SynthTraces A minimal codebase to generate synthetic coding agent session traces using Pi. Each session pairs two models working inside one of the project codebases: a remotely hosted open model (e.g. deepseek ai/DeepSeek V4 Pro , openai/gpt oss 120b , Qwen/Qwen3.6 27B ) backs the coding agent , equipped with the default Pi tools — read , write , edit , and bash ; a local model running in llama.cpp plays the user , opening with one of the starting questions and driving the conversation. The full exchange is recorded as a trace. The dataset is the cartesian product of every agent model × user model × codebase × starting question. Generation matrix Dimension Count Examples Agent models (remotely hosted, open) 20 deepseek ai/DeepSeek V4 Pro , openai/gpt oss 120b , Qwen/Qwen3.6 27B , zai org/GLM 5.1 , … User models (local, llama.cpp) 3 ggml org/Qwen3.6 27B GGUF:Q8 0 , ggml org/Qwen3.6 35B A3B MTP GGUF:Q8 0 , ggml org/gemma 4 26B A4B it GGUF:Q8 0 Project codebases 20 transformers , diffusers , lerobot , peft , candle , … Starting questions 20 "How do I run this code?" , "How is CI set up in this repo?" , "What recent changes were made and why?" , … Total sessions 24,000 20 × 3 × 20 × 20…
Runs entirely in your browser via DuckDB-Wasm — this dataset's real data file is loaded once, then queried locally. Nothing is sent to a server.
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy