root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-11-13 14:29:52 +00:00
Code Issues Actions 11 Packages Projects Releases Wiki Activity

stdout : vertical align outputs for better readibility

Browse Source
This commit is contained in:
Georgi Gerganov 2023-04-16 13:58:48 +03:00
parent 489537e6cf
commit 3173a62eb9
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
2 changed files with 10 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
convert.py
View File

@ -951,8 +951,9 @@ class OutputFile:
ndarrays = bounded_parallel_map(do_item, model.items(), concurrency=8) ndarrays = bounded_parallel_map(do_item, model.items(), concurrency=8)
for i, ((name, lazy_tensor), ndarray) in enumerate(zip(model.items(), ndarrays)): for i, ((name, lazy_tensor), ndarray) in enumerate(zip(model.items(), ndarrays)):
size = ' x '.join(map(str, lazy_tensor.shape)) size = ' x '.join(f"{dim:6d}" for dim in lazy_tensor.shape)
print(f"[{i+1}/{len(model)}] Writing tensor {name}, size {size}...") padi = len(str(len(model)))
print(f"[{i+1:{padi}d}/{len(model)}] Writing tensor {name:38s} | size {size:16} | type {lazy_tensor.data_type}")
of.write_tensor_header(name, lazy_tensor.shape, lazy_tensor.data_type) of.write_tensor_header(name, lazy_tensor.shape, lazy_tensor.data_type)
ndarray.tofile(of.fout) ndarray.tofile(of.fout)
of.fout.close() of.fout.close()

14
llama.cpp
View File

@ -262,12 +262,12 @@ static size_t checked_div(size_t a, size_t b) {
} }
static std::string llama_format_tensor_shape(const std::vector<uint32_t> & ne) { static std::string llama_format_tensor_shape(const std::vector<uint32_t> & ne) {
std::string ret = "[" + std::to_string(ne.at(0)); char buf[256];
snprintf(buf, sizeof(buf), "%5u", ne.at(0));
for (size_t i = 1; i < ne.size(); i++) { for (size_t i = 1; i < ne.size(); i++) {
ret += " x " + std::to_string(ne.at(i)); snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), " x %5u", ne.at(i));
} }
ret += "]"; return buf;
return ret;
} }
static size_t llama_calc_tensor_size(const std::vector<uint32_t> & ne, enum ggml_type type) { static size_t llama_calc_tensor_size(const std::vector<uint32_t> & ne, enum ggml_type type) {
@ -942,8 +942,8 @@ static void llama_model_load_internal(
ml->ggml_ctx = ctx; ml->ggml_ctx = ctx;
model.tok_embeddings = ml->get_tensor("tok_embeddings.weight", {n_embd, n_vocab}); model.tok_embeddings = ml->get_tensor("tok_embeddings.weight", {n_embd, n_vocab});
model.norm = ml->get_tensor("norm.weight", {n_embd}); model.norm = ml->get_tensor("norm.weight", {n_embd});
model.output = ml->get_tensor("output.weight", {n_embd, n_vocab}); model.output = ml->get_tensor("output.weight", {n_embd, n_vocab});
model.layers.resize(n_layer); model.layers.resize(n_layer);
for (uint32_t i = 0; i < n_layer; ++i) { for (uint32_t i = 0; i < n_layer; ++i) {
@ -1570,7 +1570,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
tensor.data = read_data.addr; tensor.data = read_data.addr;
model_loader->load_data_for(tensor); model_loader->load_data_for(tensor);
printf("[%zu/%zu] %36s - %s, type = %6s, ", printf("[%4zu/%4zu] %36s - %16s, type = %6s, ",
++idx, model_loader->tensors_map.tensors.size(), ++idx, model_loader->tensors_map.tensors.size(),
tensor.name.c_str(), llama_format_tensor_shape(tensor.ne).c_str(), tensor.name.c_str(), llama_format_tensor_shape(tensor.ne).c_str(),
ggml_type_name(tensor.type)); ggml_type_name(tensor.type));