Datasets:

tccoin
/

navverse-benchmark

episode_id string	scene_id string	goal_category string	goal_text string	images images list	pos_rots list	action_sequence list	action_ids list	messages list
grCommercial_scene1/episode_0	grCommercial_scene1	shelf	Slightly turn left, then move forward 4.45 m. Slightly turn right, then move forward 1.78 m. Slightly turn right, then move forward 2.05 m. Slightly turn left, then move forward 2.13 m. You will arrive at the shelf.		[ [ 3.402086402741837, 0.3712763049828966, 0.5444346901949958, -0.06211473309739659, 0.05952093628774838, 0.8063059789304041, 0.5852090963209233 ], [ 3.1477543820878076, 0.3595242860822278, 0.5006951893911826, -0.014925763969998725, 0.09520795062973127, 0.863899...	[ "TURN_LEFT", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "TURN_RIGHT", "MOVE_FORWARD", "MOVE_FORWARD",...	[ 0, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 3, 3, 3, 0 ]	[ { "content": [ { "text": "You are a vision-language navigation agent in an unknown environment.\nYou will receive a sequence of observations showing your movement history up to the current moment.\n\nInstruction: Slightly turn left, then move forward 4.45 m. Slightly turn right, then move forw...
grCommercial_scene1/episode_1	grCommercial_scene1	bookshelf	Slightly turn left, then move forward 4.11 m. Slightly turn left, then move forward 0.61 m. Slightly turn left, then move forward 1.99 m. You will arrive at the bookshelf.		[ [ 1.362065482701817, -0.9639531370341248, 0.5154626912406338, -0.03359417468176883, -0.03008937663244028, -0.08493272779889216, 0.9953655070324475 ], [ 1.3166235720203365, -0.7013896545556932, 0.4044685396645404, 0.02259738702701483, -0.007003537364985868, 0.1...	[ "TURN_LEFT", "TURN_LEFT", "TURN_LEFT", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "...	[ 0, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 1, 1, 1, 2, 0 ]	[ { "content": [ { "text": "You are a vision-language navigation agent in an unknown environment.\nYou will receive a sequence of observations showing your movement history up to the current moment.\n\nInstruction: Slightly turn left, then move forward 4.11 m. Slightly turn left, then move forwa...
grCommercial_scene1/episode_10	grCommercial_scene1	keyboard	Turn right, then move forward 2.91 m. Turn right, then move forward 5.59 m. Slightly turn left, then move forward 2.95 m. Slightly turn right, then move forward 3.44 m. You will arrive at the keyboard.		[ [ -1.1724849792911134, -3.043228551644924, 0.5071429047877382, -0.020663716592321007, -0.009384082013818525, -0.7905436423431611, 0.6119850483240244 ], [ -1.4265115291472126, -2.989518795861211, 0.4965337089809728, -0.06377522618948378, 0.0073905549613959304, ...	[ "TURN_RIGHT", "TURN_RIGHT", "TURN_RIGHT", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "TURN_LEFT", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "TURN_RIGHT", "TURN_RIGHT", "TURN_RIGHT", "TURN_RIGHT", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FOR...	[ 0, 3, 3, 3, 1, 1, 1, 2, 1, 1, 1, 1, 1, 3, 3, 3, 3, 1, 1, 1, 1, 1, 1, 3, 1, 1, 1, 3, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 3, 1, 1, 1, 1, 1, 3, 3, 2, 3, 2, 3, 1, 1, 1, 1...	[ { "content": [ { "text": "You are a vision-language navigation agent in an unknown environment.\nYou will receive a sequence of observations showing your movement history up to the current moment.\n\nInstruction: Turn right, then move forward 2.91 m. Turn right, then move forward 5.59 m. Sligh...
grCommercial_scene1/episode_11	grCommercial_scene1	basket	Start facing the large wooden cloud-shaped shelving unit. Move forward, passing the blue toy animal on your left and heading towards the center of the shelving structure. As you get closer to the shelves, specifically near the blue circular tunnel seat, turn slightly right to align with the lower shelf section. Continu...		[ [ -3.0377324045169436, 3.3101418391714974, 0.5108647551013381, -0.013569327938176397, -0.010117839042465653, -0.3735612613099692, 0.9274510697179243 ], [ -2.85685862620024, 3.505759783032562, 0.4554644379148972, -0.011787193175521416, -0.05539484080901341, -0....	[ "TURN_LEFT", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD...	[ 0, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 0 ]	[ { "content": [ { "text": "You are a vision-language navigation agent in an unknown environment.\nYou will receive a sequence of observations showing your movement history up to the current moment.\n\nInstruction: Start facing the large wooden cloud-shaped shelving unit. Move forward, passing t...
grCommercial_scene1/episode_12	grCommercial_scene1	basket	Slightly turn right, then move forward 4.27 m. Turn left, then move forward 1.55 m. Slightly turn left, then move forward 2.53 m. You will arrive at the basket.		[ [ -4.0178078312904475, -0.41048157701028376, 0.5384266293838201, 0.030833813305525738, -0.02449609184203868, 0.5395241858770113, 0.8410486729644453 ], [ -4.234507643406911, -0.29413834500455477, 0.39628509812360013, 0.04738274117316128, 0.012581554035566735, 0...	[ "TURN_LEFT", "MOVE_FORWARD", "TURN_RIGHT", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD",...	[ 0, 2, 1, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0 ]	[ { "content": [ { "text": "You are a vision-language navigation agent in an unknown environment.\nYou will receive a sequence of observations showing your movement history up to the current moment.\n\nInstruction: Slightly turn right, then move forward 4.27 m. Turn left, then move forward 1.55 ...
grCommercial_scene1/episode_13	grCommercial_scene1	shelf	Start by facing the light-colored wooden flooring and a large wooden shelving unit shaped like clouds, filled with various objects like brown teddy bears and blue circular seats. Move forward across the floor, keeping the white desk and black office chair to your right. Continue straight ahead, passing a metal rack wit...		[ [ -1.9149060128979671, 1.064483603471185, 0.5264920627854763, 0.011834019715707187, -0.03447434283027958, -0.03596781170091851, 0.9986880354671503 ], [ -1.801082521109317, 1.2986209351660678, 0.423269442423266, -0.008205744882292344, -0.009494805291438828, 0.1...	[ "TURN_LEFT", "TURN_LEFT", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "TURN_LEFT", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "...	[ 0, 2, 2, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 1, 1, 1, 3, 1, 1, 3, 3, 3, 0 ]	[ { "content": [ { "text": "You are a vision-language navigation agent in an unknown environment.\nYou will receive a sequence of observations showing your movement history up to the current moment.\n\nInstruction: Start by facing the light-colored wooden flooring and a large wooden shelving uni...
grCommercial_scene1/episode_14	grCommercial_scene1	couch	Head towards the white cabinet located between the doorway and the wooden shelving unit filled with stuffed animals. As you pass the wooden shelves on your right, proceed towards the center of the room where a girl is sitting on the rug playing with a toy. Continue moving forward, keeping the girl to your left. You wil...		[ [ -3.3667098857736555, 4.253374387340583, 0.4989753963551782, -0.008637261862445964, -0.007547041650726408, -0.6907108654401113, 0.7230400682070216 ], [ -3.131937632652874, 4.157313958689999, 0.41541477949928474, 0.031053440075683923, -0.03306306837942065, -0....	[ "TURN_LEFT", "TURN_LEFT", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", ...	[ 0, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 3, 1, 0 ]	[ { "content": [ { "text": "You are a vision-language navigation agent in an unknown environment.\nYou will receive a sequence of observations showing your movement history up to the current moment.\n\nInstruction: Head towards the white cabinet located between the doorway and the wooden shelvin...
grCommercial_scene1/episode_16	grCommercial_scene1	blanket	Turn slightly left and head towards the large wooden shelf with circular cutouts filled with toys and books. After passing a small blue cylindrical stool, turn left. Move straight across the room, keeping the wooden shelf on your left and heading towards the large window wall. As you approach the area with a large gray...		[ [ -3.109397754612685, 0.8078471572521139, 0.5331208729707497, -0.019363348422055712, -0.04150226549980563, -0.1911756791181318, 0.9804868598864499 ], [ -2.8615105635196585, 0.8340033652006517, 0.47612928560021156, -0.019279523960331325, -0.05191801422599375, -...	[ "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "TURN_LEFT", "TURN_LEFT", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "TURN_LEFT", "TURN_LEFT", "TURN_LEFT", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_F...	[ 0, 1, 1, 1, 1, 2, 2, 1, 1, 1, 1, 1, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 3, 2, 0 ]	[ { "content": [ { "text": "You are a vision-language navigation agent in an unknown environment.\nYou will receive a sequence of observations showing your movement history up to the current moment.\n\nInstruction: Turn slightly left and head towards the large wooden shelf with circular cutouts ...
grCommercial_scene1/episode_17	grCommercial_scene1	shelf	Slightly turn left, then move forward 2.83 m. Turn left, then move forward 2.95 m. Slightly turn left, then move forward 3.26 m. You will arrive at the shelf.		[ [ -3.2901936227978186, 0.8999757491887597, 0.5247505025727743, 0.0055198506043412235, -0.029589087185437766, 0.0269002674359885, 0.9991848641671424 ], [ -3.020559271474889, 0.9554055617534138, 0.48518671070044345, -0.03661380928175653, -0.04953010305291936, 0....	[ "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "TURN_LEFT", "TURN_LEFT", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", "MOVE_FORWARD", ...	[ 0, 1, 1, 1, 1, 1, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 3, 3, 3, 0 ]	[ { "content": [ { "text": "You are a vision-language navigation agent in an unknown environment.\nYou will receive a sequence of observations showing your movement history up to the current moment.\n\nInstruction: Slightly turn left, then move forward 2.83 m. Turn left, then move forward 2.95 m...
grCommercial_scene1/episode_18	grCommercial_scene1	bookshelf	"Start facing the wooden shelf on your right, filled with decorative items and bunny figures. Turn s(...TRUNCATED)	[{"src":"https://datasets-server.huggingface.co/assets/tccoin/navverse-benchmark/--/{dataset_git_rev(...TRUNCATED)	[[-0.9479613280252484,5.782564625386031,0.5036037399905974,-0.030394097332655898,-0.0019667615160647(...TRUNCATED)	["MOVE_FORWARD","MOVE_FORWARD","MOVE_FORWARD","MOVE_FORWARD","MOVE_FORWARD","TURN_LEFT","TURN_LEFT",(...TRUNCATED)	[ 0, 1, 1, 1, 1, 1, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0 ]	[{"content":[{"text":"You are a vision-language navigation agent in an unknown environment.\nYou wil(...TRUNCATED)

End of preview. Expand in Data Studio

NavVerse Benchmark

This repository hosts the NavVerse dataset.

Runnable scene release

navverse_v1_july23.tar.gz is the current runnable scene package. It contains:

52 VC+ outdoor scenes and 52 connected indoor/outdoor scenes under vc_plus/;
30 GRScenes commercial navigation scenes under grscenes_commercial/;
the corresponding prebuilt navmesh/ assets;
a relative nvidia -> vc_plus/nvidia link for the bundled CloudySky runtime lighting.

From a NavVerse-Benchmark checkout, download, verify, extract, validate, and configure the release with:

python scripts/download_navverse_data.py
python scripts/navverse_evaluation.py --episode_label vcp_london_0

The downloader installs into navverse_data/extracted by default, validates the SHA-256 file and release layout, and updates only SCENE_FOLDER in .env. The archive preserves relative symlinks and should be extracted with tar on Linux.

SFT split

The sft_data/ folder contains a per-scene parquet shard suitable for supervised fine-tuning of a vision-language navigation agent. Each row is one episode and uses a schema compatible with Aasdfip/habitat_web_pose_train:

column	dtype
`episode_id`	string
`scene_id`	string
`goal_category`	string
`goal_text`	string
`images`	list<`Image(decode=False)`> (JPEG bytes)
`pos_rots`	list<list<float64>> — `[x, y, z, qx, qy, qz, qw]` per step
`action_sequence`	list<string> — one of `MOVE_FORWARD`, `TURN_LEFT`, `TURN_RIGHT`, `STOP`
`action_ids`	list<int64> — leading STOP sentinel + encoded actions
`messages`	list of `{role, content: list of {text, type}}` — habitat-style chat

Episodes were sampled onto a habitat-style grid: each MOVE_FORWARD corresponds to roughly 0.25 m of motion; each TURN_LEFT / TURN_RIGHT to at least ~10° of yaw change (--turn-label-min-deg 10 in the converter). Exactly one STOP appears at the end of every episode.

Generated with the scripts under NavVerse-Benchmark/scripts/:

# 1. Download the raw dataset
python scripts/download_raw_data.py --output-dir .

# 2. Extract the episode_data tarball
mkdir -p extracted
tar -xzf episode_data_v1.tar.gz -C extracted

# 3. Convert episode_data → SFT parquet
python scripts/convert_episode_data_for_sft.py \
    --input extracted/episode_data \
    --output extracted/episode_data_sft \
    --turn-label-min-deg 10

A viewer notebook is included at sft_data/sft_data.ipynb.

Downloads last month: 429