A programmable Mixture-of-Models router for heterogeneous LLM inference
This report presents the forensic synthetic code analysis of vllm-project/semantic-router, a Go project with 5,388 GitHub stars. SynthScan v2.0 examined 1,358,589 lines of code across 5164 source files, recording 6271 pattern matches distributed across 24 syntactic categories. The overall adjusted score of 6.6 places this repository in the Low AI signal band.
The scanner applied 160+ deterministic lexical heuristics, multi-line block detectors, abstract syntax tree depth profilers, and a cross-file Jaccard similarity matrix to construct a statistically normalised synthetic code estimate. All matches are individually weighted by severity coefficient and contextual multiplier before summation, and the resulting headline score is temporally discounted to account for the repository's development history relative to the commercial emergence of large language model coding tooling (November 2022 onward).
This chart maps the temporal evolution of the adjusted synthetic code score across successive scan runs. An upward trajectory indicates ongoing incorporation of AI-generated code or expanding LLM-assisted scaffolding; a stable or declining trajectory may reflect active human refactoring, code removal, or the adoption of stricter authorship policies. The dashed secondary line (right axis) independently tracks total raw pattern hit count, which can diverge from the normalised score when codebase size changes significantly between scans.
Classifies detected patterns by their diagnostic confidence and structural impact. CRITICAL patterns (coefficient 10) represent definitive synthetic signatures — hallucinated imports, explicit LLM attribution metadata — virtually never produced by human authors. HIGH (5) indicates strong structural tells such as cross-file repetition or cross-linguistic idioms. MEDIUM (2) covers recognisable conversational padding and AI-specific vocabulary. LOW (1) captures subtle indicators like tautological comments and generic boilerplate that require density to carry independent signal.
This horizontal bar chart decomposes the repository's raw synthetic code score by top-level directory, allowing you to pinpoint precisely which modules or components carry the highest AI authorship density. Directories with disproportionately high scores relative to their size warrant targeted manual review: concentrated AI signatures often trace back to mass-generated configuration layers, auto-ported test suites, LLM-scaffolded boilerplate classes, or entire subsystems authored under heavy copilot assistance. Use this view to prioritise your human code-review effort.
The scanner identified 6271 distinct pattern matches across 24 syntactic categories. Each entry below represents a discrete location in the source code where the engine recorded a statistically significant AI authorship indicator. Expand any category row to inspect the individual file paths, line numbers, code snippets, and the lexical context (CODE, COMMENT, or STRING) in which each match was detected.
Reading the findings table: The Severity column indicates the diagnostic confidence level (CRITICAL / HIGH / MEDIUM / LOW). The Context column identifies whether the match occurred inside executable code, an inline comment, or a string literal — comment-context matches receive a ×1.5 weight because LLMs systematically over-annotate. The ⚡ bolt icon marks clustered matches: three or more patterns within a 10-line window, each receiving an additional ×1.5 density multiplier as dense clusters constitute far stronger evidence of synthetic authorship than isolated hits.
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | bench/agentic_routing_live_benchmark.py | 547 | def missing_router_header_counts(rows: list[dict[str, Any]]) -> dict[str, int]: | CODE |
| LOW | bench/agentic_routing_live_benchmark.py | 554 | def invalid_router_header_counts(rows: list[dict[str, Any]]) -> dict[str, int]: | CODE |
| LOW | bench/agentic_routing_live_benchmark.py | 567 | def valid_router_header_value(header: str, value: str) -> bool: | CODE |
| LOW | bench/agentic_routing_live_benchmark.py | 667 | def attach_validation_failures( | CODE |
| LOW | bench/agentic_routing_live_benchmark.py | 714 | def render_comparison_markdown(comparison: dict[str, Any]) -> str: | CODE |
| LOW | bench/plot_session_routing_figures.py | 225 | def render_synthetic_headline(overall: dict[str, Any], output: Path, dpi: int) -> None: | CODE |
| LOW | bench/plot_session_routing_figures.py | 358 | def render_ablation_cost_quality( | CODE |
| LOW | bench/plot_session_routing_figures.py | 660 | def agent_task_readiness_metrics( | CODE |
| LOW | bench/plot_session_routing_figures.py | 712 | def render_agent_task_readiness( | CODE |
| LOW | bench/agent_task_live_benchmark.py | 2337 | def missing_router_header_counts(rows: list[dict[str, Any]]) -> dict[str, int]: | CODE |
| LOW | bench/agent_task_live_benchmark.py | 2344 | def invalid_router_header_counts(rows: list[dict[str, Any]]) -> dict[str, int]: | CODE |
| LOW | bench/agent_task_live_benchmark.py | 2357 | def valid_router_header_value(header: str, value: str) -> bool: | CODE |
| LOW | bench/agent_task_live_benchmark.py | 2494 | def attach_validation_failures( | CODE |
| LOW | bench/agent_task_live_benchmark.py | 2606 | def render_comparison_markdown(comparison: dict[str, Any]) -> str: | CODE |
| LOW | bench/session_routing_branch_image_benchmark.py | 251 | def add_evidence_identity_failures( | CODE |
| LOW | bench/session_routing_branch_image_benchmark.py | 324 | def add_cache_probe_identity_failures( | CODE |
| LOW | bench/session_routing_branch_image_benchmark.py | 333 | def add_cache_probe_pair_failures( | CODE |
| LOW | bench/session_routing_branch_image_benchmark.py | 360 | def add_cache_probe_serving_path_failures( | CODE |
| LOW | bench/session_routing_branch_image_benchmark.py | 457 | def evaluate_failure_aggregate( | CODE |
| LOW | bench/session_routing_branch_image_benchmark.py | 513 | def evaluate_agent_task_summary( | CODE |
| LOW | bench/session_routing_branch_image_benchmark.py | 694 | def has_mounted_binary_marker(*values: str) -> bool: | CODE |
| LOW | bench/agentic_routing_experiment.py | 473 | def agentic_boundary_decision( | CODE |
| LOW | bench/agentic_routing_experiment.py | 805 | def summarize_baseline_policy(rows: list[dict[str, Any]]) -> dict[str, Any]: | CODE |
| LOW | bench/agentic_routing_experiment.py | 905 | def learning_architecture_fixtures() -> list[dict[str, Any]]: | CODE |
| LOW | bench/agentic_routing_experiment.py | 1078 | def summarize_learning_architecture(rows: list[dict[str, Any]]) -> dict[str, Any]: | CODE |
| LOW | bench/agentic_routing_experiment.py | 1178 | def write_learning_architecture_outputs(out_dir: Path) -> dict[str, Any]: | CODE |
| LOW | bench/agentic_routing_experiment.py | 1280 | def render_learning_architecture_report( | CODE |
| LOW | bench/test_cache_token_probe.py | 70 | def test_cached_token_extraction_handles_missing_zero_and_positive_values(): | CODE |
| LOW | bench/test_cache_token_probe.py | 104 | def test_usage_extraction_handles_chat_and_responses_token_names(): | CODE |
| LOW | bench/test_cache_token_probe.py | 127 | def test_cache_reporting_state_distinguishes_missing_zero_and_positive(): | CODE |
| LOW | bench/test_cache_token_probe.py | 138 | def test_summary_reports_cached_ratio_and_reporting_state(): | CODE |
| LOW | bench/test_cache_token_probe.py | 166 | def test_summary_records_repeated_prefix_probe_metadata(): | CODE |
| LOW | bench/test_cache_token_probe.py | 179 | def test_validate_summary_gates_positive_cache_evidence(): | CODE |
| LOW | bench/test_cache_token_probe.py | 197 | def test_validate_summary_fails_missing_or_zero_cache_evidence(): | CODE |
| LOW | bench/test_cache_token_probe.py | 226 | def test_validate_summary_fails_low_success_rate(): | CODE |
| LOW | bench/test_cache_token_probe.py | 237 | def test_validate_baseline_identity_rejects_same_base_url(): | CODE |
| LOW | bench/test_cache_token_probe.py | 259 | def test_validate_baseline_identity_accepts_direct_backend_url(): | CODE |
| LOW | bench/test_cache_token_probe.py | 279 | def test_aggregate_summary_output_matches_ga_report_shape(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 53 | def complete_branch_image_summary() -> dict: | CODE |
| LOW | bench/test_session_routing_ga_report.py | 96 | def complete_agent_task_summary() -> dict: | CODE |
| LOW | bench/test_session_routing_ga_report.py | 235 | def test_complete_evidence_passes_ga_gate(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 282 | def test_missing_positive_cache_and_branch_image_block_ga(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 332 | def test_positive_cache_without_probe_metadata_blocks_ga(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 380 | def test_positive_cache_without_direct_backend_baseline_blocks_ga(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 431 | def test_positive_cache_requires_router_baseline_probe_identity_match(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 493 | def test_positive_cache_requires_direct_backend_serving_path_identity(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 555 | def test_cache_runner_validation_failures_block_ga(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 601 | def test_agent_task_runner_validation_failures_block_ga(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 643 | def test_stale_agent_task_suite_blocks_ga(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 713 | def test_diagnostic_probe_does_not_satisfy_branch_image_benchmark(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 771 | def test_branch_image_summary_requires_all_child_checks(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 813 | def test_branch_image_summary_requires_assembler_identity_and_evidence(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 861 | def test_branch_image_summary_blocks_expected_identity_mismatch(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 910 | def test_missing_initial_policy_baseline_blocks_ga(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 959 | def test_main_writes_report_and_respects_allow_blockers(tmp_path): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 983 | def test_main_stdout_lists_blockers_for_maintainer_ops(tmp_path, capsys): | CODE |
| LOW | bench/test_session_routing_ga_report.py | 1069 | def test_markdown_renders_blockers_as_readable_sections(): | CODE |
| LOW | bench/test_session_routing_branch_image_benchmark.py | 106 | def complete_agent_task_summary( | CODE |
| LOW | bench/test_session_routing_branch_image_benchmark.py | 216 | def test_full_branch_image_summary_passes_with_complete_evidence(tmp_path): | CODE |
| LOW | bench/test_session_routing_branch_image_benchmark.py | 230 | def test_full_branch_image_summary_blocks_mounted_binary_and_stale_tasks(tmp_path): | CODE |
| 2459 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | bench/cpu-vs-gpu/bench-3way.sh | 4 | # ============================================================================= | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-3way.sh | 12 | # ============================================================================= | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-3way.sh | 34 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-3way.sh | 60 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-3way.sh | 152 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-3way.sh | 182 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-3way.sh | 224 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-3way.sh | 259 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-3way.sh | 369 | # ============================================================================= | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 4 | # ============================================================================= | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 9 | # ============================================================================= | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 32 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 35 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 193 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 195 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 229 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 231 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 292 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 294 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 330 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 332 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 353 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 355 | # --------------------------------------------------------------------------- | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 482 | # ============================================================================= | COMMENT |
| MEDIUM | bench/cpu-vs-gpu/bench-long-context.sh | 484 | # ============================================================================= | COMMENT |
| MEDIUM | tools/smoke/test-signal-api.sh | 13 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-signal-api.sh | 15 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-signal-api.sh | 78 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-signal-api.sh | 80 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-signal-api.sh | 134 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-signal-api.sh | 136 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-signal-api.sh | 186 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-signal-api.sh | 188 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-signal-api.sh | 466 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-signal-api.sh | 468 | # ============================================================================ | COMMENT |
| MEDIUM⚡ | tools/smoke/test-authz-rbac.sh | 28 | # ─── Configurable parameters (from env, with defaults) ─────────────── | COMMENT |
| MEDIUM⚡ | tools/smoke/test-authz-rbac.sh | 38 | # ─── Dynamic discovery ─────────────────────────────────────────────── | COMMENT |
| MEDIUM | tools/smoke/test-authz-rbac.sh | 103 | # ─── Helpers ────────────────────────────────────────────────────────── | COMMENT |
| MEDIUM⚡ | tools/smoke/test-authz-rbac.sh | 175 | # ─── Test 1: Admin → 14B ───────────────────────────────────────────── | COMMENT |
| MEDIUM⚡ | tools/smoke/test-authz-rbac.sh | 181 | # ─── Test 2: Alice (engineering/pro_tier) → 7B ─────────────────────── | COMMENT |
| MEDIUM⚡ | tools/smoke/test-authz-rbac.sh | 187 | # ─── Test 3: Bob (premium) + complex query → 14B ───────────────────── | COMMENT |
| MEDIUM⚡ | tools/smoke/test-authz-rbac.sh | 193 | # ─── Test 4: Bob (premium) + simple query → 7B ─────────────────────── | COMMENT |
| MEDIUM⚡ | tools/smoke/test-authz-rbac.sh | 199 | # ─── Test 5: Carol (free) → 7B ─────────────────────────────────────── | COMMENT |
| MEDIUM⚡ | tools/smoke/test-authz-rbac.sh | 205 | # ─── Test 6: Dave (contractor/free_tier) → 7B ──────────────────────── | COMMENT |
| MEDIUM⚡ | tools/smoke/test-authz-rbac.sh | 211 | # ─── Test 7: Invalid token → 401 ───────────────────────────────────── | COMMENT |
| MEDIUM⚡ | tools/smoke/test-authz-rbac.sh | 217 | # ─── Test 8: No token → 401 ────────────────────────────────────────── | COMMENT |
| MEDIUM⚡ | tools/smoke/test-authz-rbac.sh | 226 | # ─── Summary ────────────────────────────────────────────────────────── | COMMENT |
| MEDIUM⚡ | tools/smoke/test-classification-api.sh | 192 | # ============================================================================ | COMMENT |
| MEDIUM⚡ | tools/smoke/test-classification-api.sh | 194 | # ============================================================================ | COMMENT |
| MEDIUM⚡ | tools/smoke/test-classification-api.sh | 198 | # ----------------------------------------------------------------------------- | COMMENT |
| MEDIUM⚡ | tools/smoke/test-classification-api.sh | 200 | # ----------------------------------------------------------------------------- | COMMENT |
| MEDIUM | tools/smoke/test-classification-api.sh | 12 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-classification-api.sh | 14 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-classification-api.sh | 84 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-classification-api.sh | 86 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-classification-api.sh | 140 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-classification-api.sh | 142 | # ============================================================================ | COMMENT |
| MEDIUM | tools/smoke/test-classification-api.sh | 386 | # ----------------------------------------------------------------------------- | COMMENT |
| MEDIUM | tools/smoke/test-classification-api.sh | 388 | # ----------------------------------------------------------------------------- | COMMENT |
| MEDIUM | tools/smoke/test-classification-api.sh | 460 | # ----------------------------------------------------------------------------- | COMMENT |
| 587 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW⚡ | deploy/kserve/install-gpu-operator.sh | 29 | # Step 1: Install Node Feature Discovery (NFD) Operator | COMMENT |
| LOW | deploy/kserve/install-gpu-operator.sh | 81 | # Step 2: Create NFD Instance | COMMENT |
| LOW⚡ | deploy/kserve/install-gpu-operator.sh | 122 | # Step 3: Install NVIDIA GPU Operator | COMMENT |
| LOW | deploy/kserve/install-gpu-operator.sh | 175 | # Step 4: Create ClusterPolicy | COMMENT |
| LOW | deploy/kserve/install-gpu-operator.sh | 235 | # Step 5: Verify GPU Resources | COMMENT |
| LOW | website/docs/installation/k8s/agentgateway.md | 39 | ## Step 1: Create Kind Cluster (Optional) | COMMENT |
| LOW | website/docs/installation/k8s/agentgateway.md | 50 | ## Step 2: Install agentgateway | COMMENT |
| LOW | website/docs/installation/k8s/agentgateway.md | 73 | ## Step 3: Create an agentgateway proxy | COMMENT |
| LOW | website/docs/installation/k8s/agentgateway.md | 100 | ## Step 4: Deploy Demo LLM | COMMENT |
| LOW | website/docs/installation/k8s/agentgateway.md | 173 | ## Step 5: Deploy vLLM Semantic Router | COMMENT |
| LOW | website/docs/installation/k8s/agentgateway.md | 193 | ## Step 6: Create agentgateway routing resources | COMMENT |
| LOW | website/docs/installation/k8s/agentgateway.md | 231 | ## Step 7: Attach Semantic Router as ExtProc | COMMENT |
| LOW | website/docs/installation/k8s/istio.md | 42 | ## Step 1: Verify the cluster | COMMENT |
| LOW | website/docs/installation/k8s/istio.md | 48 | ## Step 2: Deploy LLM models | COMMENT |
| LOW | website/docs/installation/k8s/istio.md | 79 | ## Step 3: Install Gateway API and Istio | COMMENT |
| LOW | website/docs/installation/k8s/istio.md | 100 | ## Step 4: Update vsr config (Optional) | COMMENT |
| LOW | website/docs/installation/k8s/istio.md | 111 | ## Step 5: Deploy vLLM Semantic Router | COMMENT |
| LOW | website/docs/installation/k8s/istio.md | 138 | ## Step 6: Install additional Istio configuration | COMMENT |
| LOW | website/docs/installation/k8s/istio.md | 156 | ## Step 7: Install gateway routes | COMMENT |
| LOW | website/docs/installation/k8s/istio.md | 166 | ## Step 8: Testing the Deployment | COMMENT |
| LOW | website/docs/installation/k8s/ai-gateway.md | 47 | ## Step 1: Create Kind Cluster (Optional) | COMMENT |
| LOW | website/docs/installation/k8s/ai-gateway.md | 58 | ## Step 2: Deploy vLLM Semantic Router | COMMENT |
| LOW | website/docs/installation/k8s/ai-gateway.md | 83 | ## Step 3: Install Envoy Gateway | COMMENT |
| LOW | website/docs/installation/k8s/ai-gateway.md | 100 | ## Step 4: Install Envoy AI Gateway | COMMENT |
| LOW | website/docs/installation/k8s/ai-gateway.md | 121 | ## Step 5: Deploy Demo LLM | COMMENT |
| LOW | website/docs/installation/k8s/ai-gateway.md | 130 | ## Step 6: Create Gateway API Resources | COMMENT |
| LOW⚡ | website/blog/2026-01-23-mom-on-amd-gpu.md | 270 | ### Step 1: Install vLLM-SR | COMMENT |
| LOW⚡ | website/blog/2026-01-23-mom-on-amd-gpu.md | 278 | ### Step 2: Initialize Configuration | COMMENT |
| LOW⚡ | website/blog/2026-01-23-mom-on-amd-gpu.md | 286 | ### Step 3: Deploy vLLM on AMD GPU | COMMENT |
| LOW | website/blog/2026-01-23-mom-on-amd-gpu.md | 322 | ### Step 4: Start the Semantic Router | COMMENT |
| LOW | website/blog/2026-01-23-mom-on-amd-gpu.md | 331 | ### Step 5: Test It | COMMENT |
| LOW | …site/blog/2026-03-25-vllm-sr-on-amd-developer-cloud.md | 107 | ### Step 1: Create the Shared Docker Network | COMMENT |
| LOW | …site/blog/2026-03-25-vllm-sr-on-amd-developer-cloud.md | 117 | ### Step 2: Start the AMD ROCm vLLM Backend | COMMENT |
| LOW⚡ | …ite/versioned_docs/version-v0.1/installation/milvus.md | 356 | # Step 1: Deploy Milvus (using Helm for simplicity) | COMMENT |
| LOW⚡ | …ite/versioned_docs/version-v0.1/installation/milvus.md | 365 | # Step 2: Wait for ready | COMMENT |
| LOW⚡ | …ite/versioned_docs/version-v0.1/installation/milvus.md | 369 | # Step 3: Update router config (set backend_type: "milvus") | COMMENT |
| LOW⚡ | …ite/versioned_docs/version-v0.1/installation/milvus.md | 372 | # Step 4: Restart router | COMMENT |
| LOW | …/versioned_docs/version-v0.1/installation/k8s/llm-d.md | 35 | ## Step 1: Common Steps from Istio Guide | COMMENT |
| LOW | …/versioned_docs/version-v0.1/installation/k8s/llm-d.md | 39 | ## Step 2: Install Istio Gateway, Gateway API, Inference Extension CRDs | COMMENT |
| LOW | …/versioned_docs/version-v0.1/installation/k8s/llm-d.md | 61 | ## Step 3: Deploy LLM models | COMMENT |
| LOW | …/versioned_docs/version-v0.1/installation/k8s/llm-d.md | 100 | ## Step 4: Deploy InferencePools and LLM-D schedulers | COMMENT |
| LOW | …/versioned_docs/version-v0.1/installation/k8s/llm-d.md | 118 | ## Step 5: Additional Istio config for LLM-D connection | COMMENT |
| LOW | …/versioned_docs/version-v0.1/installation/k8s/llm-d.md | 132 | ## Step 6: Update vsr config | COMMENT |
| LOW | …/versioned_docs/version-v0.1/installation/k8s/llm-d.md | 136 | ## Step 7: Deploy vLLM Semantic Router | COMMENT |
| LOW⚡ | …/versioned_docs/version-v0.1/installation/k8s/llm-d.md | 151 | ## Step 6: Additional Istio configuration for the VSR connection | COMMENT |
| LOW⚡ | …/versioned_docs/version-v0.1/installation/k8s/llm-d.md | 160 | ## Step 7: Install gateway routes | COMMENT |
| LOW⚡ | …/versioned_docs/version-v0.1/installation/k8s/llm-d.md | 169 | ## Step 8: Testing the Deployment | COMMENT |
| LOW | …docs/version-v0.1/installation/k8s/production-stack.md | 39 | ## Step 1: Deploy the vLLM Production Stack using your Helm values | COMMENT |
| LOW | …docs/version-v0.1/installation/k8s/production-stack.md | 94 | ## Step 2: Deploy vLLM Semantic Router and point it at your vLLM router Service | COMMENT |
| LOW | …docs/version-v0.1/installation/k8s/production-stack.md | 147 | ## Step 3: Test the deployment | COMMENT |
| LOW | …versioned_docs/version-v0.1/installation/k8s/aibrix.md | 42 | ## Step 1: Create Kind Cluster (Optional) | COMMENT |
| LOW | …versioned_docs/version-v0.1/installation/k8s/aibrix.md | 53 | ## Step 2: Deploy vLLM Semantic Router | COMMENT |
| LOW | …versioned_docs/version-v0.1/installation/k8s/aibrix.md | 75 | ## Step 3: Install vLLM AIBrix | COMMENT |
| LOW | …versioned_docs/version-v0.1/installation/k8s/aibrix.md | 91 | ## Step 4: Deploy Demo LLM | COMMENT |
| LOW | …versioned_docs/version-v0.1/installation/k8s/aibrix.md | 102 | ## Step 5: Create Gateway API Resources | COMMENT |
| LOW | …/versioned_docs/version-v0.1/installation/k8s/istio.md | 27 | ## Step 1: Create Minikube Cluster | COMMENT |
| LOW | …/versioned_docs/version-v0.1/installation/k8s/istio.md | 44 | ## Step 2: Deploy LLM models | COMMENT |
| LOW | …/versioned_docs/version-v0.1/installation/k8s/istio.md | 83 | ## Step 3: Install Istio Gateway, Gateway API, Inference Extension CRDs | COMMENT |
| LOW | …/versioned_docs/version-v0.1/installation/k8s/istio.md | 105 | ## Step 4: Update vsr config | COMMENT |
| LOW | …/versioned_docs/version-v0.1/installation/k8s/istio.md | 109 | ## Step 5: Deploy vLLM Semantic Router | COMMENT |
| 446 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | bench/agentic_routing_live_benchmark.py | 12 | CODE | |
| LOW | bench/plot_session_routing_figures.py | 10 | CODE | |
| LOW | bench/agent_task_live_benchmark.py | 4 | CODE | |
| LOW | bench/session_routing_branch_image_benchmark.py | 10 | CODE | |
| LOW | bench/agentic_routing_experiment.py | 9 | CODE | |
| LOW | bench/openai_fault_proxy.py | 4 | CODE | |
| LOW | bench/session_routing_ga_report.py | 4 | CODE | |
| LOW | bench/cache_token_probe.py | 4 | CODE | |
| LOW | bench/session_routing_branch_image_probe.py | 4 | CODE | |
| LOW | bench/hallucination/datasets.py | 12 | CODE | |
| LOW | bench/hallucination/__init__.py | 7 | CODE | |
| LOW | bench/hallucination/__init__.py | 8 | CODE | |
| LOW | bench/hallucination/__init__.py | 8 | CODE | |
| LOW | bench/hallucination/__init__.py | 8 | CODE | |
| LOW | bench/hallucination/evaluate.py | 29 | CODE | |
| LOW | bench/grounded_fusion/metrics.py | 9 | CODE | |
| LOW | bench/grounded_fusion/runner.py | 9 | CODE | |
| LOW | bench/grounded_fusion/ollama_proxy.py | 17 | CODE | |
| LOW | bench/grounded_fusion/datasets.py | 18 | CODE | |
| LOW | bench/grounded_fusion/grade_only.py | 16 | CODE | |
| LOW | bench/grounded_fusion/compare_multiarm.py | 22 | CODE | |
| LOW | bench/grounded_fusion/llm_client.py | 12 | CODE | |
| LOW | bench/grounded_fusion/rubric_judge.py | 17 | CODE | |
| LOW | bench/grounded_fusion/evaluate.py | 20 | CODE | |
| LOW | bench/grounded_fusion/make_configs.py | 15 | CODE | |
| LOW | bench/grounded_fusion/sanitize_results.py | 14 | CODE | |
| LOW | bench/grounded_fusion/items.py | 13 | CODE | |
| LOW | bench/grounded_fusion/compare.py | 16 | CODE | |
| LOW | bench/reasoning/plotting.py | 1 | CODE | |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 24 | CODE | |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 27 | CODE | |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 32 | CODE | |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 32 | CODE | |
| LOW | bench/reasoning/reasoning_mode_eval.py | 16 | CODE | |
| LOW | bench/reasoning/__init__.py | 28 | CODE | |
| LOW | bench/reasoning/__init__.py | 28 | CODE | |
| LOW | bench/reasoning/__init__.py | 29 | CODE | |
| LOW | bench/reasoning/__init__.py | 29 | CODE | |
| LOW | bench/reasoning/__init__.py | 29 | CODE | |
| LOW | bench/reasoning/__init__.py | 29 | CODE | |
| LOW | bench/reasoning/__init__.py | 32 | CODE | |
| LOW | bench/reasoning/__init__.py | 32 | CODE | |
| LOW | bench/reasoning/dataset_factory.py | 8 | CODE | |
| LOW | bench/reasoning/dataset_factory.py | 18 | CODE | |
| LOW | bench/reasoning/canonical_patch.py | 1 | CODE | |
| LOW | bench/reasoning/reporting.py | 1 | CODE | |
| LOW | bench/reasoning/dataset_implementations/__init__.py | 3 | CODE | |
| LOW | bench/reasoning/dataset_implementations/__init__.py | 3 | CODE | |
| LOW | bench/reasoning/dataset_implementations/__init__.py | 3 | CODE | |
| LOW | bench/reasoning/dataset_implementations/__init__.py | 4 | CODE | |
| LOW | bench/reasoning/dataset_implementations/__init__.py | 5 | CODE | |
| LOW | bench/reasoning/dataset_implementations/__init__.py | 5 | CODE | |
| LOW | bench/reasoning/dataset_implementations/__init__.py | 5 | CODE | |
| LOW | bench/reasoning/dataset_implementations/__init__.py | 5 | CODE | |
| LOW | bench/reasoning/dataset_implementations/__init__.py | 11 | CODE | |
| LOW | bench/reasoning/dataset_implementations/__init__.py | 12 | CODE | |
| LOW | bench/reasoning/dataset_implementations/__init__.py | 12 | CODE | |
| LOW | bench/reasoning/dataset_implementations/__init__.py | 13 | CODE | |
| LOW | …/dataset_implementations/openmathrreasoning_dataset.py | 12 | CODE | |
| LOW | bench/router_flow/flow_eval.py | 10 | CODE | |
| 586 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | .pre-commit-config.yaml | 121 | # Commented out flake8 - only reports issues, doesn't auto-fix | COMMENT |
| LOW | bench/benchmark_comparison.sh | 1 | #!/bin/bash | COMMENT |
| LOW | bench/reasoning_mode_eval.sh | 1 | #!/bin/bash | COMMENT |
| LOW | bench/grounded_fusion/run_ab.sh | 1 | #!/usr/bin/env bash | COMMENT |
| LOW | bench/grounded_fusion/run_sweep.sh | 1 | #!/usr/bin/env bash | COMMENT |
| LOW | …/scripts/rust/candle-binding/continuous_batch_bench.rs | 1 | // Continuous Batch Scheduler Benchmark | COMMENT |
| LOW | …ipts/rust/candle-binding/pure_batch_inference_bench.rs | 1 | // Pure Batched Inference Benchmark (No Scheduler) | COMMENT |
| LOW | tools/smoke/test-authz-rbac.sh | 1 | #!/usr/bin/env bash | COMMENT |
| LOW | tools/mcp-classifier-server/server_generative.py | 61 | COMMENT | |
| LOW | openvino-binding/semantic-router.go | 101 | COMMENT | |
| LOW | …o-binding/cpp/src/ffi/openvino_semantic_router_ffi.cpp | 1 | /** | COMMENT |
| LOW | …nding/scripts/build_and_run_mmbert_classifier_bench.sh | 1 | #!/usr/bin/env bash | COMMENT |
| LOW | …inding/scripts/build_and_run_mmbert_embedding_bench.sh | 1 | #!/usr/bin/env bash | COMMENT |
| LOW | ml-binding/ml_binding.go | 1 | // Package ml_binding provides Go bindings for Linfa-based traditional ML algorithms. | COMMENT |
| LOW | ml-binding/src/lib.rs | 1 | //! ML Binding for Semantic Router | COMMENT |
| LOW | ml-binding/src/svm.rs | 1 | //! SVM (Support Vector Machine) inference implementation | COMMENT |
| LOW | config/runtime/vector-store/valkey.yaml | 1 | # Valkey Vector Store — Example Configuration | COMMENT |
| LOW | config/runtime/vector-store/valkey.yaml | 41 | metric_type: COSINE | COMMENT |
| LOW | config/runtime/vector-store/valkey.yaml | 61 | # valkey: | COMMENT |
| LOW | config/runtime/memory/valkey.yaml | 1 | # Valkey Memory Store Configuration for Agentic Memory | COMMENT |
| LOW | config/runtime/memory/valkey.yaml | 41 | # backend: valkey # <-- select Valkey backend | COMMENT |
| LOW | config/runtime/memory/valkey.yaml | 61 | # hybrid_mode: rerank | COMMENT |
| LOW | config/runtime/memory/valkey.yaml | 81 | # Production (Docker / Kubernetes): | COMMENT |
| LOW | config/runtime/response-api/redis-cluster.yaml | 41 | COMMENT | |
| LOW | config/runtime/response-api/redis-cluster.yaml | 61 | # - Each master should have at least 1 replica for high availability | COMMENT |
| LOW | config/runtime/response-api/redis-simple.yaml | 21 | # Timeouts (in seconds) | COMMENT |
| LOW | config/runtime/response-api/redis-simple.yaml | 41 | # response_api: | COMMENT |
| LOW | config/runtime/response-api/redis-production.yaml | 41 | # tls_enabled: true | COMMENT |
| LOW | config/runtime/response-api/redis-production.yaml | 61 | COMMENT | |
| LOW | config/runtime/response-cache/milvus.yaml | 1 | # Milvus Vector Database Configuration for Semantic Cache | COMMENT |
| LOW | config/runtime/response-cache/milvus.yaml | 141 | # Example configurations for different environments: | COMMENT |
| LOW | config/runtime/response-cache/milvus.yaml | 161 | # enabled: true | COMMENT |
| LOW | config/runtime/response-cache/valkey.yaml | 1 | # Valkey Vector Database Configuration for Semantic Cache | COMMENT |
| LOW | config/runtime/response-cache/qdrant.yaml | 1 | # Qdrant Vector Search Engine Configuration for Semantic Cache | COMMENT |
| LOW | config/runtime/response-cache/redis.yaml | 1 | # Redis Vector Database Configuration for Semantic Cache | COMMENT |
| LOW | config/runtime/response-cache/redis.yaml | 81 | drop_index_on_startup: true # Enable for development to test dynamic dimensions | COMMENT |
| LOW | config/runtime/response-cache/redis.yaml | 101 | # Production (Redis Enterprise Cloud): | COMMENT |
| LOW | config/runtime/response-cache/redis.yaml | 121 | # - For HNSW index: | COMMENT |
| LOW | deploy/openshift/demo/curl-examples.sh | 1 | #!/bin/bash | COMMENT |
| LOW | deploy/openshift/demo/live-semantic-router-logs.sh | 1 | #!/bin/bash | COMMENT |
| LOW | …s/dynamo/dynamo-resources/dynamo-graph-deployment.yaml | 1 | --- | COMMENT |
| LOW | deploy/kubernetes/dynamo/helm-chart/values.yaml | 1 | # Dynamo vLLM Helm Chart Configuration | COMMENT |
| LOW | deploy/kubernetes/dynamo/helm-chart/values.yaml | 21 | huggingface: | COMMENT |
| LOW | …etes/dynamo/helm-chart/examples/values-multi-node.yaml | 1 | # Example: Multi-Node Deployment with Node Affinity | COMMENT |
| LOW | deploy/helm/semantic-router/values.yaml | 1 | # Default values for semantic-router. | COMMENT |
| LOW | deploy/helm/semantic-router/values.yaml | 41 | # Pod security context | COMMENT |
| LOW | deploy/helm/semantic-router/values.yaml | 121 | # No initContainer is needed anymore | COMMENT |
| LOW | deploy/helm/semantic-router/values.yaml | 161 | # Affinity rules | COMMENT |
| LOW | deploy/helm/semantic-router/values.yaml | 181 | # -- Mount an existing ConfigMap over /app/config/knowledge_bases (overrides image-bundled seeds when set) | COMMENT |
| LOW | deploy/helm/semantic-router/values.yaml | 461 | # When multiple models are specified in modelRefs, use | COMMENT |
| LOW | deploy/helm/semantic-router/values.yaml | 481 | # - type: "domain" | COMMENT |
| LOW | deploy/helm/semantic-router/values.yaml | 541 | # -- Dashboard image repository | COMMENT |
| LOW | deploy/helm/semantic-router/values.yaml | 561 | # still works (the random fallback is a valid signing key, you can log in and | COMMENT |
| LOW | deploy/operator/config/crd/kustomization.yaml | 1 | # This kustomization.yaml is not intended to be run by itself, | COMMENT |
| LOW | …samples/vllm.ai_v1alpha1_semanticrouter_openshift.yaml | 361 | # Environment Variables | COMMENT |
| LOW | …rator/config/samples/vllm_v1alpha1_semanticrouter.yaml | 61 | # Readiness probe | COMMENT |
| LOW | deploy/operator/api/v1alpha1/semanticrouter_types.go | 61 | // +optional | COMMENT |
| LOW | deploy/operator/api/v1alpha1/semanticrouter_types.go | 141 | // Repository is the container image repository | COMMENT |
| LOW | deploy/operator/api/v1alpha1/semanticrouter_types.go | 201 | // Port is the service port | COMMENT |
| LOW | deploy/operator/api/v1alpha1/semanticrouter_types.go | 241 | // AccessMode is the access mode | COMMENT |
| 576 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW⚡ | deploy/kserve/install-gpu-operator.sh | 29 | # Step 1: Install Node Feature Discovery (NFD) Operator | COMMENT |
| LOW | deploy/kserve/install-gpu-operator.sh | 81 | # Step 2: Create NFD Instance | COMMENT |
| LOW⚡ | deploy/kserve/install-gpu-operator.sh | 122 | # Step 3: Install NVIDIA GPU Operator | COMMENT |
| LOW | deploy/kserve/install-gpu-operator.sh | 175 | # Step 4: Create ClusterPolicy | COMMENT |
| LOW | deploy/kserve/install-gpu-operator.sh | 235 | # Step 5: Verify GPU Resources | COMMENT |
| LOW | …ndle-binding/examples/test_classifier_compatibility.rs | 21 | // Step 1: Download Extended32K base model | COMMENT |
| LOW | …ndle-binding/examples/test_classifier_compatibility.rs | 53 | // Step 2: Load base model configuration | COMMENT |
| LOW | …ndle-binding/examples/test_classifier_compatibility.rs | 68 | // Step 3: Load base model | COMMENT |
| LOW | …ndle-binding/examples/test_classifier_compatibility.rs | 80 | // Step 4: Test with existing PII classifier | COMMENT |
| LOW | candle-binding/examples/test_full_inference_32k.rs | 22 | // Step 1: Download Extended32K base model | COMMENT |
| LOW | candle-binding/examples/test_full_inference_32k.rs | 54 | // Step 2: Load base model configuration | COMMENT |
| LOW | candle-binding/examples/test_full_inference_32k.rs | 65 | // Step 3: Load PII classifier to get classifier weights | COMMENT |
| LOW | candle-binding/examples/test_full_inference_32k.rs | 116 | // Step 4: Create combined classifier using Extended32K base model + PII classifier weights | COMMENT |
| LOW⚡ | candle-binding/examples/test_full_inference_32k.rs | 144 | // Step 9: Test inference on sample texts | COMMENT |
| LOW | …ndle-binding/examples/test_32k_classifier_inference.rs | 30 | // Step 1: Download Extended32K base model | COMMENT |
| LOW | …ndle-binding/examples/test_32k_classifier_inference.rs | 62 | // Step 2: Load base model configuration | COMMENT |
| LOW | …ndle-binding/examples/test_32k_classifier_inference.rs | 73 | // Step 3: Load PII classifier to get classifier weights | COMMENT |
| LOW | …ndle-binding/examples/test_32k_classifier_inference.rs | 131 | // Step 4: Load Extended32K base model | COMMENT |
| LOW⚡ | …ndle-binding/examples/test_32k_classifier_inference.rs | 142 | // Step 5: Load classifier weights | COMMENT |
| LOW⚡ | …ndle-binding/examples/test_32k_classifier_inference.rs | 149 | // Step 6: Load optional head (if exists in PII model) | COMMENT |
| LOW⚡ | …ndle-binding/examples/test_32k_classifier_inference.rs | 158 | // Step 7: Create classifier with Extended32K base model | COMMENT |
| LOW⚡ | …ndle-binding/examples/test_32k_classifier_inference.rs | 230 | // Step 10: Test inference on sample texts (including long texts) | COMMENT |
| LOW | candle-binding/examples/test_long_text_inference_32k.rs | 27 | // Step 1: Download and load ModernBERT-base-32k | COMMENT |
| LOW⚡ | candle-binding/examples/test_long_text_inference_32k.rs | 59 | // Step 2: Load base model | COMMENT |
| LOW⚡ | candle-binding/examples/test_long_text_inference_32k.rs | 69 | // Step 3: Load tokenizer | COMMENT |
| LOW | candle-binding/examples/test_long_text_inference_32k.rs | 82 | // Step 4: Test with different classifiers | COMMENT |
| LOW | candle-binding/examples/test_long_text_inference_32k.rs | 127 | // Step 5: Test with texts of various lengths | COMMENT |
| LOW | candle-binding/examples/test_lora_compatibility_32k.rs | 26 | // Step 1: Download and load ModernBERT-base-32k | COMMENT |
| LOW | candle-binding/examples/test_lora_compatibility_32k.rs | 55 | // Step 2: Check dimension compatibility | COMMENT |
| LOW | candle-binding/examples/test_lora_compatibility_32k.rs | 77 | // Step 3: Try to find existing LoRA adapter | COMMENT |
| LOW | candle-binding/examples/test_lora_compatibility_32k.rs | 107 | // Step 4: Check if this is a LoRA adapter or traditional classifier | COMMENT |
| LOW | candle-binding/examples/test_lora_compatibility_32k.rs | 129 | // Step 5: Try to load LoRA adapter | COMMENT |
| LOW | candle-binding/examples/test_lora_compatibility_32k.rs | 194 | // Step 5: Check for traditional classifier | COMMENT |
| LOW | candle-binding/examples/test_lora_compatibility_32k.rs | 221 | // Step 6: Summary | COMMENT |
| LOW | candle-binding/examples/test_lora_inference_32k.rs | 29 | // Step 1: Download and load ModernBERT-base-32k | COMMENT |
| LOW | candle-binding/examples/test_lora_inference_32k.rs | 61 | // Step 2: Load base model | COMMENT |
| LOW | candle-binding/examples/test_lora_inference_32k.rs | 72 | // Step 3: Load tokenizer | COMMENT |
| LOW | candle-binding/examples/test_lora_inference_32k.rs | 85 | // Step 4: Find and load LoRA adapter | COMMENT |
| LOW | candle-binding/examples/test_lora_inference_32k.rs | 214 | // Step 5: Test inference | COMMENT |
| LOW | candle-binding/examples/test_lora_inference_32k.rs | 300 | // Step 6: Summary | COMMENT |
| LOW⚡ | …e-binding/src/model_architectures/embedding/pooling.rs | 65 | // Step 1: Expand attention_mask to match hidden_states dimensions | COMMENT |
| LOW⚡ | …e-binding/src/model_architectures/embedding/pooling.rs | 71 | // Step 2: Apply mask to hidden states | COMMENT |
| LOW⚡ | …e-binding/src/model_architectures/embedding/pooling.rs | 74 | // Step 3: Sum over sequence dimension (dim=1) | COMMENT |
| LOW⚡ | …e-binding/src/model_architectures/embedding/pooling.rs | 77 | // Step 4: Count valid tokens | COMMENT |
| LOW⚡ | …e-binding/src/model_architectures/embedding/pooling.rs | 80 | // Step 5: Average (handle division by zero gracefully) | COMMENT |
| LOW⚡ | …e-binding/src/model_architectures/embedding/pooling.rs | 162 | // Step 3: Extract the last valid token for each batch | COMMENT |
| LOW⚡ | …e-binding/src/model_architectures/embedding/pooling.rs | 172 | // Step 4: Concatenate all batch embeddings: [batch_size, hidden_size] | COMMENT |
| LOW | …e-binding/src/model_architectures/embedding/pooling.rs | 130 | // Step 1: Check if left padding | COMMENT |
| LOW⚡ | …g/src/model_architectures/embedding/gemma_embedding.rs | 479 | // Step 1: Gemma3 Transformer backbone | COMMENT |
| LOW⚡ | …g/src/model_architectures/embedding/gemma_embedding.rs | 483 | // Step 2: Mean Pooling | COMMENT |
| LOW⚡ | …g/src/model_architectures/embedding/gemma_embedding.rs | 504 | // Step 3: Dense Bottleneck (768 → 3072 → 768) | COMMENT |
| LOW⚡ | …g/src/model_architectures/embedding/gemma_embedding.rs | 508 | // Step 4: L2 Normalization | COMMENT |
| LOW⚡ | …g/src/model_architectures/embedding/gemma_embedding.rs | 573 | // Step 5: Truncate to target dimension | COMMENT |
| LOW⚡ | …g/src/model_architectures/embedding/gemma_embedding.rs | 583 | // Step 6: Re-normalize after truncation | COMMENT |
| LOW⚡ | …g/src/model_architectures/embedding/qwen3_embedding.rs | 381 | // Step 2: Generate position sequence in f64 | COMMENT |
| LOW⚡ | …g/src/model_architectures/embedding/qwen3_embedding.rs | 386 | // Step 3: Compute outer product in f64: positions ⊗ inv_freq | COMMENT |
| LOW⚡ | …g/src/model_architectures/embedding/qwen3_embedding.rs | 399 | // Step 4: Expand to full head_dim by concatenating freqs with itself | COMMENT |
| LOW⚡ | …g/src/model_architectures/embedding/qwen3_embedding.rs | 407 | // Step 5: Compute cos and sin in f64, then convert to f32 | COMMENT |
| LOW⚡ | …g/src/model_architectures/embedding/qwen3_embedding.rs | 560 | // Step 3: Rotate half: rotate_half(x) = cat([-x2, x1], dim=-1) | COMMENT |
| LOW⚡ | …g/src/model_architectures/embedding/qwen3_embedding.rs | 568 | // Step 4: Apply RoPE formula: x * cos + rotate_half(x) * sin | COMMENT |
| 239 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | bench/hallucination/datasets.py | 27 | context: str # Ground truth context (tool results / RAG context) | CODE |
| MEDIUM | bench/hallucination/datasets.py | 31 | hallucination_spans: Optional[List[dict]] = None # Annotated hallucination spans | CODE |
| MEDIUM⚡ | bench/hallucination/evaluate.py | 77 | true_positives: int # Correctly detected hallucinations | CODE |
| MEDIUM⚡ | bench/hallucination/evaluate.py | 78 | false_positives: int # Incorrectly flagged as hallucination | CODE |
| MEDIUM⚡ | bench/hallucination/evaluate.py | 80 | false_negatives: int # Missed hallucinations | CODE |
| MEDIUM | bench/hallucination/evaluate.py | 183 | # Add tool result with context (enables hallucination detection) | COMMENT |
| MEDIUM | bench/hallucination/evaluate.py | 217 | # Extract hallucination detection headers | COMMENT |
| MEDIUM | bench/hallucination/evaluate.py | 276 | # Print hallucination detection immediately when found | COMMENT |
| MEDIUM | bench/grounded_fusion/make_configs.py | 133 | # embedding model and otherwise fatal at startup) -- the benchmark routes | COMMENT |
| MEDIUM | …/dataset_implementations/openmathrreasoning_dataset.py | 39 | return True # Has detailed chain-of-thought solutions | CODE |
| MEDIUM | config/config.yaml | 946 | # (hallucination detector against context, else cross-model NLI) and | COMMENT |
| MEDIUM | config/config.yaml | 1838 | # winning candidate once with the hallucination explainer NLI model and | COMMENT |
| MEDIUM | config/fragments/signal/embedding/image-routing.yaml | 3 | # the multimodal embedding model. | COMMENT |
| MEDIUM | config/runtime/memory/valkey.yaml | 3 | # module) as the agentic memory backend. | COMMENT |
| MEDIUM | config/runtime/response-cache/milvus.yaml | 50 | dimension: 384 # This value is ignored - dimension is auto-detected from the embedding model | CODE |
| MEDIUM | config/runtime/response-cache/valkey.yaml | 47 | dimension: 384 # This value is ignored - dimension is auto-detected from the embedding model | CODE |
| MEDIUM | config/runtime/response-cache/redis.yaml | 47 | dimension: 384 # This value is ignored - dimension is auto-detected from the embedding model | CODE |
| MEDIUM | deploy/kserve/deploy.sh | 82 | # Deploy with custom storage class and a canonical embedding model | COMMENT |
| MEDIUM | deploy/operator/main.go | 40 | // +kubebuilder:scaffold:imports | COMMENT |
| MEDIUM | deploy/operator/main.go | 54 | // +kubebuilder:scaffold:scheme | COMMENT |
| MEDIUM | deploy/operator/main.go | 98 | // +kubebuilder:scaffold:builder | COMMENT |
| MEDIUM⚡ | deploy/operator/config/crd/kustomization.yaml | 6 | # +kubebuilder:scaffold:crdkustomizeresource | COMMENT |
| MEDIUM⚡ | deploy/operator/config/crd/kustomization.yaml | 12 | # +kubebuilder:scaffold:crdkustomizewebhookpatch | COMMENT |
| MEDIUM⚡ | deploy/operator/config/crd/kustomization.yaml | 17 | # +kubebuilder:scaffold:crdkustomizecainjectionpatch | COMMENT |
| MEDIUM | …amples/vllm.ai_v1alpha1_semanticrouter_complexity.yaml | 3 | # This example demonstrates few-shot complexity classification for intelligent routing | COMMENT |
| MEDIUM | deploy/operator/config/samples/kustomization.yaml | 4 | #+kubebuilder:scaffold:manifestskustomizesamples | COMMENT |
| MEDIUM | website/docs/training/model-performance-eval.md | 116 | ## Generate a configuration scaffold | COMMENT |
| MEDIUM | website/docs/training/model-performance-eval.md | 175 | ## Turn the scaffold into a routing policy | COMMENT |
| MEDIUM | website/docs/installation/valkey-memory.md | 156 | dimension: 384 # Must match your embedding model | CODE |
| MEDIUM | website/docs/installation/qdrant.md | 7 | This guide covers deploying [Qdrant](https://qdrant.tech/) as a backend for the Semantic Router. Qdrant can serve as the | CODE |
| MEDIUM | website/docs/installation/qdrant.md | 192 | dimension: 384 # Must match your embedding model | CODE |
| MEDIUM | website/docs/troubleshooting/common-errors.md | 315 | ## A classifier or embedding model cannot load | COMMENT |
| MEDIUM | website/docs/tutorials/plugin/rag.md | 1 | # RAG | COMMENT |
| MEDIUM | website/docs/tutorials/signal/heuristic/conversation.md | 107 | [`config/fragments/signal/conversation/agentic-shape.yaml`](https://github.com/vllm-project/semantic-router/blob/main/co | CODE |
| MEDIUM | website/docs/overview/use-cases.md | 145 | ### Bounded recovery and orchestration | COMMENT |
| MEDIUM | website/blog/2026-06-29-micro-agent-frontier-models.md | 333 | [Agentic Intelligence Lab](https://agentic-in.ai/), especially | CODE |
| MEDIUM | …log/2026-06-18-agentic-intelligent-routing-amd-rocm.md | 49 | | Recipe | [`config/recipes/agent/config.yaml`](https://github.com/vllm-project/semantic-router/blob/main/config/recipes | CODE |
| MEDIUM | …log/2026-06-18-agentic-intelligent-routing-amd-rocm.md | 50 | | Inferoa | [github.com/agentic-in/inferoa](https://github.com/agentic-in/inferoa) | Optional agent client with native v | CODE |
| MEDIUM | …log/2026-06-18-agentic-intelligent-routing-amd-rocm.md | 396 | [Inferoa](https://github.com/agentic-in/inferoa) is an inference-native agent | CODE |
| MEDIUM | …log/2026-06-18-agentic-intelligent-routing-amd-rocm.md | 402 | [inferoa.agentic-in.ai](https://inferoa.agentic-in.ai/docs/intro). | CODE |
| MEDIUM | …bsite/blog/2026-06-02-session-aware-agentic-routing.md | 10 | source_url: "https://github.com/vllm-project/vllm-project.github.io/blob/main/_posts/2026-06-02-session-aware-agentic-ro | CODE |
| MEDIUM | website/blog/2025-12-14-halugate.md | 223 | ### The `hallucination` Plugin | COMMENT |
| MEDIUM | website/blog/2025-12-14-halugate.md | 234 | # Action when hallucination detected | COMMENT |
| MEDIUM | website/blog/2025-12-14-halugate.md | 430 | # Decision with hallucination plugin | COMMENT |
| MEDIUM | website/blog/2026-03-10-v0.2-vllm-sr-athena-release.md | 153 | ### 4. Memory, RAG, and Response State Move Into the Core Runtime | COMMENT |
| MEDIUM | …lettucedetect-v2-generative-hallucination-detection.md | 83 | The `endpoint` backend (implemented by community contributor [@pranavthakur0-0](https://github.com/pranavthakur0-0) in [ | CODE |
| MEDIUM | website/blog/2026-03-12-v0-3-themis-roadmap.md | 64 | The first track is **NL-to-DSL authoring** in the dashboard, tracked in [#1511](https://github.com/vllm-project/semantic | CODE |
| MEDIUM | website/blog/2026-03-12-v0-3-themis-roadmap.md | 64 | The first track is **NL-to-DSL authoring** in the dashboard, tracked in [#1511](https://github.com/vllm-project/semantic | CODE |
| MEDIUM | website/blog/2026-03-12-v0-3-themis-roadmap.md | 98 | That broader observability push now also includes ClawOS-specific visibility into model and tool behavior through [#1523 | CODE |
| MEDIUM | website/blog/2026-03-12-v0-3-themis-roadmap.md | 98 | That broader observability push now also includes ClawOS-specific visibility into model and tool behavior through [#1523 | CODE |
| MEDIUM | …/version-v0.1/tutorials/semantic-cache/milvus-cache.md | 65 | dimension: 384 # Must match embedding model dimension | CODE |
| MEDIUM | …s/version-v0.1/tutorials/semantic-cache/redis-cache.md | 70 | dimension: 384 # Must match embedding model dimension | CODE |
| MEDIUM | ….1/tutorials/content-safety/hallucination-detection.md | 32 | # Disabled by default - enable in decisions via hallucination plugin | COMMENT |
| MEDIUM | ….1/tutorials/content-safety/hallucination-detection.md | 80 | # Action when hallucination detected: "header", "body", "block", or "none" | COMMENT |
| MEDIUM | ….1/tutorials/content-safety/hallucination-detection.md | 126 | ### RAG (Retrieval Augmented Generation) | COMMENT |
| MEDIUM | …n-v0.1/proposals/hallucination-mitigation-milestone.md | 1080 | # Global hallucination detection settings | COMMENT |
| MEDIUM | …sioned_docs/version-v0.2/installation/configuration.md | 40 | # Vector Store — local document ingestion and search (RAG) | COMMENT |
| MEDIUM | …rsioned_docs/version-v0.2/installation/k8s/operator.md | 944 | dimension: 384 # Match your embedding model | CODE |
| MEDIUM | …rsioned_docs/version-v0.2/installation/k8s/operator.md | 1024 | dimension: 384 # Match your embedding model | CODE |
| MEDIUM | …/version-v0.2/tutorials/semantic-cache/milvus-cache.md | 65 | dimension: 384 # Must match embedding model dimension | CODE |
| 84 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | bench/agentic_routing_live_benchmark.py | 248 | except Exception as exc: # pragma: no cover - network errors vary by platform | CODE |
| LOW | bench/agentic_routing_live_benchmark.py | 636 | except Exception as exc: # pragma: no cover - network errors vary by platform | CODE |
| LOW | bench/agent_task_live_benchmark.py | 2040 | except Exception as exc: # pragma: no cover - network errors vary by platform | CODE |
| LOW | bench/openai_fault_proxy.py | 151 | except Exception as exc: # pragma: no cover - network errors vary | CODE |
| LOW | bench/benchmark_comparison.sh | 168 | except Exception as e: | CODE |
| LOW | bench/benchmark_comparison.sh | 193 | except Exception as e: | CODE |
| LOW | bench/plot_comprehensive_results.py | 61 | except Exception as e: | CODE |
| LOW | bench/comprehensive_bench.sh | 299 | except Exception as e: | CODE |
| MEDIUM | bench/comprehensive_bench.sh | 300 | print(f'Error processing $target_file: {e}', file=sys.stderr) | CODE |
| LOW | bench/router_reason_bench.py | 155 | except Exception as e: | CODE |
| MEDIUM | bench/router_reason_bench.py | 156 | print(f"Error communicating with endpoint to list models: {e}") | CODE |
| LOW | bench/router_reason_bench.py | 310 | except Exception: | CODE |
| LOW | bench/router_reason_bench.py | 344 | except Exception as e: | CODE |
| LOW | bench/router_reason_bench.py | 685 | except Exception: | CODE |
| LOW | bench/router_reason_bench.py | 1005 | except Exception as e: | CODE |
| LOW | bench/router_reason_bench.py | 1093 | except Exception as e: | CODE |
| LOW | bench/cache_token_probe.py | 166 | except Exception as exc: # pragma: no cover - network errors vary by platform | CODE |
| LOW | bench/session_routing_branch_image_probe.py | 108 | except Exception as exc: # pragma: no cover - platform network errors vary | CODE |
| LOW | bench/hallucination/datasets.py | 67 | except Exception as e: | CODE |
| MEDIUM | bench/hallucination/datasets.py | 68 | print(f"Error loading HaluEval: {e}") | CODE |
| LOW | bench/hallucination/datasets.py | 115 | except Exception as e: | CODE |
| LOW | bench/hallucination/evaluate.py | 238 | except Exception as e: | CODE |
| LOW | bench/hallucination/evaluate.py | 629 | except Exception as e: | CODE |
| LOW | bench/hallucination/evaluate_detectors.py | 110 | except Exception as e: | CODE |
| LOW | bench/grounded_fusion/ollama_proxy.py | 94 | except Exception as e: | CODE |
| MEDIUM⚡ | bench/reasoning/router_reason_bench_multi_dataset.py | 290 | print(f"Error communicating with endpoint to list models: {e}") | CODE |
| MEDIUM | bench/reasoning/router_reason_bench_multi_dataset.py | 1104 | print(f"Error loading dataset '{args.dataset}': {e}") | CODE |
| LOW⚡ | bench/reasoning/router_reason_bench_multi_dataset.py | 289 | except Exception as e: | CODE |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 329 | except Exception: | CODE |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 619 | except Exception as e: | CODE |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 795 | except Exception: | CODE |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 919 | except Exception: | CODE |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 1103 | except Exception as e: | CODE |
| LOW | bench/reasoning/reasoning_mode_eval.py | 213 | except Exception as exc: | CODE |
| MEDIUM | bench/reasoning/reasoning_mode_eval.py | 214 | print(f"Error fetching models: {exc}") | CODE |
| LOW | bench/reasoning/reasoning_mode_eval.py | 624 | except Exception as exc: | CODE |
| LOW | bench/reasoning/dataset_factory.py | 144 | except Exception as e: | CODE |
| MEDIUM | bench/reasoning/cli.py | 321 | print(f"Error importing dataset factory: {e}") | CODE |
| LOW | bench/reasoning/dataset_implementations/gpqa_dataset.py | 57 | except Exception as e: | CODE |
| LOW | …oning/dataset_implementations/commonsenseqa_dataset.py | 53 | except Exception as e: | CODE |
| LOW | …easoning/dataset_implementations/truthfulqa_dataset.py | 47 | except Exception as e: | CODE |
| LOW | bench/reasoning/dataset_implementations/math_dataset.py | 46 | except Exception: | CODE |
| LOW | bench/reasoning/dataset_implementations/math_dataset.py | 49 | except Exception: | CODE |
| LOW | …reasoning/dataset_implementations/hellaswag_dataset.py | 53 | except Exception as e: | CODE |
| LOW | bench/router_flow/flow_eval.py | 268 | except Exception as exc: | CODE |
| LOW | bench/router_flow/real_eval/run_evalscope_suite.py | 254 | except Exception: | CODE |
| LOW | bench/router_flow/real_eval/test_run_evalscope_suite.py | 736 | except Exception: | STRING |
| MEDIUM | tools/security/scan_malicious_code.py | 250 | print(f"Error: cannot load {SIGNATURES_CONFIG}: {exc}", file=sys.stderr) | CODE |
| MEDIUM | tools/security/scan_malicious_code.py | 547 | print(f"Error: {root} is not a directory", file=sys.stderr) | CODE |
| MEDIUM | tools/security/ast_security_scanner.py | 1925 | print(f"Error getting diff: {e}", file=sys.stderr) | CODE |
| MEDIUM | tools/security/ast_security_scanner.py | 2068 | print(f"Error: {root} is not a directory", file=sys.stderr) | CODE |
| LOW | tools/security/ast_security_scanner.py | 2075 | except Exception as exc: | CODE |
| LOW | tools/agent/scripts/router_calibration_loop.py | 319 | except Exception as exc: | CODE |
| LOW | tools/agent/scripts/tuning/client.py | 47 | except Exception: | CODE |
| LOW | tools/agent/scripts/tuning/client.py | 87 | except Exception as exc: | CODE |
| LOW | tools/models/generate_gemma_reference.py | 329 | except Exception as e: | CODE |
| LOW | tools/mcp-classifier-server/server_generative.py | 292 | except Exception: | CODE |
| LOW | tools/mcp-classifier-server/server_generative.py | 352 | except Exception as e: | CODE |
| LOW | tools/mcp-classifier-server/server_generative.py | 599 | except Exception as e: | CODE |
| LOW | tools/mcp-classifier-server/server_generative.py | 801 | except Exception as e: | CODE |
| 236 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | bench/agentic_routing_live_benchmark.py | 514 | CODE | |
| LOW | bench/agent_task_live_benchmark.py | 2104 | CODE | |
| LOW | bench/agentic_routing_experiment.py | 855 | CODE | |
| LOW | bench/router_reason_bench.py | 276 | CODE | |
| LOW | bench/hallucination/evaluate.py | 309 | CODE | |
| LOW | bench/grounded_fusion/evaluate.py | 221 | CODE | |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 186 | CODE | |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 294 | CODE | |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 448 | CODE | |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 666 | CODE | |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 742 | CODE | |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 805 | CODE | |
| LOW | bench/reasoning/dataset_interface.py | 124 | CODE | |
| LOW | bench/reasoning/cli.py | 186 | CODE | |
| LOW | bench/reasoning/dataset_implementations/gpqa_dataset.py | 68 | CODE | |
| LOW | bench/reasoning/dataset_implementations/gpqa_dataset.py | 84 | CODE | |
| LOW | bench/router_flow/render_report.py | 101 | CODE | |
| LOW | …nch/router_flow/real_eval/collect_evalscope_results.py | 322 | CODE | |
| LOW | …nch/router_flow/real_eval/collect_evalscope_results.py | 353 | CODE | |
| LOW | tools/security/scan_malicious_code.py | 311 | CODE | |
| LOW | tools/security/scan_malicious_code.py | 401 | CODE | |
| LOW | tools/security/ast_security_scanner.py | 626 | CODE | |
| LOW | tools/security/ast_security_scanner.py | 716 | CODE | |
| LOW | tools/security/ast_security_scanner.py | 978 | CODE | |
| LOW | tools/security/ast_security_scanner.py | 1261 | CODE | |
| LOW | tools/security/ast_security_scanner.py | 1395 | CODE | |
| LOW | tools/security/ast_security_scanner.py | 1499 | CODE | |
| LOW | tools/security/ast_security_scanner.py | 1538 | CODE | |
| LOW | tools/security/ast_security_scanner.py | 1716 | CODE | |
| LOW | tools/security/ast_security_scanner.py | 1873 | CODE | |
| LOW | tools/security/ast_security_scanner.py | 1729 | CODE | |
| LOW | tools/ci/community_lifecycle.py | 70 | CODE | |
| LOW | tools/ci/classify_pr_changes.py | 459 | CODE | |
| LOW | tools/hf-playground/app.py | 163 | CODE | |
| LOW | tools/hf-playground/app.py | 604 | CODE | |
| LOW | tools/agent/scripts/router_calibration_image.py | 114 | CODE | |
| LOW | tools/agent/scripts/maintainer_board.py | 230 | CODE | |
| LOW | tools/agent/scripts/maintainer_board.py | 602 | CODE | |
| LOW | tools/agent/scripts/router_calibration_fixture_test.py | 69 | CODE | |
| LOW | tools/agent/scripts/router_calibration_fixture_test.py | 197 | CODE | |
| LOW | tools/agent/scripts/structure_check.py | 423 | CODE | |
| LOW | tools/agent/scripts/tuning/engine_selection.py | 533 | CODE | |
| LOW | tools/agent/scripts/tuning/engine_selection.py | 556 | CODE | |
| LOW | tools/agent/scripts/tuning/analyzer.py | 41 | CODE | |
| LOW | tools/agent/scripts/tuning/engine.py | 160 | CODE | |
| LOW | tools/agent/scripts/tuning/engine.py | 216 | CODE | |
| LOW | tools/agent/scripts/tuning/engine.py | 540 | CODE | |
| LOW | tools/agent/scripts/tuning/engine.py | 586 | CODE | |
| LOW | …s/tuning/verify_results/run_confidence_verification.py | 35 | CODE | |
| LOW | tools/mcp-classifier-server/server_generative.py | 583 | CODE | |
| LOW | tools/mcp-classifier-server/server_generative.py | 661 | CODE | |
| LOW | tools/mcp-classifier-server/server_generative.py | 681 | CODE | |
| LOW | tools/mcp-classifier-server/server_embedding.py | 580 | CODE | |
| LOW | tools/mcp-classifier-server/server_embedding.py | 604 | CODE | |
| LOW | tools/mcp-classifier-server/server_keyword.py | 360 | CODE | |
| LOW | tools/mcp-classifier-server/server_keyword.py | 374 | CODE | |
| LOW | deploy/openshift/demo/demo-semantic-router.py | 221 | CODE | |
| LOW | deploy/openshift/demo/demo-semantic-router.py | 494 | CODE | |
| LOW | onnx-binding/ort-ck-flash-attn/scripts/rewrite_graph.py | 35 | CODE | |
| LOW | onnx-binding/ort-ck-flash-attn/scripts/rewrite_graph.py | 300 | CODE | |
| 148 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | bench/plot_comprehensive_results.py | 312 | # Check if we're dealing with a single dataset | COMMENT |
| LOW | bench/comprehensive_bench.sh | 434 | # Check if plots already exist | COMMENT |
| LOW | bench/hallucination/evaluate.py | 641 | # Print results | COMMENT |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 826 | # Check if dataset has actual CoT content by examining sample questions | COMMENT |
| LOW | tools/smoke/test-image-gen.sh | 53 | # Check if response contains image | COMMENT |
| LOW | tools/smoke/test-classification-api.sh | 62 | # Check if expected domain is in matched domains | COMMENT |
| LOW | tools/smoke/test-classification-api.sh | 364 | # Check if expected domain is in matched domains | COMMENT |
| LOW | tools/smoke/test-classification-api.sh | 629 | # Check if expected signal is in matched fact check signals | COMMENT |
| LOW | tools/smoke/test-classification-api.sh | 769 | # Check if expected signal is in matched user feedback signals | COMMENT |
| LOW | tools/kind/generate-kind-config.sh | 23 | # Check if template exists | COMMENT |
| LOW | tools/kind/generate-kind-config.sh | 29 | # Check if models directory exists | COMMENT |
| LOW | tools/mcp-classifier-server/server_generative.py | 325 | # Check if it's a local path that exists | COMMENT |
| LOW | tools/mcp-classifier-server/server_generative.py | 329 | # Check if it looks like a HuggingFace model ID (contains /) | COMMENT |
| LOW⚡ | tools/mcp-classifier-server/server_embedding.py | 254 | # Check if collection exists | COMMENT |
| LOW⚡ | tools/mcp-classifier-server/server_embedding.py | 258 | # Check if we need to rebuild (verify count matches CSV) | COMMENT |
| LOW | tools/dev/local-up-router.sh | 228 | # Check if all processes are still running. Prints a warning once each time | COMMENT |
| LOW | tools/milvus/test-milvus-deployment.sh | 103 | # Check if release exists | COMMENT |
| LOW | tools/milvus/test-milvus-deployment.sh | 256 | # Check if Prometheus Operator is installed | COMMENT |
| LOW | openvino-binding/convert_modernbert_models.py | 86 | # Check if model exists | COMMENT |
| LOW | openvino-binding/convert_modernbert_models.py | 91 | # Check if already converted | COMMENT |
| LOW | openvino-binding/scripts/convert_test_tokenizers.py | 118 | # Check if the model directory exists (should be created by optimum-cli) | COMMENT |
| LOW | openvino-binding/scripts/convert_test_tokenizers.py | 128 | # Check if tokenizer already exists | COMMENT |
| LOW | deploy/kserve/deployment.yaml | 74 | # Check if model weights actually exist (not just the directory) | COMMENT |
| LOW | deploy/kserve/deploy.sh | 302 | # Check if logged in | COMMENT |
| LOW | deploy/kserve/deploy.sh | 319 | # Check if namespace exists | COMMENT |
| LOW | deploy/kserve/install-kserve.sh | 336 | # Check if both KServe CRDs are already installed | COMMENT |
| LOW | deploy/kserve/install-gpu-operator.sh | 14 | # Check if running on OpenShift | COMMENT |
| LOW⚡ | deploy/kserve/install-gpu-operator.sh | 32 | # Check if NFD operator is already installed | COMMENT |
| LOW⚡ | deploy/kserve/install-gpu-operator.sh | 126 | # Check if GPU operator is already installed | COMMENT |
| LOW | deploy/openshift/deployment.yaml | 245 | # Check if model already exists | COMMENT |
| LOW | deploy/openshift/deployment-simulator.yaml | 180 | # Check if model already exists | COMMENT |
| LOW | deploy/openshift/deploy-to-openshift.sh | 94 | # Check if logged in to OpenShift | COMMENT |
| LOW | deploy/openshift/deploy-to-openshift.sh | 636 | # Check if dashboard imagestream exists | COMMENT |
| LOW | deploy/openshift/deploy-to-openshift.sh | 642 | # Check if dashboard buildconfig exists | COMMENT |
| LOW | deploy/openshift/cleanup-openshift.sh | 150 | # Check if oc is installed | COMMENT |
| LOW | deploy/openshift/demo/curl-examples.sh | 19 | # Check if logged into OpenShift | COMMENT |
| LOW | deploy/openshift/demo/live-classifier-logs.sh | 60 | # Increment counter | COMMENT |
| LOW | deploy/kubernetes/istio/deployment.yaml | 28 | # Check if all required models already exist in PVC; if yes, skip downloads entirely | COMMENT |
| LOW | deploy/kubernetes/llm-katan/base/deployment.yaml | 44 | # Check if model already exists in PVC | COMMENT |
| LOW⚡ | deploy/helm/validate-chart.sh | 264 | # Check if labels helper exists | COMMENT |
| LOW⚡ | deploy/helm/validate-chart.sh | 272 | # Check if selector labels helper exists | COMMENT |
| LOW⚡ | deploy/helm/validate-chart.sh | 280 | # Check if NOTES.txt exists | COMMENT |
| LOW | onnx-binding/scripts/run_benchmark_rocm.sh | 14 | # Check if binary exists | COMMENT |
| LOW | onnx-binding/ort-ck-flash-attn/scripts/rewrite_graph.py | 516 | # Check if any output is consumed (by existing nodes OR new nodes) | COMMENT |
| LOW | .github/workflows/integration-test-k8s.yml | 200 | # Check if there are more lines | COMMENT |
| LOW | e2e/testing/03-classification-api-test.py | 147 | # Check if classification is correct | COMMENT |
| LOW | e2e/testing/05-jailbreak-test.py | 221 | # Check if jailbreak was properly detected | COMMENT |
| LOW | e2e/testing/05-jailbreak-test.py | 417 | # Check if the request went through (indicating security bypass) | COMMENT |
| LOW | e2e/testing/05-jailbreak-test.py | 433 | # Check if response contains harmful instructions | COMMENT |
| LOW | e2e/testing/start-llm-katan.sh | 63 | # Check if ports are available | COMMENT |
| LOW | e2e/testing/02-router-classification-test.py | 257 | # Check if classification and routing metrics exist | COMMENT |
| LOW | e2e/testing/06-pii-detection-test.py | 220 | # Check if PII was properly detected | COMMENT |
| LOW | e2e/testing/06-pii-detection-test.py | 409 | # Check if the request went through | COMMENT |
| LOW | e2e/testing/06-pii-detection-test.py | 425 | # Check if response potentially exposes PII patterns | COMMENT |
| LOW | e2e/testing/run_all_tests.py | 104 | # Check if services are running | COMMENT |
| LOW | e2e/testing/04-cache-test.py | 90 | # Check if cache is enabled in metrics | COMMENT |
| LOW | e2e/testing/04-cache-test.py | 177 | # Check if cache hits increased | COMMENT |
| LOW | e2e/testing/vllm-sr-cli/test_unit_serve.py | 243 | # Check if the default image exists | COMMENT |
| LOW | e2e/testing/vllm-sr-cli/test_unit_serve.py | 441 | # Check if models directory was created | COMMENT |
| LOW | e2e/testing/hallucination-demo/mock_vllm_toolcall.py | 105 | # Check if this is a follow-up with tool results | COMMENT |
| 44 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | bench/reasoning/dataset_interface.py | 0 | get list of all available categories in the dataset. | STRING |
| HIGH | bench/reasoning/dataset_implementations/drop_dataset.py | 0 | get list of all available categories in the dataset. | STRING |
| HIGH | …easoning/dataset_implementations/openbookqa_dataset.py | 0 | get list of all available categories in the dataset. | STRING |
| HIGH | …easoning/dataset_implementations/strategyqa_dataset.py | 0 | get list of all available categories in the dataset. | STRING |
| HIGH | bench/reasoning/dataset_implementations/math_dataset.py | 0 | get list of all available categories in the dataset. | STRING |
| HIGH | …/reasoning/dataset_implementations/aqua_rat_dataset.py | 0 | get list of all available categories in the dataset. | STRING |
| HIGH | bench/reasoning/dataset_implementations/sciq_dataset.py | 0 | get list of all available categories in the dataset. | STRING |
| HIGH | …nch/reasoning/dataset_implementations/gsm8k_dataset.py | 0 | get list of all available categories in the dataset. | STRING |
| HIGH | …/dataset_implementations/openmathrreasoning_dataset.py | 0 | get list of all available categories in the dataset. | STRING |
| HIGH | tools/mcp-classifier-server/server_generative.py | 0 | you are a history expert. when answering historical questions: - provide accurate dates, names, and historical context - | STRING |
| HIGH | tools/mcp-classifier-server/server_embedding.py | 0 | you are a history expert. when answering historical questions: - provide accurate dates, names, and historical context - | STRING |
| HIGH | tools/mcp-classifier-server/server_keyword.py | 0 | you are a history expert. when answering historical questions: - provide accurate dates, names, and historical context - | STRING |
| HIGH | tools/mcp-classifier-server/server_generative.py | 0 | you are a mathematics expert. when answering math questions: - show step-by-step solutions with clear explanations - use | STRING |
| HIGH | tools/mcp-classifier-server/server_embedding.py | 0 | you are a mathematics expert. when answering math questions: - show step-by-step solutions with clear explanations - use | STRING |
| HIGH | tools/mcp-classifier-server/server_keyword.py | 0 | you are a mathematics expert. when answering math questions: - show step-by-step solutions with clear explanations - use | STRING |
| HIGH | tools/mcp-classifier-server/server_generative.py | 0 | you are a knowledgeable assistant. when answering general questions: - provide balanced, well-rounded responses - draw f | STRING |
| HIGH | tools/mcp-classifier-server/server_embedding.py | 0 | you are a knowledgeable assistant. when answering general questions: - provide balanced, well-rounded responses - draw f | STRING |
| HIGH | tools/mcp-classifier-server/server_keyword.py | 0 | you are a knowledgeable assistant. when answering general questions: - provide balanced, well-rounded responses - draw f | STRING |
| HIGH | e2e/testing/05-jailbreak-test.py | 0 | check if the services are running before running tests. | STRING |
| HIGH | e2e/testing/06-pii-detection-test.py | 0 | check if the services are running before running tests. | STRING |
| HIGH | e2e/testing/02-router-classification-test.py | 0 | check if the services are running before running tests. | STRING |
| HIGH | e2e/testing/04-cache-test.py | 0 | check if the services are running before running tests. | STRING |
| HIGH | …ck_fine_tuning_lora/fact_check_bert_finetuning_lora.py | 0 | create tokenizer with model-specific configuration. args: model_path: path to load tokenizer from base_model_name: optio | STRING |
| HIGH | …ng_classifier/modality_routing_bert_finetuning_lora.py | 0 | create tokenizer with model-specific configuration. args: model_path: path to load tokenizer from base_model_name: optio | STRING |
| HIGH | …ard_fine_tuning_lora/jailbreak_bert_finetuning_lora.py | 0 | create tokenizer with model-specific configuration. args: model_path: path to load tokenizer from base_model_name: optio | STRING |
| HIGH | …pii_model_fine_tuning_lora/pii_bert_finetuning_lora.py | 0 | create tokenizer with model-specific configuration. args: model_path: path to load tokenizer from base_model_name: optio | STRING |
| HIGH | …er/classifier_model_fine_tuning_lora/ft_linear_lora.py | 0 | create tokenizer with model-specific configuration. args: model_path: path to load tokenizer from base_model_name: optio | STRING |
| HIGH | …ard_fine_tuning_lora/jailbreak_bert_finetuning_lora.py | 0 | merge lora adapter with base model to create a complete model for rust inference. this function is automatically called | STRING |
| HIGH | …pii_model_fine_tuning_lora/pii_bert_finetuning_lora.py | 0 | merge lora adapter with base model to create a complete model for rust inference. this function is automatically called | STRING |
| HIGH | …er/classifier_model_fine_tuning_lora/ft_linear_lora.py | 0 | merge lora adapter with base model to create a complete model for rust inference. this function is automatically called | STRING |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | bench/comprehensive_bench.sh | 352 | # Generate updated comprehensive plots for current dataset | COMMENT |
| MEDIUM | bench/comprehensive_bench.sh | 361 | # Function to generate comprehensive plot with all completed datasets (called after each dataset completes) | COMMENT |
| MEDIUM | bench/router_reason_bench.py | 277 | # Normalize non-string responses into a string to be robust to providers | COMMENT |
| MEDIUM | bench/grounded_fusion/make_configs.py | 134 | # purely on the sentinel the harness prepends to every prompt. | COMMENT |
| MEDIUM⚡ | bench/reasoning/router_reason_bench_multi_dataset.py | 296 | # Normalize non-string responses into a string to be robust to providers | COMMENT |
| MEDIUM | bench/reasoning/cli.py | 109 | """Register the 'comprehensive' command - run full research benchmark.""" | STRING |
| MEDIUM | bench/reasoning/cli.py | 348 | """Run comprehensive benchmark.""" | STRING |
| MEDIUM | bench/reasoning/reporting.py | 9 | """Generate a comprehensive markdown report.""" | STRING |
| MEDIUM | tools/linter/go/.golangci.yml | 20 | - testifylint # checks for best practices when using the testify testing framework | CODE |
| MEDIUM⚡ | deploy/helm/validate-chart.sh | 260 | # Test 9: Check for common Helm best practices | COMMENT |
| MEDIUM | …ig/samples/vllm.ai_v1alpha1_semanticrouter_simple.yaml | 4 | # For a comprehensive example with all features, see vllm.ai_v1alpha1_semanticrouter_openshift.yaml | COMMENT |
| MEDIUM | website/package-lock.json | 20228 | "resolved": "https://registry.npmjs.org/robust-predicates/-/robust-predicates-3.0.3.tgz", | CODE |
| MEDIUM | candle-binding/Cargo.toml | 84 | # Example testing ModernBERT-base-32k comprehensive validation | COMMENT |
| MEDIUM | candle-binding/src/model_architectures/model_factory.rs | 4 | //! Traditional and LoRA models through a unified interface, enabling seamless | COMMENT |
| MEDIUM | dashboard/frontend/src/utils/mlPipelineApi.ts | 103 | // on connect, so reconnection is seamless. | COMMENT |
| MEDIUM | .github/workflows/performance-test.yml | 76 | # the regular HF CDN. The gate depends on this step, so keep it robust. | COMMENT |
| MEDIUM | e2e/testing/run_memory_integration.sh | 191 | # Ensure the mmbert model used by the CI harness is available. Tests and | COMMENT |
| MEDIUM | e2e/testing/run_memory_integration.sh | 193 | # the CI stable and compatible with the rest of the harness (collection dims, etc.). | COMMENT |
| MEDIUM | src/semantic-router/pkg/cache/benchmark.go | 113 | // to leverage continuous batching for faster population | COMMENT |
| MEDIUM | src/semantic-router/pkg/looper/confidence.go | 931 | // Uses JSON marshaling for robust extraction across SDK versions | COMMENT |
| MEDIUM | src/semantic-router/pkg/selection/hybrid.go | 69 | // HybridSelector combines multiple selection methods for robust routing | COMMENT |
| MEDIUM | src/fleet-sim/fleet_sim/routing/compress_route.py | 86 | # Definitely short | COMMENT |
| MEDIUM | src/fleet-sim/fleet_sim/routing/compress_route.py | 117 | # Definitely long | COMMENT |
| MEDIUM | …r/prompt_guard_fine_tuning_lora/train_cpu_optimized.sh | 20 | LEARNING_RATE=3e-5 # Optimized learning rate based on PEFT best practices | CODE |
| MEDIUM | …pii_model_fine_tuning_lora/pii_bert_finetuning_lora.py | 281 | # Use more robust tokenization that preserves character positions | COMMENT |
| MEDIUM | …pii_model_fine_tuning_lora/pii_bert_finetuning_lora.py | 324 | # Add comprehensive data validation and quality analysis | COMMENT |
| MEDIUM | …pii_model_fine_tuning_lora/pii_bert_finetuning_lora.py | 566 | # Use robust tokenization | COMMENT |
| MEDIUM | …pii_model_fine_tuning_lora/pii_bert_finetuning_lora.py | 1224 | # Training arguments optimized for LoRA token classification based on PEFT best practices | COMMENT |
| MEDIUM | …pii_model_fine_tuning_lora/pii_bert_finetuning_lora.py | 1131 | learning_rate: float = 3e-5, # Optimized for LoRA based on PEFT best practices | CODE |
| MEDIUM | …pii_model_fine_tuning_lora/pii_bert_finetuning_lora.py | 354 | """Validate BIO label consistency and report comprehensive statistics.""" | STRING |
| MEDIUM | …edback_classifier/data_processing/download_datasets.py | 861 | # Extract label using robust extraction | COMMENT |
| MEDIUM | …_classifier/user_feedback_classifier/configs/config.py | 50 | max_seq_length: int = 1024 # Increased from 512 to leverage longer context | CODE |
| MEDIUM | …assifier_model_fine_tuning_lora/train_cpu_optimized.sh | 20 | LEARNING_RATE=3e-5 # Optimized learning rate based on PEFT best practices | CODE |
| MEDIUM | …er/classifier_model_fine_tuning_lora/ft_linear_lora.py | 556 | # Training arguments optimized for LoRA sequence classification based on PEFT best practices | COMMENT |
| MEDIUM | …ment/mmlu_pro_solver_lora/ft_qwen3_mmlu_solver_lora.py | 1028 | # Save comprehensive results | COMMENT |
| LOW | src/training/model_eval/signal_eval.py | 44 | # To add new datasets (e.g., multilingual), just add new entries here | COMMENT |
| MEDIUM | src/training/model_eval/mom_collection_eval.py | 364 | """Load model and tokenizer with robust handling using base config.""" | STRING |
| MEDIUM | src/training/model_eval/mom_collection_eval.py | 428 | """Evaluate a single model with comprehensive error handling.""" | STRING |
| MEDIUM | …raining/model_selection/ml_model_selection/validate.go | 246 | // Uses huggingface-cli for robust downloads (handles auth, LFS, caching, parquet, etc.) | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | bench/router_flow/real_eval/test_run_amd_eval_matrix.py | 121 | assert "docker rm -f vllm-sr-router-container >/dev/null 2>&1 || true" in script | CODE |
| HIGH⚡ | bench/router_flow/real_eval/test_run_amd_eval_matrix.py | 151 | assert "curl -fsS http://127.0.0.1:8899/v1/models 2>/dev/null || true" in script | CODE |
| HIGH | bench/router_flow/real_eval/run_amd_eval_matrix.py | 519 | " docker rm -f vllm-sr-router-container >/dev/null 2>&1 || true", | CODE |
| HIGH | bench/router_flow/real_eval/run_amd_eval_matrix.py | 540 | " models_json=$(curl -fsS http://127.0.0.1:8899/v1/models 2>/dev/null || true)", | CODE |
| HIGH | bench/router_flow/real_eval/run_amd_eval_matrix.py | 541 | f' if [ -n "$models_json" ] && printf \'%s\\n\' "$models_json" | grep -F {quoted_model} >/dev/null; then', | CODE |
| HIGH | openvino-binding/scripts/convert_test_tokenizers.py | 157 | print(" cd openvino-binding && make test") | CODE |
| HIGH⚡ | e2e/testing/hallucination-demo/web_client.py | 311 | const factCheckNeeded = data.headers && data.headers['x-vsr-fact-check-needed'] === 'true'; | CODE |
| HIGH⚡ | e2e/testing/hallucination-demo/web_client.py | 315 | const unverifiedResponse = data.headers && data.headers['x-vsr-unverified-factual-response'] === 'true'; | CODE |
| HIGH⚡ | e2e/testing/hallucination-demo/web_client.py | 316 | const hallucinationDetected = data.headers && data.headers['x-vsr-hallucination-detected'] === 'true'; | CODE |
| HIGH | e2e/testing/hallucination-demo/web_client.py | 215 | if (typeEffect && role === 'assistant') { | CODE |
| HIGH | e2e/testing/hallucination-demo/web_client.py | 239 | if (detail && detailEl) detailEl.textContent = detail; | CODE |
| HIGH | e2e/testing/hallucination-demo/web_client.py | 353 | addMessage(data.error ? 'error' : 'assistant', data.response || data.error, true); | CODE |
| HIGH | src/vllm-sr/tests/test_dashboard_dockerfile_surface.py | 68 | '"$CONTAINER_SOCKET_PATH" 2>/dev/null); then' in content | CODE |
| HIGH⚡ | src/vllm-sr/cli/container_log_spool.py | 65 | if [ -n "$relay_pid" ]; then kill "$relay_pid" 2>/dev/null || true; fi | CODE |
| HIGH⚡ | src/vllm-sr/cli/container_log_spool.py | 66 | if [ -n "$child_pid" ]; then kill "$child_pid" 2>/dev/null || true; fi | CODE |
| HIGH⚡ | src/vllm-sr/cli/container_log_spool.py | 70 | if [ -n "$child_pid" ]; then kill -TERM "$child_pid" 2>/dev/null || true; fi | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | bench/router_reason_bench.py | 525 | client = OpenAI(base_url=endpoint, api_key=api_key or "dummy-key") | CODE |
| LOW | bench/reasoning/router_reason_bench_multi_dataset.py | 823 | client = OpenAI(base_url=endpoint, api_key=api_key or "dummy-key", timeout=300.0) | CODE |
| LOW | …odel_architectures/classification/mmbert_classifier.rs | 1170 | text: "test@test.com".to_string(), | CODE |
| LOW | …version-v0.1/tutorials/content-safety/pii-detection.md | 22 | | `EMAIL_ADDRESS` | Email addresses | "user@example.com" | | CODE |
| LOW | …version-v0.2/tutorials/content-safety/pii-detection.md | 22 | | `EMAIL_ADDRESS` | Email addresses | "user@example.com" | | CODE |
| LOW | …version-v0.1/tutorials/content-safety/pii-detection.md | 29 | | `EMAIL_ADDRESS` | 电子邮件地址 | "user@example.com" | | CODE |
| LOW | …lugin-content-docs/version-v0.2/cookbook/pii-policy.md | 62 | | `EMAIL_ADDRESS`| 电子邮件地址 | "user@example.com" | | CODE |
| LOW | …lugin-content-docs/version-v0.2/cookbook/pii-policy.md | 66 | | `ORGANIZATION` | 公司/组织名称 | "Acme Corp" | | CODE |
| LOW | …version-v0.2/tutorials/content-safety/pii-detection.md | 29 | | `EMAIL_ADDRESS` | 电子邮件地址 | "user@example.com" | | CODE |
| LOW⚡ | candle-binding/examples/test_full_inference_32k.rs | 154 | "Lorem ipsum dolor sit amet. ".repeat(50) | CODE |
| LOW⚡ | candle-binding/examples/test_full_inference_32k.rs | 154 | "Lorem ipsum dolor sit amet. ".repeat(50) | CODE |
| LOW⚡ | …ndle-binding/examples/test_32k_classifier_inference.rs | 240 | "Lorem ipsum dolor sit amet. ".repeat(50) | CODE |
| LOW⚡ | …ndle-binding/examples/test_32k_classifier_inference.rs | 240 | "Lorem ipsum dolor sit amet. ".repeat(50) | CODE |
| LOW | candle-binding/src/ffi/classify_test.rs | 87 | let text_cstring = CString::new("John Doe").expect("Failed to create CString"); | CODE |
| LOW | candle-binding/src/ffi/classify_test.rs | 116 | assert_eq!(text_str, "John Doe"); | CODE |
| LOW | dashboard/frontend/e2e/playground.spec.ts | 512 | await expect(dialog).toContainText('admin@example.com') | CODE |
| LOW | dashboard/frontend/e2e/auth-flow.spec.ts | 194 | email: 'admin@example.com', | CODE |
| LOW | dashboard/frontend/e2e/auth-flow.spec.ts | 213 | email: 'admin@example.com', | CODE |
| LOW | dashboard/frontend/e2e/auth-flow.spec.ts | 283 | await page.getByPlaceholder('you@example.com').fill('admin@example.com') | CODE |
| LOW | dashboard/frontend/e2e/auth-flow.spec.ts | 609 | email: 'admin@example.com', | CODE |
| LOW | dashboard/frontend/e2e/auth-flow.spec.ts | 628 | email: 'admin@example.com', | CODE |
| LOW | dashboard/frontend/e2e/auth-flow.spec.ts | 680 | await page.getByPlaceholder('you@example.com').fill('admin@example.com') | CODE |
| LOW | dashboard/frontend/e2e/auth-flow.spec.ts | 956 | email: 'admin@example.com', | CODE |
| LOW | dashboard/frontend/e2e/auth-flow.spec.ts | 987 | email: 'admin@example.com', | CODE |
| LOW | dashboard/frontend/e2e/auth-flow.spec.ts | 1022 | email: 'admin@example.com', | CODE |
| LOW | dashboard/frontend/e2e/public-responsive.spec.ts | 206 | await page.getByPlaceholder('you@example.com').fill('admin@example.com') | CODE |
| LOW | dashboard/frontend/e2e/public-responsive.spec.ts | 251 | email: 'admin@example.com', | CODE |
| LOW | dashboard/frontend/e2e/support/auth.ts | 30 | email: 'admin@example.com', | CODE |
| LOW | dashboard/frontend/src/pages/UsersPageUserDialog.tsx | 147 | placeholder="Jane Doe" | CODE |
| LOW | dashboard/backend/auth/handlers_test.go | 66 | user := newTestUser(t, svc, "admin@example.com", "admin", "active") | CODE |
| LOW | dashboard/backend/auth/store_test.go | 19 | admin := newTestUser(t, NewService(store, "test-secret", 1), "admin@example.com", RoleAdmin, "active") | CODE |
| LOW | dashboard/backend/auth/invitation_test.go | 11 | admin := newTestUser(t, svc, "admin@example.com", RoleAdmin, "active") | CODE |
| LOW | dashboard/backend/auth/invitation_test.go | 44 | admin := newTestUser(t, svc, "admin@example.com", RoleAdmin, "active") | CODE |
| LOW | dashboard/backend/auth/invitation_test.go | 70 | admin := newTestUser(t, svc, "admin@example.com", RoleAdmin, "active") | CODE |
| LOW | dashboard/backend/auth/invitation_test.go | 83 | admin := newTestUser(t, svc, "admin@example.com", RoleAdmin, "active") | CODE |
| LOW | dashboard/backend/auth/password_test.go | 63 | admin := newTestUser(t, svc, "admin@example.com", RoleAdmin, "active") | CODE |
| LOW | dashboard/backend/auth/password_test.go | 98 | admin := newTestUser(t, svc, "admin@example.com", RoleAdmin, "active") | CODE |
| LOW | dashboard/backend/auth/password_test.go | 131 | admin := newTestUser(t, svc, "admin@example.com", RoleAdmin, "active") | CODE |
| LOW | dashboard/backend/auth/bootstrap_gate_test.go | 83 | newTestUser(t, svc, "admin@example.com", RoleAdmin, "active") | CODE |
| LOW | dashboard/backend/auth/bootstrap_gate_test.go | 91 | rec := postRegister(svc, "admin@example.com") | CODE |
| LOW | dashboard/backend/auth/bootstrap_gate_test.go | 105 | newTestUser(t, svc, "admin@example.com", RoleAdmin, "active") | CODE |
| LOW | dashboard/backend/auth/bootstrap_gate_test.go | 114 | rec := postRegister(svc, "admin@example.com") | CODE |
| LOW | dashboard/backend/auth/bootstrap_gate_test.go | 125 | rec := postRegister(svc, "admin@example.com") | CODE |
| LOW | dashboard/backend/auth/bootstrap_gate_test.go | 136 | newTestUser(t, svc, "admin@example.com", RoleAdmin, "active") | CODE |
| LOW | dashboard/backend/auth/bootstrap_gate_test.go | 178 | newTestUser(t, svc, "admin@example.com", RoleAdmin, "active") | CODE |
| LOW | dashboard/backend/auth/bootstrap_gate_test.go | 198 | rec := postRegister(svc, "admin@example.com") | CODE |
| LOW | dashboard/backend/auth/bootstrap_gate_test.go | 331 | if rec := postRegister(svc, "admin@example.com"); rec.Code != http.StatusOK { | CODE |
| LOW | dashboard/backend/router/mcp_routes_test.go | 210 | BootstrapAdminEmail: "admin@example.com", | CODE |
| LOW | dashboard/backend/router/mcp_routes_test.go | 250 | body := bytes.NewBufferString(`{"email":"admin@example.com","password":"secret-password"}`) | CODE |
| LOW | e2e/scripts/generate_test_data.py | 122 | "PERSON": ["Jane Smith", "John Doe", "Michael Johnson"], | CODE |
| LOW | …ntic-router/pkg/classification/vllm_classifier_test.go | 18 | ModelEndpoint: config.ClassifierVLLMEndpoint{Address: "placeholder", Port: 1}, | CODE |
| LOW | …antic-router/pkg/classification/classifier_pii_test.go | 218 | piiEntity("PERSON", "John Doe", 0, 8, 0.9), | CODE |
| LOW | …antic-router/pkg/classification/classifier_pii_test.go | 232 | piiEntity("PERSON", "John Doe", 0, 8, 0.9), | CODE |
| LOW | …ntic-router/pkg/classification/http_classifier_test.go | 141 | ModelEndpoint: config.ClassifierVLLMEndpoint{Address: "placeholder", Port: 1}, | CODE |
| LOW | src/vllm-sr/tests/test_runtime_dashboard_bootstrap.py | 48 | monkeypatch.setenv("DASHBOARD_ADMIN_EMAIL", "admin@example.com") | CODE |
| LOW | src/vllm-sr/tests/test_runtime_dashboard_bootstrap.py | 77 | ("DASHBOARD_ADMIN_EMAIL", "admin@example.com"), | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | bench/hallucination/__init__.py | 10 | __all__ = [ | CODE |
| LOW | bench/reasoning/__init__.py | 38 | __all__ = [ | CODE |
| LOW | bench/reasoning/dataset_implementations/__init__.py | 15 | __all__ = [ | CODE |
| LOW | tools/ci/community_lifecycle.py | 34 | __all__ = [ | CODE |
| LOW | tools/agent/scripts/router_calibration_manifest.py | 33 | __all__ = [ | CODE |
| LOW | tools/agent/scripts/router_calibration_fixture_test.py | 153 | def _update_probe_digest(digest: Any, probe_id: str, payload: bytes) -> None: | CODE |
| LOW | tools/agent/scripts/agent_resolution.py | 34 | __all__ = [ | CODE |
| LOW | tools/agent/scripts/router_calibration_support.py | 50 | __all__ = [ | CODE |
| LOW⚡ | tools/agent/scripts/tuning/engine_selection.py | 510 | def _update_mapping_outputs(m: dict, fix: Fix) -> None: | CODE |
| LOW | tools/agent/scripts/tuning/engine_selection.py | 525 | def _update_score_weight(score: dict, signal_name: str, new_weight: float) -> None: | CODE |
| LOW | tools/agent/scripts/tuning/__init__.py | 18 | __all__ = [ | CODE |
| LOW | tools/agent/scripts/tuning/scenarios/__init__.py | 12 | __all__ = ["CalibrationScenario", "PrivacyScenario"] | CODE |
| LOW | tools/mcp-classifier-server/server_generative.py | 68 | logger = logging.getLogger(__name__) | CODE |
| LOW | tools/mcp-classifier-server/server_embedding.py | 68 | logger = logging.getLogger(__name__) | CODE |
| LOW | tools/mcp-classifier-server/server_keyword.py | 59 | logger = logging.getLogger(__name__) | CODE |
| LOW | e2e/testing/llm-katan/llm_katan/server.py | 30 | logger = logging.getLogger(__name__) | CODE |
| LOW | e2e/testing/llm-katan/llm_katan/__init__.py | 24 | __all__ = ["create_app", "ModelBackend", "main"] | CODE |
| LOW | e2e/testing/llm-katan/llm_katan/model.py | 17 | logger = logging.getLogger(__name__) | CODE |
| LOW | e2e/testing/llm-katan/llm_katan/cli.py | 30 | logger = logging.getLogger(__name__) | CODE |
| LOW | e2e/testing/memory_tests/__init__.py | 21 | __all__ = [ | CODE |
| LOW | src/fleet-sim/fleet_sim/__init__.py | 97 | __all__ = [ | CODE |
| LOW | src/fleet-sim/fleet_sim/hardware/__init__.py | 21 | __all__ = [ | CODE |
| LOW | src/fleet-sim/fleet_sim/core/__init__.py | 8 | __all__ = [ | CODE |
| LOW | src/fleet-sim/fleet_sim/optimizer/__init__.py | 35 | __all__ = [ | CODE |
| LOW | src/fleet-sim/fleet_sim/models/__init__.py | 23 | __all__ = [ | CODE |
| LOW | src/fleet-sim/fleet_sim/workload/__init__.py | 6 | __all__ = ["CdfWorkload", "PoissonWorkload", "TraceWorkload"] | CODE |
| LOW | src/fleet-sim/fleet_sim/routing/spillover.py | 43 | def set_pools(self, live_pools) -> None: | CODE |
| LOW | src/fleet-sim/fleet_sim/routing/__init__.py | 11 | __all__ = [ | CODE |
| LOW | src/fleet-sim/fleet_sim/routing/least_loaded.py | 25 | def set_pools(self, live_pools) -> None: | CODE |
| LOW | src/fleet-sim/fleet_sim/gpu_profiles/__init__.py | 9 | __all__ = [ | CODE |
| LOW | src/fleet-sim/fleet_sim/gpu_profiles/profiles.py | 184 | __all__ = [ | CODE |
| LOW | src/training/model_classifier/common_lora_utils.py | 16 | logger = logging.getLogger(__name__) | CODE |
| LOW | …odel_classifier/verify_text_classification_datasets.py | 90 | logger = logging.getLogger(__name__) | CODE |
| LOW | …/model_classifier/user_feedback_classifier/__init__.py | 42 | __all__ = [ | CODE |
| LOW | …edback_classifier/data_processing/download_datasets.py | 47 | logger = logging.getLogger(__name__) | CODE |
| LOW | …ck_classifier/data_processing/verify_dataset_labels.py | 33 | logger = logging.getLogger(__name__) | CODE |
| LOW | …lassifier/user_feedback_classifier/configs/__init__.py | 15 | __all__ = [ | CODE |
| LOW | …model_experiment/multitask_wiki_classifier_training.py | 21 | logger = logging.getLogger(__name__) | CODE |
| LOW | …/multitask_bert_fine_tuning/multitask_bert_training.py | 21 | logger = logging.getLogger(__name__) | CODE |
| LOW | …training/model_embeddings/cache_embeddings/__init__.py | 30 | __all__ = ["setup_logging"] | CODE |
| LOW | …ning/model_embeddings/cache_embeddings/lora_trainer.py | 49 | logger = logging.getLogger(__name__) | CODE |
| LOW | …/model_selection/rl_model_selection/train_gmtrouter.py | 48 | logger = logging.getLogger(__name__) | CODE |
| LOW | …model_selection/rl_model_selection/automix_verifier.py | 33 | logger = logging.getLogger(__name__) | CODE |
| LOW | …/model_selection/rl_model_selection/train_router_r1.py | 52 | logger = logging.getLogger(__name__) | CODE |
| LOW | …model_selection/rl_model_selection/router_r1_server.py | 30 | logger = logging.getLogger(__name__) | CODE |
| LOW | src/vllm-sr/cli/container_cli.py | 37 | __all__ = [ | CODE |
| LOW | src/vllm-sr/cli/model_catalog.py | 58 | __all__ = [ | CODE |
| LOW | src/vllm-sr/cli/config_migration.py | 780 | def _set_if_missing(target: dict[str, Any], key: str, value: Any) -> None: | CODE |
| LOW | src/vllm-sr/cli/storage_backends.py | 454 | __all__ = [ | CODE |
| LOW | src/vllm-sr/cli/commands/__init__.py | 10 | __all__ = ["config_command"] | CODE |
| LOW | src/vllm-sr/cli/commands/recipe_learning_metrics.py | 169 | def update_decision_tier_metrics(metrics: DecisionMetrics, decision_tier: int) -> None: | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | deploy/kubernetes/llm-katan/base/deployment.yaml | 17 | # Create a non-root user for security (matching Dockerfile) | COMMENT |
| MEDIUM | deploy/operator/config/crd/kustomizeconfig.yaml | 1 | # This file is for teaching kustomize how to substitute name and namespace reference in CRD | COMMENT |
| MEDIUM⚡ | e2e/testing/08-rag-openai-test.py | 42 | # Create a request with RAG enabled (direct_search mode) | COMMENT |
| MEDIUM⚡ | e2e/testing/08-rag-openai-test.py | 98 | # Create a request with file_search tool (Responses API format) | COMMENT |
| MEDIUM⚡ | e2e/testing/run_all_tests.py | 153 | # Create a test suite from the module | COMMENT |
| MEDIUM | e2e/scripts/generate_test_data.py | 190 | # Define the PII types we need (matching semantic-router config) | COMMENT |
| MEDIUM | …ng/model_experiment/dual_classifier/dual_classifier.py | 165 | # Create a new instance | COMMENT |
| MEDIUM | …/multitask_bert_fine_tuning/multitask_bert_training.py | 471 | # Create a temporary trainer to load datasets and determine configurations | COMMENT |
| MEDIUM | …/multitask_bert_fine_tuning/multitask_bert_training.py | 514 | # Create the real trainer | COMMENT |
| MEDIUM | src/training/model_eval/plot_category_accuracies.py | 108 | # Create a new dict to avoid modifying the original during iteration | COMMENT |
| MEDIUM | …model_embeddings/cache_embeddings/domains/prompts.yaml | 3 | # This file defines how the LLM should generate paraphrases and hard negatives | COMMENT |
| MEDIUM⚡ | …model_selection/ml_model_selection/models.example.yaml | 3 | # This file defines multiple LLM models with different endpoints and authentication. | COMMENT |
| MEDIUM | …/training/model_selection/ml_model_selection/models.py | 473 | # Create a fitted SVM using sklearn's internal attributes | COMMENT |
| MEDIUM | src/vllm-sr/cli/templates/grafana-datasource.serve.yaml | 2 | # This file is provisioned automatically when Grafana starts | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | bench/reasoning/dataset_factory.py | 58 | Create a dataset instance by name. Args: name: Name of the dataset to create Returns: | STRING |
| HIGH | src/training/model_classifier/common_lora_utils.py | 20 | Get appropriate target_modules for LoRA based on model architecture. Args: model_name: Name of the mod | STRING |
| HIGH | src/training/model_classifier/common_lora_utils.py | 91 | Validate and normalize LoRA configuration parameters. Args: lora_config: Dictionary containing LoRA pa | STRING |
| HIGH | …edback_classifier/data_processing/download_datasets.py | 1329 | Label a list of examples using the LLM API with production features. Features: - Retry logic for failed AP | STRING |
| HIGH | src/vllm-sr/cli/parser.py | 440 | Parse and validate user configuration file. Args: config_path: Path to config.yaml log_summary | STRING |
| HIGH | src/vllm-sr/cli/parser.py | 529 | Load configuration file as dictionary. Args: config_path: Path to configuration file Returns: | STRING |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | bench/reasoning_mode_eval.sh | 11 | # Usage: | COMMENT |
| LOW | bench/grounded_fusion/run_ab.sh | 9 | # Usage: | COMMENT |
| LOW⚡ | tools/smoke/test-authz-rbac.sh | 22 | # Usage: | COMMENT |
| LOW | tools/milvus/test-milvus-deployment.sh | 5 | # Usage: | COMMENT |
| LOW | …tes/dynamo/helm-chart/examples/values-multi-model.yaml | 5 | # Usage: | COMMENT |
| LOW | …etes/dynamo/helm-chart/examples/values-multi-node.yaml | 5 | # Usage: | COMMENT |
| LOW | …classifier/modality_routing_classifier/run_training.sh | 5 | # Usage: | COMMENT |
| LOW | …_feedback_classifier/scripts/label_balanced_dataset.sh | 18 | # Usage: | COMMENT |
| LOW⚡ | …lu_pro_solver_lora/train_all_specialists_no_leakage.sh | 8 | # Usage: | COMMENT |
| LOW | src/training/model_eval/arc_challenge_vllm_eval.py | 3 | # Usage example: | COMMENT |
| LOW | src/training/model_eval/mmlu_pro_vllm_eval.py | 3 | # Sample usage: | COMMENT |
| LOW⚡ | …model_selection/ml_model_selection/models.example.yaml | 6 | # Usage: | COMMENT |
| LOW | …tion/rl_model_selection/scripts/train_router_r1_gpu.sh | 9 | # Usage: | COMMENT |
| LOW | …tion/rl_model_selection/scripts/train_gmtrouter_gpu.sh | 9 | # Usage: | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | e2e/testing/llm-katan/llm_katan/model.py | 713 | CODE | |
| MEDIUM | e2e/testing/llm-katan/llm_katan/model.py | 714 | CODE | |
| MEDIUM | e2e/testing/llm-katan/llm_katan/model.py | 722 | CODE | |
| MEDIUM | e2e/testing/llm-katan/llm_katan/model.py | 725 | CODE | |
| MEDIUM | e2e/testing/llm-katan/llm_katan/model.py | 740 | CODE | |
| MEDIUM | e2e/testing/llm-katan/llm_katan/model.py | 745 | CODE | |
| MEDIUM | e2e/testing/llm-katan/llm_katan/model.py | 751 | CODE | |
| MEDIUM | e2e/testing/llm-katan/llm_katan/model.py | 756 | CODE | |
| MEDIUM | e2e/testing/llm-katan/llm_katan/model.py | 763 | CODE | |
| MEDIUM | e2e/testing/llm-katan/llm_katan/model.py | 768 | CODE | |
| MEDIUM | e2e/testing/llm-katan/llm_katan/model.py | 775 | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | src/fleet-sim/fleet_sim/optimizer/disagg.py | 124 | CODE | |
| LOW | src/fleet-sim/examples/routing_comparison.py | 70 | CODE | |
| LOW | …ck_fine_tuning_lora/fact_check_bert_finetuning_lora.py | 1208 | CODE | |
| LOW | …ng_classifier/modality_routing_bert_finetuning_lora.py | 2201 | CODE | |
| LOW | …ard_fine_tuning_lora/jailbreak_bert_finetuning_lora.py | 481 | CODE | |
| LOW | …pii_model_fine_tuning_lora/pii_bert_finetuning_lora.py | 1124 | CODE | |
| LOW | …ssifier/hallucination_detection_classifier/finetune.py | 358 | CODE | |
| LOW | …ier_model_fine_tuning_lora/ft_qwen3_generative_lora.py | 393 | CODE | |
| LOW | …er/classifier_model_fine_tuning_lora/ft_linear_lora.py | 481 | CODE | |
| LOW | …r_model_fine_tuning_lora/ft_linear_lora_consistency.py | 485 | CODE | |
| LOW | …ro_solver_lora/ft_qwen3_mmlu_solver_lora_no_leakage.py | 1235 | CODE | |
| LOW | …ment/mmlu_pro_solver_lora/ft_qwen3_mmlu_solver_lora.py | 687 | CODE | |
| LOW | …aining/model_selection/ml_model_selection/benchmark.py | 1037 | CODE | |
| LOW | …c/training/model_selection/ml_model_selection/train.py | 86 | CODE | |
| LOW | …c/training/model_selection/ml_model_selection/train.py | 207 | CODE | |
| LOW | src/vllm-sr/cli/config_translator.py | 39 | CODE | |
| LOW | src/vllm-sr/cli/container_backend.py | 20 | CODE | |
| LOW | src/vllm-sr/cli/core.py | 146 | CODE | |
| LOW | src/vllm-sr/cli/k8s_backend.py | 141 | CODE | |
| LOW | src/vllm-sr/cli/container_start.py | 62 | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | …tend/src/pages/ConfigPageMoMWorkspace.contract.test.ts | 57 | expect(usage).not.toContain('YOUR_API_KEY') | CODE |
| HIGH | e2e/testing/09-openai-api-validation-test.py | 47 | "your-api-key", | CODE |
| HIGH | …mantic-router/pkg/openai/openai_validation_e2e_test.go | 18 | var placeholderKeys = []string{"sk-your-key", "sk-your-real-key", "sk-...", "your-api-key"} | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | …oard/frontend/src/pages/ConfigPageModelUsageDialog.tsx | 51 | print(chunk.choices[0].delta.content or "", end="")`, | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | onnx-binding/semantic-router.go | 601 | Probabilities: []float32{}, // TODO: implement probability extraction | CODE |
| LOW | onnx-binding/semantic-router.go | 798 | Probabilities: []float32{}, // TODO: implement probability extraction | CODE |
| LOW | onnx-binding/semantic-router.go | 819 | Probabilities: []float32{}, // TODO: implement probability extraction | CODE |
| LOW | onnx-binding/semantic-router.go | 840 | Probabilities: []float32{}, // TODO: implement probability extraction | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | website/src/pages/community/contributing.tsx | 17 | description={<Translate id="contributing.subtitle">We welcome contributions from the community! Here's how you c | CODE |