| Name | Last commit | Updated |
|---|
|
..
|
|
bench_decode.py
|
H4: fused up+gate option (bit-identical, no end-to-end gain), D-011, LOG |
|
|
bench_distributed.py
|
Gate E: authenticated TCP transport, expert worker, remote executor with local... |
|
|
collect_traces.py
|
Gate B/C: sync-free fast path, bench_decode, capacity sweep, routing traces,... |
|
|
power_sampler.py
|
Gate E: authenticated TCP transport, expert worker, remote executor with local... |
|
|
profile_layer.py
|
Cold sweep complete, host-time attribution, Gate D profiling (latency-bound),... |
|
|
run_cold_nvme_laptop.sh
|
Qwen3-30B parity on M3U96a (bit-identical, 121.7 vs 121.5 tok/s),... |
|
|
simulate_cache.py
|
Probe predictor (fixed) results, depth variants, attribution figure/table, H4... |
|
|
ssd_bandwidth.py
|
Gate A/B: scaffold, mandate, inventory tooling, prior art (59 refs),... |
|