bstein/titan-iac
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

hermes: strengthen Switchyard classifier
All checks were successful
Tests / Declarative: Post Actions passed: 244
Details

Browse Source
This commit is contained in:
jenkins 2026-08-11 23:29:04 -03:00
parent c8409b8f3a
commit 85b78405dd
3 changed files with 5 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
services/hermes/switchyard-configmap.yaml
View File

@ -79,7 +79,7 @@ data:
max_retries = 0 max_retries = 0
[targets.classifier] [targets.classifier]
id = "qwen2.5:3b-instruct-q4_0" id = "qwen2.5:14b-instruct-q4_0"
llm_client = "classifier" llm_client = "classifier"
[targets.local_qwen_low] [targets.local_qwen_low]

2
services/hermes/switchyard-deployment.yaml
View File

@ -19,7 +19,7 @@ spec:
labels: labels:
app: hermes-switchyard app: hermes-switchyard
annotations: annotations:
ai.bstein.dev/config-rev: "20260811-switchyard-authority-v7" ai.bstein.dev/config-rev: "20260811-switchyard-authority-v8"
prometheus.io/scrape: "true" prometheus.io/scrape: "true"
prometheus.io/port: "9005" prometheus.io/port: "9005"
prometheus.io/path: /metrics prometheus.io/path: /metrics

4
testing/tests/test_hermes_chat_quality.py
View File

@ -766,6 +766,8 @@ def test_local_flux_runtime_and_gpu_handoff_are_flux_managed():
switchyard = _documents(HERMES / "switchyard-configmap.yaml")[0]["data"][ switchyard = _documents(HERMES / "switchyard-configmap.yaml")[0]["data"][
"routes.toml" "routes.toml"
] ]
assert 'id = "qwen2.5:14b-instruct-q4_0"' in switchyard
assert "qwen2.5:3b-instruct-q4_0" not in switchyard
assert "route/local/qwen2.5-14b/medium" in switchyard assert "route/local/qwen2.5-14b/medium" in switchyard
model_gate = _documents(HERMES / "model-gate-configmap.yaml")[0]["data"][ model_gate = _documents(HERMES / "model-gate-configmap.yaml")[0]["data"][
"model_gate.py" "model_gate.py"
@ -774,7 +776,7 @@ def test_local_flux_runtime_and_gpu_handoff_are_flux_managed():
def test_titan20_serializes_classifier_and_local_chat_model_residency(): def test_titan20_serializes_classifier_and_local_chat_model_residency():
"""The two Qwen weights must not overcommit Xavier unified memory.""" """Classifier and local chat share one serialized resident Qwen weight."""
deployment = _documents( deployment = _documents(
Path(__file__).parents[2] / "services/ai-llm/deployment.yaml" Path(__file__).parents[2] / "services/ai-llm/deployment.yaml"
)[0] )[0]