kobzaond/RLVRAMBench
RLVRAMBench Which language-model training configurations can I use with the memory I have, and how much testing does that decision require? RLVRAMBench is a measurement dataset with open evaluation tasks for a specific language-model training system. It measures memory feasibility when response generation and reinforcement-learning updates share the same graphics processors. It provides measured outcomes, fixed prediction tasks, a budgeted decision replay, reference methods, and… See the full description on the dataset page: https://huggingface.co/datasets/kobzaond/RLVRAMBench.
0223
1{2 "release": "submission-20260915-6a7606f",3 "source_commit": "6a7606fa72a162b9929d2e73a3175f1ac3ac91a4",4 "huggingface_repository": "kobzaond/RLVRAMBench",5 "code_repository": "kobzaond/RLVRAMBench-reproduce",6 "archive": {7 "path": "releases/RLVRAMBench-review-artifact-submission-20260915-6a7606f.tar.zst",8 "sha256": "b9b62e1d94da603ba96ddc89dc362d661ca7366c3cfe798dc0a63a4fa8fc26fb",9 "size_bytes": 18147757710 },11 "manifest_entries": 21374,12 "archive_members": 21376,13 "expected_derived_files": 40,14 "principal_processes": 588,15 "supporting_source_processes": 16,16 "model_checkpoint_payloads_included": false,17 "doi": null,18 "visibility_at_initial_deposit": "private",19 "visibility": "public",20 "public_release_date": "2026-09-14",21 "original_material_license": "MIT",22 "expected_review_derived_files": 27,23 "review_control_completed_processes": 24,24 "review_control_attempted_processes": 26,25 "review_control_usable_blocks": 6,26 "original_recorded_attempts": 657,27 "prior_immutable_release_preserved": "admission-20260915-7b534ef",28 "manuscript_editing_authority": "direct_tex",29 "benchmark_protocol_version": "1.0",30 "benchmark_tasks": 52,31 "benchmark_queries": 400,32 "distinct_benchmark_target_configurations": 94,33 "curated_eligible_processes": 612,34 "curated_recorded_attempts": 683,35 "license_selection_date": "2026-09-15",36 "benchmark_evaluator_version": "1.1",37 "decision_protocol_version": "1.0",38 "decision_cases": 4,39 "decision_candidates": 12,40 "decision_planned_slots": 48,41 "decision_recorded_attempts": 48,42 "decision_validated_memory_outcomes": 48,43 "expected_decision_derived_files": 11,44 "expected_curated_benchmark_files": 14,45 "estimation_protocol_version": "estimation-1.0",46 "expected_estimation_derived_files": 13,47 "estimation_counts": {48 "planned_processes": 36,49 "planned_pairs": 18,50 "planned_configurations": 12,51 "eligible_processes": 23,52 "completed_processes": 0,53 "unresolved_processes": 13,54 "known_launched_processes": 34,55 "unknown_launch_status_processes": 0,56 "resolved_configurations": 6,57 "unresolved_configurations": 658 },59 "estimation_protocol_sha256": "1694e447e842600612604891e7b3d036ff0791fa84c4bdbeba54ed093feefdeb",60 "estimation_prediction_seal_sha256": "5ed714caff922a5109346a4242545d19f2824636d0f52bb0bbd2a09b40ed4666",61 "estimation_execution_commits": {62 "est-qwen25-7b-code_heavy_tail-c2-s151": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",63 "est-qwen25-7b-code_heavy_tail-c2-s152": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",64 "est-qwen25-7b-code_heavy_tail-c2-s153": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",65 "est-qwen25-7b-code_heavy_tail-c3-s151": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",66 "est-qwen25-7b-code_heavy_tail-c3-s152": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",67 "est-qwen25-7b-code_heavy_tail-c3-s153": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",68 "est-qwen25-7b-gsm8k-c2-s151": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",69 "est-qwen25-7b-gsm8k-c2-s152": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",70 "est-qwen25-7b-gsm8k-c2-s153": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",71 "est-qwen25-7b-gsm8k-c3-s151": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",72 "est-qwen25-7b-gsm8k-c3-s152": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",73 "est-qwen25-7b-gsm8k-c3-s153": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",74 "est-qwen25-7b-math-c2-s151": "0bd659b3f69e2f779dbea84bb2da221bd0ff52db",75 "est-qwen25-7b-math-c2-s152": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",76 "est-qwen25-7b-math-c2-s153": "0bd659b3f69e2f779dbea84bb2da221bd0ff52db",77 "est-qwen25-7b-math-c3-s151": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",78 "est-qwen25-7b-math-c3-s152": "d8e0c3551d8572fb64e3b7126fbd7569a987973f",79 "est-qwen25-7b-math-c3-s153": "d8e0c3551d8572fb64e3b7126fbd7569a987973f"80 },81 "estimation_execution_commits_sha256": "cf70aa7c361f114b2d06a5da5fbbbf5021a13b7d19c24a55023647f5b3c8301a",82 "estimation_allocation_accounting_sha256": "1a91571d5bc83040c72a507602c8640faac7d6197608fcac4d17596c7ffc8483",83 "expected_host_capacity_derived_files": 13,84 "host_capacity_protocol_version": "host-capacity-1.0",85 "host_capacity_counts": {86 "planned_processes": 18,87 "planned_pairs": 9,88 "planned_configurations": 6,89 "eligible_processes": 9,90 "completed_processes": 0,91 "unresolved_processes": 9,92 "known_launched_processes": 18,93 "unknown_launch_status_processes": 0,94 "resolved_configurations": 3,95 "unresolved_configurations": 396 },97 "host_capacity_protocol_sha256": "f3f49617a7fca63ddac8923bccb83a15eaf52646e5f3db5df4aa716bb25c1ef6",98 "host_capacity_study_freeze_sha256": "80559199c59246d0e4ce52bf268d0375c90fc5168a5246ed900c9266abd951cd",99 "host_capacity_original_protocol_sha256": "1694e447e842600612604891e7b3d036ff0791fa84c4bdbeba54ed093feefdeb",100 "host_capacity_prediction_seal_sha256": "5ed714caff922a5109346a4242545d19f2824636d0f52bb0bbd2a09b40ed4666",101 "host_capacity_execution_commits": {102 "host-qwen25-7b-code_heavy_tail-c2-s161": "120e29a146294722a6668f55a1e7fed2878eb62a",103 "host-qwen25-7b-code_heavy_tail-c2-s162": "120e29a146294722a6668f55a1e7fed2878eb62a",104 "host-qwen25-7b-code_heavy_tail-c2-s163": "120e29a146294722a6668f55a1e7fed2878eb62a",105 "host-qwen25-7b-gsm8k-c2-s161": "120e29a146294722a6668f55a1e7fed2878eb62a",106 "host-qwen25-7b-gsm8k-c2-s162": "120e29a146294722a6668f55a1e7fed2878eb62a",107 "host-qwen25-7b-gsm8k-c2-s163": "120e29a146294722a6668f55a1e7fed2878eb62a",108 "host-qwen25-7b-math-c2-s161": "120e29a146294722a6668f55a1e7fed2878eb62a",109 "host-qwen25-7b-math-c2-s162": "120e29a146294722a6668f55a1e7fed2878eb62a",110 "host-qwen25-7b-math-c2-s163": "120e29a146294722a6668f55a1e7fed2878eb62a"111 },112 "host_capacity_execution_commits_sha256": "117b2449b2365db03f5af05aeae2c2943421ab739e5fd3c0ae16e3ce67ac5fcc",113 "host_capacity_allocation_accounting_sha256": "8df8040c1cd1608abd10759dd42448d734aa1b501733c926978247d2177f12ee"114}115 