add citation count

1 files changed, 48 insertions, 0 deletions

diff --git a/sources.md b/sources.md
index 6406a2b..08bbf2a 100644
--- a/sources.md
+++ b/sources.md
@@ -1,4 +1,6 @@
 # The Probabilistic Relevance Framework: BM25 and Beyond
+**Citations**: 4117
+
 **Domain**: RAG
 
 **Relevance Score**: 4
@@ -77,6 +79,8 @@ explores the theoretical underpinnings, development, and extensions of the Proba
 
 
 # Dense Passage Retrieval for Open-Domain Question Answering
+**Citations**: 3548
+
 **Domain**: RAG
 
 **Relevance Score**: 6
@@ -175,6 +179,8 @@ The paper contributes to the field by providing a robust and efficient dense ret
 - **Integration with Other Models**: The DPR can be combined with other models, such as generative models or more complex reader architectures, to create hybrid systems that leverage the strengths of both retrieval and generation.
 
 # Learning Transferable Visual Models From Natural Language Supervision 
+**Citations**: 28769
+
 **Domain**: OCR
 
 **Relevance Score**: 4
@@ -254,6 +260,8 @@ The paper introduces CLIP (Contrastive Language-Image Pretraining), a scalable f
   - The need for standardized benchmarks to evaluate zero-shot transfer capabilities and broader task learning in computer vision is emphasized, which could help in assessing the true performance of models like CLIP.
 
 # C-RAG: Certified Generation Risks for Retrieval-Augmented Language Models
+**Citations**: 11
+
 **Domain**: RAG
 
 **Relevance Score**: 6.5
@@ -347,6 +355,8 @@ The paper introduces C-RAG, a framework designed to certify and provide theoreti
 - **Future Work on Time-Series Data**: The paper highlights the potential for future research to extend conformal risk analysis to time-series data, which remains an unexplored area but is crucial for practical deployments.
 
 # Atlas: Few-shot Learning with Retrieval Augmented Language Models
+**Citations**: 327
+
 **Domain**: RAG
 
 **Relevance Score**: 6.5
@@ -482,6 +492,8 @@ The paper presents Atlas, a retrieval-augmented language model designed to excel
 - **Exploration of Temporal Knowledge**: The ability to update the model's knowledge base in real-time opens up opportunities for research into temporal knowledge representation and reasoning, which could be crucial for applications requiring up-to-date information.
 
 # REST: Retrieval-Based Speculative Decoding
+**Citations**: 72
+
 **Domain**: RAG
 
 **Relevance Score**: 7
@@ -568,6 +580,8 @@ The paper introduces Retrieval-Based Speculative Decoding (REST), a novel algori
 - **Exploration of Datastore Construction**: Future work could focus on constructing datastores from content generated by the LLM itself to improve alignment and relevance.
 
 # Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
+**Citations**: 5758
+
 **Domain**: RAG
 
 **Relevance Score**: 7
@@ -677,6 +691,8 @@ The paper presents Retrieval-Augmented Generation (RAG), a novel approach that c
 - **Broader Applications**: The RAG framework can be extended to various NLP tasks beyond those evaluated in the paper, such as dialogue systems, summarization, and other forms of question answering.
 
 # REALM: retrieval-augmented language model pre-training
+**Citations**: 2074
+
 **Domain**: Foundation model + RAG
 
 **Relevance Score**: 7
@@ -763,6 +779,8 @@ The paper details the architecture of REALM, which consists of a neural knowledg
 
 
 # Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering 
+**Citations**: 1140
+
 **Domain**: RAG
 
 **Relevance Score**: 7
@@ -863,6 +881,8 @@ The performance of the Fusion-in-Decoder model improves significantly with an in
 - **Combining with Other Modalities**: Future work could explore the integration of multimodal data (e.g., images, videos) alongside text to enhance the model's capabilities in answering questions that require diverse forms of evidence.
 
 # Retrieval Augmented Code Generation and Summarization
+**Citations**: 165
+
 **Domain**: RAG
 
 **Relevance Score**: 8
@@ -991,6 +1011,8 @@ The paper presents REDCODER, a retrieval-augmented framework designed to enhance
 - **Broader Application**: The framework could be adapted for use in various software engineering applications beyond code generation and summarization, such as code review, debugging, and documentation generation.
 
 # DocPrompting: Generating Code by Retrieving the Docs
+**Citations**: 136
+
 **Domain**: RAG
 
 **Relevance Score**: 8
@@ -1090,6 +1112,8 @@ The paper introduces a novel approach called DocPrompting, which enhances natura
 The paper provides a comprehensive survey of Retrieval-Augmented Generation (RAG) techniques for enhancing Large Language Models (LLMs). It discusses the limitations of LLMs, such as hallucination and outdated knowledge, and presents RAG as a solution that integrates external knowledge sources to improve the accuracy and credibility of generated content. The authors categorize RAG into three paradigms: Naive RAG, Advanced RAG, and Modular RAG, each representing a progression in methodology and effectiveness. The paper meticulously examines the core components of RAG, including retrieval, generation, and augmentation techniques, while also highlighting state-of-the-art technologies and evaluation frameworks.
 
 # Document Language Models, Query Models, and Risk Minimization for Information Retrieval
+**Citations**: 1132
+
 **Domain**: RAG
 
 **Relevance Score**: 7
@@ -1204,6 +1228,8 @@ The paper presents a novel framework for information retrieval that integrates d
   - Expanding the framework to accommodate more complex query types and longer queries.
 
 # A Neural Corpus Indexer for Document Retrieval
+**Citations**: 123
+
 **Domain**: RAG
 
 **Relevance Score**: 7
@@ -1312,6 +1338,8 @@ The paper presents the Neural Corpus Indexer (NCI), an innovative end-to-end dee
 - **Semantic Clustering for Efficient Retrieval**: Investigating methods to group documents into semantic clusters could allow NCI to retrieve relevant cluster identifiers, improving efficiency in document retrieval.
 
 # TIARA: Multi-grained Retrieval for Robust Question Answering over Large Knowledge Base
+**Citations**: 81
+
 **Domain**: RAG
 
 **Relevance Score**: 6
@@ -1456,6 +1484,8 @@ The experimental results demonstrate that TIARA significantly outperforms previo
   - Investigating more efficient retrieval mechanisms to enhance the speed and scalability of the model for practical applications.
 
 # Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection
+**Citations**: 493
+
 **Domain**: RAG
 
 **Relevance Score**: 9
@@ -1561,6 +1591,8 @@ The paper introduces a novel framework called Self-Reflective Retrieval-Augmente
 - **Integration with Other Techniques**: There is scope for integrating Self-RAG with other advancements in natural language processing, such as reinforcement learning from human feedback (RLHF) or multi-modal models, to further enhance its capabilities.
 
 # Precise Zero-Shot Dense Retrieval without Relevance Labels
+**Citations**: 247
+
 **Domain**: RAG
 
 **Relevance Score**: 9
@@ -1715,6 +1747,8 @@ The paper introduces Corrective Retrieval Augmented Generation (CRAG), a novel a
   - The execution time may increase, especially when processing multiple retrieval actions.
 
 # Re2G: Retrieve, Rerank, Generate
+**Citations**: 115
+
 **Domain**: RAG
 
 **Relevance Score**: 7
@@ -1859,6 +1893,8 @@ The paper presents a novel approach called Re2G (Retrieve, Rerank, Generate), wh
 - **Future Research Directions**: The paper suggests that further experiments could be conducted on domain adaptation of the Re²G framework for specific tasks like question answering or dialog, which may provide insights into its application in real-world scenarios.
 
 # Active Retrieval Augmented Generation
+**Citations**: 421
+
 **Domain**: RAG
 
 **Relevance Score**: 9
@@ -1954,6 +1990,8 @@ The paper presents a novel approach called Forward-Looking Active Retrieval Augm
 - **Future Research Directions**: The paper suggests exploring better strategies for active retrieval, such as refining the mechanisms for determining when to retrieve and improving the efficiency of the integration of retrieval and generation.
 
 # Fine Tuning vs. Retrieval Augmented Generation for Less Popular Knowledge
+**Citations**: 20
+
 **Domain**: PEFT + RAG
 
 **Relevance Score**:
@@ -2055,6 +2093,7 @@ The research emphasizes the importance of customizing LMs for less-resourced dom
 - **Exploration of Other Domains**: Testing the proposed methods on a wider range of datasets and domains to assess generalizability and effectiveness in different contexts.
 
 # Evaluating Retrieval Quality in Retrieval-Augmented Generation
+**Citations**: 61
 
 **Domain**: RAG
 
@@ -2142,6 +2181,7 @@ The authors conducted extensive experiments across various datasets, demonstrati
 - **Future Research Directions**: The paper opens avenues for further research in improving the evaluation of retrieval models, particularly in exploring different LLM architectures and their impact on retrieval performance.
 
 # Benchmarking Large Language Models in Retrieval-Augmented Generation
+**Citations**: 346
 
 **Domain**: RAG
 
@@ -2269,6 +2309,8 @@ The paper does not utilize a pre-existing dataset; instead, it creates a new dat
 - **Benchmarking Framework**: The introduction of the RGB benchmark provides a structured framework for evaluating RAG capabilities in LLMs, which can be utilized in future studies to assess new models or improvements in existing ones.
 
 # How Much Knowledge Can You Pack Into the Parameters of a Language Model?
+**Citations**: 900
+
 **Domain**: Foundation
 
 **Relevance Score**: 
@@ -2353,6 +2395,8 @@ The study highlights the trade-offs of storing knowledge within model parameters
 - **Efficiency Improvements**: Future research could focus on developing more efficient language models that maintain high performance while reducing computational requirements.
 
 # Retrieval-Enhanced Machine Learning
+**Citations**: 64
+
 **Domain**: Retrieval
 
 **Relevance Score**:
@@ -2458,6 +2502,8 @@ The authors outline the core principles of REML, including querying, retrieval,
 - **Research Agenda for Future Work**: The authors outline a comprehensive research agenda that includes exploring optimization strategies, feedback mechanisms, and evaluation methodologies, paving the way for further advancements in retrieval-enhanced machine learning.
 
 # Can Knowledge Graphs Reduce Hallucinations in LLMs
+**Citations**: 86
+
 **Domain**: Knowledge Graph
 
 **Relevance Score**:
@@ -2554,6 +2600,8 @@ The effectiveness of the various KG-augmented methods is assessed through empiri
   - Investigating the synergistic relationship between LLMs and KGs for mutual enhancement.
 
 # Retrieval Augmentation Reduces Hallucination in Conversation
+**Citations**: 663
+
 **Domain**: RAG
 
 **Relevance Score**: