#CodeSearch

1 month ago

For code-related RAG, BM25 and hybrid search often outperform pure vector databases. Commenters noted faster, more accurate results when combining BM25 with vector embeddings. Pure vector search can struggle with code's unique structure. #CodeSearch 2/6

0 0 1 0

Ioannis Polyzos

@ipolyzos.com

2 months ago

Repogrep ultra fast codebase search

As codebases and AI-generated code scale, text-based search breaks down.

RepoGrep explores semantic-aware code search — a key building block for better reviews, refactors, and AI-assisted engineering.

app.ami.dev/repogrep

#DeveloperTools #CodeSearch

1 0 0 0

@bloomville.bsky.social

3 months ago

GitHub - sourcegraph/zoekt: Fast trigram based code search Fast trigram based code search . Contribute to sourcegraph/zoekt development by creating an account on GitHub.

Need to turbocharge your code search? Zoekt's fast trigram search, now on @sourcegraph's GitHub, could be a game-changer for dev teams aiming to boost productivity. Curious how it stacks up against other tools? Dive in: github #CodeSearch https://github.com/sourcegraph/zoekt

0 0 0 0

FreelanceBar.me

@freelancebarme.bsky.social

5 months ago

GitHub Copilot gets smarter at finding your code: Inside our new embedding model Learn about a new Copilot-Embedding model that makes code search in VS Code faster, lighter on memory, and far more accurate.

Introducing a new embedding model for GitHub Copilot, enhancing code search in VS Code with faster, lighter memory usage, and improved accuracy. #GitHub #CodeSearch github.blog/news-insights/product-ne...

1 0 0 0

GetNews.me

@getnews-me.bsky.social

5 months ago

LLM-as-a-Judge Enhances Code Search Across Five Programming Languages

The study of C, Java, JavaScript, Go and Python found semantic retrieval excels for Python, while sparse retrieval works better for C and Go. getnews.me/llm-as-a-judge-enhances-... #codesearch #llm #programming

1 0 0 0

EveryDev AI

@everydevai.bsky.social

6 months ago

Sourcegraph Cody - AI Tool for Devs | EveryDev.ai Cody is an AI coding assistant that helps developers understand, write, and fix code faster by leveraging advanced search capabilities and…

Sourcegraph Cody
Cody by Sourcegraph is now listed. Great when you need an assistant that actually understands your codebase.
www.everydev.ai/tools/sourc...

#CodeSearch #AICompanion

0 0 0 0

AI Coding Weekly

@handle.invalid

8 months ago

But fast search isn’t enough — codebases change constantly. So we built a smart indexing pipeline that matches a live snapshot of your codebase… …and transparently falls back to full search for new/changed code while a new index is built. Augment Code has developed a smart indexing pipeline that significantly enhances search speed for large codebases, achieving over a 40% reduction in retrieval latency while maintaining 99.9% accuracy. The system utilizes approximate nearest neighbor (ANN) search through quantization, allowing for faster context retrieval in codebases exceeding 100 million lines of code. This innovative approach ensures that users receive relevant search results in real-time, even as codebases evolve, by seamlessly falling back to full search for new or changed code during indexing. As stated, 'AI for software doesn’t just mean “smarter” — it has to mean faster, too.'

Augment Code's smart indexing pipeline is a game-changer for large codebases! 40% faster retrieval with 99.9% accuracy, plus seamless fallback for new/changed code. As @augmentcode says: "AI for software doesn't just mean 'smarter' — it has to mean faster, too."

#DevTools #AI #CodeSearch

1 0 0 0

Linghua Jin

@badmonster0.bsky.social

9 months ago

Real-time updates, syntax-aware chunking.
Production-ready. Ultra-performant. Fully #OpenSource.

#LLM #RAG #AIForCode #CodeSearch #DevTools #RealtimeAI #OpenSource #AIInfra #Rust #Python #TreeSitter #VectorDB #Embeddings #AIEngineering #GenerativeAI #Codex #CodingAgents #Claude #Cursor

5 1 0 0

School of Computer Science, University of Sydney

@sydneycompsci.bsky.social

10 months ago

MM-SCS: Leveraging Multimodal Features to Enhance Smart Contract Code Search Semantic code search technology allows searching for existing code snippets through natural language, which can greatly improve programming efficiency. Smart contracts, programs that run on the blockchain, have a code reuse rate of more than 79%, which means developers have a great demand for semantic code search tools. However, the existing code search models still have a semantic gap between code and query and perform poorly on specialized queries of smart contracts. In this paper, we propose a Multi-Modal Smart contract Code Search (MM-SCS) model. Specifically, we construct a Contract Elements Dependency Graph (CEDG) for MM-SCS as an additional modality to capture the data flow and control flow information of the code. To make the model more focused on the key contextual information, we use a multi-head attention network to generate embeddings for code features. In addition, we use a fine-tuned pretrained model to ensure the model's effectiveness when the training data is small. We compared MM-SCS with four state-of-the-art models on a dataset with 470K (code, docstring) pairs collected from Github and Etherscan. Experimental results show that MM-SCS achieves an MRR (Mean Reciprocal Rank) of 0.572, outperforming four state-of-the-art models UNIF, DeepCS, CARLCS-CNN, and TAB-CS by 34.2%, 59.3%, 36.8%, and 14.1%, respectively. Additionally, the search speed of MM-SCS is second only to UNIF, reaching 0.34s/query.

🎉 Jiangshan Yu's paper "MM-SCS: Leveraging Multimodal Features to Enhance Smart Contract Code Search" has been accepted at #IEEETSE! Check it out here: ieeexplore.ieee.org/stamp/stamp....

🔗 More about IEEE TSE: ieeexplore.ieee.org/xpl/RecentIs...
#Research #smartcontract #codesearch #blockchain

3 0 0 0

Nimbus

@nimbusai.dev

1 year ago

From Code to Context: Feature Location for Modern Development Every software developer knows the frustration: you're tasked with fixing a bug or adding a feature, but first, you have to find the relevant code. In today's complex, distributed systems, this…

Navigating complex codebases shouldn't feel like searching for a needle in a haystack. We're working on better ways to locate features across distributed systems. Read more: buff.ly/433jrz3
#CodeSearch

1 0 1 0

Renzume.

@renzume.com

1 year ago

searchcode.com now runs on a 6.4TB SQLite DB - possibly the world's largest SQLite implementation
boyter.org/posts/searchcode-bigger-...
#databasemigration #sqlite #infrastructure #performance #codesearch

0 0 0 0

Paul Keen

@pftg.ruby.social.ap.brid.gy

1 year ago

Phind 2 is here! Visual answers + live code execution = a dev's dream come true. No more context switching.

#AI #DevTools #CodeSearch https://www.phind.com/blog/phind-2

0 0 0 0