The primary goal of the IGLU project is to approach the problem of how to develop interactive agents that learn to solve a task while provided with grounded natural language instructions in a collaborative environment.
Following this objective, the project has collected a multi-modal dataset with different types of interactions during a block building task. The dataset consists of two parts, the multi-turn or seed dataset for the single-turn dataset.
Due to the complexity of a general interactive task, the problem is simplified to interactions inside of a finite, Minecraft-like world of blocks. The goal of the interaction is to build a structure using a limited number of block types, which can vary in complexity. Examples of possible target structures to build are:
*** Seed or Multi-turn data: *** The Architect is provided with a target structure that needs to be built by the Builder. The Architect provides instructions to the Builder on how to create the target structure and the Builder can ask clarifying questions to the Architect if an instruction is unclear.
The progression of each game is recorded, corresponding to the construction of a target structure by an Architect and Builder pair, as a discrete sequence of game observations. Each observation contains the following information: 1) a time stamp, 2) the chat history up until that point in time, 3) the Builder’s position (a tuple of real-valued x, y, z coordinates as well as pitch and yaw angles, representing the orientation of their camera), 4) the Builder’s block inventory, 5) the locations of the blocks in the build region.
*** Single-turn data: *** This is the main IGLU dataset which comprises of participants performing free-form building actions, while providing instructions that should allow another participant to rebuild the same structure. These single-turn task segments enable asynchronous collaboration between participants We utilized the Seed dataset to provide diverse starting canvases for participants.
To install the stable version of the IGLU-datasets API, please use the following command:
pip install git+https://github.com/microsoft/iglu-datasets.git@masterAlternatively, you can download the repo and install it manually:
git clone https://github.com/microsoft/iglu-datasets.git
cd iglu-datasets && python setup.py installThe requirements to install the API are the followings:
tqdm
PILLOW
opencv-python==4.5.5.64
pandas
numpy
filelock
requests
The IGLU-datasets library provides an easy and flexible way to download the Single and Multi turn datasets. Here is an example of how to use it:
from iglu_datasets import MultiturnDataset, SingleturnDataset
# leave dataset_version empty to access the most recent version of the dataset.
# create a multiturn dataset instance
dataset = MultiturnDataset(dataset_version='v1.0')
# create a singleturn dataset instance
dataset = SingleturnDataset(dataset_version='v1.0') On creation, this class will automatically download the dataset from this repository and parse (might take a few minutes) the raw data to store in a format that is fast and convienient to load.
There are two ways for accessing the underlying data.
First, the .sample() method can be used. This method simply retrieves one random example from the dataset. The example is
wrapped into Task format which has the following fields:
sample = dataset.sample() # here a `Task` instance will be returned
print('Previous dialog\n', sample.dialog) # the whole prior conversation EXCLUDING the current instruction.
print('Instruction\n', sample.instruction) # the current instruction to execute by the builder.
# If builder asks for clarifying questions, this will be a tuple (instruction, question, answer)
# concatenated to a string.
print('Target grid\n', sample.target_grid) # 3D numpy array of shape (9, 11, 11).
# Represents the volume snapshot of the target blocks world that correspond to the builder's
# result in responce to the instruction.
print('Starting grid\n', sample.starting_grid) # 3D numpy array of shape (9, 11, 11).
# Represents the volume snapshot of the starting blocks world with which the builder
# starts executing the instruction.The multiturn dataset consists of structures that represent overall collaboration goals between an architect and a builder. The architect provides instructions to the builder to complete the target structure. The builder either performs the instruction by placing blocks according to the instruction or issues a question to clarify the instruction if it is unclear.
For each structure, we have several collaboration sessions, reffered to as games, that pair architects with builders to build each particular structure. Each session consists of a sequence of "turns". Each turn represents an atomic instruction and corresponding changes of the blocks in the world. The structure of a Task object is following:
target_grid- target blocks configuration that needs to be builtstarting_grid- optional, blocks for the environment to begin the game with.dialog- full conversation between the architect and builder, including the most recent instructioninstruction- last utterance of the architect
Sometimes, the instructions can be ambiguous and the builder asks a clarifying question which the architect answers. In the latter case, instruction will contain three utterances: an instruction, a clarifying question, and an answer to that question. Otherwise, instruction is just one utterance of the architect.
Here is an example of task (the target structure is shown on the left and blocks to add are on the right):
To represent collaboration sessions, the Subtasks class is used. This class represents a sequence of dialog utterances and their corresponding goals (each of which is a partially completed structure). On .sample() call, it picks a random turn and returns a Task object, where starting and target grids are consecutive partial structures and the dialog contains all utterances up until the one corresponding to the target grid.
In the example above, the dataset object is structured as follows:
# .tasks is a dict mapping from structure to a list of sessions of interaction
dataset.tasks
# each value contains a list corresponding to collaboration sessions.
dataset.tasks['c73']
# Each element of this list is an instance of `Subtasks` class
dataset.tasks['c73'][0]The .sample() method of MultiturnDataset does effectively the following:
def sample(dataset):
task_id = random.choice(dataset.tasks.keys())
session = random.choice(dataset.tasks[task_id])
subtask = session.sample() # Task object is returned
return subtaskThis behavior can be customized simply by overriding the reset method in a subclass:
from iglu_datasets import MultiturnDataset
class MyDataset(MultiturnDataset):
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.my_task_id = 'c73'
self.my_session = 0
def sample(self):
return self.tasks[self.my_task_id][self.my_session].sample()
my_dataset = MyDataset(dataset_version='v1.0')
# do training/samplingOn the first creation, the dataset is downloaded and parsed automatically. Below you will find the structure of the dataset:
dialogs.csv
builder-data/
...
1-c118/ # session id - structure_id
step-2
...
9-c118/
step-2
step-4
step-6
1-c120/
step-2
...
23-c126/
step-2
step-4
step-6
step-8
Here, dialog.csv contains the utterances of architects and builders solving different tasks in
different sessions. The builder-data/ directory contains builder behavior recorded by the voxel.js engine. Right now we extract only the resulting grids and use them as targets.
This dataset consists of
clarifying_questions_train.csvquestion_bank.csvinitial_world_statesfolder: Contains states of starting world for participants to build on.target_world_statesfolder: actionHit folder contains final state of the world after performing the free-form building task.
clarifying_questions_train.csv has the following columns:
GameId- Id of the game session.InitializedWorldPath- Path to the file underinitial_world_statesthat contains state of the world intialized to the architect. The architect provides an instruction to build based on this world state. More information to follow on how the world state can be parsed/ visualized.InputInstruction- Instruction provided by the architect.IsInstructionClear- Specifies whether the instruction provided by architect is clear or ambiguous. This has been marked by another annotator who is not the architect.ClarifyingQuestion- Question asked by annotator upon marking instruction as being ambiguous.qrel- Question id (qid) of the relevant clarifying question for the current instruction.qbank- List of clarifying question ids that need to be ranked for each unclear instruction. The mapping between clarifying questions and ids is present in thequestion_bank.csv.
Merged list of ids in the qrel and qbank columns will give you the list of all qids to be ranked for each ambiguous instruction.
question_bank.csv: This file contains mapping between qids mentioned in qrel and qbank columns of the clarifying_questions_train.csv to the bank of clarifying questions issued by annotators.
We have released the data collection and evaluation tool in this repo.
We have also released a human-in-the-loop interactive evaluation platform. Check out this [repo] (https://github.com/microsoft/greenlands).
We have released the data collection tool used for collecting these datasets. For more information check out this repo.
Given a predicted grid and a target one, the intersection score is calculated based on their similarity. The score is determined regardless of global spatial position of currently placed blocks, it only takes into account how much the built blocks are similar to the target structure. To make it possible, at each step we calculate the intersection between the built and the target structures for each spatial translation within the horizontal plane and rotation around the vertical axis. Then we take the maximal intersection value among all translation and rotations. To calculate the score, we compare the maximal intersection size from the current step with the one from the previous step. The resulting intersection size can serve as a true positive rate for the f1/precision/recall score calculations, also it can be used as a reward function for a reinforcement learning agent. A visual example is shown below.
Specifically, we run the code that is equivalent to the following one. Note that our version is much more optimized, while the version is given for reference:
def maximal_intersection(grid, target_grid):
"""
Args:
grid (np.ndarray[Y, X, Z]): numpy array snapshot of a built structure
target_grid (np.ndarray[Y, X, Z]): numpy array snapshot of the target structure
"""
maximum = 0
# iterate over orthogonal rotations
for i in range(4):
# iterate over translations
for dx in range(-X, X + 1):
for dz in range(-Z, Z + 1):
shifted_grid = translate(grid, dx, dz)
intersection = np.sum( (shifted_grid == target) & (target != 0) )
maximum = max(maximum, intersection)
grid = rotate_y_axis(grid)
return maximumIn practice, a more optimized version is used. There is a way to convert this score into a reward function for a reinforcement learning agent. To do that, we can calculate the reward based on the temporal difference between maximal intersection of the two consecutive grids. Formally, suppose grids[t] is a built structure at timestep t. The reward is then calculated as:
def calc_reward(prev_grid, grid, target_grid, , right_scale=2, wrong_scale=1):
prev_max_int = maximal_intersection(prev_grid, target_grid)
max_int = maximal_intersection(grid, target_grid)
diff = max_int - prev_max_int
prev_grid_size = num_blocks(prev_grid)
grid_size = num_blocks(grid)
if diff == 0:
return wrong_scale * np.sign(grid_size - prev_grid_size)
else:
return right_scale * np.sign(diff)In other words, if a recently placed block strictly increases or decreases the maximal intersection, the reward is positive or negative and is equal to +/-right_scale. Otherwise, its absolute value is equal to wrong_scale and the sign is positive if a block was removed or negative if added. This reward function is implemented in the embodied IGLU environment.
This repository also contains the human evaluation data for the top two agents of the IGLU 2022 challenge. This was obtained in collaboration with the Greenlands project. The data is stored in the datasets/human_evaluation directory and contains the following files:
human_agent_feedback.csv— Raw feedback/comments from human evaluators about each of the agents.gamedata.json— List of events that happened in each game as reported by the Greenlands platform.taskdata.json— Definition for each task in the format used by the Greenlands platform. These contain descriptions of the initial blocks in the world as well as the target structure that the agent needs to build.
The described datasets are collected as a part of IGLU:Interactive Grounded Language Understanding in a Collaborative Environment, which is described in the following papers:
@article{mohanty2022collecting,
title={Collecting Interactive Multi-modal Datasets for Grounded Language Understanding},
author={Mohanty, Shrestha and Arabzadeh, Negar and Teruel, Milagro and Sun, Yuxuan and Zholus, Artem and Skrynnik, Alexey and Burtsev, Mikhail and Srinet, Kavya and Panov, Aleksandr and Szlam, Arthur and others},
journal={arXiv preprint arXiv:2211.06552},
year={2022}
}
@inproceedings{kiseleva2022interactive,
title={Interactive grounded language understanding in a collaborative environment: Iglu 2021},
author={Kiseleva, Julia and Li, Ziming and Aliannejadi, Mohammad and Mohanty, Shrestha and ter Hoeve, Maartje and Burtsev, Mikhail and Skrynnik, Alexey and Zholus, Artem and Panov, Aleksandr and Srinet, Kavya and others},
booktitle={NeurIPS 2021 Competitions and Demonstrations Track},
pages={146--161},
year={2022},
organization={PMLR}
}
@article{kiseleva2022iglu,
title={Iglu 2022: Interactive grounded language understanding in a collaborative environment at neurips 2022},
author={Kiseleva, Julia and Skrynnik, Alexey and Zholus, Artem and Mohanty, Shrestha and Arabzadeh, Negar and C{\^o}t{\'e}, Marc-Alexandre and Aliannejadi, Mohammad and Teruel, Milagro and Li, Ziming and Burtsev, Mikhail and others},
journal={arXiv preprint arXiv:2205.13771},
year={2022}
}
Consider citing the papers above if you use the assets for your research.
This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com.
When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.
This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact [email protected] with any additional questions or comments.
This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow Microsoft's Trademark & Brand Guidelines. Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos are subject to those third-party's policies.
See here for the MIT License.