# Research

A selection of research papers published by the WRITER team. Read key findings and takeaways, and click through to see the complete papers.

## [How personalized context quietly degrades AI accuracy: a deeper look](/content/engineering/personalized-context-degrades-ai-accuracy/index.html)

**Writer Team**

## [Palmyra-mini: Small models, big throughput, powerful reasoning](/content/engineering/palmyra-mini-open-source-models/index.html)

**Rakshith Vasudev**

## [Reflect, retry, reward: Self-improving LLMs via reinforcement learning](/content/engineering/self-reflection-llm-reinforcement-learning/index.html)

**Shelly Bensal**

## [Expecting the unexpected: A new benchmark for LLM resilience in finance — FailSafeQA](/content/engineering/failsafeqa-benchmark/index.html)

**Writer Team**

## [Introducing self-evolving models](/content/engineering/self-evolving-models/index.html)

**Waseem AlShikh**

## [Introducing Writing in the Margins](/content/engineering/writing-in-the-margins/index.html)

**Writer Team**

## [Comparative analysis of retrieval systems in the real world](/content/engineering/robustqa-graph-rag-benchmarking/index.html)

**Writer Team**

## [OmniACT: A dataset and benchmark for enabling multimodal generalist autonomous agents for desktop and web](/content/engineering/omniact-dataset-benchmark-multimodal-autonomous-agents/index.html)

**Writer Team**

## [Palmyra Med: instruction-based fine-tuning of LLMs enhancing medical domain performance](/content/engineering/palmyra-med-instruction-based-fine-tuning-medical-domain-performance/index.html)

**Writer Team**

## [Becoming self-instruct: introducing early stopping criteria for minimal instruct tuning](/content/engineering/becoming-self-instruct-early-stopping-criteria/index.html)

**Writer Team**

## [Grammatical error correction: a survey of the state of the art](/content/engineering/research-grammatical-error-correction/index.html)

**Writer Team**

## [Fusion-in-Decoder: Achieving state-of-the-art open-domain QA performance](/content/engineering/fusion-in-decoder-open-domain-qa/index.html)

**Writer Team**