tzhouam 9df5ac8836 Run InternVLA synthetic policy with Radeon Cosmos handoff

2026-09-24 01:12:27 +08:00

..

local

Run InternVLA synthetic policy with Radeon Cosmos handoff

2026-09-24 01:12:27 +08:00

__init__.py

[Edge][WP4] Add hardware adaptation: probe, derived overrides, auto profile, calibration

2026-09-09 11:01:40 +08:00

adapt.py

[Edge][WP8] Per-step overhead attribution, edge scheduling helpers, CPU thread sweep

2026-09-09 14:55:55 +08:00

auto_profile.py

[Edge][WP8] Per-step overhead attribution, edge scheduling helpers, CPU thread sweep

2026-09-09 14:55:55 +08:00

calibrate.py

[Edge][WP4] Add hardware adaptation: probe, derived overrides, auto profile, calibration

2026-09-09 11:01:40 +08:00

decoder_export.py

[Edge][WP4] Windowed Code2Wav decoder export (torch.export / ExecuTorch) + client backend; results summarizer

2026-09-09 15:02:19 +08:00

hardware_probe.py

[edge-infer] WIP snapshot: uncommitted M0/W1/W4/W5 edge work as of 2026-09-18

2026-09-18 09:47:29 +08:00

kv_budget.py

[edge-infer] WIP snapshot: uncommitted M0/W1/W4/W5 edge work as of 2026-09-18

2026-09-18 09:47:29 +08:00

memory.py

[Edge][CPU] Return the weight-loading churn to the OS: 7338 -> 3510 MB

2026-09-12 14:27:11 +08:00

npu_ryzenai.py

[edge-infer] WIP snapshot: uncommitted M0/W1/W4/W5 edge work as of 2026-09-18

2026-09-18 09:47:29 +08:00

placement.py

[edge-infer] WIP snapshot: uncommitted M0/W1/W4/W5 edge work as of 2026-09-18

2026-09-18 09:47:29 +08:00

qnn_export.py

[edge-infer] WIP snapshot: uncommitted M0/W1/W4/W5 edge work as of 2026-09-18

2026-09-18 09:47:29 +08:00

scheduling.py

[Edge][WP8] Per-step overhead attribution, edge scheduling helpers, CPU thread sweep

2026-09-09 14:55:55 +08:00

spark_export.py

[Edge][Mobile] Ring cache keeps the fused softmax so it survives w4a16

2026-09-11 17:41:29 +08:00

step_stats.py

[edge-infer] vllm_omni.windows: host-OS compatibility plugin for native Windows

2026-09-21 14:02:01 +08:00

topology.py

Integrate portable edge graph backends with Omni runtime

2026-09-22 13:12:33 +08:00

weight_path.py

edge: measure cpu_gemm_wna16 instead of asserting it away (I2)

2026-09-13 18:53:31 +08:00