __func__, n_dims, length, ftype);
fin.read(reinterpret_cast<char *>(&ttype), sizeof(ttype));
if (fin.eof()) {
break;
@ -408,7 +376,7 @@ bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & mod
std::string name(length, 0);
fin.read(&name[0], length);
fprintf(stderr, "%s: %s\n", __func__, name.data());
if (model.tensors.find(name.data()) == model.tensors.end()) {
fprintf(stderr, "%s: unknown tensor '%s' in model file\n", __func__, name.data());
return false;
@ -421,29 +389,17 @@ bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & mod
}
if (tensor->ne[0] != ne[0] || tensor->ne[1] != ne[1]) {
fprintf(stderr, "%s: tensor '%s' has wrong shape in model file: got [%lu, %lu], expected [%d, %d]\n",
__func__, name.data(), tensor->ne[0], tensor->ne[1], ne[0], ne[1]);
fprintf(stderr, "%s: tensor '%s' has wrong shape in model file: got [%d, %d], expected [%d, %d]\n",
__func__, name.data(), (int) tensor->ne[0], (int) tensor->ne[1], ne[0], ne[1]);
return false;
}
// for debugging
if (0) {
static const char * ftype_str[] = { "f32", "f16", "q4_0", "q4_1", };
printf("%24s - [%5d, %5d], type = %6s, %6.2f MB, %9zu bytes\n", name.data(), ne[0], ne[1], ftype_str[ftype], ggml_nbytes(tensor)/1024.0/1024.0, ggml_nbytes(tensor));
printf("%24s - [%5d, %5d], type = %6s, %6.2f MB, %9zu bytes\n", name.data(), ne[0], ne[1], ggml_type_name(ggml_type(ttype)), ggml_nbytes(tensor)/1024.0/1024.0, ggml_nbytes(tensor));
}
size_t bpe = 0;
switch (ftype) {
case 0: bpe = ggml_type_size(GGML_TYPE_F32); break;
case 1: bpe = ggml_type_size(GGML_TYPE_F16); break;
case 2: bpe = ggml_type_size(GGML_TYPE_Q4_0); assert(ne[0] % 64 == 0); break;
case 3: bpe = ggml_type_size(GGML_TYPE_Q4_1); assert(ne[0] % 64 == 0); break;
default:
{
fprintf(stderr, "%s: unknown ftype %d in model file\n", __func__, ftype);
return false;
}
};
const size_t bpe = ggml_type_size(ggml_type(ttype));
if ((nelements*bpe)/ggml_blck_size(tensor->type) != ggml_nbytes(tensor)) {
fprintf(stderr, "%s: tensor '%s' has wrong size in model file: got %zu, expected %zu\n",
@ -453,8 +409,7 @@ bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & mod
fin.read(reinterpret_cast<char *>(tensor->data), ggml_nbytes(tensor));
fprintf(stderr, "%42s - [%5d, %5d], type = %6s, %6.2f MB\n", name.data(), ne[0], ne[1], ftype == 0 ? "float" : "f16", ggml_nbytes(tensor)/1024.0/1024.0);
//printf("%42s - [%5d, %5d], type = %6s, %6.2f MB\n", name.data(), ne[0], ne[1], ttype == 0 ? "float" : "f16", ggml_nbytes(tensor)/1024.0/1024.0);
total_size += ggml_nbytes(tensor);
if (++n_tensors % 8 == 0) {
printf(".");
@ -468,7 +423,6 @@ bool mpt_model_load(const std::string &fname, std::istream &fin, mpt_model & mod
}
return true;
}
}
@ -502,29 +456,30 @@ bool mpt_eval(
const int n_ctx = hparams.n_ctx;
const int n_head = hparams.n_head;
const int n_vocab = hparams.n_vocab;
const int n_rot = hparams.n_rot;
const int expand = hparams.expand;
const int d_key = n_embd/n_head;
static size_t buf_size = 1024u*MB;
if (!model.buf.addr || model.buf.size < buf_size)
model.buf.resize(buf_size);
static size_t buf_size = 256u*1024*1024;
static void * buf = malloc(buf_size);
if (mem_per_token > 0 && mem_per_token*N > model.buf.size) {
if (mem_per_token > 0 && mem_per_token*N > buf_size) {
const size_t buf_size_new = 1.1*(mem_per_token*N); // add 10% to account for ggml object overhead
printf("\n%s: reallocating buffer from %zu to %zu bytes\n", __func__, model.buf.size, buf_size_new);
//printf("\n%s: reallocating buffer from %zu to %zu bytes\n", __func__, buf_size, buf_size_new);
// reallocate
model.buf.resize(buf_size_new);
if (model.buf.addr == nullptr) {
fprintf(stderr, "%s: failed to allocate %zu bytes\n", __func__, model.buf.size);
buf_size = buf_size_new;
buf = realloc(buf, buf_size);
if (buf == nullptr) {
fprintf(stderr, "%s: failed to allocate %zu bytes\n", __func__, buf_size);
return false;
}
}
struct ggml_init_params params = {
.mem_size = model.buf.size,
.mem_buffer = model.buf.addr,
.mem_size = buf_size,
.mem_buffer = buf,
.no_alloc = false,
};
struct ggml_context * ctx0 = ggml_init(params);
@ -537,27 +492,26 @@ bool mpt_eval(
struct ggml_tensor * inpL = ggml_get_rows(ctx0, model.wte, embd);
for (int il = 0; il < n_layer; ++il) {
struct ggml_tensor * cur;
// norm
struct ggml_tensor * inpSA = inpL;
struct ggml_tensor * cur = inpSA;
// self-attention
{
cur = ggml_norm(ctx0, inpL);
// cur = ln_1_g*cur + ln_1_b
cur = ggml_add(ctx0,
ggml_mul(ctx0,
ggml_repeat(ctx0, model.layers[il].norm_1_g, cur),
cur),
ggml_repeat(ctx0, model.layers[il].norm_1_b, cur));
}
struct ggml_tensor * inpSA = cur;
// norm1
cur = ggml_norm(ctx0, cur);
cur = ggml_mul(ctx0,
ggml_repeat(ctx0, model.layers[il].norm_1_w, cur),
cur);
// compute QKV
cur = ggml_mul_mat(ctx0,
model.layers[il].attn_Wqkv_w,
cur);
// self-attention
{
struct ggml_tensor * Qcur = ggml_mul_mat(ctx0, model.layers[il].c_attn_q_proj_w, cur);
struct ggml_tensor * Kcur = ggml_mul_mat(ctx0, model.layers[il].c_attn_k_proj_w, cur);
struct ggml_tensor * Vcur = ggml_mul_mat(ctx0, model.layers[il].c_attn_v_proj_w, cur);
// TODO: clip_qkv
struct ggml_tensor * Qcur = ggml_cont(ctx0, ggml_view_2d(ctx0, cur, n_embd, N, cur->nb[1], 0*ggml_element_size(cur)*n_embd));
struct ggml_tensor * Kcur = ggml_cont(ctx0, ggml_view_2d(ctx0, cur, n_embd, N, cur->nb[1], 1*ggml_element_size(cur)*n_embd));
struct ggml_tensor * Vcur = ggml_cont(ctx0, ggml_view_2d(ctx0, cur, n_embd, N, cur->nb[1], 2*ggml_element_size(cur)*n_embd));
// store key and value to memory
if (N >= 1) {
@ -567,21 +521,17 @@ bool mpt_eval(
ggml_build_forward_expand(&gf, ggml_cpy(ctx0, Kcur, k));
ggml_build_forward_expand(&gf, ggml_cpy(ctx0, Vcur, v));
}
// Q = Qcur.contiguous().view(n_embd/n_head, n_head, N).permute(0, 2, 1, 3)
struct ggml_tensor * Q =
ggml_permute(ctx0,
ggml_cpy(ctx0,
Qcur,
ggml_new_tensor_3d(ctx0, GGML_TYPE_F32, n_embd/n_head, n_head, N)),
ggml_reshape_3d(ctx0, Qcur, n_embd/n_head, n_head, N),
0, 2, 1, 3);
// K = Kmem.view(n_embd/n_head, n_head, n_past + N).permute(0, 2, 1, 3)
struct ggml_tensor * K =
ggml_permute(ctx0,
ggml_reshape_3d(ctx0,
ggml_view_1d(ctx0, model.kv_self.k, (n_past + N)*n_embd, il*n_ctx*ggml_element_size(model.kv_self.k)*n_embd),
n_embd/n_head, n_head, n_past + N),
ggml_permute(ctx0,
ggml_reshape_3d(ctx0,
ggml_view_1d(ctx0, model.kv_self.k, (n_past + N)*n_embd, il*n_ctx*ggml_element_size(model.kv_self.k)*n_embd),
n_embd/n_head, n_head, n_past + N),
0, 2, 1, 3);
// K * Q
@ -594,26 +544,26 @@ bool mpt_eval(
ggml_new_f32(ctx0, 1.0f/sqrt(float(n_embd)/n_head))
);
struct ggml_tensor * KQ_scaled_alibi = ggml_alibi(ctx0, KQ_scaled, n_past, n_head);
// Alibi
struct ggml_tensor * KQ_scaled_biased = ggml_alibi(ctx0, ggml_cont(ctx0, KQ_scaled), n_past, n_head);
// KQ_masked = mask_past(KQ_scaled)
struct ggml_tensor * KQ_masked = ggml_diag_mask_inf(ctx0, KQ_scaled_alibi, n_past);
struct ggml_tensor * KQ_masked = ggml_diag_mask_inf(ctx0, KQ_scaled_biased, n_past);
// KQ = soft_max(KQ_masked)
struct ggml_tensor * KQ_soft_max = ggml_soft_max(ctx0, KQ_masked);
// V_trans = Vmem.view(n_embd/n_head, n_head, n_past + N).permute(1, 2, 0, 3).contiguous()
struct ggml_tensor * V_trans =
ggml_cpy(ctx0,
ggml_permute(ctx0,
ggml_reshape_3d(ctx0,
ggml_view_1d(ctx0, model.kv_self.v, (n_past + N)*n_embd, il*n_ctx*ggml_element_size(model.kv_self.v)*n_embd),
n_embd/n_head, n_head, n_past + N),
1, 2, 0, 3),
ggml_new_tensor_3d(ctx0, model.kv_self.v->type, n_past + N, n_embd/n_head, n_head));
struct ggml_tensor * V =
ggml_view_3d(ctx0, model.kv_self.v,
n_past + N, n_embd/n_head, n_head,
n_ctx*ggml_element_size(model.kv_self.v),
n_ctx*ggml_element_size(model.kv_self.v)*n_embd/n_head,
il*n_ctx*ggml_element_size(model.kv_self.v)*n_embd);
// KQV = transpose(V) * KQ_soft_max
struct ggml_tensor * KQV = ggml_mul_mat(ctx0, V_trans, KQ_soft_max);
struct ggml_tensor * KQV = ggml_mul_mat(ctx0, V, KQ_soft_max);
// KQV_merged = KQV.permute(0, 2, 1, 3)
struct ggml_tensor * KQV_merged = ggml_permute(ctx0, KQV, 0, 2, 1, 3);
@ -625,93 +575,55 @@ bool mpt_eval(
// projection (no bias)
cur = ggml_mul_mat(ctx0,
model.layers[il].c_attn_proj_w,
model.layers[il].attn_out_proj_w,
cur);
}
// norm 2
{
cur = ggml_norm(ctx0, cur);
// cur = ln_1_g*cur + ln_1_b
cur = ggml_add(ctx0,
ggml_mul(ctx0,
ggml_repeat(ctx0, model.layers[il].norm_2_g, cur),
cur),
ggml_repeat(ctx0, model.layers[il].norm_2_b, cur));
}
struct ggml_tensor * inpFF = cur;
// residual
struct ggml_tensor * resSA = ggml_add(ctx0, cur, inpSA);
// feed-forward network
// this is independent of the self-attention result, so it could be done in parallel to the self-attention
{
// note here we pass inpSA instead of cur
cur = resSA;
// norm2
cur = ggml_norm(ctx0, cur);
cur = ggml_mul(ctx0,
ggml_repeat(ctx0, model.layers[il].norm_2_w, cur),
cur);
// ffn
cur = ggml_mul_mat(ctx0,
model.layers[il].up_proj_w,
inpSA);
cur = ggml_add(ctx0,
ggml_repeat(ctx0, model.layers[il].up_proj_b, cur),
model.layers[il].ffn_up_proj_w,
cur);
// RELU activation
cur = ggml_relu(ctx0, cur);
// projection
// cur = proj_w*cur + proj_b
cur = ggml_gelu(ctx0, cur);
cur = ggml_mul_mat(ctx0,
model.layers[il].down_proj_w,
model.layers[il].ffn_down_proj_w,
cur);
cur = ggml_add(ctx0,
ggml_repeat(ctx0, model.layers[il].down_proj_b, cur),
cur);
}
// self-attention + FF
cur = ggml_add(ctx0, cur, inpFF);
// input for next layer
inpL = ggml_add(ctx0, cur, inpL);
}
// norm
{
inpL = ggml_norm(ctx0, inpL);
// inpL = ln_f_g*inpL + ln_f_b
inpL = ggml_add(ctx0,
ggml_mul(ctx0,
ggml_repeat(ctx0, model.ln_f_g, inpL),
inpL),
ggml_repeat(ctx0, model.ln_f_b, inpL));
inpL = ggml_add(ctx0, cur, resSA);
}
// lm_head with weight tying
struct ggml_tensor * out = inpL;
// -> logits
{
inpL = ggml_mul_mat(ctx0, model.wte, inpL);
out = ggml_norm(ctx0, out);
out = ggml_mul(ctx0,
ggml_repeat(ctx0, model.norm_f_w, out),
out);
out = ggml_mul_mat(ctx0, model.wte, out);
}
// logits -> probs
//inpL = ggml_soft_max(ctx0, inpL);
// run the computation
ggml_build_forward_expand(&gf, inpL);
ggml_build_forward_expand(&gf, out);
ggml_graph_compute (ctx0, &gf);
//if (n_past%100 == 0) {
// ggml_graph_print (&gf);
// ggml_graph_dump_dot(&gf, NULL, "gpt-2.dot");
//}
//embd_w.resize(n_vocab*N);
//memcpy(embd_w.data(), ggml_get_data(inpL), sizeof(float)*n_vocab*N);
// return result for just the last token
embd_w.resize(n_vocab);
memcpy(embd_w.data(), (float *) ggml_get_data(inpL) + (n_vocab*(N-1)), sizeof(float)*n_vocab);
memcpy(embd_w.data(), (float *) ggml_get_data(out) + (n_vocab*(N-1)), sizeof(float)*n_vocab);
if (mem_per_token == 0) {
mem_per_token = ggml_used_mem(ctx0)/N;

Write Preview
Loading…
Cancel
Save