code generation
PulseAugur coverage of code generation — every cluster mentioning code generation across labs, papers, and developer communities, ranked by signal.
4 day(s) with sentiment data
-
New research advances on-policy distillation for LLM training · 6 sources tracked
Researchers are developing advanced techniques for on-policy distillation (OPD), a method used to improve large language models. New approaches like $\gamma$OPD and STRIDE aim to enhance optimization stability and effic…
-
New CS-Guard benchmark reveals LLM guardrails fail to prevent malicious code generation
A new benchmark called CS-Guard has been developed to evaluate the effectiveness of Large Language Model (LLM) guardrails in preventing the generation of malicious code. The benchmark includes over 1000 prompts for text…
-
RISE method enhances language model training via self-extrapolation
Researchers have introduced RISE, a novel method for improving language model post-training through self-extrapolating policy distillation. This technique constructs a synthetic teacher from the model's own reinforcemen…
-
Research reveals covert bias injection risk in synthetic LLM data
A new research paper published on arXiv details a method for covertly injecting social biases into large language models (LLMs) through synthetic data. The study demonstrates that even seemingly benign text used in trai…
-
AI code generation shifts focus to programming language design
This article explores the evolving criteria for evaluating programming languages in the age of AI-driven code generation. As artificial intelligence increasingly contributes to software development, the qualities that d…
-
Google Explains Why Go is Ideal for AI-Assisted Software Engineering · 4 sources tracked
Google Developers Blog has published an article detailing why the Go programming language is well-suited for AI-assisted software engineering. The post highlights Go's strengths in areas such as code generation and debu…
-
RAG vs. Fine-Tuning: Choosing the Right LLM Approach for Your Business
Developers and businesses face a key decision when building AI applications: whether to use Retrieval-Augmented Generation (RAG) or fine-tuning. RAG combines a large language model with an external knowledge source, all…
-
New method improves LLM code generation uncertainty estimation
Researchers have developed a new method for estimating uncertainty in code generated by large language models, addressing the risks associated with silently incorrect code. The approach, detailed in a new paper, recogni…
-
New methods refine LLM fine-tuning for better performance
Researchers have developed new methods to improve supervised fine-tuning (SFT) for large language models. One approach, FisherAdapTune, uses the Fisher information geometry to dynamically select parameter groups for ada…
-
New research explores advanced RL for LLM code generation
Three new research papers explore advanced reinforcement learning techniques for improving large language models (LLMs) in code generation. One paper introduces offline reinforcement learning to leverage existing code d…
-
AI-generated code security remains a concern despite advanced prompting
New research indicates that while advanced prompting techniques can influence the types of security vulnerabilities present in AI-generated code, they do not reliably reduce the overall number or severity of these issue…
-
Luddite computer scientist questions AI code generation methods
An organic computer scientist and self-proclaimed Luddite, Anthony, argues that gradient ascent is a superior method to generate-and-test for software development, especially when dealing with potentially harmful outcom…
-
Pass-rate rewards fail to boost AI code generation, study finds
A new research paper explores the effectiveness of using pass-rate rewards in reinforcement learning for code generation tasks. The study found that while pass-rate rewards can alleviate the issue of sparse rewards, the…
-
New research tackles LLM factuality, architecture inference, and specialized evaluation
Researchers are developing new methods to improve the accuracy and reliability of large language models (LLMs). Google Research has introduced SLED (Self Logits Evolution Decoding), a technique that leverages all layers…