• Joined on 2026-02-17
grenade created branch feat/11-prefix-kv-cache in helexa/helexa 2026-06-12 14:14:12 +00:00
grenade pushed to feat/11-prefix-kv-cache at helexa/helexa 2026-06-12 14:14:12 +00:00
c5378d532d feat(neuron): prefix KV caching across requests — single-GPU + CPU paths (#11)
grenade pushed to main at gongfoo/gongfoo 2026-06-12 13:31:08 +00:00
168a57b3ff feat(runner): upgrade to gitea-runner 1.0.8 and repair auto-update
grenade pushed to main at gongfoo/gongfoo 2026-06-12 13:11:48 +00:00
4013c51429 fix(dashboard): host status shows stale when heartbeat exceeds 2 minutes
grenade deleted branch feat/gateway-24-anthropic-sse from helexa/helexa 2026-06-12 12:57:11 +00:00
grenade merged pull request helexa/helexa#33 2026-06-12 12:57:10 +00:00
feat(gateway): Anthropic streaming SSE translation (#24)
grenade closed issue helexa/helexa#24 2026-06-12 12:57:10 +00:00
feat(gateway): Anthropic streaming SSE translation
grenade pushed to main at helexa/helexa 2026-06-12 12:57:10 +00:00
9f383e7bc7 Merge pull request 'feat(gateway): Anthropic streaming SSE translation (#24)' (#33) from feat/gateway-24-anthropic-sse into main
569c528c4b feat(gateway): Anthropic streaming SSE translation (#24)
Compare 2 commits »
grenade pushed to main at gongfoo/gongfoo 2026-06-12 12:49:49 +00:00
eb8ae53c8f fix(deploy): start cert-watch path units at deploy, not next boot
grenade created pull request helexa/helexa#33 2026-06-12 12:47:57 +00:00
feat(gateway): Anthropic streaming SSE translation (#24)
grenade created branch feat/gateway-24-anthropic-sse in helexa/helexa 2026-06-12 12:47:34 +00:00
grenade pushed to feat/gateway-24-anthropic-sse at helexa/helexa 2026-06-12 12:47:34 +00:00
569c528c4b feat(gateway): Anthropic streaming SSE translation (#24)
grenade deleted branch feat/22-benchmark-harness from helexa/helexa 2026-06-12 12:46:37 +00:00
grenade pushed to main at helexa/helexa 2026-06-12 12:46:35 +00:00
06e4ffc25c Merge pull request 'feat(bench): reproducible benchmark harness + first fleet numbers (#22)' (#32) from feat/22-benchmark-harness into main
a2e73a8907 feat(bench): reproducible batch-1 benchmark harness + first fleet numbers (#22)
Compare 2 commits »
grenade merged pull request helexa/helexa#32 2026-06-12 12:46:34 +00:00
feat(bench): reproducible benchmark harness + first fleet numbers (#22)
grenade commented on issue helexa/helexa#22 2026-06-12 12:39:54 +00:00
bench: reproducible benchmark harness + published numbers

PR #32 lands the harness (script/bench.py) and the first published numbers (doc/benchmarks.md, 2026-06-12): 1.7B@3060 81 tok/s, 8B@4090 62 tok/s, 27B@2×5090 Q6K TP=2 at a steady 35 tok/s with…

grenade created pull request helexa/helexa#32 2026-06-12 12:39:34 +00:00
feat(bench): reproducible benchmark harness + first fleet numbers (#22)
grenade created branch feat/22-benchmark-harness in helexa/helexa 2026-06-12 12:39:16 +00:00
grenade pushed to feat/22-benchmark-harness at helexa/helexa 2026-06-12 12:39:16 +00:00
a2e73a8907 feat(bench): reproducible batch-1 benchmark harness + first fleet numbers (#22)
grenade opened issue helexa/helexa#31 2026-06-12 12:37:56 +00:00
feat(neuron): emit final usage frame on SSE streams (stream_options.include_usage)