520 MB

Ctrl+K

1 contributor

History: 144 commits

ProbioticFarmer

Clean up ephemeral _ephemeral_combo_exp4_priming_results.tar.gz

302f2ce verified about 16 hours ago

results
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
submissions
Add clean 240x <think> Chain-of-Thought evaluation results (HumanEval+ 83.54%, MBPP 95.24%, MBPP+ 79.89%, BCB-Hard 30.41%) 2 days ago
.gitattributes

1.52 kB
initial commit 5 days ago
COMPETITIONS_LEADERBOARD_REPORT.md

2.04 kB
Upload COMPETITIONS_LEADERBOARD_REPORT.md from A100-80GB 4-tier competition evaluation 5 days ago
FORENSIC_AUDIT_AND_REMEDIATION_HANDOFF.md

24.7 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
FULL_UPSTREAM_LEADERBOARD_REPORT.md

1.12 kB
Upload FULL_UPSTREAM_LEADERBOARD_REPORT.md from Full Upstream EvalPlus A100-80GB evaluation 5 days ago
HELDOUT_CODE_MATH_REPORT.md

1.29 kB
Upload HELDOUT_CODE_MATH_REPORT.md from Held-Out Code & Math A100-80GB evaluation 4 days ago
INDEPENDENT_AGENT_AUDIT_AND_TECHNICAL_REPORT.md

38.2 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
LEADERBOARD_ALL3_REPORT.md

5.78 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
PROJECT_README.md

17.8 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
README.md

10.8 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
SIDE_BY_SIDE_BENCHMARK_REPORT.md

6.53 kB
Upload SIDE_BY_SIDE_BENCHMARK_REPORT.md 3 days ago
adapter_config.json

660 Bytes
Upload verified 64-layer Qwen3.8-27B-BarronPolyStep-RealCoder-v3 (256 PEFT LoRA tensors, audited) 5 days ago
adapter_model.safetensors

69.2 MB
xet

Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
audit_realcoder_v3.py

16.9 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
benchmark_external_humaneval_ds.py

46.4 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
benchmark_report.json

4.31 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
clean_240x_think_eval_results.json

5.4 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
competition_requirements.txt

116 Bytes
Upload competition_requirements.txt from A100-80GB 4-tier competition evaluation 5 days ago
competitions_suite_hf_results.json

29.6 kB
Upload competitions_suite_hf_results.json from A100-80GB 4-tier competition evaluation 5 days ago
cuda_qwen38_gpu.py

29.9 kB
Upload cuda_qwen38_gpu.py from A100-80GB 4-tier competition evaluation 5 days ago
external_humaneval_ds1000_report.json

31.3 kB
Update RealCoder-v3 adapter and 40-function external open-ended benchmark results 5 days ago
extract_real_qwen38_features.py

20.5 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
full_upstream_evalplus_hf_results.json

75.5 kB
Upload full_upstream_evalplus_hf_results.json from Full Upstream EvalPlus A100-80GB evaluation 5 days ago
heldout_code_math_hf_results.json

27.8 kB
Upload heldout_code_math_hf_results.json from Held-Out Code & Math A100-80GB evaluation 4 days ago
hf_clean_240x_think_eval.py

72.3 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
leaderboard_all3_hf_results.json

46.1 kB
Upload leaderboard_all3_hf_results.json from 3-Step Leaderboard A100-80GB pipeline 3 days ago
mlx_qwen38_gpu.py

60.3 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
piqwen_a100_bundle.tar.gz

39.5 MB
xet

Upload PiQwen + SPI Experiment A bundle for A100-80GB execution 5 days ago
piqwen_a100_results.tar.gz

142 kB
xet

Upload PiQwen + SPI Experiment A A100-80GB rollout results 5 days ago
piqwen_competitions_suite_bundle.tar.gz

39.5 MB
xet

Upload 4-tier competitions & leaderboards bundle for A100-80GB execution 5 days ago
piqwen_competitions_suite_results.tar.gz

12.4 kB
xet

Upload 4-Tier Coding Competitions & Leaderboards A100-80GB results and official submissions 5 days ago
piqwen_full_upstream_evalplus_bundle.tar.gz

39.5 MB
xet

Upload Full Upstream EvalPlus (164 HumanEval+ & 60 MBPP+) bundle for A100-80GB execution 5 days ago
piqwen_full_upstream_evalplus_results.tar.gz

72.1 kB
xet

Upload Full Upstream EvalPlus (164 HumanEval+ + 60 MBPP+) A100-80GB results and submissions 5 days ago
piqwen_heldout_code_math_bundle.tar.gz

39.5 MB
xet

Upload Held-Out Code & Math Test-Time Strategy bundle for A100-80GB execution 4 days ago
piqwen_heldout_code_math_results.tar.gz

5.52 kB
xet

Upload Held-Out Code & Math Test-Time Strategy A100-80GB results and submissions 4 days ago
piqwen_leaderboard_all3_bundle.tar.gz

39.8 MB
xet

Upload 3-Step Public Leaderboard & Merge Verification bundle for A100-80GB execution 4 days ago
piqwen_leaderboard_all3_results.tar.gz

39 MB
xet

Upload 3-Step Public Leaderboard A100-80GB results, BCB-Hard 148, MBPP+ 378, and Merge Verification 3 days ago
piqwen_side_by_side_bundle.tar.gz

153 MB
xet

Upload side-by-side Hard-27B benchmark bundle for A100-80GB execution 5 days ago
piqwen_side_by_side_results.tar.gz

6.91 kB
xet

Upload A100-80GB side-by-side benchmark results (PolyStep-RealCoder-v3 vs Qwen3.8-27B Base) 3 days ago
piqwen_v4_patch.tar.gz

58.6 MB
xet

Upload lightweight v5 280x CUDA SVD patch for 3-Step Leaderboard A100-80GB execution 3 days ago
run_hf_clean_240x_think_eval.py

8.2 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
run_polystep_240x_clean.py

56.3 kB
Publish remediated strict Train-40-isolated 240x checkpoint (max_abs_diff=0.0 noise invariance) & A100-80GB zero-helper-leakage <think> evaluation results (93.39% MBPP / 78.57% MBPP+ / 79.88%-83.54% HE+) 2 days ago
script.py

7.68 kB
Upload script.py from A100-80GB 4-tier competition evaluation 5 days ago
side_by_side_hf_results.json

31 kB
Upload side_by_side_hf_results.json 3 days ago