Commit Graph

Select branches

Hide Pull Requests

main

Mono Color

• b1fdbde895 build: Rewrite artifacts/ when it is missing, and stop failing re-runs main MOVZXandHinata 2026-10-18 08:38:14 +00:00
• fe3cd9247b build: Reuse the extensions for a Python-only patch series MOVZXandHinata 2026-10-17 23:29:02 +00:00
• 50dabc9989 ple: drop the deferred state on a prefill-only step (avoid crashing) MOVZXandHinata 2026-10-18 06:23:27 +07:00
• d9ca3bc4f2 feat: Set concurrency and the CUDA graph list from .env MOVZXandHinata 2026-10-18 02:04:50 +07:00
• fab0eb9fc7 feat: Offload the vision tower to host RAM MOVZXandHinata 2026-10-18 01:49:04 +07:00
• 3a0a5d1510 feat: Set the engine stats interval to 1 s MOVZXandHinata 2026-10-16 06:01:27 +00:00
• c4d88acb4d feat: Accept any model name in a request MOVZXandHinata 2026-10-16 05:40:19 +00:00
• af239c4229 docs: Document the KV scale sidecar and the model variables MOVZXandHinata 2026-10-15 18:52:18 +00:00
• b89e85e99f feat: Mount the KV scale sidecar from any host path MOVZXandHinata 2026-10-15 18:52:18 +00:00
• 1281bf9c03 fix: Load the FP8 KV scales from the shipped file name MOVZXandHinata 2026-10-15 18:52:04 +00:00
• d6062059b4 docs: Add the Indonesian README and sync the guide with the code MOVZXandHinata 2026-10-15 07:52:20 +00:00
• 27f50debe0 feat: Add benchmark wrapper script MOVZXandHinata 2026-10-15 07:39:39 +00:00
• d365f4586f docs: Add power-limit benchmark results MOVZXandHinata 2026-10-15 07:26:46 +00:00
• 5cc9ee2505 chore: Reset vLLM logging interval back to default MOVZXandHinata 2026-10-15 06:37:23 +00:00
• 7e80cc67ee feat: Add pretty vLLM logging scripts MOVZXandHinata 2026-10-15 06:31:31 +00:00
• eb8c302bb3 fix: Adjust start script to avoid crashing on high pressure KV cache MOVZX 2026-10-15 06:18:06 +00:00
• 777e910dd0 Initial commit MOVZX 2026-10-15 01:01:52 +07:00