grenade

grenade pushed to main at helexa/helexa

2026-05-26 13:10:45 +00:00

24e20dcb5c feat(catalogue,gateway): model aliases (helexa/small, helexa/balanced, helexa/large)

grenade pushed to main at helexa/helexa

2026-05-26 12:26:25 +00:00

becf61b9c1 feat(script): validate-neuron.sh waits for /health activation=ready

b9e7a76a7a feat(gateway): surface mid-prewarm models as Loading on /v1/models

Compare 2 commits »

grenade pushed to main at helexa/helexa

2026-05-26 12:18:07 +00:00

800498f530 feat(neuron): bind listener before pre-warm, surface activation in /health

grenade pushed to main at helexa/helexa

2026-05-26 11:05:57 +00:00

d3f2d50749 feat(deploy): per-host neuron config + pre-warm headline models

grenade pushed to main at helexa/helexa

2026-05-26 09:37:36 +00:00

2740e61a23 fix(neuron,candle): name lifetime on acquire_pool_lock

grenade pushed to main at helexa/helexa

2026-05-26 09:30:17 +00:00

67f79c868f fix(neuron,shutdown): time-bound unloads, fast-exit past tokio drain

fc6ef0ee0f feat(neuron,candle): detect CUDA context poisoning and refuse follow-ups

1385979e3d feat(neuron,candle): log per-device VRAM at chat_completion start

0a1cfcd4d0 feat(neuron,candle): req_id spans, terminal failure logs, pool-lock warnings

ea0e0f7911 fix(neuron,tp): log leader forward errors with full context

Compare 5 commits »

grenade pushed to main at helexa/helexa

2026-05-22 04:17:29 +00:00

aa88d37509 fix(gateway): full observability + stop leaking upstream bodies

grenade pushed to main at helexa/helexa

2026-05-22 04:10:44 +00:00

0f00f72b47 fix(router,handlers): strip trailing slash from rewritten URL + log upstream failures

grenade pushed to main at helexa/helexa

2026-05-22 03:23:52 +00:00

9b0ed0b57f fix(router): rewrite loopback inference URLs to use neuron's host

grenade pushed to main at helexa/helexa

2026-05-22 03:12:54 +00:00

dc2a803266 fix(rpm): migrate legacy helexa-cortex firewalld service to cortex

grenade pushed to main at helexa/helexa

2026-05-21 18:53:18 +00:00

e71181499e feat(stage-8e-3): quantize lm_head in TP Qwen3-Next

grenade pushed to main at helexa/helexa

2026-05-21 18:50:49 +00:00

ee663e5e99 fix(stage-8e-2e): bump quant prefill threshold to M > 64

grenade pushed to main at helexa/helexa

2026-05-21 18:15:36 +00:00

34f9b77d9d feat(stage-8e-2d): route quantized matmul by M (prefill vs decode)

grenade pushed to main at helexa/helexa

2026-05-21 17:05:23 +00:00

f084aaab8e fix(stage-8e-2c): cast bf16/f16 activations to f32 around QMatMul

grenade pushed to main at helexa/helexa

2026-05-21 16:17:17 +00:00

68a606a79c fix(stage-8e-2b): allow quant on the TP load path

grenade pushed to main at helexa/helexa

2026-05-21 15:03:41 +00:00

4aa71902d0 feat(stage-8e-2): plumb quant config from ModelSpec to TP load path

grenade pushed to main at helexa/helexa

2026-05-21 14:55:29 +00:00

bef159b21c feat(stage-8e-1): MaybeQuantLinear primitive + parallel-linear quant variants

grenade pushed to main at helexa/helexa

2026-05-21 14:49:38 +00:00

8d7b099b36 feat(stage-8d-7): direct safetensors fused-region loader

grenade pushed to main at helexa/helexa

2026-05-21 09:54:08 +00:00

89d98d1fb2 diag(stage-8d-6): per-layer VRAM logging in TP load path

grenade pushed to main at helexa/helexa

2026-05-21 08:52:41 +00:00

cc95fe28d9 feat(stage-8d-5b): wire fused_gdn_gating CUDA kernel