Skip to content

code : normalize enum names #5697

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Feb 25, 2024
Merged

code : normalize enum names #5697

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
0aead81
coda : normalize enum names
ggerganov Feb 24, 2024
cb24663
code : cont
ggerganov Feb 24, 2024
0db4cb0
code : cont
ggerganov Feb 24, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
18 changes: 9 additions & 9 deletions common/common.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -295,9 +295,9 @@ bool gpt_params_parse_ex(int argc, char ** argv, gpt_params & params) {
break;
}
std::string value(argv[i]);
/**/ if (value == "none") { params.rope_scaling_type = LLAMA_ROPE_SCALING_NONE; }
else if (value == "linear") { params.rope_scaling_type = LLAMA_ROPE_SCALING_LINEAR; }
else if (value == "yarn") { params.rope_scaling_type = LLAMA_ROPE_SCALING_YARN; }
/**/ if (value == "none") { params.rope_scaling_type = LLAMA_ROPE_SCALING_TYPE_NONE; }
else if (value == "linear") { params.rope_scaling_type = LLAMA_ROPE_SCALING_TYPE_LINEAR; }
else if (value == "yarn") { params.rope_scaling_type = LLAMA_ROPE_SCALING_TYPE_YARN; }
else { invalid_param = true; break; }
} else if (arg == "--rope-scale") {
if (++i >= argc) {
Expand Down Expand Up @@ -630,11 +630,11 @@ bool gpt_params_parse_ex(int argc, char ** argv, gpt_params & params) {
}
std::string arg_next = argv[i];
if (arg_next == "none") {
params.split_mode = LLAMA_SPLIT_NONE;
params.split_mode = LLAMA_SPLIT_MODE_NONE;
} else if (arg_next == "layer") {
params.split_mode = LLAMA_SPLIT_LAYER;
params.split_mode = LLAMA_SPLIT_MODE_LAYER;
} else if (arg_next == "row") {
params.split_mode = LLAMA_SPLIT_ROW;
params.split_mode = LLAMA_SPLIT_MODE_ROW;
} else {
invalid_param = true;
break;
Expand Down Expand Up @@ -837,15 +837,15 @@ bool gpt_params_parse_ex(int argc, char ** argv, gpt_params & params) {
sep++;
if (strncmp(sep, "int:", 4) == 0) {
sep += 4;
kvo.tag = LLAMA_KV_OVERRIDE_INT;
kvo.tag = LLAMA_KV_OVERRIDE_TYPE_INT;
kvo.int_value = std::atol(sep);
} else if (strncmp(sep, "float:", 6) == 0) {
sep += 6;
kvo.tag = LLAMA_KV_OVERRIDE_FLOAT;
kvo.tag = LLAMA_KV_OVERRIDE_TYPE_FLOAT;
kvo.float_value = std::atof(sep);
} else if (strncmp(sep, "bool:", 5) == 0) {
sep += 5;
kvo.tag = LLAMA_KV_OVERRIDE_BOOL;
kvo.tag = LLAMA_KV_OVERRIDE_TYPE_BOOL;
if (std::strcmp(sep, "true") == 0) {
kvo.bool_value = true;
} else if (std::strcmp(sep, "false") == 0) {
Expand Down
4 changes: 2 additions & 2 deletions common/common.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -61,7 +61,7 @@ struct gpt_params {
float p_split = 0.1f; // speculative decoding split probability
int32_t n_gpu_layers = -1; // number of layers to store in VRAM (-1 - use default)
int32_t n_gpu_layers_draft = -1; // number of layers to store in VRAM for the draft model (-1 - use default)
llama_split_mode split_mode = LLAMA_SPLIT_LAYER; // how to split the model across GPUs
llama_split_mode split_mode = LLAMA_SPLIT_MODE_LAYER; // how to split the model across GPUs
int32_t main_gpu = 0; // the GPU that is used for scratch and small tensors
float tensor_split[128] = {0}; // how split tensors should be distributed across GPUs
int32_t n_beams = 0; // if non-zero then use beam search of given width.
Expand All @@ -75,7 +75,7 @@ struct gpt_params {
float yarn_beta_fast = 32.0f; // YaRN low correction dim
float yarn_beta_slow = 1.0f; // YaRN high correction dim
int32_t yarn_orig_ctx = 0; // YaRN original context length
int32_t rope_scaling_type = LLAMA_ROPE_SCALING_UNSPECIFIED;
int32_t rope_scaling_type = LLAMA_ROPE_SCALING_TYPE_UNSPECIFIED;
ggml_numa_strategy numa = GGML_NUMA_STRATEGY_DISABLED;

// // sampling parameters
Expand Down
10 changes: 5 additions & 5 deletions common/train.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,7 +31,7 @@ struct train_state * init_train_state() {

state->opt = new struct ggml_opt_context;
state->opt->ctx = NULL;
state->opt->params = ggml_opt_default_params(GGML_OPT_ADAM);
state->opt->params = ggml_opt_default_params(GGML_OPT_TYPE_ADAM);
state->opt->params.graph_size = LLAMA_TRAIN_MAX_NODES;
state->opt->loss_after = 0.0f;

Expand Down Expand Up @@ -556,7 +556,7 @@ void load_opt_context_gguf(struct gguf_context * fctx, struct ggml_context * f_g
std::string opt_type;
GGUF_GET_KEY(fctx, opt_type, gguf_get_val_str, GGUF_TYPE_STRING, true, LLM_KV_OPTIMIZER_TYPE);
if (opt_type == LLM_KV_OPTIMIZER_TYPE_ADAM) {
opt->params.type = GGML_OPT_ADAM;
opt->params.type = GGML_OPT_TYPE_ADAM;

GGUF_GET_KEY(fctx, opt->adam.fx_best, gguf_get_val_f32, GGUF_TYPE_FLOAT32, true, LLM_KV_OPTIMIZER_ADAM_BEST_LOSS);
GGUF_GET_KEY(fctx, opt->adam.fx_prev, gguf_get_val_f32, GGUF_TYPE_FLOAT32, true, LLM_KV_OPTIMIZER_ADAM_PREVIOUS_LOSS);
Expand All @@ -568,7 +568,7 @@ void load_opt_context_gguf(struct gguf_context * fctx, struct ggml_context * f_g
copy_tensor_by_name(opt->adam.v, f_ggml_ctx, LLM_TENSOR_OPTIMIZER_ADAM_SECOND_MOMENTS);
copy_tensor_by_name(opt->adam.pf, f_ggml_ctx, LLM_TENSOR_OPTIMIZER_ADAM_PAST_LOSS_VALUES);
} else if (opt_type == LLM_KV_OPTIMIZER_TYPE_LBFGS) {
opt->params.type = GGML_OPT_LBFGS;
opt->params.type = GGML_OPT_TYPE_LBFGS;

GGUF_GET_KEY(fctx, opt->params.lbfgs.m, gguf_get_val_u32, GGUF_TYPE_UINT32, true, LLM_KV_OPTIMIZER_LBFGS_APPROX_HESSIAN_COUNT);
GGUF_GET_KEY(fctx, opt->lbfgs.fx_best, gguf_get_val_f32, GGUF_TYPE_FLOAT32, true, LLM_KV_OPTIMIZER_LBFGS_BEST_LOSS);
Expand Down Expand Up @@ -603,7 +603,7 @@ void save_opt_context_gguf(struct gguf_context * fctx, struct ggml_opt_context *
gguf_set_val_bool(fctx, LLM_KV_OPTIMIZER_JUST_INITIALIZED, opt->just_initialized);

switch (opt->params.type) {
case GGML_OPT_ADAM:
case GGML_OPT_TYPE_ADAM:
{
gguf_set_val_str(fctx, LLM_KV_OPTIMIZER_TYPE, LLM_KV_OPTIMIZER_TYPE_ADAM);
gguf_set_val_f32(fctx, LLM_KV_OPTIMIZER_ADAM_BEST_LOSS, opt->adam.fx_best);
Expand All @@ -622,7 +622,7 @@ void save_opt_context_gguf(struct gguf_context * fctx, struct ggml_opt_context *
gguf_add_tensor(fctx, opt->adam.pf);
}
} break;
case GGML_OPT_LBFGS:
case GGML_OPT_TYPE_LBFGS:
{
gguf_set_val_str(fctx, LLM_KV_OPTIMIZER_TYPE, LLM_KV_OPTIMIZER_TYPE_LBFGS);
gguf_set_val_u32(fctx, LLM_KV_OPTIMIZER_LBFGS_APPROX_HESSIAN_COUNT, opt->params.lbfgs.m);
Expand Down
2 changes: 1 addition & 1 deletion examples/baby-llama/baby-llama.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1547,7 +1547,7 @@ int main(int argc, char ** argv) {

float error_before_opt = ggml_get_f32_1d(e, 0);

struct ggml_opt_params opt_params_lbfgs = ggml_opt_default_params(GGML_OPT_LBFGS);
struct ggml_opt_params opt_params_lbfgs = ggml_opt_default_params(GGML_OPT_TYPE_LBFGS);
opt_params_lbfgs.print_forward_graph = false;
opt_params_lbfgs.print_backward_graph = false;
opt_params_lbfgs.lbfgs.n_iter = 16;
Expand Down
2 changes: 1 addition & 1 deletion examples/finetune/finetune.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1531,7 +1531,7 @@ int main(int argc, char ** argv) {
lora.hparams.n_rank_output = n_rank_output;

// set opt params from command line
opt->params = ggml_opt_default_params(GGML_OPT_ADAM);
opt->params = ggml_opt_default_params(GGML_OPT_TYPE_ADAM);
opt->params.print_forward_graph = false;
opt->params.print_backward_graph = false;
opt->params.graph_size = LLAMA_TRAIN_MAX_NODES;
Expand Down
14 changes: 7 additions & 7 deletions examples/llama-bench/llama-bench.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -157,9 +157,9 @@ static const char * output_format_str(output_formats format) {

static const char * split_mode_str(llama_split_mode mode) {
switch (mode) {
case LLAMA_SPLIT_NONE: return "none";
case LLAMA_SPLIT_LAYER: return "layer";
case LLAMA_SPLIT_ROW: return "row";
case LLAMA_SPLIT_MODE_NONE: return "none";
case LLAMA_SPLIT_MODE_LAYER: return "layer";
case LLAMA_SPLIT_MODE_ROW: return "row";
default: GGML_ASSERT(!"invalid split mode");
}
}
Expand Down Expand Up @@ -193,7 +193,7 @@ static const cmd_params cmd_params_defaults = {
/* type_v */ {GGML_TYPE_F16},
/* n_threads */ {get_num_physical_cores()},
/* n_gpu_layers */ {99},
/* split_mode */ {LLAMA_SPLIT_LAYER},
/* split_mode */ {LLAMA_SPLIT_MODE_LAYER},
/* main_gpu */ {0},
/* no_kv_offload */ {false},
/* mul_mat_q */ {true},
Expand Down Expand Up @@ -358,11 +358,11 @@ static cmd_params parse_cmd_params(int argc, char ** argv) {
for (const auto & m : p) {
llama_split_mode mode;
if (m == "none") {
mode = LLAMA_SPLIT_NONE;
mode = LLAMA_SPLIT_MODE_NONE;
} else if (m == "layer") {
mode = LLAMA_SPLIT_LAYER;
mode = LLAMA_SPLIT_MODE_LAYER;
} else if (m == "row") {
mode = LLAMA_SPLIT_ROW;
mode = LLAMA_SPLIT_MODE_ROW;
} else {
invalid_param = true;
break;
Expand Down
2 changes: 1 addition & 1 deletion examples/llava/llava.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -152,7 +152,7 @@ static bool clip_llava_handle_patches(clip_ctx * ctx_clip, std::vector<float *>

ggml_tensor * newline_tmp = clip_get_newline_tensor(ctx_clip);
model.newline = ggml_new_tensor_1d(model.ctx, GGML_TYPE_F32, newline_tmp->ne[0]);
if (newline_tmp->backend != GGML_BACKEND_CPU) {
if (newline_tmp->backend != GGML_BACKEND_TYPE_CPU) {
if (newline_tmp->buffer == NULL) {
printf("newline_tmp tensor buffer is NULL\n");
}
Expand Down
18 changes: 9 additions & 9 deletions examples/server/server.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2086,9 +2086,9 @@ static void server_params_parse(int argc, char **argv, server_params &sparams,
break;
}
std::string value(argv[i]);
/**/ if (value == "none") { params.rope_scaling_type = LLAMA_ROPE_SCALING_NONE; }
else if (value == "linear") { params.rope_scaling_type = LLAMA_ROPE_SCALING_LINEAR; }
else if (value == "yarn") { params.rope_scaling_type = LLAMA_ROPE_SCALING_YARN; }
/**/ if (value == "none") { params.rope_scaling_type = LLAMA_ROPE_SCALING_TYPE_NONE; }
else if (value == "linear") { params.rope_scaling_type = LLAMA_ROPE_SCALING_TYPE_LINEAR; }
else if (value == "yarn") { params.rope_scaling_type = LLAMA_ROPE_SCALING_TYPE_YARN; }
else { invalid_param = true; break; }
}
else if (arg == "--rope-freq-base")
Expand Down Expand Up @@ -2212,15 +2212,15 @@ static void server_params_parse(int argc, char **argv, server_params &sparams,
std::string arg_next = argv[i];
if (arg_next == "none")
{
params.split_mode = LLAMA_SPLIT_NONE;
params.split_mode = LLAMA_SPLIT_MODE_NONE;
}
else if (arg_next == "layer")
{
params.split_mode = LLAMA_SPLIT_LAYER;
params.split_mode = LLAMA_SPLIT_MODE_LAYER;
}
else if (arg_next == "row")
{
params.split_mode = LLAMA_SPLIT_ROW;
params.split_mode = LLAMA_SPLIT_MODE_ROW;
}
else {
invalid_param = true;
Expand Down Expand Up @@ -2447,15 +2447,15 @@ static void server_params_parse(int argc, char **argv, server_params &sparams,
sep++;
if (strncmp(sep, "int:", 4) == 0) {
sep += 4;
kvo.tag = LLAMA_KV_OVERRIDE_INT;
kvo.tag = LLAMA_KV_OVERRIDE_TYPE_INT;
kvo.int_value = std::atol(sep);
} else if (strncmp(sep, "float:", 6) == 0) {
sep += 6;
kvo.tag = LLAMA_KV_OVERRIDE_FLOAT;
kvo.tag = LLAMA_KV_OVERRIDE_TYPE_FLOAT;
kvo.float_value = std::atof(sep);
} else if (strncmp(sep, "bool:", 5) == 0) {
sep += 5;
kvo.tag = LLAMA_KV_OVERRIDE_BOOL;
kvo.tag = LLAMA_KV_OVERRIDE_TYPE_BOOL;
if (std::strcmp(sep, "true") == 0) {
kvo.bool_value = true;
} else if (std::strcmp(sep, "false") == 0) {
Expand Down
2 changes: 1 addition & 1 deletion examples/train-text-from-scratch/train-text-from-scratch.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -960,7 +960,7 @@ int main(int argc, char ** argv) {
struct ggml_opt_context * opt = train->opt;

// set opt params from command line
opt->params = ggml_opt_default_params(GGML_OPT_ADAM);
opt->params = ggml_opt_default_params(GGML_OPT_TYPE_ADAM);
opt->params.print_forward_graph = false;
opt->params.print_backward_graph = false;
opt->params.graph_size = LLAMA_TRAIN_MAX_NODES;
Expand Down
Loading