task_id string | name string | domain string | skill_type string | task_complexity string | command_complexity string | scenario string | language string | anchor string | verifier_kind string | fixture_kind string | corpus_kind string | base_image string | ablation string | description string | truth string | primitive_skills list | metadata_json string | setup string | container_def string | test_initial_state string | test_final_state string |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
task_000000_0362c25b | task_000000_0362c25b | data_querying | Query Construction | null | null | compliance officer auditing systems | Python | null | exact_text | text_only | legacy | null | no_difficulty | You are an AI assistant assisting a Compliance Officer with an internal audit. The organization suspects that a highly central figure in their communication network is engaged in anomalous financial activities.
You have been provided with two raw data sources in the directory `/home/user/audit/`:
1. `comms_export.jso... | The test suite must ensure the environment is prepared by running the following Python setup script *before* the agent starts. This script creates the required `/home/user/audit/` directory, the JSONL NoSQL dump, and the undocumented SQLite database.
```python
import os
import json
import sqlite3
os.makedirs("/home/u... | [
"Data model reverse engineering",
"Graph analytics (centrality, clustering)",
"Window functions and analytical aggregation",
"NoSQL aggregation pipelines"
] | {"ablation": "no_difficulty", "anchor": null, "base_image": null, "command_complexity": null, "corpus_kind": "legacy", "description": "You are an AI assistant assisting a Compliance Officer with an internal audit. The organization suspects that a highly central figure in their communication network is engaged in anomal... | #!/bin/bash
set -euo pipefail
mkdir -p /home/user/.setup_tmp
export PYTHONPATH=/home/user/.local/lib/python3/dist-packages:${PYTHONPATH:-}
apt-get update && apt-get install -y python3 python3-pip sqlite3
pip3 install --target /home/user/.local/lib/python3/dist-packages pytest networkx pandas
python3 << 'EOF... | Bootstrap: docker
From: ubuntu:22.04
%post
export DEBIAN_FRONTEND=noninteractive
apt-get update && apt-get install -y python3 python3-pip sqlite3
pip3 install pytest networkx pandas
useradd -m -s /bin/bash user || true
python3 << 'EOF'
import os
import json
import sqlite3
os.makedirs("/home/user... | # test_initial_state.py
import os
import json
import sqlite3
import pytest
def test_audit_dir_exists():
assert os.path.isdir("/home/user/audit"), "/home/user/audit directory is missing"
def test_comms_export_file():
filepath = "/home/user/audit/comms_export.jsonl"
assert os.path.isfile(filepath), f"{filep... | # test_final_state.py
import os
import json
import sqlite3
import math
import pytest
def get_betweenness_centrality(nodes, edges):
"""
Computes betweenness centrality for all nodes in an undirected graph
using Brandes' algorithm.
"""
cb = {v: 0.0 for v in nodes}
for s in nodes:
S = []
... |
task_000000_06df5c65 | task_000000_06df5c65 | data_science | Systems | null | null | researcher organizing datasets | Go | null | exact_text | text_only | legacy | null | no_difficulty | You are acting as an automated Systems Researcher. We are analyzing a newly collected dataset of telemetry logs from our compute cluster to determine the relationship between CPU utilization and thermal output, and we need you to build a small ETL pipeline and analysis tool in **Go**.
We have raw telemetry data arrivi... | The setup requires generating the raw CSV dataset before the agent starts.
Execute the following Python script during the environment setup to create `/home/user/system_metrics.csv`.
```python
# Create setup script
cat << 'EOF' > /tmp/setup_data.py
import random
import csv
random.seed(42) # Ensuring reproducibility
... | [
"Analysis environment setup",
"Model training and evaluation",
"Correlation and covariance analysis",
"ETL pipeline construction"
] | {"ablation": "no_difficulty", "anchor": null, "base_image": null, "command_complexity": null, "corpus_kind": "legacy", "description": "You are acting as an automated Systems Researcher. We are analyzing a newly collected dataset of telemetry logs from our compute cluster to determine the relationship between CPU utiliz... | #!/bin/bash
set -euo pipefail
mkdir -p /home/user/.setup_tmp
export PYTHONPATH=/home/user/.local/lib/python3/dist-packages:${PYTHONPATH:-}
apt-get update && apt-get install -y python3 python3-pip golang-go
cat << 'EOF' > /home/user/.setup_tmp/setup_data.py
import random
import csv
random.seed(42) # Ensuring reproducib... | Bootstrap: docker
From: ubuntu:22.04
%post
export DEBIAN_FRONTEND=noninteractive
apt-get update && apt-get install -y python3 python3-pip golang-go
pip3 install pytest
useradd -m -s /bin/bash user || true
cat << 'EOF' > /tmp/setup_data.py
import random
import csv
random.seed(42) # Ensuring reproducibility
with open... | # test_initial_state.py
import os
import csv
def test_system_metrics_file_exists():
"""Test that the raw telemetry data file exists."""
file_path = '/home/user/system_metrics.csv'
assert os.path.exists(file_path), f"File {file_path} does not exist."
assert os.path.isfile(file_path), f"Path {file_path}... | # test_final_state.py
import os
import csv
import json
import math
import pytest
def get_expected_metrics(csv_path):
"""
Recompute the ground truth metrics directly from the CSV file
according to the strict ETL filtering rules.
"""
x, y = [], []
with open(csv_path, 'r', encoding='utf-8') as f:
... |
task_000000_2ef27705 | task_000000_2ef27705 | data_science | Algorithmic | null | null | data scientist cleaning datasets | Rust | null | exact_text | text_only | legacy | null | no_difficulty | We are implementing a predictive maintenance system for our factory's CNC machines, and we need to build a rock-solid, reproducible data cleaning and dimensionality reduction pipeline. Our raw sensor data is spread across multiple files and contains both missing readings and extreme outliers.
I need you to write a Ru... | The task expects the initial CSVs to be deterministically generated before the agent starts.
**Setup Script (Run before task starts):**
```python
import os
import random
import math
random.seed(42)
def generate_normal(mean, std):
u1 = random.random()
u2 = random.random()
if u1 == 0.0: u1 = 1e-7
z0 = ... | [
"Reproducible pipeline construction",
"Multi-source data joining",
"Missing value and outlier handling",
"Pipeline reproducibility testing",
"Dimensionality reduction"
] | {"ablation": "no_difficulty", "anchor": null, "base_image": null, "command_complexity": null, "corpus_kind": "legacy", "description": "We are implementing a predictive maintenance system for our factory's CNC machines, and we need to build a rock-solid, reproducible data cleaning and dimensionality reduction pipeline. ... | #!/bin/bash
set -euo pipefail
mkdir -p /home/user/.setup_tmp
export PYTHONPATH=/home/user/.local/lib/python3/dist-packages:${PYTHONPATH:-}
apt-get update && apt-get install -y python3 python3-pip cargo rustc
cat << 'EOF' > /home/user/.setup_tmp/setup.py
import os
import random
import math
random.seed(42)
def generate... | Bootstrap: docker
From: ubuntu:22.04
%post
export DEBIAN_FRONTEND=noninteractive
apt-get update && apt-get install -y python3 python3-pip cargo rustc
pip3 install pytest
useradd -m -s /bin/bash user || true
cat << 'EOF' > /tmp/setup.py
import os
import random
import math
random.seed(42)
def generate_normal(mean, s... | # test_initial_state.py
import os
import pytest
def test_data_directory_exists():
"""Verify that the data directory exists."""
assert os.path.isdir('/home/user/data'), "The directory /home/user/data is missing."
def test_registry_csv_exists_and_valid():
"""Verify that registry.csv exists and has the corr... | # test_final_state.py
import os
import math
import glob
import subprocess
import pytest
def test_rust_project_exists():
"""Verify that the Cargo project was created in the correct directory."""
assert os.path.isdir('/home/user/sensor_pipeline'), "Directory /home/user/sensor_pipeline does not exist."
asser... |
task_000000_3fdd596c | task_000000_3fdd596c | data_querying | Result Processing | null | null | compliance officer auditing systems | Python | null | metric_threshold | audio | rl_v2 | intricate | no_difficulty | You are an AI assistant acting as a technical compliance officer. We are investigating a potential access-control violation internally known as "Project Chimera."
A whistleblower left an audio dictation containing the scope of the audit. You can find this recording at `/app/whistleblower.wav`.
Our IT landscape and ... | The task requires the agent to correctly transcribe the target entities from the audio, perform recursive CTEs on both `employees` and `systems` tables, join them with `access_logs`, filter by `access_count > 100`, sort the results, and export to JSON.
### Setup Process
The following operations must be executed to pr... | [
"Result sorting, pagination, and filtering",
"Query result export and format conversion",
"Recursive and hierarchical queries",
"Graph projection and materialization",
"Complex joins and subqueries"
] | {"ablation": "no_difficulty", "anchor": null, "base_image": "intricate", "command_complexity": null, "corpus_kind": "rl_v2", "description": "You are an AI assistant acting as a technical compliance officer. We are investigating a potential access-control violation internally known as \"Project Chimera.\" \n\nA whistleb... | #!/bin/bash
set -euo pipefail
mkdir -p /home/user/.setup_tmp
export PYTHONPATH=/home/user/.local/lib/python3/dist-packages:${PYTHONPATH:-}
apt-get update && apt-get install -y python3 python3-pip espeak sqlite3
mkdir -p /app
espeak -w /app/whistleblower.wav "We need to audit the following targets. For manag... | Bootstrap: docker
From: ubuntu:22.04
%files
/gpfs/scrubbed/hamishiv/tmax-ablation-generation-20261003/runs/no_difficulty/chunks/rl_v2_0005/task_000000_3fdd596c/fixtures/audio.wav /app/fixtures/audio.wav
/gpfs/scrubbed/hamishiv/tmax-ablation-generation-20261003/runs/no_difficulty/chunks/rl_v2_0005/task_000000_3... | # test_initial_state.py
import os
import sqlite3
def test_audio_file_exists():
"""Verify that the whistleblower audio file is present."""
audio_path = "/app/whistleblower.wav"
assert os.path.exists(audio_path), f"{audio_path} is missing."
assert os.path.isfile(audio_path), f"{audio_path} is not a valid... | # test_final_state.py
import os
import json
import sqlite3
import pytest
def get_ground_truth():
"""
Dynamically recalculate the expected output based on the intent of the rubric.
This ensures we don't just hardcode the expected JSON but actually verify the logic.
"""
db_path = '/app/audit_data.db'... |
task_000000_43bea0b1 | task_000000_43bea0b1 | data_processing | Systems | null | null | log analyst investigating patterns | Python | null | exact_text | text_only | legacy | null | no_difficulty | You are a log analyst investigating anomalous thermal spikes and CPU bottlenecks in a distributed server farm. Our telemetry aggregation system experienced network drops, resulting in interleaved logs and corrupted data payloads.
Your task is to write a Python pipeline that parses the raw log file, reconstructs the ti... | The setup process requires generating the `/home/user/raw_telemetry.log` file.
Run this exact Python script to create the initial environment state:
```python
import os
import random
from datetime import datetime, timedelta
os.makedirs('/home/user', exist_ok=True)
start_time = datetime(2024, 1, 1, 12, 0, 0)
server... | [
"Feature extraction transforms",
"Large-scale sorting and grouping",
"Multi-stage pipeline orchestration",
"Interpolation and imputation"
] | {"ablation": "no_difficulty", "anchor": null, "base_image": null, "command_complexity": null, "corpus_kind": "legacy", "description": "You are a log analyst investigating anomalous thermal spikes and CPU bottlenecks in a distributed server farm. Our telemetry aggregation system experienced network drops, resulting in i... | #!/bin/bash
set -euo pipefail
mkdir -p /home/user/.setup_tmp
export PYTHONPATH=/home/user/.local/lib/python3/dist-packages:${PYTHONPATH:-}
apt-get update && apt-get install -y python3 python3-pip
pip3 install --target /home/user/.local/lib/python3/dist-packages pytest pandas
cat << 'EOF' > /home/user/.setup_tmp/setup.p... | Bootstrap: docker
From: ubuntu:22.04
%post
export DEBIAN_FRONTEND=noninteractive
apt-get update && apt-get install -y python3 python3-pip
pip3 install pytest pandas
useradd -m -s /bin/bash user || true
cat << 'EOF' > /tmp/setup.py
import os
import random
from datetime import datetime, timedelta
os.makedirs('/home/u... | # test_initial_state.py
import os
import pytest
def test_raw_telemetry_log_exists():
log_file = '/home/user/raw_telemetry.log'
assert os.path.exists(log_file), f"The file {log_file} is missing. Please ensure the setup script ran correctly."
assert os.path.isfile(log_file), f"The path {log_file} exists but... | # test_final_state.py
import os
import csv
import re
from datetime import datetime
import pytest
RAW_LOG_PATH = '/home/user/raw_telemetry.log'
PROCESSED_CSV_PATH = '/home/user/processed_stats.csv'
@pytest.fixture(scope="module")
def expected_data():
"""
Derives the expected aggregated statistics directly fro... |
task_000000_47a2847a | task_000000_47a2847a | system_administration | Filesystem | null | null | kubernetes operator managing manifests | Python | a Docker Compose setup where services can't reach each other due to network misconfiguration | exact_text | text_only | legacy | null | no_difficulty | You are tasked with fixing a broken, custom-built Python GitOps operator that manages Kubernetes manifests. The system is designed to simulate a GitOps pipeline where pushing to a local Git repository automatically deploys manifests to a mock Kubernetes API.
Currently, the system is completely broken due to missing TL... | The test suite will check that the Git hook exists and works, that `api_mock.py` received the exact payload and logged it, and that `final.json` contains the expected output from `operator.py`'s status endpoint.
**Pre-requisite Setup (Run before agent starts):**
```bash
#!/bin/bash
# Run as the 'user' user (no root re... | [
"User account and group administration",
"Git server and hook configuration",
"Web server setup and TLS configuration",
"Process monitoring and control"
] | {"ablation": "no_difficulty", "anchor": "a Docker Compose setup where services can't reach each other due to network misconfiguration", "base_image": null, "command_complexity": null, "corpus_kind": "legacy", "description": "You are tasked with fixing a broken, custom-built Python GitOps operator that manages Kubernete... | #!/bin/bash
set -euo pipefail
mkdir -p /home/user/.setup_tmp
export PYTHONPATH=/home/user/.local/lib/python3/dist-packages:${PYTHONPATH:-}
apt-get update && apt-get install -y python3 python3-pip git openssl curl
mkdir -p /home/user/operator_system/certs
mkdir -p /home/user/operator_system/manifests
mkd... | Bootstrap: docker
From: ubuntu:22.04
%post
export DEBIAN_FRONTEND=noninteractive
apt-get update && apt-get install -y python3 python3-pip git openssl curl
pip3 install pytest
mkdir -p /home/user/operator_system/certs
mkdir -p /home/user/operator_system/manifests
mkdir -p /home/user/operator_sy... | # test_initial_state.py
import os
import json
import pytest
BASE_DIR = "/home/user/operator_system"
def test_directories_exist():
expected_dirs = [
f"{BASE_DIR}/certs",
f"{BASE_DIR}/manifests",
f"{BASE_DIR}/workspace",
f"{BASE_DIR}/repo.git"
]
for d in expected_dirs:
... | # test_final_state.py
import os
import json
import pytest
BASE_DIR = "/home/user/operator_system"
def test_git_hook_configured():
"""Verify that the Git post-receive hook was created and is executable."""
hook_path = f"{BASE_DIR}/repo.git/hooks/post-receive"
assert os.path.exists(hook_path), f"Git hook n... |
task_000000_58fde94a | task_000000_58fde94a | scientific_computing | Testing | null | null | researcher running simulations | Bash | a simulation that produces non-reproducible results due to floating-point reduction order | exact_text | text_only | legacy | null | no_difficulty | "You are an AI assistant helping a bioinformatics researcher debug a scientific computing pipeline.\(...TRUNCATED) | "The setup requires creating the initial files in `/home/user/primer_sim`. This must be done before (...TRUNCATED) | ["Monte Carlo simulation","Primer design and sequence alignment","Reference dataset comparison","Num(...TRUNCATED) | "{\"ablation\": \"no_difficulty\", \"anchor\": \"a simulation that produces non-reproducible results(...TRUNCATED) | "#!/bin/bash\nset -euo pipefail\nmkdir -p /home/user/.setup_tmp\nexport PYTHONPATH=/home/user/.local(...TRUNCATED) | "Bootstrap: docker\nFrom: ubuntu:22.04\n\n%post\n export DEBIAN_FRONTEND=noninteractive\n apt-(...TRUNCATED) | "# test_initial_state.py\n\nimport os\nimport stat\n\nBASE_DIR = \"/home/user/primer_sim\"\n\ndef te(...TRUNCATED) | "# test_final_state.py\n\nimport os\nimport subprocess\nimport re\nimport pytest\n\nBASE_DIR = \"/ho(...TRUNCATED) |
task_000000_5ac9aa24 | task_000000_5ac9aa24 | debugging | Algorithmic | null | null | operations engineer triaging incidents | Rust | null | exact_text | text_only | legacy | null | no_difficulty | "**Subject: Urgent Incident - Telemetry Ingestion Service Down**\n\nWe have an ongoing incident with(...TRUNCATED) | "The task requires resolving a missing cargo feature (`macros` and `rt-multi-thread` for `tokio`), t(...TRUNCATED) | ["Regression test construction","Intermediate state tracing","Dependency conflict resolution","Forma(...TRUNCATED) | "{\"ablation\": \"no_difficulty\", \"anchor\": null, \"base_image\": null, \"command_complexity\": n(...TRUNCATED) | "#!/bin/bash\nset -euo pipefail\nmkdir -p /home/user/.setup_tmp\nexport PYTHONPATH=/home/user/.local(...TRUNCATED) | "Bootstrap: docker\nFrom: ubuntu:22.04\n\n%post\nexport DEBIAN_FRONTEND=noninteractive\napt-get upda(...TRUNCATED) | "# test_initial_state.py\n\nimport os\n\ndef test_project_structure():\n assert os.path.isdir(\"/(...TRUNCATED) | "# test_final_state.py\n\nimport os\nimport re\n\ndef test_bug_line_txt():\n log_path = \"/home/u(...TRUNCATED) |
task_000000_5d0bda94 | task_000000_5d0bda94 | scientific_computing | Algorithmic | null | null | researcher running simulations | C++ | null | exact_text | text_only | legacy | null | no_difficulty | "You are an astrobiologist and bioinformatician simulating the directed evolution of synthetic DNA s(...TRUNCATED) | "This task requires setting up an HDF5 dataset prior to the agent's run and provides a canonical Pyt(...TRUNCATED) | ["Primer design and sequence alignment","Scientific data format I/O (NetCDF, HDF5, FITS)","Reproduci(...TRUNCATED) | "{\"ablation\": \"no_difficulty\", \"anchor\": null, \"base_image\": null, \"command_complexity\": n(...TRUNCATED) | "#!/bin/bash\nset -euo pipefail\nmkdir -p /home/user/.setup_tmp\nexport PYTHONPATH=/home/user/.local(...TRUNCATED) | "Bootstrap: docker\nFrom: ubuntu:22.04\n\n%post\nexport DEBIAN_FRONTEND=noninteractive\napt-get upda(...TRUNCATED) | "# test_initial_state.py\n\nimport os\nimport subprocess\nimport pytest\n\ndef test_sim_data_file_ex(...TRUNCATED) | "# test_final_state.py\n\nimport os\nimport subprocess\nimport json\nimport math\nimport pytest\n\nd(...TRUNCATED) |
task_000000_62c4d9d0 | task_000000_62c4d9d0 | data_processing | Data I/O | null | null | configuration manager tracking changes | Rust | null | exact_text | text_only | legacy | null | no_difficulty | "You are a Data Engineer tasked with building a robust, streaming ETL pipeline in Rust to process a (...TRUNCATED) | "# SETUP SCRIPT (Run this before the agent starts to create the pre-existing environment)\n# This Py(...TRUNCATED) | ["Timestamp alignment and parsing","Pipeline logging and monitoring","Large-file streaming","Unicode(...TRUNCATED) | "{\"ablation\": \"no_difficulty\", \"anchor\": null, \"base_image\": null, \"command_complexity\": n(...TRUNCATED) | "#!/bin/bash\nset -euo pipefail\nmkdir -p /home/user/.setup_tmp\nexport PYTHONPATH=/home/user/.local(...TRUNCATED) | "Bootstrap: docker\nFrom: ubuntu:22.04\n\n%post\nexport DEBIAN_FRONTEND=noninteractive\napt-get upda(...TRUNCATED) | "# test_initial_state.py\nimport os\nimport json\n\ndef test_raw_configs_initial_state():\n file_(...TRUNCATED) | "# test_final_state.py\nimport json\nimport os\nimport hashlib\nimport unicodedata\nfrom datetime im(...TRUNCATED) |
End of preview. Expand in Data Studio
TMax Ablation: No Difficulty (5K)
This dataset contains 5,000 selected initial-generation terminal tasks for the no_difficulty ablation. Both task complexity and command complexity prompt blocks are omitted. Their metadata values are null. All remaining generation axes retain the original sampling configuration, with 3,333 legacy tasks and 1,667 rl_v2 tasks.
Tasks were generated with gemini/gemini-3.1-pro-preview using the direct Gemini API, with axes seed 42. This export contains the selected corpus, preserving the original task IDs, metadata and task file bytes.
Generation stage
These tasks passed the generation pipeline's build and initial-state checks. No rubric labeling, repair pass, or agent rollout verification has been applied. The generated final-state verifiers and reference answers may still contain defects. The truth field is the generator's reference answer; it is not an independently validated solution.
This arm requested 5,850 candidates across 39 completed chunks. The survivor target is exactly 5,000, selected to satisfy the corpus quotas. provenance.json records the configuration and generator source checksums; export-report.json records counts and artifact checksums.
Files and schema
data/train-00000-of-00001.parquet: thetrainsplit, one row per task.tasks.zip: task directories containing the originaltask.json,setup.sh,container.def,test_initial_state.py,test_final_state.py, and other task-local files included in the file manifest.task-ids.json: the 5,000 canonical task directory IDs.task-files.jsonl: archive paths, byte counts and SHA-256 checksums for every included task file.export-report.jsonandprovenance.json: export validation and generation provenance.
The canonical task_id matches the directory in tasks.zip; name preserves the original generated name. The Parquet columns include all sampled metadata: domain, skill_type, task_complexity, command_complexity, scenario, language, anchor, verifier_kind, fixture_kind, corpus_kind, base_image, ablation, and primitive_skills. They also contain description, truth, the complete original metadata as metadata_json, and the source text of setup, container_def, test_initial_state, and test_final_state.
metadata_json preserves original JSON types and values. Nullable complexity values remain null. Task-local fixture files are preserved in the ZIP. Binary container images, execution logs, and API completion caches are omitted. The cached validation base was used only during generation validation; saved task definitions are unchanged.
Loading
from datasets import load_dataset
tasks = load_dataset("TMaxxx/TMax-Ablation-No-Difficulty-5K", split="train")
from huggingface_hub import hf_hub_download
from zipfile import ZipFile
archive = hf_hub_download("TMaxxx/TMax-Ablation-No-Difficulty-5K", "tasks.zip", repo_type="dataset")
with ZipFile(archive) as z:
z.extractall("tmax-no_difficulty")
The task archive provides environment definitions and tests for further verification and repair. It does not include prebuilt container images or an agent solve pass.
- Downloads last month
- 124