datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
hermes-function-calling-v1
Hermes Function-Calling V1
This dataset is the compilation of structured output and function calling data used in the Hermes 2 Pro series of models.
This repository contains a structured output dataset with function-calling conversations, json-mode, agentic json-mode and structured extraction samples, designed to train LLM models in performing function calls and returning structured output based on natural language instructions. The dataset features various conversational scenarios… See the full description on the dataset page: https://huggingface.co/datasets/NousResearch/hermes-function-calling-v1.xlam-function-calling-60k
APIGen Function-Calling Datasets
Paper | Website | Models
This repo contains 60,000 data collected by APIGen, an automated data generation pipeline designed to produce verifiable high-quality datasets for function-calling applications. Each data in our dataset is verified through three hierarchical stages: format checking, actual function executions, and semantic verification, ensuring its reliability and correctness.
We conducted human evaluation over 600 sampled data points, and… See the full description on the dataset page: https://huggingface.co/datasets/lockon/xlam-function-calling-60k.glaive-function-calling-v2xlam-function-calling-60k
APIGen Function-Calling Datasets
Paper | Website | Models
This repo contains 60,000 data collected by APIGen, an automated data generation pipeline designed to produce verifiable high-quality datasets for function-calling applications. Each data in our dataset is verified through three hierarchical stages: format checking, actual function executions, and semantic verification, ensuring its reliability and correctness.
We conducted human evaluation over 600 sampled data points… See the full description on the dataset page: https://huggingface.co/datasets/Salesforce/xlam-function-calling-60k.apigen-function-calling
Dataset card for argilla/apigen-function-calling
This dataset is a merge of argilla/Synth-APIGen-v0.1
and Salesforce/xlam-function-calling-60k, making
over 100K function calling examples following the APIGen recipe.
Prepare for training
This version is not ready to do fine tuning, but you can run a script like prepare_for_sft.py
to prepare it, and run the same recipe that can be found in
argilla/Llama-3.2-1B-Instruct-APIGen-FC-v0.1#training-procedure.
Modify the prompt… See the full description on the dataset page: https://huggingface.co/datasets/argilla/apigen-function-calling.xlam-function-calling-60k-parsed
[PARSED] APIGen Function-Calling Datasets (xLAM)
This dataset contains the full data from the original Salesforce/xlam-function-calling-60k
Subset name
multi-turn
parallel
multiple definition
Last turn type
number of dataset
xlam-function-calling-60k
no
yes
yes
tool_calls
60000
This is a re-parsing formatting dataset for the xLAM official dataset.
Load the dataset
from datasets import load_dataset
ds =… See the full description on the dataset page: https://huggingface.co/datasets/minpeter/xlam-function-calling-60k-parsed.glaive-function-calling-v2-sharegptThe glaive-function-calling-v2 dataset in sharegpt format.
You can use it in LLaMA Factory by specifying --dataset glaive_toolcall_100k.
function-calling-sharegptThis is a dataset for finetuning models on function calling based on glaiveai/glaive-function-calling-v2.
The dataset includes 86,864 examples of chats that include function calling as part of the conversation. The system prompt includes either 0, 1, or 2 functions that the assistant can use, and instructions on how the agent can use it.
Changes include:
Using ShareGPT format for chats
Adding "function_response" as a role
Removing code examples
Removing examples with invalid JSON as function… See the full description on the dataset page: https://huggingface.co/datasets/hypervariance/function-calling-sharegpt.glaive-function-callingThis dataset consists of 52k samples generated through Glaive for the task of function calling, in the following format-
SYSTEM: You are an helpful assistant who has access to the following functions to help the user, you can use the functions if needed-
{
JSON function definiton
}
USER: user message
ASSISTANT: assistant message
Function call invocations are formatted as-
ASSISTANT: <functioncall> {json function call}
Response to the function call is formatted as-
FUNCTION RESPONSE: {json… See the full description on the dataset page: https://huggingface.co/datasets/glaiveai/glaive-function-calling.hermes-function-calling-v1-parsed
[PARSED] Hermes Function-Calling V1
The data in this dataset is a subset of the original NousResearch/hermes-function-calling-v1
Subset name
multi-turn
parallel
multiple definition
Last turn type
number of dataset
func-calling
yes
yes
yes
complex
1.8k
func-calling-singleturn
no
yes
yes
tool_calls
1.8k
glaive-function-calling-5k
yes
?
yes
complex
5k
func-calling-singleturn: Single turn function calls
func-calling: Multi-turn conversation function calls… See the full description on the dataset page: https://huggingface.co/datasets/minpeter/hermes-function-calling-v1-parsed.hermes-function-calling-v1
Hermes Function-Calling V1
This dataset is the compilation of structured output and function calling data used in the Hermes 2 Pro series of models.
This repository contains a structured output dataset with function-calling conversations, json-mode, agentic json-mode and structured extraction samples, designed to train LLM models in performing function calls and returning structured output based on natural language instructions. The dataset features various conversational scenarios… See the full description on the dataset page: https://huggingface.co/datasets/interstellarninja/hermes-function-calling-v1.glaive-function-calling-v2-openai-native
glaive-function-calling-v2-openai-native
glaiveai/glaive-function-calling-v2 restructured into the native OpenAI / TRL
format: tools is a typed column and tool_calls[].function.arguments is a
real object — not JSON inside a string.
The original is widely used (69k downloads/month) but inactive for ~3 years, and
ships tool calls as <functioncall> text blobs with Python-quoted arguments.
Existing repackagings either keep ShareGPT with tools as a string, or carry
no license at all.… See the full description on the dataset page: https://huggingface.co/datasets/Archangel-system/glaive-function-calling-v2-openai-native.function_calling_extended
Trelis Function Calling Dataset
UPDATE: As of Dec 5th 2023, there is a v3 of this dataset now available from here.
Allows models to be fine-tuned for function-calling.
The dataset is human generated and does not make use of Llama 2 or OpenAI!
Contains 59 training and 17 test rows
Based on eight functions: search_bing, search_arxiv, save_chat, read_json_file, list_files, get_current_weather, delete_file, clear_chat
Access this dataset by purchasing a license HERE.
Alternatively… See the full description on the dataset page: https://huggingface.co/datasets/Trelis/function_calling_extended.gemma-function-calling
👉🏽 Important
This dataset is adapted from hypervariance/function-calling-sharegpt to fine-tune the Google gemma-2-2b-it model for function calling.
🔀 Changes Made
Merged consecutive "GPT" responses into single responses (affected 8.49% of examples, 7372 out of 86864).
Updated role names:
"system" → Removed (function usage instructions moved to separate column)
"human" → "user"
"gpt" → "assistant"
"function_response" → Unchanged
Changed message keys from ["from"… See the full description on the dataset page: https://huggingface.co/datasets/dinushiTJ/gemma-function-calling.fiftyone-function-calling-14k
FiftyOne Function Calling 14k Dataset
Overview
This dataset is derived from the FiftyOne documentation and is designed to train AI assistants to understand and answer questions about FiftyOne's functionality. The dataset follows the format specified in the APIGen paper, structuring the data to map natural language queries to appropriate API tools and their usage.
Purpose
Train AI models to understand FiftyOne-related queries
Provide structured examples of… See the full description on the dataset page: https://huggingface.co/datasets/Voxel51/fiftyone-function-calling-14k.Funcdex-MT-Function-Calling
Funcdex-MT-Function-Calling Dataset
Funcdex-MT-Function-Calling is a multi-turn function calling dataset designed for training language models to interact with real-world tools and APIs. The dataset contains 1,787 conversations covering 10 individual toolkits and 5 multi-toolkit bundles, with comprehensive system prompts and realistic multi-turn interactions.The code used to generate the dataset can be found here.
Models trained on this dataset have excellent… See the full description on the dataset page: https://huggingface.co/datasets/prem-research/Funcdex-MT-Function-Calling.function_calling_v3
Trelis Function Calling Dataset - VERSION 3
Access this dataset by purchasing a license HERE.
Allows models to be fine-tuned for function-calling.
The dataset is human generated and does not make use of Llama 2 or OpenAI!
The dataset includes 66 training rows, 19 validation rows and 5 test rows (for manual evaluation).
Based on eight functions: search_bing, search_arxiv, save_chat, read_json_file, list_files, get_current_weather, delete_file, clear_chat
Alternatively, you can find… See the full description on the dataset page: https://huggingface.co/datasets/Trelis/function_calling_v3.assist-llm-function-calling
Function Calling dataset for Assist LLM for Home Assistant
This dataset is generated by using other conversation agent pipelines as teachers
from the deivce-actions-v2 dataset.
This dataset is used to support fine tuning of llama based models.
See Device Actions for a notebook for construction of this dataset and the device-actions dataset.
narrative-function-calling-v1
Narrative Function Calling v1
Welcome to Narrative Function Calling v1! This dataset is purpose-built for training (or fine-tuning) models that produce consistent, structured function calls in conversation-like settings. The dataset integrates and normalizes data from both Glaive Function Calling v2 (Apache License 2.0) and Salesforce XLAM function calling data (CC-BY-4.0)[^liu2024apigen]. It provides a clean, rich, and comprehensive set of examples that guide large language models… See the full description on the dataset page: https://huggingface.co/datasets/narrative-io/narrative-function-calling-v1.hibo-function-calling-v1
hibo-function-calling-v1
📖 Dataset Description
This dataset, named "hibo-function-calling-v1", is designed to facilitate the fine-tuning of Large Language Models (LLMs) for function calling tasks. It comprises a single 'train' split containing 323,271 data points across three columns: 'dataset_origin', 'system', and 'chat'.
The dataset is a result of merging two distinct sources: gathnex/Gath_baize and glaiveai/glaive-function-calling-v2, with an aim to provide… See the full description on the dataset page: https://huggingface.co/datasets/thibaud-perrin/hibo-function-calling-v1.Arabic_Function_Calling
Arabic Function Calling Dataset (50K+ Samples)
مجموعة بيانات استدعاء الدوال العربية
أول وأكبر مجموعة بيانات عربية متخصصة في استدعاء الدوال (Function Calling) تغطي جميع اللهجات العربية الرئيسية والمجالات الحياتية المهمة.
Dataset Description
This is the first comprehensive Arabic function calling dataset designed for training and evaluating LLMs on Arabic tool use capabilities. The dataset covers:
5 Arabic Dialects: MSA (Modern Standard Arabic), Egyptian… See the full description on the dataset page: https://huggingface.co/datasets/HeshamHaroon/Arabic_Function_Calling.function-calling-dataset
Function-Calling Dataset
High-quality synthetic dataset for training function-calling / tool-use capabilities in LLMs.
Dataset Details
9723 examples across 6 domains
30 unique function definitions with JSON Schema parameters
5 complexity levels: simple, moderate, complex, multi-tool, conversational
Generated using Gemini 2.0 Flash with careful prompt engineering and validation
Domains
Domain
Description
Functions
E-commerce
Shopping… See the full description on the dataset page: https://huggingface.co/datasets/Johin/function-calling-dataset.function-calling-ja-trial
🛑 Most Function-Calling Failures Are Not Schema Failures. They Are State Failures.
The model believes the world is still valid — when reality has already changed.
エージェントの事故は「ツールの失敗」ではなく「古い状態を信じたまま正常終了する」ことで起きます。本データセットは多輪ツール呼び出し・エラー復帰・検証ステップを日本語で学習させます。
Japanese Function Calling Dataset — Official Open‑Source Evaluation Package (50 Rows Subset) by springofwindslabs
Full production volumes (1,000-row standard and 2,300+ row non-overlapping extended lots), commercial licensing… See the full description on the dataset page: https://huggingface.co/datasets/springofwindslabs/function-calling-ja-trial.function-calling-en-trial
🛑 Most Function-Calling Failures Are Not Schema Failures. They Are State Failures.
The model believes the world is still valid — when reality has already changed.
It calls the right tool with stale beliefs, gets "success", and moves on. That is how agents silently corrupt production state.
This dataset trains argument-level correctness under changing state: multi-turn tool use, error recovery, and explicit verification steps — not just happy-path calls.
Function Calling Dataset… See the full description on the dataset page: https://huggingface.co/datasets/springofwindslabs/function-calling-en-trial.function-calling-training-pool
Function calling training pool
Public function-calling data from five datasets, read at the pinned revisions named below and laid
out twice. Train on either layer or on both.
pool.jsonl
Every source rewritten into one shape, 195624 rows, one JSON object per line, with these fields.
Field
What it holds
id
a row identifier unique within this file
query
the user's request, as its source publishes it
functions
the declarations offered with the request… See the full description on the dataset page: https://huggingface.co/datasets/Emulated-Inc/function-calling-training-pool.turkish-hermes-function-calling
turkish-hermes-function-calling
NousResearch/hermes-function-calling-v1 datasetinin Türkçe çevirisi — Hermes 2 Pro modelinin araç kullanımı ve yapılandırılmış çıktı yeteneklerini kazandıran orijinal veri seti.
Genel Bakış
Satır sayısı
11.567
Dil
Türkçe (tr)
Lisans
Apache 2.0
Kaynak dataset
NousResearch/hermes-function-calling-v1
Çeviri modeli
DeepSeek V4 Flash (deepseek-chat)
Ort. tur / konuşma
5,6
Çok turlu konuşma
6.120 (%52,9)
Araç… See the full description on the dataset page: https://huggingface.co/datasets/Tuguberk/turkish-hermes-function-calling.hermes-function-calling-v1-jsonl
Hermes Function-Calling V1
This dataset is the compilation of structured output and function calling data used in the Hermes 2 Pro series of models.
This repository contains a structured output dataset with function-calling conversations, json-mode, agentic json-mode and structured extraction samples, designed to train LLM models in performing function calls and returning structured output based on natural language instructions. The dataset features various conversational scenarios… See the full description on the dataset page: https://huggingface.co/datasets/minpeter/hermes-function-calling-v1-jsonl.openai-function-calling-5k
OpenAI Function Calling Format (5K)
Synthetic function-calling conversations in the OpenAI messages format (tool_calls / tool role).
Why This Dataset
Compatible with GPT-4, Mistral, Llama-3.1, Qwen2.5, and any model trained on the OpenAI chat format. Most existing function-calling datasets use abstract schemas — this uses the exact wire format models see in production.
Dataset Description
5,000 conversations across 10 tool types:
Tool… See the full description on the dataset page: https://huggingface.co/datasets/stindardlogic/openai-function-calling-5k.function_calling_v3_SAMPLE
Trelis Function Calling Dataset - VERSION 3 - SAMPLE
This is a SAMPLE of the v3 dataset available for purchase here.
Features:
Allows models to be fine-tuned for function-calling.
The dataset is human generated and does not make use of Llama 2 or OpenAI!
The dataset includes 66 training rows, 19 validation rows and 5 test rows (for manual evaluation).
Based on eight functions: search_bing, search_arxiv, save_chat, read_json_file, list_files, get_current_weather, delete_file… See the full description on the dataset page: https://huggingface.co/datasets/Trelis/function_calling_v3_SAMPLE.turkish-function-calling-20kUsed argilla-warehouse/python-seed-tools to sample tools.
Preprocessing
Since some answers might not contain a valid JSON schema, ensure that you preprocess and validate the answer to check if it satisfies the query using the given tools. You can use the preprocessing code below:
import json
from datasets import Dataset, load_dataset
def validate_answers(sample):
if sample["answers"] is None:
return True
try:
tools = json.loads(sample["tools"])… See the full description on the dataset page: https://huggingface.co/datasets/atasoglu/turkish-function-calling-20k.