port of generator

.gitignore

@@ -0,0 +1 @@
+.DS_Store

Dockerfile

@@ -0,0 +1,23 @@
+# -------- base image --------
+FROM python:3.11-slim
+
+ENV PYTHONUNBUFFERED=1 \
+    OMP_NUM_THREADS=1 \
+    TOKENIZERS_PARALLELISM=false 
+    #GRADIO_MCP_SERVER=True         
+
+# -------- install deps --------
+WORKDIR /app
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+
+# -------- copy source --------
+COPY app ./app
+COPY params.cfg .
+COPY .env* ./
+
+# Ports:
+#  • 7860 → Gradio UI (HF Spaces standard)
+EXPOSE 7860
+
+CMD ["python", "-m", "app.main"]

README.md

@@ -1,10 +1,20 @@
 ---
-title: Chatfed Reranker
-emoji: 🌍
-colorFrom: yellow
-colorTo: red
+title: ChatFed Re-Ranker Service
+emoji: 🤖
+colorFrom: blue
+colorTo: purple
 sdk: docker
 pinned: false
+license: mit
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# ReRanker Module
+
+This is an LLM-based generation service designed to be deployed as a modular component of a broader RAG system. The service runs on a docker container and exposes a gradio UI on port 7860 as well as an MCP endpoint.
+
+## Configuration
+
+1. The module requires an API key (set as an environment variable) for a model provider to run. Make sure to set the appropriate environment variables:
+- HuggingFace: `HF_TOKEN` 
+
+2. Inference provider and model settings are accessible via params.cfg

app/main.py

@@ -0,0 +1,39 @@
+import gradio as gr
+from .reranker import rerank_context
+
+# ---------------------------------------------------------------------
+# Gradio Interface with MCP support
+# ---------------------------------------------------------------------
+ui = gr.Interface(
+    fn=rerank_context,
+    inputs=[
+        gr.Textbox(
+            label="Query", 
+            lines=2, 
+            placeholder="Paste user query here",
+            info="Enter user query"
+        ),
+        gr.Textbox(
+            label="Retrieved Context", 
+            lines=8, 
+            placeholder="Paste retrieved context here",
+            info="Provide the context/documents to use for reranking"
+        ),
+    ],
+    outputs=gr.Textbox(
+        label="Reranked Context",
+        lines=6,
+        show_copy_button=True
+    ),
+    title="RAG Reranking Service UI",
+    description="Reranks previously retrieved context. Intended for use in RAG pipelines (i.e. context supplied by semantic retriever service) as an MCP server.",
+)
+
+# Launch with MCP server enabled
+if __name__ == "__main__":
+    ui.launch(
+        server_name="0.0.0.0",
+        server_port=7860,
+        mcp_server=True,
+        show_error=True
+    )

app/reranker.py

@@ -0,0 +1,41 @@
+from typing import List, Dict, Any
+from langchain.retrievers.document_compressors import CrossEncoderReranker
+from langchain_community.cross_encoders import HuggingFaceCrossEncoder
+from langchain.schema import Document
+from .utils import getconfig
+
+config = getconfig("params.cfg")
+
+# load ranker settings from your existing config
+
+RANKER_MODEL = config.get("reranker", "MODEL")
+RANKER_TOP_K  = int(config.get("reranker", "TOP_K"))
+
+def rerank_context(
+    query: str,
+    contexts: List[Dict[str, Any]],
+    top_n: int = None
+) -> List[Dict[str, Any]]:
+    """
+    Re-ranks a list of context dicts (each with 'page_content' & 'metadata')
+    using a cross-encoder and returns the top_n sorted results.
+    """
+    # wrap into LangChain Documents
+    docs = [
+        Document(page_content=c["page_content"], metadata=c.get("metadata", {}))
+        for c in contexts
+    ]
+
+    # instantiate reranker
+    n = top_n or RANKER_TOP_K
+    model    = HuggingFaceCrossEncoder(model_name=RANKER_MODEL)
+    reranker = CrossEncoderReranker(model=model, top_n=n)
+
+    # perform reranking
+    reranked: List[Document] = reranker.rerank(query, docs)
+
+    # return as plain dicts
+    return [
+        {"page_content": d.page_content, "metadata": d.metadata}
+        for d in reranked
+    ]

app/utils.py

@@ -0,0 +1,16 @@
+import configparser
+import logging
+
+def getconfig(configfile_path: str):
+    """
+    Read the config file
+    Params
+    ----------------
+    configfile_path: file path of .cfg file
+    """
+    config = configparser.ConfigParser()
+    try:
+        config.read_file(open(configfile_path))
+        return config
+    except:
+        logging.warning("config file not found")

params.cfg

@@ -0,0 +1,4 @@
+[reranker]
+MODEL = BAAI/bge-reranker-v2-m3  
+TOP_K = 5  
+

requirements.txt

@@ -0,0 +1,21 @@
+# Core dependencies
+gradio>=4.0.0
+gradio[mcp]
+python-dotenv>=1.0.0
+
+# LangChain core
+langchain-core>=0.1.0
+langchain-community>=0.0.1
+
+# Provider-specific LangChain packages
+langchain-openai>=0.1.0
+langchain-anthropic>=0.1.0
+langchain-cohere>=0.1.0
+langchain-together>=0.1.0
+langchain-huggingface>=0.0.1
+
+# Additional dependencies that might be needed
+requests>=2.31.0
+pydantic>=2.0.0
+sentence-transformers>=2.2.2  # Required for cross-encoders
+torch>=2.0.0  # Required for HuggingFace models