examples(gguf): GGUF example outputs

[gabe-l-hart]

Description

This PR is extracted from #16982 since it's an isolated change that's not strictly related to implementing the SSD algorithm.

The changes in this PR add some extra output to the llama-gguf tool to show each tensor's type and element count.

Example Output

...
gguf_ex_read_1: tensor[0]: name = token_embd.weight, size = 513802240, offset = 0, type = bf16, n_elts = 256901120
gguf_ex_read_1: tensor[1]: name = blk.0.attn_norm.weight, size = 10240, offset = 513802240, type = f32, n_elts = 2560
gguf_ex_read_1: tensor[2]: name = blk.0.ffn_norm.weight, size = 10240, offset = 513812480, type = f32, n_elts = 2560
gguf_ex_read_1: tensor[3]: name = blk.0.attn_k.weight, size = 2621440, offset = 513822720, type = bf16, n_elts = 1310720
gguf_ex_read_1: tensor[4]: name = blk.0.attn_output.weight, size = 13107200, offset = 516444160, type = bf16, n_elts = 6553600
gguf_ex_read_1: tensor[5]: name = blk.0.attn_q.weight, size = 13107200, offset = 529551360, type = bf16, n_elts = 6553600
gguf_ex_read_1: tensor[6]: name = blk.0.attn_v.weight, size = 2621440, offset = 542658560, type = bf16, n_elts = 1310720
gguf_ex_read_1: tensor[7]: name = blk.0.ffn_gate.weight, size = 41943040, offset = 545280000, type = bf16, n_elts = 20971520
...

[ggerganov]

valign:

Suggested change

	const auto type = gguf_get_tensor_type(ctx, i);
	const char * type_name = ggml_type_name(type);
	const size_t type_size = ggml_type_size(type);
	const size_t n_elements = size / type_size;
	const auto type = gguf_get_tensor_type (ctx, i);
	const char * type_name = ggml_type_name(type);
	const size_t type_size = ggml_type_size(type);
	const size_t n_elements = size / type_size;