code wiki / (root) / nx_f32_llama_stack_v4.nx

nx_f32_llama_stack_v4.nx

buildroot/runtime/nx_f32_llama_stack_v4.nx

3270 B96 linesdepth 11pulls 41 transitivereach 45 importersview sourcekind librarytopic f32

docs dependencies structs consts functions

about

nx_f32_llama_stack_v4.nx -- N-layer transformer stack using v4 lazy layers. Mirrors nx_f32_llama_stack but accepts an array of *NxF32LlamaLayerLazy and dispatches each layer through nx_f32_llama_block_forward_v4. genealogy_id: standard_transformer_stack + lazy_dispatch lineage_id: substrate_f32_llama_stack_v4

dependencies 6 imports · 35 importers

diagram shows first 10 each side; +0 more imports, +25 more importers in the complete lists below.

imports: nx_syscalls.nxnx_tier.nxnx_f32.nxnx_f32_kv_cache.nxnx_f32_lazy_weight.nxnx_f32_llama_block_v4.nx

imported by: nx_batched_gate.nxnx_batched_tput.nxnx_chat_warmcache.nxnx_coder_gen_f32.nxnx_companion_chat.nxnx_companion_voice.nxnx_f32_llama_stack_v4_test.nxnx_f32_llama_v4b.nxnx_f32_llama_v4p.nxnx_f32_llm_bench_test.nxnx_f32_llm_chat_test.nxnx_f32_llm_cohere_test.nxnx_f32_llm_live_gen_test.nxnx_f32_llm_live_load_test.nxnx_f32_llm_run_v3.nxnx_f32_llm_run_v3_test.nxnx_f32_llm_serve.nxnx_f32_llm_v4.nxnx_f32_llm_v4_test.nxnx_f32_qwen_capital_gate.nxnx_llm_argmax_probe.nxnx_llm_batch_serve.nxnx_llm_diff_probe.nxnx_llm_forward_profile.nxnx_llm_france_probe.nxnx_llm_layerlens_probe.nxnx_llm_sched_gate.nxnx_lowrank_kv_real.nxnx_paged_fwd_gate.nxnx_qwen_native_gate.nxnx_qwen_wsl_timing_gate.nxnx_reason_paged_probe.nxnx_reason_probe.nxnx_reasoning_gate.nxnx_specdec_gate.nx

structs

none

consts

16	const NX_STK4_OK: nx_int = 0
17	const NX_STK4_ERR_BAD_DIM: nx_int = 1
18	const NX_STK4_ERR_NULL: nx_int = 2
19	const NX_STK4_ERR_LAYER: nx_int = 3
20	const NX_STK4_N_VERDICTS: nx_int = 4

functions

22	func nx_stk4_verdict_is_valid(v: nx_int) -> nx_int called by 1: main
30	func nx_f32_llama_stack_forward_v4( called by 2: main _flv4_core calls 3: sys_mmap nx_f32_llama_block_forward_v4 nx_f32_kv_cache_advance