Dataset · Salesforce
5000 multi-step agent trajectories
Salesforce/APIGen-MT-5k
5000 multi-step agent trajectories for training and evaluating agent models in English.
What's inside
The dataset contains 5,000 multi-step trajectories for agent applications generated by the APIGen-MT pipeline. This is a subset of data used to train xLAM-2; the language is English. The format is not specified in the card.
What you can do with it
- Train agentic LLMs on multi-step trajectories
- Use for SFT or RL of agents with API calls
- Evaluate models on planning and tool calling
- Tasks
- Question answering, Text generation, Reinforcement learning
- Languages
- English
- Size
- 1K–10K rows
- Downloads this month
- 4.7K
- Likes
- 118
Models trained on it
- Salesforce/Llama-xLAM-2-8b-fc-rText generation 193.1K
- Salesforce/Llama-xLAM-2-8b-fc-r-ggufText generation 4.5K
- Salesforce/xLAM-2-3b-fc-r-ggufText generation 4.3K
- Salesforce/xLAM-2-1b-fc-rText generation 3.9K
- Salesforce/xLAM-2-3b-fc-rText generation 2.4K
- Salesforce/Llama-xLAM-2-70b-fc-rText generation 2.3K
- Salesforce/xLAM-2-1b-fc-r-ggufText generation 1.7K
- mradermacher/Llama-xLAM-2-70b-fc-r-i1-GGUF 1.6K
How to load it
from datasets import load_dataset
ds = load_dataset("Salesforce/APIGen-MT-5k")