code wiki / (root) / nx_gguf_load_model_test.nx

nx_gguf_load_model_test.nx

buildroot/runtime/nx_gguf_load_model_test.nx

7488 B185 linesdepth 7pulls 9 transitivereach 0 importersview sourcekind gate/prooftopic gguf
docsdependenciesstructsconstsfunctions

about

nx_gguf_load_model_test.nx -- smoke for nx_gguf_load_model.nx. Builds two synthetic GGUFs: A) Full model (3 tensors): token_embd + output_norm + output B) Tied-embedding model (2 tensors): token_embd + output_norm (no output.weight; caller expected to reuse token_embd) Verifies: - Bundle struct populated correctly when all 3 tensors present - is_output_tied flag and null output_weight when output.weight is missing - Verdict-range gate

dependencies 7 imports · 0 importers

nx_syscalls.nx nx_tier.nx nx_le.nx nx_tensor.nx nx_gguf.nx nx_gguf_load.nx nx_gguf_load_model.nx nx_gguf_load_model_test.nx

imports: nx_syscalls.nxnx_tier.nxnx_le.nxnx_tensor.nxnx_gguf.nxnx_gguf_load.nxnx_gguf_load_model.nx

imported by: nobody (leaf or entry point)

call flow from main pre-order; caps 40 nodes / depth 6 declared; ↻ = already shown

main nx_gml_verdict_is_valid sys_mmap nx_le_write_u64 nx_le_write_u32 _write_ti nx_le_write_u64 ↻ nx_le_write_u32 ↻ nx_le_write_u32 ↻ nx_gguf_parse nx_le_read_u32 nx_le_read_u64 nx_le_read_u32 ↻ _gguf_skip_value nx_le_read_u64 ↻ nx_le_read_u32 ↻ _gguf_skip_value ↻ sys_mmap ↻ nx_gguf_load_model_weights sys_mmap ↻ nx_gguf_load_tensor nx_gguf_find_tensor nx_gguf_tensor_at nx_gguf_name_equals nx_gguf_tensor_at ↻ nx_gguf_tensor_n_values sys_mmap ↻ nx_t_alloc nx_dt_is_valid nx_dt_element_bytes sys_mmap ↻ nx_t_compute_strides_rowma nx_gguf_dequant_f32 nx_le_read_u32 ↻ _gguf_f32_to_q10 nx_gguf_dequant_f16 nx_le_read_u16 _gguf_f16_to_q10 nx_gguf_dequant_q8_0 nx_le_read_u16 ↻

structs

none

consts

none

functions

24func _write_ti(buf: *u8, off: i64,
called by 1: main calls 2: nx_le_write_u64nx_le_write_u32
41func main() -> i64
calls 7: nx_gml_verdict_is_validsys_mmapnx_le_write_u64_write_tinx_le_write_u32nx_gguf_parse+1