• Joined on 2026-02-17
grenade opened issue helexa/helexa#99 2026-06-27 08:17:28 +00:00
F4e — cortex: vision cold-swap policy (27B-VL ⇄ frontier)
grenade opened issue helexa/helexa#98 2026-06-27 08:17:22 +00:00
F4d — perf(neuron): continuous batching for agentic fan-out
grenade opened issue helexa/helexa#97 2026-06-27 08:17:18 +00:00
F4c — perf(neuron): KV-cache quantization (measurement-gated)
grenade opened issue helexa/helexa#96 2026-06-27 08:17:13 +00:00
F4b — perf(neuron): wire speculative decoding into generation
grenade opened issue helexa/helexa#95 2026-06-27 08:17:08 +00:00
F4a — perf(neuron): enable flash-attention on full-attn layers
grenade opened issue helexa/helexa#94 2026-06-27 08:17:04 +00:00
F3 — bench: frontier A/B decision gate, pick resident model
grenade opened issue helexa/helexa#93 2026-06-27 08:17:00 +00:00
F2 — neuron/cortex: catalogue + load Qwen3-Next-80B-A3B family @ Q4_K_M on beast
grenade opened issue helexa/helexa#92 2026-06-27 08:16:54 +00:00
F1 — neuron: MoE FFN into the qwen3_5 hybrid layer stack + TP expert sharding
grenade opened issue helexa/helexa#91 2026-06-27 08:16:43 +00:00
O7 — helexa-bench: capability probe scenario (quality, not just speed)
grenade opened issue helexa/helexa#90 2026-06-27 08:16:38 +00:00
O6 — helexa-bench: cold-load + model-swap cost scenario
grenade opened issue helexa/helexa#89 2026-06-27 08:16:34 +00:00
O5 — helexa-bench: concurrency / agentic-load scenario
grenade opened issue helexa/helexa#88 2026-06-27 08:16:29 +00:00
O4 — helexa-bench: context-length scaling scenario (128 → 128k)
grenade opened issue helexa/helexa#87 2026-06-27 08:16:24 +00:00
O3 — helexa-bench: per-run VRAM high-water + GPU util/temp
grenade opened issue helexa/helexa#86 2026-06-27 08:16:19 +00:00
O2 — helexa-bench: raw per-sample retention + p50/p95/p99 (schema migration)
grenade opened issue helexa/helexa#85 2026-06-27 08:16:13 +00:00
O1 — neuron: emit prefill/decode timing breakdown in wire Finish event
grenade opened issue helexa/helexa#84 2026-06-27 08:15:59 +00:00
[epic] Reasoning frontier: 80B-A3B MoE on beast (qwen3_5 hybrid + high-sparsity MoE)
grenade opened issue helexa/helexa#83 2026-06-27 08:15:48 +00:00
[epic] Performance observability: per-phase profiling baseline before engine work
grenade deleted branch fix/responses-usage-reasoning-tokens from helexa/helexa 2026-06-26 18:30:10 +00:00
grenade pushed to main at helexa/helexa 2026-06-26 18:30:08 +00:00
66eb9f558f Merge pull request 'fix(neuron): surface reasoning_tokens in non-streaming /v1/responses usage' (#82) from fix/responses-usage-reasoning-tokens into main
f96a2e7ed3 fix(neuron): surface reasoning_tokens in non-streaming /v1/responses usage
Compare 2 commits »
grenade merged pull request helexa/helexa#82 2026-06-26 18:30:06 +00:00
fix(neuron): surface reasoning_tokens in non-streaming /v1/responses usage