datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
arm1-eval-rollouts
arm1 eval rollouts
Eval rollouts (32 samples/problem) for ReasoningRegisters/arm1.
Split folders named step{k}_{benchmark}[_variant]; JSONL per shard: prompt, generation, correctness.
qwen14b-math500-eval-rollouts
Qwen3-14B GRPO — MATH-500 eval rollout shards
Raw evaluation rollouts behind the MATH-500 curves in the two Qwen3-14B GRPO runs
(vanilla qwen14b-MATH and the cue-forced qwen14b-MATH-cue branch of
avdravid/reasoning_registers_grpo). Checkpoints: model repos
ReasoningRegisters/qwen14b and
ReasoningRegisters/qwen14b_cue.
Layout
Two runs, one folder per curve point, 8 shards each (the eval ran 8-way sharded over MATH-500):
vanilla/step{0,50,...,300}/ — Qwen3-14B-Base… See the full description on the dataset page: https://huggingface.co/datasets/ReasoningRegisters/qwen14b-math500-eval-rollouts.arm3-eval-rollouts
arm3 eval rollouts
Eval rollouts (32 samples/problem) for ReasoningRegisters/arm3.
Split folders named step{k}_{benchmark}[_variant]; JSONL per shard: prompt, generation, correctness.
base-eval-rollouts
base eval rollouts
Eval rollouts (32 samples/problem) for ReasoningRegisters/base (untrained base model).
Split folders named step{k}_{benchmark}[_variant]; JSONL per shard: prompt, generation, correctness.
arm2-eval-rollouts
arm2 eval rollouts
Eval rollouts (32 samples/problem) for ReasoningRegisters/arm2.
Split folders named step{k}_{benchmark}[_variant]; JSONL per shard: prompt, generation, correctness.
arm4-eval-rollouts
arm4 eval rollouts
Eval rollouts (32 samples/problem) for ReasoningRegisters/arm4.
Split folders named step{k}_{benchmark}[_variant]; JSONL per shard: prompt, generation, correctness.
olmo32b-math500-eval-rollouts
olmo32b — MATH-500 eval rollouts
MATH-500 evaluation rollouts for the Olmo-3-32B GRPO run (base allenai/Olmo-3-1125-32B,
RL-Zero prompt, LoRA r=64, 300 steps). Adapters: ReasoningRegisters/olmo32b.
Layout (same as qwen14b-math500-eval-rollouts):
vanilla/step{0,50,100,150,200,250,300}/ — plain prompt, 500 problems × 8 rollouts,
T=0.6, top-p 0.95, 8,192-token cap. step0 = base model; step50+ scored on merged weights.
vanilla/step300_roll32/ — final row: 500 × 32 rollouts, 16… See the full description on the dataset page: https://huggingface.co/datasets/ReasoningRegisters/olmo32b-math500-eval-rollouts.olmo32b-aime-eval-rollouts
olmo32b — AIME 2024/2025 eval rollouts
AIME evaluation rollouts for the Olmo-3-32B GRPO run (adapters:
ReasoningRegisters/olmo32b; MATH-500 rollouts:
ReasoningRegisters/olmo32b-math500-eval-rollouts).
aime2024/step{0,300}/, aime2025/step{0,300}/ — 30 problems × 32 rollouts,
T=0.6, top-p 0.95, 16,384-token cap. step0 = base allenai/Olmo-3-1125-32B,
step300 = merged final checkpoint.
Each directory holds 8 shard_XX.json (per-problem counts) and 8
text_shard_XX.jsonl (one line… See the full description on the dataset page: https://huggingface.co/datasets/ReasoningRegisters/olmo32b-aime-eval-rollouts.olmo32b-cue-rollouts
olmo32b-cue — GRPO training rollouts
Training rollout logs from the cue-forced GRPO run on Olmo-3-1125-32B
(adapters: ReasoningRegisters/olmo32b-cue; 300 steps, every rollout
prefilled with the ".\n\nOkay," opener).
rollout_log/rollouts_rank{0..15}.jsonl — one line per rollout:
gstep, reward, capped, n_tokens, opener, register flag.
rollout_log/text_rank{0..15}.jsonl.gz — same rollouts with full
generated text (gzip).
Step-0 evaluation rollouts for the cue arm live under… See the full description on the dataset page: https://huggingface.co/datasets/ReasoningRegisters/olmo32b-cue-rollouts.