feat: GPU context 256K→128K fleet-wide + Genesis Hermes V3 on Strix Halo #20

Merged

jerome merged 5 commits from feat/gpu-128k-genesis-hermes-v3-20260717 into master 2026-07-17 11:21:32 +00:00

Conversation 0 Commits 5 Files Changed 5

+304 -155

abiba-bot commented 2026-07-17 11:09:27 +00:00

Owner

Copy Link

Copy Source

Lint passed. See commit b9149bce for full details.

Lint passed. See commit b9149bce for full details.

abiba-bot added 3 commits 2026-07-17 11:09:27 +00:00

Merge branch 'master' into feat/zulip-v3-resilience 24cd7f72ac

contract: update litellm-api-keys to v2026-07-17 — fleet standardization ... 65c99dab50

- Document canonical systemd drop-in pattern (ExecStart= reset + wrapper)
- Hardcode venv paths — never use variables in single-quoted bash -c
- Update migration status: all 4 agents on while-true wrapper + st.8e848433
- Infisical CLI update procedure (0.38.0 → 0.43.109 via artifacts-cli)
- Service token inventory + .env fallback inventory
- Key rotation log: fleet standardize + tanko fix-zulip entries
- Remove git merge conflict artifacts
- Add fleet-wide standardization lessons section

feat: GPU context 256K→128K fleet-wide + Genesis Hermes V3 on Strix Halo ... b9149bce47

GPU Changes:
- All 3 GPUs reduced from 256K (-c 262144) to 128K (-c 131072) for stability
- Observed instability near 100K at 256K — 128K is the stable ceiling
- VRAM improved: RTX 3090 ~70% (was 90%), RTX 5070 ~65% (was 88%)
- Strix Halo swapped to LuffyTheFox/Genesis Hermes V3 APEX
  - Hermes agent fine-tune, tensor repair (3 SSM layers, 76% W1 improvement)
  - Uncensored (0/465 refusals), multimodal (mmproj F16)
  - Speed: 65 tok/s gen, 140 tok/s prompt
  - Alias strix-moe maintained

Agent Updates:
- Mumuni: max_context_window 262144→131072, already aligned on strix-moe/0.65
- Tanko: max_context_window 262144→131072
- Koonimo: max_context_window + context_length 262144→131072
- CT114 SSH access confirmed (was 'Zulip only')

LiteLLM (CT116):
- Updated backend model references qwen3.6-35B-udq4→strix-moe
- Removed stale ornith-1.0-35b from model_cost
- Fallback chains updated

Contracts Updated:
- gpu-fleet.prose.md: topology, VRAM, benchmarks, config lines, model assignments
- gpu-self-heal.prose.md: Rule 9/10 context targets
- hermes-config-template.prose.md: template values, Rules 7-9, compression thresholds
- inference-optimization.prose.md: added to repo, 128K recommendation

Compression: 0.65 fires at ~85K (~43K headroom before 128K ceiling)
For >128K workloads: route to external providers (deepseek)

abiba-bot referenced this issue from a commit 2026-07-17 11:10:32 +00:00

ci: re-trigger pipeline for PR #20

abiba-bot added 1 commit 2026-07-17 11:10:32 +00:00

ci: re-trigger pipeline for PR #20 ba2c55c7e6

abiba-bot added 1 commit 2026-07-17 11:11:29 +00:00

fix: add missing description to inference-optimization frontmatter 4a1f476623

jerome merged commit 33cb88d571 into master 2026-07-17 11:21:32 +00:00

Sign in to join this conversation.

Reviewers

No Reviewers

Labels

No items

No labels

Milestone

No items

No Milestone

Projects

Clear projects

No projects

Assignees

Clear assignees

abiba-bot jerome (Jerome Kwame Tabiri) kagentz-bot kwame (Kwame Tabiri) mumuni-bot (Mumuni 🦅 (Syslog Falcon)) tanko-bot theo

No Assignees

1 Participants

Notifications

Subscribe

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: SyslogSolution/prose-contracts#20