Smriti AI live dashboard

{% if dashboard.live_model.enabled %}Live Gemma 4 with external memory.{% else %}External memory for small models.{% endif %}

Inject facts, distract the model, inspect retrieved memories, trace graph facts, verify identity governance, and prove privacy deletion from one presentation-ready cockpit.

Run demo flow View benchmark evidence
Baseline recall {{ dashboard.major_results.baseline_recall }} Frozen Gemma 4, no memory
Smriti recall {{ dashboard.major_results.smriti_recall }} {{ dashboard.major_results.recall_lift }} over baseline
Broader rerun {{ dashboard.major_results.broader_delta }} {{ dashboard.major_results.historical_note }}
Training status Frozen {{ dashboard.training_status.detail }}

Capability map

Every upgrade, visible in one pass

The live UI exercises memory paths directly; the evidence table reads current Gemma 4 CSV artifacts.

{% for card in dashboard.feature_cards %}
{{ card.status }}

{{ card.title }}

{{ card.detail }}

{% endfor %}

Release and production readiness

Today’s hardening, visible

This panel surfaces the agent hardening layer, production-break suite, file audit, security acceptance, and release packaging state without exposing secrets or private memory.

Production guard Fail closed

No fake/mock/tiny model, synthetic official benchmark, unsafe harness, secret leak, or cross-user memory path is allowed in production.

{% for card in dashboard.readiness.cards %}
{{ card.status }}

{{ card.title }}

{{ card.detail }}

{% endfor %}
{% for item in dashboard.readiness.fail_closed %} {{ item }} {% endfor %}

Agentic harness evolution

Observable memory knobs

Active: {{ dashboard.harness.active_harness }}
Config
{{ dashboard.harness.config_path }}
Manifest
{{ dashboard.harness.manifest_path }}
Entries
{{ dashboard.harness.manifest_count }}
Registered
{{ dashboard.harness.registry|length }} harnesses
Production gate
{{ "passed" if dashboard.readiness.production_gates.passed else "review required" }}
Breaker suite
{{ dashboard.readiness.production_break.verdict or "missing" }}
Export report Compare with seed
{% for row in dashboard.harness.manifest %} {% else %} {% endfor %}
#ComponentChangeStatusRollback
{{ row.index }} {{ row.component }} {{ row.previous_value }} -> {{ row.new_value }} {{ row.status }}
No evolution entries yet. Run python benchmarks/collect_evidence.py then python evolve_harness.py.

Each automated or manual parameter change records before/after config snapshots, a reason, and predicted impact. The base model remains frozen.

Harness benchmark verdict {{ dashboard.harness.benchmark_readiness.verdict.replace("_", " ") }}

{{ dashboard.harness.benchmark_readiness.verdict_detail }}

{% for card in dashboard.harness.benchmark_readiness.cards %}
{{ card.label }} {{ card.value }} {{ card.detail }}
{% endfor %}
{% if dashboard.harness.benchmark_readiness.warnings %}
Industry-standard benchmark check

Internal production gates pass, but these results should not be positioned as full enterprise/industry proof until the warnings below are resolved with larger real-model, real-backend, and user-like benchmark runs.

    {% for warning in dashboard.harness.benchmark_readiness.warnings %}
  • {{ warning }}
    • {% endfor %}
{% endif %}
{% for row in dashboard.harness.validation_rows %} {% else %} {% endfor %}
SystemRecallPrecision@Kp50 msp95 msToken overheadDelete
{{ row.system }} {{ row.memory_recall_accuracy }} {{ row.retrieval_precision_at_k }} {{ row.p50_latency_ms }} {{ row.p95_latency_ms }} {{ row.token_overhead }} {{ row.privacy_delete_passed }}
No validation table yet. Run quick evaluation.
{% for row in dashboard.harness.cross_rows %} {% else %} {% endfor %}
Model familySeed recallEvolved recallp95 deltaToken deltaGate
{{ row.model_name }} {{ row.seed_harness_recall }} {{ row.evolved_harness_recall }} {{ row.p95_latency_change }} {{ row.token_overhead_change }} {{ row.production_threshold }}
No cross-model harness report yet.

Live memory flow

Inject, distract, recall

Ready
1 Inject 2 Distract 3 Recall 4 Inspect 5 Delete

Privacy and ops

Backend, deletion, API

Model mode
{{ dashboard.live_model.status }} / {{ dashboard.live_model.model }}
Device
{{ dashboard.live_model.device }}
Training
{{ dashboard.training_status.status }}
Release
{{ dashboard.release.pypi_target }}
PyPI
{{ dashboard.release.pypi_status }}
Guards
{{ dashboard.readiness.result_claims.valid and "claim provenance valid" or "claim provenance review" }}
Backend
{{ dashboard.backend.name }}
Storage
{{ dashboard.backend.storage }}
Encrypted
{{ "yes" if dashboard.backend.encrypted else "ready; set SMRITI_MEMORY_KEY" }}
Routes
{{ ", ".join(dashboard.api_routes) }}
HF repo
luciferai-devil/smriti-ai

Deletion clears the in-memory user store, durable backend record, and demo identity state.

{% if dashboard.public_demo %}

Public demo safety: do not enter real PII. Demo memory is automatically deleted after {{ dashboard.memory_ttl_seconds }} seconds of inactivity.

{% endif %}

Gemma 4 benchmark evidence

Saved comparison table

CSV-backed
{% for row in dashboard.benchmark_rows %} {% endfor %}
ConfigRecallLatencyTokens
{{ row.configuration_label }} {{ row.facts_recalled }}/3 {{ "%.0f" | format(row.average_latency_s | float * 1000) }} ms {{ "%.1f" | format(row.average_token_usage | float) }}

Historical +31.2% is not claimed as current. Current broader-protocol rerun is shown separately in the README/model card.