datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
deepseek-v4-pro-agent-tool-calling-trajectory
DeepSeek V4 Pro ToolScale Agent SFT Dataset
A curated subset of multi-turn tool-calling trajectories generated by DeepSeek V4 Pro on ToolScale. The dataset is filtered by action-match score against ground-truth trajectories and is designed for supervised fine-tuning of agentic models on realistic, multi-step tool use.
Each conversation includes natural-language user requests, tool calls, tool observations, assistant reasoning traces, and grounded final responses across five… See the full description on the dataset page: https://huggingface.co/datasets/zake7749/deepseek-v4-pro-agent-tool-calling-trajectory.Qwen-3.6-plus-agent-tool-calling-trajectory
Qwen 3.6 Plus: ToolScale Agent SFT Dataset
Multi-turn tool-calling trajectories generated by Qwen 3.6 Plus via OpenRouter on ToolScale. Both passing and near-passing rollouts are included, allowing users to choose their own quality threshold using reward and score.
Each row is a flattened conversation prefix ending at one assistant turn, ready for next-token SFT. Assistant turns include a reasoning_content field containing the model’s reasoning.
What's inside… See the full description on the dataset page: https://huggingface.co/datasets/zake7749/Qwen-3.6-plus-agent-tool-calling-trajectory.2026-07-31-toolcalling-tulu-sft-run
Run record — Qwen3.6-27B tool-calling 20/80 SFT
Everything the training run produced except the weights: the TRL log history, the resolved
config, the environment, the loss/accuracy figure and its greppable markdown mirror.
The adapter is at LASR-Callum/2026-07-31-wrongly-trained-qwen36-toolcalling-tulu-lora-20-80; the training data is
at LASR-Callum/2026-07-31-toolcalling-tulu-20-80-mixture.
Required metadata
field
value
experiment
One bf16 LoRA SFT… See the full description on the dataset page: https://huggingface.co/datasets/dougalldeepmind/2026-07-31-toolcalling-tulu-sft-run.reasoning-tool-calling-v1
Reasoning and Tool Calling
This dataset converts tagged reasoning and tool-use conversations into typed
messages and tool definitions. It also includes calibration data derived from
those converted conversations.
The source is
Mustafaege/qwen3.5-toolcalling-v2
at revision 8f0343a5613879fefda0eb002d10ff7150a2c588.
How this differs from the source
The source stores 92,153 train conversations and 10,240 test conversations in
a tagged message format. Protocol… See the full description on the dataset page: https://huggingface.co/datasets/steadfastgaze/reasoning-tool-calling-v1.turkish-tool-calling
Türkçe Tool-Calling Veri Seti
56.247 kayıt. xLAM/APIGen 60k ve NVIDIA When2Call'dan türetilmiş,
üç davranış sınıfı içeren Türkçe function-calling veri seti.
from datasets import load_dataset
ds = load_dataset("bilalabic/turkish-tool-calling") # mesaj listesi
ds = load_dataset("bilalabic/turkish-tool-calling", "table") # düz tablo
ds = load_dataset("bilalabic/turkish-tool-calling", "sharegpt") # ShareGPT
İçerik
Kayıt
56.247… See the full description on the dataset page: https://huggingface.co/datasets/bilalabic/turkish-tool-calling.pvp-tool-calling-sft
PvP tool-calling SFT cold-start data
Claude-vs-Claude games played through the G.O.D PvP tool-calling harness. Each row is one model turn (or post-game reflection): the system+user prompt the harness built, the assistant response (content + tool_calls), and the tools schemas — i.e. the OpenAI messages+tools format consumed by tokenizer.apply_chat_template(messages, tools=tools). On a move turn the assistant co-emits any memory-tool edits and a game_action committing a legal… See the full description on the dataset page: https://huggingface.co/datasets/gradients-io-tournaments/pvp-tool-calling-sft.qwen-3.6-plus-agent-tool-calling
Qwen 3.6 Plus: ToolScale Agent SFT Dataset
Multi-turn tool-calling trajectories generated by Qwen 3.6 Plus via OpenRouter on ToolScale. Both passing and near-passing rollouts are included, allowing users to choose their own quality threshold using reward and score.
Each row is a flattened conversation prefix ending at one assistant turn, ready for next-token SFT. Assistant turns include a reasoning_content field containing the model’s reasoning.
What's inside… See the full description on the dataset page: https://huggingface.co/datasets/ansulev/qwen-3.6-plus-agent-tool-calling.adaptive-curriculum-tool-calling-pool
Adaptive Curriculum Tool-Calling Pool — v2.1
A gated snapshot of the tool-calling training-data pool produced by the
Adaptive Curriculum for Tool Calling sub-experiment. This is the additive v2.1
revision: it keeps the entire v1 + v2 payload and adds the six per-campaign
partition manifests under manifests/partitions/. Nothing from v1 or v2 was
re-encoded, recompressed, moved, or rewritten.
Access is gated. The repository uses manual gating. You must be granted access
by the… See the full description on the dataset page: https://huggingface.co/datasets/kesava89/adaptive-curriculum-tool-calling-pool.arxiv-to-code-agentic-tool-calling
arxiv-to-code-agentic-tool-calling
Multi-turn tool-calling dataset where an assistant implements ML papers in PyTorch through file-creation and command-execution tool calls.
Built from lucidrains' (Phil Wang) open-source paper implementations. There are ~217 repositories on Codeberg, each implementing a different ML paper. This dataset reverse-engineers those into synthetic coding conversations.
What's in it
199 conversations, each covering one repository. Every… See the full description on the dataset page: https://huggingface.co/datasets/SultanR/arxiv-to-code-agentic-tool-calling.Llama-3.2-1B-Instruct-magpie-tool-callingtool-calling-dsapex-multiturn-toolcallingtaubench-tool-calling-Qwen2.5-7B-Instruct-0.0_range_0-10_user-gpt-4o-llm_1116210635sft_w_think_and_tool_calling_prism_claude_250625_uniquehistory-search-tool-calling
History Search – Evaluation Datasets
This repository contains evaluation-only datasets for history search tool usage.
Datasets
intent (canonical)
Evaluates whether the model should invoke history search in a multi-turn conversation.
One row per conversation
Preserves full context
Used by eval code
Key fields
id
turns[] → { turn_index, user, expect_search_history }
intent_turns (derived)
Flattened, one row per turn view of intent.
For HF Dataset… See the full description on the dataset page: https://huggingface.co/datasets/Mozilla/history-search-tool-calling.sft_w_think_and_tool_calling_prism_claude_250625_unique_max_10tool-calling-datasetLlama-3.2-1B-Instruct-magpie-tool-calling