whn09 0cae424dc4 Neuron: one KV gather per chunk for dense attention; no layer_idx guards (#26) ... - nki_attention._chunk_attention: a prompt chunk run as a decode step gathers a dense compressed-KV scope once (every token's list is a prefix of the longest) and masks it per row; top-k and SWA scopes stay per row. The per-row gather was [128, 8192, 512] for a ratio-2 layer of the full model: 3.8 GB moved per call and a 3 GB NEFF (model code) that ran HBM out. - decode_part_timer: the decode timers are no-ops inside region graphs. Their f-string labels read layer_idx, so Dynamo guarded on it and retraced every layer instead of every layer kind. Tiny V4.1 on a NeuronCore, bucket 128: short 9/9, long 9/9 (logit cos >= 0.997). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>		2026-10-23 04:32:20 +08:00
..
eplb	initial commit	2026-10-20 12:09:42 +08:00
input_metadata	initial commit	2026-10-20 12:09:42 +08:00
layers	Neuron: full DeepSeek-V4.1-Flash load + decode region graphs at EP=64 (issues #21, #25)	2026-10-23 01:39:22 +08:00
models	Neuron: one KV gather per chunk for dense attention; no layer_idx guards (#26)	2026-10-23 04:32:20 +08:00
parallel_utils	initial commit	2026-10-20 12:09:42 +08:00
__init__.py	initial commit	2026-10-20 12:09:42 +08:00
layer_utils.py	initial commit	2026-10-20 12:09:42 +08:00
model_loader.py	Neuron: full DeepSeek-V4.1-Flash load + decode region graphs at EP=64 (issues #21, #25)	2026-10-23 01:39:22 +08:00
model_metadata_utils.py	Neuron: prompt chunks through the decode region graphs (issue #26)	2026-10-23 03:19:56 +08:00
model_output.py	initial commit	2026-10-20 12:09:42 +08:00
sampling_inputs.py	initial commit	2026-10-20 12:09:42 +08:00
utils.py	Neuron: static decode graphs, static-shape MoE, trn2 e4m3 fp8 (issues #9, #12, #24)	2026-10-22 11:59:37 +08:00