364 kB

Ctrl+K

1 contributor

History: 26 commits

ryugyosoft

v0.7.1: model list notes and catalog memory updated for onw 0.7

88cdcee verified about 9 hours ago

onw
v0.7.1: model list notes and catalog memory updated for onw 0.7 about 9 hours ago
.gitattributes

1.52 kB
initial commit 2 days ago
LICENSE

11.3 kB
npue: engine for LLMs entirely on the Intel NPU 2 days ago
README.md

24.8 kB
v0.7.0: 20-60% less memory (own blob store, vision loaded on demand and freed, host tables mapped, direct-I/O expert load); REAP streams too; first-on-Intel-NPU SSD streaming note about 9 hours ago
README_en.md

17.8 kB
v0.7.0: 20-60% less memory (own blob store, vision loaded on demand and freed, host tables mapped, direct-I/O expert load); REAP streams too; first-on-Intel-NPU SSD streaming note about 9 hours ago
bench_ssd.py

1.62 kB
v0.6.0: stream MoE experts from the SSD (LRU cache, direct I/O) when the model does not fit; expert memory setting about 10 hours ago
check_ref.py

3.02 kB
onw 0.2: renamed from npue; onw command; Qwen3.5 (dense) / Gemma 4 / vision support; LM head segment 2 days ago
check_vision.py

2.14 kB
onw 0.2: renamed from npue; onw command; Qwen3.5 (dense) / Gemma 4 / vision support; LM head segment 2 days ago
diag_long.py

1.01 kB
v0.4.7: no NPUW for MoE models, bounded hold on broken characters, clear-cache button, save button on top 2 days ago
diag_pld.py

1.53 kB
onw 0.2: renamed from npue; onw command; Qwen3.5 (dense) / Gemma 4 / vision support; LM head segment 2 days ago
diag_states.py

989 Bytes
v0.4.8: NPUW fixes (no SLICE_OUT, bool constants), NaN check, ONW_SIZES, Linux driver 1.38 note, diagnostic tools 1 day ago
exp_npuw.py

1.86 kB
v0.4.5: NPUW weight sharing (auto on NPU 4000+, experimental), taller list rows on Ubuntu 2 days ago
find_python.bat

2.22 kB
onw 0.4: API-first server (/ info, /health, CORS, --api-key, /chat, /check), onw window (model download manager, server control, settings), setup finds / installs Python 2 days ago
install.ps1

1.68 kB
v0.5.0: installers retry each download and normalise line endings 1 day ago
install.sh

1.61 kB
v0.5.0: installers retry each download and normalise line endings 1 day ago
mem_all.py

2.03 kB
v0.7.0: 20-60% less memory (own blob store, vision loaded on demand and freed, host tables mapped, direct-I/O expert load); REAP streams too; first-on-Intel-NPU SSD streaming note about 9 hours ago
mem_breakdown.py

2.9 kB
v0.7.0: 20-60% less memory (own blob store, vision loaded on demand and freed, host tables mapped, direct-I/O expert load); REAP streams too; first-on-Intel-NPU SSD streaming note about 9 hours ago
mem_footprint.py

460 Bytes
onw 0.3: --context at load time, top_k / penalties, tray app + setup scripts, --compact / --prune 2 days ago
mem_import.py

3.22 kB
v0.7.0: 20-60% less memory (own blob store, vision loaded on demand and freed, host tables mapped, direct-I/O expert load); REAP streams too; first-on-Intel-NPU SSD streaming note about 9 hours ago
mem_stages.py

2.92 kB
v0.7.0: 20-60% less memory (own blob store, vision loaded on demand and freed, host tables mapped, direct-I/O expert load); REAP streams too; first-on-Intel-NPU SSD streaming note about 9 hours ago
mem_vision.py

1.4 kB
v0.7.0: 20-60% less memory (own blob store, vision loaded on demand and freed, host tables mapped, direct-I/O expert load); REAP streams too; first-on-Intel-NPU SSD streaming note about 9 hours ago
probe_segments.py

1.71 kB
v0.4.8: NPUW fixes (no SLICE_OUT, bool constants), NaN check, ONW_SIZES, Linux driver 1.38 note, diagnostic tools 1 day ago
prof_block.py

1.56 kB
onw 0.2: renamed from npue; onw command; Qwen3.5 (dense) / Gemma 4 / vision support; LM head segment 2 days ago
prof_head.py

1.75 kB
onw 0.3: --context at load time, top_k / penalties, tray app + setup scripts, --compact / --prune 2 days ago
prof_segments.py

1.23 kB
onw 0.2: renamed from npue; onw command; Qwen3.5 (dense) / Gemma 4 / vision support; LM head segment 2 days ago
pyproject.toml

704 Bytes
v0.7.1: model list notes and catalog memory updated for onw 0.7 about 9 hours ago
requirements.txt

520 Bytes
onw 0.2: renamed from npue; onw command; Qwen3.5 (dense) / Gemma 4 / vision support; LM head segment 2 days ago
setup.bat

2.18 kB
v0.5.0: setup no longer upgrades pip (a new pip failed to install packages with scripts), rebuilds the venv once on failure, installs the web view package on upgrade 1 day ago
setup.sh

2.58 kB
v0.5.0: restore setup.sh / start.sh (they were uploaded empty) 1 day ago
sim_expert_cache.py

2.99 kB
v0.6.0: stream MoE experts from the SSD (LRU cache, direct I/O) when the model does not fit; expert memory setting about 10 hours ago
start.bat

1.63 kB
onw 0.4.1: Japanese README (README.md) + English (README_en.md); setup/start call find_python.bat by path 2 days ago
start.sh

1.56 kB
v0.5.0: restore setup.sh / start.sh (they were uploaded empty) 1 day ago
test_chat.py

2.98 kB
onw 0.2: renamed from npue; onw command; Qwen3.5 (dense) / Gemma 4 / vision support; LM head segment 2 days ago
test_context.py

1.39 kB
onw 0.3: --context at load time, top_k / penalties, tray app + setup scripts, --compact / --prune 2 days ago
test_image.py

1.54 kB
onw 0.2: renamed from npue; onw command; Qwen3.5 (dense) / Gemma 4 / vision support; LM head segment 2 days ago
test_layer_major.py

1.83 kB
v0.6.1: layer-major prompt processing when experts stream from the SSD (each layer's experts read once per prompt) about 10 hours ago
test_moe_npu.py

3.92 kB
v0.4.8: NPUW fixes (no SLICE_OUT, bool constants), NaN check, ONW_SIZES, Linux driver 1.38 note, diagnostic tools 1 day ago
test_npuw_compile.py

1.39 kB
v0.4.8: NPUW fixes (no SLICE_OUT, bool constants), NaN check, ONW_SIZES, Linux driver 1.38 note, diagnostic tools 1 day ago
test_stream.py

2.16 kB
v0.6.0: stream MoE experts from the SSD (LRU cache, direct I/O) when the model does not fit; expert memory setting about 10 hours ago
trace_experts.py

3.97 kB
v0.6.0: stream MoE experts from the SSD (LRU cache, direct I/O) when the model does not fit; expert memory setting about 10 hours ago