Skip to content

Dataset · kai-os

GLM-5 AI agent execution traces in Hermes environment

kai-os/carnice-glm5-hermes-traces

1k to 10k task execution logs of the GLM-5 model using terminal, code, and browser tools.

What's inside

Contains 1,000 to 10,000 entries with execution traces of the z-ai/glm-5 model within the Hermes Agent wrapper on prompts from the carnice_trace_prompt_bank_v4 dataset. During data collection, the model operated in isolated environments and called file, terminal, and code tools, as well as a browser and Tavily web search. Data is presented in text and tabular formats.

What you can do with it

  • SFT fine-tuning of AI agents for terminal, file, and code execution work
  • Training models in tool use and web information search
Tasks
Text generation, Other
Size
1K–10K rows
Downloads this month
237
Likes
59

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("kai-os/carnice-glm5-hermes-traces")