✳flâneur — a map of the web's best reading
Composer2.pdf
cursor.com · 10,939 words · saved by 10 readers
N/A
# link_2g9ydvw35a7.pdf ## Metadata - PDFFormatVersion=1.5 - IsLinearized=true - IsAcroFormPresent=false - IsXFAPresent=false - IsCollectionPresent=false - IsSignaturesPresent=false - CreationDate=D:20260324205803Z - Creator=LaTeX with hyperref - ModDate=D:20260324205803Z - Custom.PTEX.Fullbanner=This is pdfTeX, Version 3.141592653-2.6-1.40.27 (TeX Live 2025) kpathsea version 6.4.1 - Producer=pdfTeX-1.40.27 - Trapped=False ## Contents ### Page 1 Composer 2 Technical ReportCursor Research Team1 IntroductionComposer 2 is a specialized model designed for agentic software engineering. The model
Explore this link on the map →saved by
- Ratan Kaliani
- Aryan Naik
- Karthik Suresh
- Aaron Pham
- Yudhister Joel Kumar
- Vincent Huang
- Davis Brown
- Devin Plumb
- Tanvi Reddy
- Alp Niksarli
related reading
- Improving Composer through real-time RL · Cursorcursor.com
- PostTrainBenchposttrainbench.com
- Quantifying infrastructure noise in agentic coding evals \ Anthropicanthropic.com
- Coding vs thinking — Paradigm 3paradigm3.org
- AI in 2025: gestalt — LessWronglesswrong.com
- IsoCompute Playbook: Optimally Scaling Sampling Compute for RL Training of LLMscompute-optimal-rl-llm-scaling.github.io
- Andrej Karpathy — AGI is still a decade awaydwarkesh.com
- 2025: The year in LLMssimonwillison.net
- A Taxonomy of RL Environments for LLM Agentsleehanchung.github.io
- Kevin-32B: Multi-Turn RL for Writing CUDA Kernels | Cognitioncognition.ai
- Sonya Huang 🐥 on X: "Today's Training Data episode takes us BTS on the infrastructure challenges required to do large RL runs at scale, featuring @ellev3n11 (Composer Lead at @cursor_ai) and @dzhulgakov (Co-Founder at @FireworksAI_HQ). The Cursor team trained Composer 2 on Fireworks by starting with https://t.co/6LLlJlyl8Q" / Xx.com
- Training Composer for longer horizons · Cursorcursor.com