Publish AgentWeave Router MiniLM from e4e9dce

README.md

@@ -1,8 +1,5 @@
 ---
 license: apache-2.0
-library_name: sentence-transformers
-pipeline_tag: feature-extraction
-base_model: sentence-transformers/all-MiniLM-L6-v2
 tags:
   - agentweave
   - agentic-ai
@@ -13,8 +10,11 @@ tags:
   - minilm
   - sentence-transformers
   - pre-inference-routing
+  - arxiv:2608.23078
 language:
   - en
+datasets:
+  - sauravsingla08/AgentWeave-Tool-Routing
 ---
 
 # AgentWeave Router MiniLM 🧭
@@ -25,6 +25,17 @@ A lightweight, CPU-first semantic capability router for **AgentWeave**. It uses
 
 This repository is intentionally small: it publishes the AgentWeave routing configuration, route prototypes, and executable router code while reusing the upstream MiniLM encoder at runtime instead of copying its weights.
 
+## Artifact type and dependency boundary
+
+This Hub repository is a **routing artifact**, not a derived MiniLM checkpoint.
+
+- No MiniLM weights are included in this repository.
+- No fine-tuning, adapter training, quantization, merge, or weight transformation is claimed.
+- `sentence-transformers/all-MiniLM-L6-v2` is downloaded separately at runtime and used as a **frozen encoder dependency**.
+- AgentWeave-specific behavior comes from the route taxonomy, human-readable prototypes, and cosine-ranking logic published here.
+
+For that reason the Hub metadata intentionally does not declare this repository as a `base_model` derivative or as a standalone SentenceTransformers checkpoint.
+
 ## Why this exists
 
 Tool-rich agents can expose large action spaces to a language model. AgentWeave explores a complementary systems strategy: reduce the candidate action space *before* model reasoning. This model repository provides an experimental semantic routing companion to AgentWeave's default deterministic routing path.
@@ -47,7 +58,7 @@ Task / user request
         │
         ▼
 all-MiniLM-L6-v2
-  384-d embedding
+frozen 384-d encoder
         │
         ├──────────────┐
         ▼              ▼
@@ -66,6 +77,8 @@ query vector     route prototypes
 
 ## Quick start
 
+Clone or download the files in this repository, then run:
+
 ```bash
 pip install -r requirements.txt
 python router.py "research the latest protocol changes, verify the sources, and summarize the findings"
@@ -120,7 +133,7 @@ Relevance routing also does **not** grant permission to execute a tool. Policy f
 |---|---|
 | `router.py` | CPU semantic router implementation |
 | `route_prototypes.json` | Human-readable capability prototypes |
-| `config.json` | Base model and routing configuration |
+| `config.json` | Frozen encoder dependency and routing configuration |
 | `requirements.txt` | Minimal runtime dependencies |
 
 ## Intended use
@@ -138,6 +151,7 @@ Not intended as:
 - a calibrated confidence model
 - an authorization engine
 - a safety classifier
+- a standalone MiniLM or SentenceTransformers checkpoint
 - a replacement for downstream function-calling evaluation
 
 ## Limitations
@@ -148,13 +162,12 @@ Not intended as:
 - the route taxonomy is intentionally compact
 - domain-specific tools may need custom prototypes
 
-## Source
-
-AgentWeave source code and research artifacts are maintained at:
-
-`https://github.com/sauravsingla/AgentWeave`
+## Related research artifacts
 
-The Hugging Face Space provides an interactive companion experience under the same project name.
+- Paper: https://arxiv.org/abs/2608.23078
+- Dataset: https://huggingface.co/datasets/sauravsingla08/AgentWeave-Tool-Routing
+- Interactive Space: https://huggingface.co/spaces/sauravsingla08/AgentWeave
+- Source: https://github.com/sauravsingla/agentweave
 
 ## License