Update README.md

README.md

@@ -8,7 +8,6 @@ tags:
 - dynamic-scaling
 - api-retrieval
 - tool-discovery
-- library-management
 license: mit
 ---
 
@@ -20,29 +19,24 @@ A sophisticated query embedding model designed to act as an intelligent "librari
 
 This model serves as a **Librarian of Tools** - an AI system that understands user intentions and finds the most appropriate tools, APIs, or functions to accomplish their tasks. It's particularly effective for:
 
-- **API Discovery**: Finding relevant APIs from large collections
-- **Tool Recommendation**: Suggesting appropriate tools for specific tasks
-- **Function Retrieval**: Matching queries to available functions or methods
-- **Library Management**: Organizing and retrieving from tool libraries
+- **API Discovery**: Finding relevant rapid APIs from large collections
 - **Iterative Search**: Progressive tool discovery with residual-based refinement
 
 ## 🏗️ Model Architecture
 
 - **Base Model**: ToolBench/ToolBench_IR_bert_based_uncased
-- **Architecture**: Dynamic direction-focused query embedding with scale prediction
+- **Architecture**: Query embedding with scale prediction
 - **Special Features**: 
-  - **Dynamic Scale Prediction**: Each query gets its own optimal scaling factor
-  - **Direction-Focused Training**: Prioritizes semantic alignment over magnitude
   - **Query-Specific Adaptation**: Tailors embeddings to individual query characteristics
   - **Balanced Magnitude Handling**: Maintains appropriate scaling for retrieval tasks
-  - **Residual-Based Iteration**: Supports iterative greedy search for comprehensive tool discovery
+  - **Residual-Based Iteration**: Supports iterative search for comprehensive tool discovery
 
 ## 🎓 Training Strategy
 
 - **Training Approach**: Dynamic direction-focused with AdamW optimizer
 - **Loss Function**: Combined MSE, direction loss, and magnitude loss
-- **Scale Prediction**: Uses softplus + 1 activation for 1+ scale factors
-- **Dataset**: Trained on normalized query-API pairs for optimal tool retrieval
+- **Scale Prediction**: Uses softplus + 1 activation for 1 + scale factors
+- **Dataset**: Trained on query-(sum vector of relevant API embeddings) pairs from ToolBench(https://github.com/OpenBMB/ToolBench)
 
 ## 🚀 Usage
 
@@ -93,9 +87,9 @@ print("Query similarity matrix:")
 print(similarity_matrix)
 ```
 
-### Iterative Greedy Search for Comprehensive Tool Discovery
+### Iterative Search for Comprehensive Tool Discovery
 
-The model supports iterative greedy search, which progressively discovers tools by removing found APIs from the query representation and continuing the search:
+The model supports iterative search, which progressively discovers tools by removing found APIs from the query representation and continuing the search:
 
 ```python
 import torch
@@ -103,14 +97,14 @@ import torch.nn.functional as F
 import numpy as np
 
 class LibrarianSearch:
-    def __init__(self, model, tokenizer, vector_db_index, documents, threshold=0.6):
+    def __init__(self, model, tokenizer, vector_db_index, documents, threshold=0.5):
         self.model = model
         self.tokenizer = tokenizer
         self.index = vector_db_index
         self.documents = documents
         self.threshold = threshold
     
-    def get_query_embedding(self, query, normalize=True):
+    def get_query_embedding(self, query, normalize=False):
         """Get query embedding using the Librarian model"""
         inputs = self.tokenizer(
             query, 
@@ -245,77 +239,23 @@ def beam_search_iterative(self, query, beam_size=5):
     pass
 ```
 
-## 🎯 Use Cases
-
-### 1. API Discovery
-```python
-# Find APIs for specific tasks
-query = "Convert image to different formats"
-# Model will help find relevant image processing APIs
-```
-
-### 2. Tool Recommendation
-```python
-# Recommend tools for data analysis
-query = "Analyze time series data and create visualizations"
-# Model will suggest appropriate data analysis tools
-```
-
-### 3. Function Retrieval
-```python
-# Find functions in code libraries
-query = "Calculate distance between two geographic coordinates"
-# Model will help locate relevant geospatial functions
-```
-
-### 4. Multi-Tool Discovery
-```python
-# Find multiple tools for complex workflows
-query = "Send email notifications, process uploaded files, and generate reports"
-# Iterative search will find email, file processing, and reporting tools
-```
-
-## 📊 Performance
-
-This model excels at:
-- **Semantic Understanding**: Captures nuanced differences between similar tool requests
-- **Direction Alignment**: Ensures embeddings point in the right semantic direction
-- **Magnitude Optimization**: Maintains appropriate scaling for retrieval systems
-- **Query Adaptation**: Tailors responses to specific query characteristics
-- **Iterative Discovery**: Progressively finds multiple relevant tools through residual-based search
-- **Beam Search Optimization**: Finds optimal combinations of tools for complex queries
-
-## 🔧 Integration
-
-The model is designed to work seamlessly with:
-- Vector databases (FAISS, Pinecone, Weaviate)
-- Retrieval systems
-- Recommendation engines
-- Tool discovery platforms
-- API marketplaces
-- Iterative search frameworks
-
 ## 📚 Citation
 
 If you use this model in your research or applications, please cite:
 
 ```bibtex
 @misc{librarian_of_tools,
-  title={Librarian of Tools: Dynamic Direction-Focused Query Embedding for Tool Discovery},
+  title={Librarian of Tools},
   author={jhleepidl},
-  year={2024},
-  url={https://huggingface.co/jhleepidl/librarian}
+  year={2025},
+  url={https://github.com/jhleepidl/librarian}
 }
 ```
 
-## 🤝 Contributing
-
-This model is part of a larger effort to improve tool discovery and API retrieval. Contributions and feedback are welcome!
-
 ## 📄 License
 
 This model is released under the MIT License, making it suitable for both research and commercial applications.
 
 ---
 
-**The Librarian of Tools** - Your intelligent assistant for discovering the right tools for any task! 🛠️📚 
+**The Librarian of Tools** - Your intelligent assistant for discovering the right tools for any task! 🛠️📚