Dataset · Salesforce
xLAM: 60k function calling examples
Salesforce/xlam-function-calling-60k
Verified English function-calling examples for training and evaluating LLM agents.
What's inside
The dataset contains 60k examples for function-calling tasks collected via the APIGen automated pipeline. Each example passed three levels of validation: format, real function execution, and semantic verification, with manual evaluation of 600 examples. Language is English, modality is text, tasks include question-answering, text-generation, and reinforcement-learning.
What you can do with it
- SFT models for API and function calling
- Train LoRA for a function-calling agent
- Evaluate agent model reliability on verified examples
- Tasks
- Question answering, Text generation, Reinforcement learning
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 17K
- Likes
- 766
- Access
- on request on Hugging Face
Models trained on it
- Salesforce/Llama-xLAM-2-8b-fc-rText generation 193.1K
- Salesforce/xLAM-1b-fc-rText generation 8.6K
- Manojb/Qwen3-4B-toolcalling-gguf-codexText generation 6.9K
- legraphista/xLAM-8x7b-r-IMat-GGUFText generation 4.8K
- ermiaazarkhalili/Qwen3.8-9B-Function-Calling-xLAM-Unsloth-GGUFText generation 4.8K
- Salesforce/Llama-xLAM-2-8b-fc-r-ggufText generation 4.5K
- legraphista/xLAM-8x22b-r-IMat-GGUFText generation 4.3K
- Salesforce/xLAM-2-3b-fc-r-ggufText generation 4.3K
How to load it
from datasets import load_dataset
ds = load_dataset("Salesforce/xlam-function-calling-60k")