metadata
dataset_info:
- config_name: base
features:
- name: data_source
dtype: string
- name: prompt
list:
- name: role
dtype: string
- name: content
dtype: string
- name: ability
dtype: string
- name: reward_model
struct:
- name: style
dtype: string
- name: ground_truth
dtype: large_string
- name: reward_weights
list:
- name: key
dtype: string
- name: weight
dtype: float64
- name: rubrics
list:
- name: component_id
dtype: string
- name: criterion
dtype: string
- name: tags
list:
- name: key
dtype: string
- name: value
dtype: string
- name: criteria_weights
list:
- name: key
dtype: string
- name: weight
dtype: float64
- name: extra_info
struct:
- name: split
dtype: string
- name: index
dtype: int64
- name: num_generations
dtype: int64
- name: validation
struct:
- name: core
list: string
- name: pass_at_k
list: string
- name: macro
struct:
- name: task
dtype: string
- name: metric
dtype: string
- name: question
dtype: string
- name: answer
dtype: string
- name: task
dtype: string
- name: length_refs
list: 'null'
- name: rm_scales
list: 'null'
- name: origin_dataset
dtype: string
- name: origin_revision
dtype: string
- name: origin_subset
dtype: string
- name: num_tests
dtype: int64
- name: source
dtype: string
- name: num_tests_original
dtype: int64
- name: source_id
dtype: string
- name: reference_checked
dtype: bool
- name: origin_source
dtype: string
- name: constraint_count
dtype: int64
- name: rule_count
dtype: int64
splits:
- name: train
num_bytes: 2183355318
num_examples: 197832
download_size: 2158936857
dataset_size: 2183355318
- config_name: olmo-3-7b-it-sft
features:
- name: data_source
dtype: string
- name: prompt
list:
- name: role
dtype: string
- name: content
dtype: string
- name: ability
dtype: string
- name: reward_model
struct:
- name: style
dtype: string
- name: ground_truth
dtype: large_string
- name: reward_weights
list:
- name: key
dtype: string
- name: weight
dtype: float64
- name: rubrics
list:
- name: component_id
dtype: string
- name: criterion
dtype: string
- name: tags
list:
- name: key
dtype: string
- name: value
dtype: string
- name: criteria_weights
list:
- name: key
dtype: string
- name: weight
dtype: float64
- name: extra_info
struct:
- name: split
dtype: string
- name: index
dtype: int64
- name: num_generations
dtype: int64
- name: validation
struct:
- name: core
list: string
- name: pass_at_k
list: string
- name: macro
struct:
- name: task
dtype: string
- name: metric
dtype: string
- name: question
dtype: string
- name: answer
dtype: string
- name: task
dtype: string
- name: length_refs
list:
- name: model
dtype: string
- name: length_ref
dtype: float64
- name: source
dtype: string
- name: quality
dtype: float64
- name: rm_scales
list:
- name: model
dtype: string
- name: rm
dtype: string
- name: lo
dtype: float64
- name: hi
dtype: float64
- name: origin_dataset
dtype: string
- name: origin_revision
dtype: string
- name: origin_subset
dtype: string
- name: num_tests
dtype: int64
- name: source
dtype: string
- name: num_tests_original
dtype: int64
- name: source_id
dtype: string
- name: reference_checked
dtype: bool
- name: origin_source
dtype: string
- name: constraint_count
dtype: int64
- name: rule_count
dtype: int64
- name: base_levels
list:
- name: model
dtype: string
- name: rm
dtype: string
- name: pairs
list:
- name: child
dtype: string
- name: parent
dtype: string
- name: level
dtype: float64
- name: at_max
dtype: float64
- name: slope
dtype: float64
- name: slope_lo
dtype: float64
- name: slope_hi
dtype: float64
- name: slope_all
dtype: float64
- name: difficulty
dtype: float64
- name: difficulty_lo
dtype: float64
- name: difficulty_hi
dtype: float64
- name: summary
dtype: string
- name: prompt_id
dtype: int64
- name: generations
list:
- name: text
dtype: string
- name: tokens
dtype: int64
- name: completed
dtype: bool
- name: score
dtype: float64
splits:
- name: train
num_bytes: 1316794795
num_examples: 43400
download_size: 1307667986
dataset_size: 1316794795
- config_name: qwen3-4b-base
features:
- name: data_source
dtype: string
- name: prompt
list:
- name: role
dtype: string
- name: content
dtype: string
- name: ability
dtype: string
- name: reward_model
struct:
- name: style
dtype: string
- name: ground_truth
dtype: large_string
- name: reward_weights
list:
- name: key
dtype: string
- name: weight
dtype: float64
- name: rubrics
list:
- name: component_id
dtype: string
- name: criterion
dtype: string
- name: tags
list:
- name: key
dtype: string
- name: value
dtype: string
- name: criteria_weights
list:
- name: key
dtype: string
- name: weight
dtype: float64
- name: extra_info
struct:
- name: split
dtype: string
- name: index
dtype: int64
- name: num_generations
dtype: int64
- name: validation
struct:
- name: core
list: string
- name: pass_at_k
list: string
- name: macro
struct:
- name: task
dtype: string
- name: metric
dtype: string
- name: question
dtype: string
- name: answer
dtype: string
- name: task
dtype: string
- name: length_refs
list:
- name: model
dtype: string
- name: length_ref
dtype: float64
- name: source
dtype: string
- name: quality
dtype: float64
- name: rm_scales
list:
- name: model
dtype: string
- name: rm
dtype: string
- name: lo
dtype: float64
- name: hi
dtype: float64
- name: origin_dataset
dtype: string
- name: origin_revision
dtype: string
- name: origin_subset
dtype: string
- name: num_tests
dtype: int64
- name: source
dtype: string
- name: num_tests_original
dtype: int64
- name: source_id
dtype: string
- name: reference_checked
dtype: bool
- name: origin_source
dtype: string
- name: constraint_count
dtype: int64
- name: rule_count
dtype: int64
- name: base_levels
list:
- name: model
dtype: string
- name: rm
dtype: string
- name: pairs
list:
- name: child
dtype: string
- name: parent
dtype: string
- name: level
dtype: float64
- name: at_max
dtype: float64
- name: slope
dtype: float64
- name: slope_lo
dtype: float64
- name: slope_hi
dtype: float64
- name: slope_all
dtype: float64
- name: difficulty
dtype: float64
- name: difficulty_lo
dtype: float64
- name: difficulty_hi
dtype: float64
- name: summary
dtype: string
- name: prompt_id
dtype: int64
- name: generations
list:
- name: text
dtype: string
- name: tokens
dtype: int64
- name: completed
dtype: bool
- name: score
dtype: float64
splits:
- name: train
num_bytes: 1803969541
num_examples: 39352
download_size: 1795962302
dataset_size: 1803969541
- config_name: qwen3-8b-base
features:
- name: data_source
dtype: string
- name: prompt
list:
- name: role
dtype: string
- name: content
dtype: string
- name: ability
dtype: string
- name: reward_model
struct:
- name: style
dtype: string
- name: ground_truth
dtype: large_string
- name: reward_weights
list:
- name: key
dtype: string
- name: weight
dtype: float64
- name: rubrics
list:
- name: component_id
dtype: string
- name: criterion
dtype: string
- name: tags
list:
- name: key
dtype: string
- name: value
dtype: string
- name: criteria_weights
list:
- name: key
dtype: string
- name: weight
dtype: float64
- name: extra_info
struct:
- name: split
dtype: string
- name: index
dtype: int64
- name: num_generations
dtype: int64
- name: validation
struct:
- name: core
list: string
- name: pass_at_k
list: string
- name: macro
struct:
- name: task
dtype: string
- name: metric
dtype: string
- name: question
dtype: string
- name: answer
dtype: string
- name: task
dtype: string
- name: length_refs
list:
- name: model
dtype: string
- name: length_ref
dtype: float64
- name: source
dtype: string
- name: quality
dtype: float64
- name: rm_scales
list:
- name: model
dtype: string
- name: rm
dtype: string
- name: lo
dtype: float64
- name: hi
dtype: float64
- name: origin_dataset
dtype: string
- name: origin_revision
dtype: string
- name: origin_subset
dtype: string
- name: num_tests
dtype: int64
- name: source
dtype: string
- name: num_tests_original
dtype: int64
- name: source_id
dtype: string
- name: reference_checked
dtype: bool
- name: origin_source
dtype: string
- name: constraint_count
dtype: int64
- name: rule_count
dtype: int64
- name: base_levels
list:
- name: model
dtype: string
- name: rm
dtype: string
- name: pairs
list:
- name: child
dtype: string
- name: parent
dtype: string
- name: level
dtype: float64
- name: at_max
dtype: float64
- name: slope
dtype: float64
- name: slope_lo
dtype: float64
- name: slope_hi
dtype: float64
- name: slope_all
dtype: float64
- name: difficulty
dtype: float64
- name: difficulty_lo
dtype: float64
- name: difficulty_hi
dtype: float64
- name: summary
dtype: string
- name: prompt_id
dtype: int64
- name: generations
list:
- name: text
dtype: string
- name: tokens
dtype: int64
- name: completed
dtype: bool
- name: score
dtype: float64
splits:
- name: train
num_bytes: 2120113992
num_examples: 46820
download_size: 2110646814
dataset_size: 2120113992
configs:
- config_name: base
data_files:
- split: train
path: base/train-*
- config_name: olmo-3-7b-it-sft
data_files:
- split: train
path: olmo-3-7b-it-sft/train-*
- config_name: qwen3-4b-base
data_files:
- split: train
path: qwen3-4b-base/train-*
- config_name: qwen3-8b-base
data_files:
- split: train
path: qwen3-8b-base/train-*
MTMR (Multi-Task Multi-Reward) base
base/train is the shared, unbalanced prompt pool for model-specific sampling and calibration.
Sources
- agentica-org/DeepScaleR-Preview-Dataset, revision
b6ae8c60f5c1f2b594e2140b91c49c9ad0949e29. - open-r1/DAPO-Math-17k-Processed, revision
31dd309567e3da778038cc87d868b6097a3ccf68. - PRIME-RL/Eurus-2-RL-Data, revision
9776b13264b5aaa0b16495fcf086a0a8d86fd655. - allenai/Dolci-RL-Zero-Code-7B, revision
054c0a1f9c5fd52566ad9124aa2e702924803cd3. - open-r1/verifiable-coding-problems-python_decontaminated-tested, revision
77ace849f7cc8f10d23945c748f9ff464427f83b. - sojuL/RubricHub_v1, revision
3837d55971473a872e84879c88f708b8da3ec2ef. Math uses DeepScaleR and DAPO-Math-17k-Processed (all/train, English and Chinese). Code uses Eurus, Dolci and Open-R1's tested Python problems; IF/Writing use RubricHub'srurbichub_v1_Instruction_Followingandrurbichub_v1_Writingfiles (notrurbichub_v1_Chat). Rows retain source repository, revision and subset.
Preparation
- Validate prompts, Math gold parsing, rubric weights and executable IF rules.
- Remove provably contradictory IF rules and explicit label-only Writing tasks. Deduplicate identical
Writing criteria with equal weights; reject conflicting duplicate weights. Ambiguous context/NAME
placeholders are flagged for review, not filtered. Cleaning reports are saved locally in
cleaning/. - Normalize Code interfaces/tests; check available Dolci and required Open-R1 reference programs with the training executor (5s per-test timeout, 120s total execution budget). Keep at most 20 uniformly sampled, prompt-seeded tests per question.
- Render every Code prompt as
### Question / ### Formatfrom the bare statement (chat, Eurus and Open-R1 wrappers stripped for all sources); drop questions with fewer than 15 words before their first input/example block. - Deduplicate normalized prompts across all four tasks; remove exact evaluation-question matches and embedded copies of evaluation questions >=80 characters. Semantic paraphrases are not detected.
- Require the complete chat prompt to be <=2048 tokens under all three tokenizers: RLLab/olmo-3-1025-7b, Qwen/Qwen3-4B-Base, Qwen/Qwen3.5-2B-Base.
- Store outer reward weights separately from rubric-criterion weights; length/RM references are filled by the model-specific sampler.
Overlap sources (upstream, not local evaluation files): Eurus, eval-set/aime24, eval-set/aime25, eval-set/amc23, eval-set/minervamath, eval-set/olympiadbench, eval-set/math500, code_generation_lite/v6, MMLU-Pro/default, ZebraLogic/grid_mode, humanevalplus, mbppplus, IFEval, IFBench, WritingBench, WritingBench-validation, Arena-Hard-v2, GPQA-Diamond, ResearchQA, Creative-Writing-v3.
Task counts: {"code": 35642, "if": 93340, "math": 52981, "writing": 15869}.
Refer to the source repositories for dataset licenses.