datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
GSM-Symbolic
GSM-Symbolic
This project accompanies the research paper, GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models.
Getting Started
In our first release, we provide data for GSM-Symbolic, GSM-Symbolic-P1, and GSM-Symbolic-P2 variants. For each variant, we have released both templates (available on Github), and a sample of generated data that can be used for evaluation.
To load the data, you can use the following code. Note that in… See the full description on the dataset page: https://huggingface.co/datasets/apple/GSM-Symbolic.declref-01-declref_01_symbols-2B
declref-01-declref_01_symbols-2B
Procedurally generated decl-ref-01 documents — the scoped declare/reference language of declref, with noisy references, sparse part-transition chains, interleaved openings, and periodic topic shifts tuned so that a model trained on it matches natural-language / code entropy dynamics (positional entropy profile, its fluctuation texture, and the entropy-quantile distribution), and holds that match as training doubles. Each document is a random… See the full description on the dataset page: https://huggingface.co/datasets/alexkstern/declref-01-declref_01_symbols-2B.bfs-optimal-symbolic-algebra-simplification-trajectories
ISRE v7 BFS Trajectories
This dataset contains BFS-optimal symbolic algebra simplification trajectories
for the ISRE research project.
Each row is one trajectory. Nested fields are stored as JSON strings to preserve
the original AST and step structure without lossy flattening.
What This Dataset Is
This is not a natural-language instruction dataset. It is a symbolic algebra
policy-learning dataset.
Each trajectory starts from a deliberately scrambled algebraic… See the full description on the dataset page: https://huggingface.co/datasets/NecroMOnk/bfs-optimal-symbolic-algebra-simplification-trajectories.gsm_infinite_symbolic_2kgsm_infinite_symbolic_4kgsm_infinite_symbolic_example_0GSM-Symbolic_self_doubt_preferencestateful-gsm-symbolic-questionnanochat-depo-l0-symbolic-20260715
Nanochat Depo-L0: symbolic
This is a diagnostic, separately versioned Depo source. Each row contains one
16-node cycle and eight queries at depths 1, 2, 4, and 8. Only the eight
single-letter answers and terminal token are supervised. It is designed for a
one-document-per-sequence training protocol and must not be treated as public
Depo v3 data.
stateful-gsm-symbolicgsm_infinite_symbolic_1kGSM-Symbolic-TTT
GSM-Symbolic
Dataset Description
This dataset contains symbolic variations of grade-school math word problems.The dataset is constructed by merging multiple generated datasets where each instance corresponds to a symbolic template used to produce variations of a math reasoning problem.
Each instance contains a math word problem along with its corresponding solution and final numeric answer.
Dataset Structure
Data Instances
Each row in the dataset is… See the full description on the dataset page: https://huggingface.co/datasets/nafisehNik/GSM-Symbolic-TTT.gsm_infinite_symbolic_4k_test_seed1000gsm_infinite_symbolic_1k_test_seed1000unsymbols-lg-sim-symbols-lgdeduprlvr_task086_translated_symbol_arithmeticbatman-random-symbol-mapping-questionunsymbols-lg-sim-symbolsgsm_infinite_symbolic_2k_test_seed1000GSM-Symbolic_model_generationsfinal_symbolic_dataset_harmfulstocks-symbolsunsymbols-sim-symbolsstateful-gsm-symbolic-question-trainstateful-gsm-symbolic-question-valunsymbols-X-symbols_lgGSM-Symbolic_model_generations_errorsqwen3_0.6b-rlvr_task086_translated_symbol_arithmeticunsymbols-X-symbolsfinal_symbolic_dataset_harmful_NEW