Datasets:
The dataset viewer is not available for this subset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
for split_generator in builder._split_generators(
~~~~~~~~~~~~~~~~~~~~~~~~~^
StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/harbor/harbor.py", line 171, in _split_generators
raise DataFilesNotFoundError("No task.toml or instruction.md files found")
datasets.exceptions.DataFilesNotFoundError: No task.toml or instruction.md files found
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
for split in get_dataset_split_names(
~~~~~~~~~~~~~~~~~~~~~~~^
path=dataset,
^^^^^^^^^^^^^
config_name=config,
^^^^^^^^^^^^^^^^^^^
token=hf_token,
^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
info = get_dataset_config_info(
path,
...<6 lines>...
**config_kwargs,
)
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Arxh-DeepLogic v1.3
Deep Learning × Advanced Logical Reasoning
Research corpus by Arxh AI.
What it is
A 75M-instance deterministic research corpus designed to teach and evaluate structured reasoning over deep-learning problems.
Scale
| Metric | Value |
|---|---|
| Total instances | 75,000,000 |
| L7/L8 research+ | 58,000,000 |
| L0-L6 | 17,000,000 |
| Languages | 75 |
| Task families | 60 |
| Shards | 16 |
The 58M research+ block is exactly the L7/L8 portion of v1.3.
Why recipes?
Each JSONL row is a deterministic recipe rather than a giant duplicated prompt. This gives Arxh AI a reproducible way to sample, shard, render, and verify millions of scenarios while keeping the release practical.
Use the renderer in scripts/render.py to expand recipes into conventional prompt/answer records.
Core capabilities
- tensor/parameter/FLOP reasoning
- CNN/Transformer/attention geometry
- optimizer and gradient reasoning
- memory and distributed-training accounting
- MoE, FlashAttention, RoPE and long-context constraints
- data quality, deduplication and evaluation contamination
- calibration and uncertainty
- causal ablation and counterfactual reasoning
- hypothesis testing and experiment design
- out-of-distribution reasoning
Provenance and research basis
Arxh-DeepLogic does not copy long passages from papers or existing datasets. It reconstructs task patterns from public research methodology and benchmarks into original deterministic recipes.
Status
v1.3 is a scalable research-data release, not a claim that every generated natural-language surface is human-perfect. Downstream users should run their own validation and ablations before treating it as a benchmark standard.
- Downloads last month
- 118