Dataset · Team-ACE
Dialogues for training models on API usage
Team-ACE/ToolACE
Agent dialogues with tool calls from a pool of 26,507 APIs — for training function calling models.
What's inside
The dataset contains English and Chinese dialogues generated by the automated ToolACE agent: multiple agents reason and call tools from a shared pool of 26,507 diverse APIs. Data accuracy is verified by a two-tier system — rules and a model. Volume ranges from 10 to 100 thousand examples.
What you can do with it
- SFT models on tool calls and function calling
- Training LoRA for an agent model on a custom API set
- Testing an agent system on complex multi-step scenarios
- Tasks
- Text generation
- Languages
- English, Chinese
- Size
- 10K–100K rows
- Downloads this month
- 13K
- Likes
- 212
Models trained on it
- bartowski/RinggAI_ringg-router-e2b-GGUFMultimodal 3.8K
- bartowski/INTELLECT-1-Instruct-GGUFText generation 2.7K
- ewin-reg/MiniCPM5-1B-Agentic-Tooluse-v3-GGUFText generation 1.7K
- ewin-reg/MiniCPM5-1B-Agentic-Tooluse-GGUFText generation 1.5K
- mradermacher/ALIA-40b-fc-2606-i1-GGUF 1.3K
- mradermacher/ALIA-40b-fc-2605-i1-GGUF 1.2K
- mradermacher/INTELLECT-1-Instruct-i1-GGUF 1.1K
- mradermacher/ToolACE-2.5-Llama-3.1-8B-i1-GGUF 984
How to load it
from datasets import load_dataset
ds = load_dataset("Team-ACE/ToolACE")