vllm-sr 0.1.0b2.dev20260202073049__py3-none-any.whl → 0.1.0b2.dev20260202094318__py3-none-any.whl

cli/templates/router-defaults.yaml

@@ -1,17 +1,3 @@
-# You can override by specifying your own mappings below:
-# Uncomment and customize if you need different model mappings:
-# mom_registry:
-#   "models/mom-domain-classifier": "LLM-Semantic-Router/lora_intent_classifier_bert-base-uncased_model"
-#   "models/mom-pii-classifier": "LLM-Semantic-Router/lora_pii_detector_bert-base-uncased_model"
-#   "models/mom-jailbreak-classifier": "LLM-Semantic-Router/lora_jailbreak_classifier_bert-base-uncased_model"
-#   "models/mom-halugate-detector": "KRLabsOrg/lettucedect-base-modernbert-en-v1"
-#   "models/mom-halugate-sentinel": "LLM-Semantic-Router/halugate-sentinel"
-#   "models/mom-halugate-explainer": "tasksource/ModernBERT-base-nli"
-#   "models/mom-feedback-detector": "llm-semantic-router/feedback-detector"
-#   "models/mom-embedding-pro": "Qwen/Qwen3-Embedding-0.6B"
-#   "models/mom-embedding-flash": "google/embeddinggemma-300m"
-#   "models/mom-embedding-ultra": "llm-semantic-router/mmbert-embed-32k-2d-matryoshka"
-
 # Response API Configuration
 # Enables OpenAI Response API support with conversation chaining
 response_api:
@@ -59,25 +45,26 @@ tools:
 
 prompt_guard:
   enabled: true  # Global default - can be overridden per category with jailbreak_enabled
-  use_modernbert: false
-  model_id: "models/mom-jailbreak-classifier"
-  jailbreak_mapping_path: "models/mom-jailbreak-classifier/jailbreak_type_mapping.json"
+  use_mmbert_32k: true
+  model_id: "models/mmbert32k-jailbreak-detector-merged"
+  jailbreak_mapping_path: "models/mmbert32k-jailbreak-detector-merged/jailbreak_type_mapping.json"
   threshold: 0.7
   use_cpu: true
 
 # Classifier configuration
 classifier:
   category_model:
-    model_id: "models/mom-domain-classifier"
+    model_id: "models/mmbert32k-intent-classifier-merged"
+    use_mmbert_32k: true
     threshold: 0.6
     use_cpu: true
-    category_mapping_path: "models/mom-domain-classifier/category_mapping.json"
+    category_mapping_path: "models/mmbert32k-intent-classifier-merged/category_mapping.json"
   pii_model:
-    model_id: "models/mom-pii-classifier"
-    use_modernbert: false
+    model_id: "models/mmbert32k-pii-detector-merged"
+    use_mmbert_32k: true
     threshold: 0.9
     use_cpu: true
-    pii_mapping_path: "models/mom-pii-classifier/pii_type_mapping.json"
+    pii_mapping_path: "models/mmbert32k-pii-detector-merged/pii_type_mapping.json"
 
 # Hallucination mitigation configuration
 # Disabled by default - enable in decisions via hallucination plugin
@@ -85,9 +72,10 @@ hallucination_mitigation:
   enabled: false
   # Fact-check classifier: determines if a prompt needs fact verification
   fact_check_model:
-    model_id: "models/mom-halugate-sentinel"
+    model_id: "models/mmbert32k-factcheck-classifier-merged"
     threshold: 0.6
     use_cpu: true
+    use_mmbert_32k: true
   # Hallucination detector: verifies if LLM response is grounded in context
   hallucination_model:
     model_id: "models/mom-halugate-detector"
@@ -109,9 +97,10 @@ hallucination_mitigation:
 # Classifies user feedback into 4 types: satisfied, need_clarification, wrong_answer, want_different
 feedback_detector:
   enabled: true
-  model_id: "models/mom-feedback-detector"
+  model_id: "models/mmbert32k-feedback-detector-merged"
   threshold: 0.7
   use_cpu: true
+  use_mmbert_32k: true
 
 # External models configuration
 # Used for advanced routing signals like preference-based routing via external LLM

{vllm_sr-0.1.0b2.dev20260202073049.dist-info → vllm_sr-0.1.0b2.dev20260202094318.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: vllm-sr
-Version: 0.1.0b2.dev20260202073049
+Version: 0.1.0b2.dev20260202094318
 Summary: vLLM Semantic Router - Intelligent routing for Mixture-of-Models
 Author: vLLM-SR Team
 License: Apache-2.0
@@ -54,7 +54,8 @@ pip install -e .
 vllm-sr init
 
 # Start the router (includes dashboard)
-vllm-sr serve
+# Provide your HF_TOKEN to run the evaluation tests; this is required for downloading the necessary datasets
+HF_TOKEN=hf_xxx vllm-sr serve
 
 # Open dashboard in browser
 vllm-sr dashboard

{vllm_sr-0.1.0b2.dev20260202073049.dist-info → vllm_sr-0.1.0b2.dev20260202094318.dist-info}/RECORD

@@ -28,10 +28,10 @@ cli/templates/grafana-datasource.serve.yaml,sha256=Cxjz1zVWoUdSzbSsS_iJhMHRrmRi6
 cli/templates/grafana.serve.ini,sha256=x9bCkzxqm5gC4fKToY2lhNPdWhwAaJGVe5ABMW6Dv-c,1674
 cli/templates/llm-router-dashboard.serve.json,sha256=pwnTjUh7z3_3LnIwtaLXjDWH4aHd2Mc57z0oekgt-Bk,60903
 cli/templates/prometheus.serve.yaml,sha256=MGYq8dlRq_i2m5sogQ--kwTvJpkf44QQoCNoI7oyVT8,270
-cli/templates/router-defaults.yaml,sha256=crPnhOGAQYMgnIjHJEU8aNtlplau8wjrvGLrjqPsnwY,8647
+cli/templates/router-defaults.yaml,sha256=ar-tv3fI8nCind6Ci58gYO_ZbxNtl9BawFfeDpiYXws,7851
 cli/templates/tools_db.json,sha256=CPqPBkd5nc966m1YEozz06frrmv3Pd5rrkxKkO3rTiA,4537
-vllm_sr-0.1.0b2.dev20260202073049.dist-info/METADATA,sha256=yWySmdgOvteP453pmYwdRlqWQWyQVnmo-OTpYUaeVkY,7173
-vllm_sr-0.1.0b2.dev20260202073049.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
-vllm_sr-0.1.0b2.dev20260202073049.dist-info/entry_points.txt,sha256=WhlBPbLHUpWUsMuUQX9cnvsYMf0ih5i57vvJ1jJNi0k,42
-vllm_sr-0.1.0b2.dev20260202073049.dist-info/top_level.txt,sha256=2ImG917oaVHlm0nP9oJE-Qrgs-fq_fGWgba2H1f8fpE,4
-vllm_sr-0.1.0b2.dev20260202073049.dist-info/RECORD,,
+vllm_sr-0.1.0b2.dev20260202094318.dist-info/METADATA,sha256=iaG3L_p5hWOvRaCgZNIcjrzbao7Kw9dUYS-sz-J5j_M,7298
+vllm_sr-0.1.0b2.dev20260202094318.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+vllm_sr-0.1.0b2.dev20260202094318.dist-info/entry_points.txt,sha256=WhlBPbLHUpWUsMuUQX9cnvsYMf0ih5i57vvJ1jJNi0k,42
+vllm_sr-0.1.0b2.dev20260202094318.dist-info/top_level.txt,sha256=2ImG917oaVHlm0nP9oJE-Qrgs-fq_fGWgba2H1f8fpE,4
+vllm_sr-0.1.0b2.dev20260202094318.dist-info/RECORD,,