From b3a94dd9e0a538e48e15af175535aaaa5f16d3e6 Mon Sep 17 00:00:00 2001 From: Pierrick HYMBERT Date: Mon, 18 Mar 2024 17:51:20 +0100 Subject: [PATCH] gguf-split: rename `--split-tensors-size` to `--split-max-tensors`. Set general.split_count KV to all split --- examples/gguf-split/README.md | 2 +- examples/gguf-split/gguf-split.cpp | 6 +++--- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/examples/gguf-split/README.md b/examples/gguf-split/README.md index c1a5dab1d..ddb1f7649 100644 --- a/examples/gguf-split/README.md +++ b/examples/gguf-split/README.md @@ -5,5 +5,5 @@ CLI to split / merge GGUF files. **Command line options:** - `--split`: split GGUF to multiple GGUF, default operation. -- `--split-tensors-size`: size of tensors in each split: default(128) +- `--split-max-tensors`: maximum tensors in each split: default(128) - `--merge`: merge multiple GGUF to a single GGUF. diff --git a/examples/gguf-split/gguf-split.cpp b/examples/gguf-split/gguf-split.cpp index 063461365..aa618d9a1 100644 --- a/examples/gguf-split/gguf-split.cpp +++ b/examples/gguf-split/gguf-split.cpp @@ -48,7 +48,7 @@ static void split_print_usage(const char * executable) { printf(" -h, --help show this help message and exit\n"); printf(" --version show version and build info\n"); printf(" --split split GGUF to multiple GGUF (default)\n"); - printf(" --split-tensors-size size of tensors in each split: default(%d)\n", default_params.n_split_tensors); + printf(" --split-max-tensors max tensors in each split: default(%d)\n", default_params.n_split_tensors); printf(" --merge merge multiple GGUF to a single GGUF\n"); printf(" --upload upload GGUF to an URL\n"); printf("\n"); @@ -85,7 +85,7 @@ static bool split_params_parse_ex(int argc, const char ** argv, split_params & p arg_found = true; params.operation = SPLIT_OP_SPLIT; } - if (arg == "--split-tensors-size") { + if (arg == "--split-max-tensors") { if (++arg_idx >= argc) { invalid_param = true; break; @@ -188,9 +188,9 @@ struct split_strategy { // Save all metadata in first split only if (i_split == 0) { gguf_set_kv(ctx_out, ctx_gguf); - gguf_set_val_u8(ctx_out, LLM_KV_GENERAL_SPLIT_N_SPLIT, n_split); } gguf_set_val_u8(ctx_out, LLM_KV_GENERAL_SPLIT_I_SPLIT, i_split); + gguf_set_val_u8(ctx_out, LLM_KV_GENERAL_SPLIT_N_SPLIT, n_split); // populate the original tensors, so we get an initial metadata for (int i = i_split * params.n_split_tensors; i < n_tensors