Add files using upload-large-folder tool

.gitattributes

@@ -64,3 +64,140 @@ eval_results/vllm/qwen3-1.7B-it-nnm/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.9_K
 eval_results/vllm/qwen2.5-1.5b-tsd/Minsang__TSD-KD_Qwen2.5-1.5B/samples_gsm_plus_2026-05-20T09-44-19.696932.jsonl filter=lfs diff=lfs merge=lfs -text
 eval_results/vllm/gemma-2-2b-it/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/google__gemma-2-2b-it/samples_gsm8k_2026-05-20T05-33-07.171742.jsonl filter=lfs diff=lfs merge=lfs -text
 eval_results/vllm/qwen3-1.7B-it-amid/results__qwen3-1.7B\#amid__ab_pr_0.5_0.5_4_1e-4/samples_gsm8k_2026-05-18T05-31-42.614039.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-it-nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0-1246/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0__1246/samples_gsm_plus_2026-05-16T05-48-29.086091.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm_plus_2026-05-19T16-02-01.567296.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/samples_gsm_plus_2026-05-21T17-26-49.376807.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/Qwen__Qwen3-1.7B/samples_gsm_plus_2026-05-19T16-00-54.182264.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm_plus_2026-05-20T05-14-37.474356.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lw__nnm0.2_K128_L4_epoch1_lr1e-4_kdr1.0/samples_gsm_plus_2026-05-20T10-06-20.055408.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm_plus_2026-05-21T19-04-38.994347.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/gemma-2-2b-it/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/google__gemma-2-2b-it/samples_gsm_plus_2026-05-20T05-58-40.592353.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/gemma-2-2b-it/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/google__gemma-2-2b-it/samples_gsm_plus_2026-05-20T11-11-49.737097.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/gemma-2-2b-it/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/google__gemma-2-2b-it/samples_gsm_plus_2026-05-20T03-58-28.135424.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen3-1.7B-it-amid/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/samples_gsm_plus_2026-05-18T06-09-50.831279.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/samples_mbpp_2026-05-19T04-59-33.294652.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-19T04-46-37.564280.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/samples_minerva_math_counting_and_prob_2026-05-21T20-09-58.294511.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/samples_minerva_math_num_theory_2026-05-21T20-09-58.294511.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/samples_minerva_math_intermediate_algebra_2026-05-21T20-09-58.294511.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/samples_gsm8k_2026-05-21T19-56-04.497992.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/samples_mmlu_pro_math_2026-05-21T20-33-27.843841.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/samples_minerva_math_prealgebra_2026-05-21T20-09-58.294511.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/samples_minerva_math_geometry_2026-05-21T20-09-58.294511.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/samples_minerva_math_precalc_2026-05-21T20-09-58.294511.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/samples_mmlu_pro_math_2026-05-21T21-15-17.902035.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/samples_minerva_math_intermediate_algebra_2026-05-21T20-52-01.134693.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_algebra_2026-05-21T23-19-57.890601.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/samples_minerva_math_precalc_2026-05-21T20-52-01.134693.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/samples_mmlu_pro_math_2026-05-21T23-53-19.245282.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_prealgebra_2026-05-21T23-19-57.890601.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/samples_gsm8k_2026-05-21T20-38-57.497114.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_num_theory_2026-05-21T23-19-57.890601.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/samples_minerva_math_algebra_2026-05-21T20-52-01.134693.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_precalc_2026-05-21T23-19-57.890601.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_counting_and_prob_2026-05-21T23-19-57.890601.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_geometry_2026-05-21T23-19-57.890601.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-20T04-30-26.192040.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-20T04-45-38.942117.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/erank0.2_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__erank0.2_K128_L4_epoch2_lr1e-4_kdr1.0/samples_mmlu_pro_math_2026-05-20T10-44-29.653753.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/erank0.2_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__erank0.2_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-20T10-27-27.647641.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-21T22-15-42.779604.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/samples_gsm8k_2026-05-21T21-21-27.176019.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/samples_minerva_math_precalc_2026-05-21T17-47-51.336222.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/samples_minerva_math_prealgebra_2026-05-21T17-47-51.336222.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/samples_minerva_math_intermediate_algebra_2026-05-21T17-47-51.336222.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_intermediate_algebra_2026-05-21T21-35-33.295225.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_geometry_2026-05-21T22-29-17.792749.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/samples_mmlu_pro_math_2026-05-21T22-09-52.686827.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_geometry_2026-05-21T21-35-33.295225.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_precalc_2026-05-21T22-29-17.792749.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_num_theory_2026-05-21T22-29-17.792749.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_prealgebra_2026-05-21T22-29-17.792749.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_algebra_2026-05-21T22-29-17.792749.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_intermediate_algebra_2026-05-21T22-29-17.792749.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_mmlu_pro_math_2026-05-21T23-00-14.570401.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/samples_gsm8k_2026-05-21T17-34-22.668800.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_counting_and_prob_2026-05-21T22-29-17.792749.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/samples_minerva_math_algebra_2026-05-21T17-47-51.336222.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/samples_mmlu_pro_math_2026-05-21T17-28-31.668635.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_intermediate_algebra_2026-05-21T23-19-57.890601.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_num_theory_2026-05-21T16-54-41.262068.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_precalc_2026-05-21T16-54-41.262068.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_geometry_2026-05-21T16-54-41.262068.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_prealgebra_2026-05-21T16-54-41.262068.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_intermediate_algebra_2026-05-21T16-54-41.262068.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_counting_and_prob_2026-05-21T16-54-41.262068.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-21T23-05-45.631858.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/samples_gsm8k_2026-05-21T16-40-30.991516.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/gemma-2-2b-it/nnm0.1_K128_L4_epoch2_lr1e-4_kdr0.9/google__gemma-2-2b-it/samples_gsm8k_2026-05-22T00-00-05.536749.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/samples_gsm8k_2026-05-21T16-34-13.857266.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/gemma-2-2b-it/nnm0.1_K128_L4_epoch2_lr1e-4_kdr0.9_ckpt2492/google__gemma-2-2b-it/samples_gsm8k_2026-05-22T02-39-46.573021.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/gemma-2-2b-it/nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0_ckpt2492/google__gemma-2-2b-it/samples_gsm8k_2026-05-22T03-44-09.453955.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_precalc_2026-05-21T18-35-42.244057.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_counting_and_prob_2026-05-21T18-35-42.244057.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_algebra_2026-05-21T16-54-41.262068.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_mmlu_pro_math_2026-05-22T02-24-55.845923.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_geometry_2026-05-21T18-35-42.244057.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_prealgebra_2026-05-21T18-35-42.244057.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_mmlu_pro_math_2026-05-21T19-06-40.893218.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_algebra_2026-05-22T02-46-47.903666.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_intermediate_algebra_2026-05-22T02-46-47.903666.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_intermediate_algebra_2026-05-21T18-35-42.244057.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5b-tsd/Minsang__TSD-KD_Qwen2.5-1.5B/samples_sciq_2026-05-20T07-36-21.609469.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5b-tsd/Minsang__TSD-KD_Qwen2.5-1.5B/samples_gsm8k_2026-05-20T08-39-25.330595.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/Qwen__Qwen3-1.7B/samples_minerva_math_counting_and_prob_2026-05-19T15-31-31.443457.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/Qwen__Qwen3-1.7B/samples_minerva_math_geometry_2026-05-20T13-59-52.863729.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/Qwen__Qwen3-1.7B/samples_minerva_math_precalc_2026-05-19T15-31-31.443457.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5b-tsd/Minsang__TSD-KD_Qwen2.5-1.5B/samples_gsm8k_2026-05-20T09-24-44.698376.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.3_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.3_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-19T05-03-19.631875.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.3_K128_L4_epoch2_lr1e-4_kdr1.0/Qwen__Qwen3-1.7B/samples_gsm8k_2026-05-19T14-42-25.634092.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/samples_mmlu_pro_math_2026-05-19T16-03-33.329973.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-19T15-47-08.528666.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-20T04-02-44.931627.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-22T02-32-56.858752.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-22T02-02-38.373102.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_num_theory_2026-05-21T18-35-42.244057.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_num_theory_2026-05-21T21-35-33.295225.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_algebra_2026-05-21T18-35-42.244057.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_counting_and_prob_2026-05-21T21-35-33.295225.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_precalc_2026-05-21T21-35-33.295225.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_prealgebra_2026-05-21T21-35-33.295225.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-21T18-21-31.796596.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm1.0_K128_L4_epoch1_lr1e-4_kdr1.0/samples_minerva_math_algebra_2026-05-21T21-35-33.295225.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr0.75/samples_mmlu_pro_math_2026-05-21T18-15-32.438748.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/gemma-2-2b-it/nnm0.3_K128_L4_epoch2_lr1e-4_kdr1.0/google__gemma-2-2b-it/samples_gsm8k_2026-05-22T01-36-44.654190.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/gemma-2-2b-it/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/google__gemma-2-2b-it/samples_gsm8k_2026-05-22T02-08-03.041526.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-it-amid/results__qwen2.5-1.5B-Instruct\#amid__ab_pr_0.5_0.5_4_1e-4/samples_gsm8k_2026-05-17T08-02-58.299918.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/gemma-2-2b-it/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/google__gemma-2-2b-it/samples_gsm_plus_2026-05-22T00-56-12.500640.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-it-amid/results__qwen2.5-1.5B-Instruct\#csd__ab_pr_0.5_0.5_8_1e-4/samples_gsm8k_2026-05-18T01-45-38.483394.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm_new0.3_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm_plus_2026-05-21T23-50-13.541723.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/gemma-2-2b-it/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0_ckpt1246/google__gemma-2-2b-it/samples_gsm_plus_2026-05-22T03-36-20.052365.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr0.75/samples_minerva_math_algebra_2026-05-21T20-09-58.294511.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_precalc_2026-05-21T19-25-36.790826.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/samples_mmlu_pro_math_2026-05-21T19-50-55.324036.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_counting_and_prob_2026-05-21T19-25-36.790826.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_prealgebra_2026-05-21T19-25-36.790826.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_algebra_2026-05-21T19-25-36.790826.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_geometry_2026-05-21T19-25-36.790826.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/samples_minerva_math_counting_and_prob_2026-05-21T20-52-01.134693.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/samples_minerva_math_prealgebra_2026-05-21T20-52-01.134693.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/samples_minerva_math_intermediate_algebra_2026-05-21T19-25-36.790826.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-21T19-11-58.605949.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5b-tsd/Minsang__TSD-KD_Qwen2.5-1.5B/samples_gsm8k_2026-05-20T07-25-44.261824.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.9_K128_L4_epoch2_lr1e-4_kdr0.75/samples_gsm_plus_2026-05-21T21-13-42.194632.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-20T09-37-48.075763.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5b-tsd/Minsang__TSD-KD_Qwen2.5-1.5B/samples_gsm8k_2026-05-20T07-41-10.416362.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-Instruct\#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-20T04-59-22.053427.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen3-1.7B\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm_plus_2026-05-22T02-23-27.038737.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/Qwen__Qwen3-1.7B/samples_minerva_math_intermediate_algebra_2026-05-20T13-59-52.863729.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/Qwen__Qwen3-1.7B/samples_gsm8k_2026-05-19T15-05-24.735051.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/Qwen__Qwen3-1.7B/samples_gsm8k_2026-05-19T14-47-12.766152.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/Qwen__Qwen3-1.7B/samples_gsm8k_2026-05-19T15-11-29.808755.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/Qwen__Qwen3-1.7B/samples_minerva_math_algebra_2026-05-20T13-59-52.863729.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-it-nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0-1246/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0__1246/samples_gsm8k_2026-05-16T05-36-28.370914.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-it-nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0-1246/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0__1246/samples_gsm8k_2026-05-16T04-45-16.283740.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen2.5-1.5B-it-nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0-1246/results__qwen2.5-1.5B-Instruct\#sfkl_nnm_lora__nnm0.1_K128_L4_epoch1_lr1e-4_kdr1.0__1246/samples_gsm8k_2026-05-16T04-47-56.106645.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/Qwen__Qwen3-1.7B/samples_minerva_math_counting_and_prob_2026-05-20T13-59-52.863729.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/gemma-2-2b-it/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/google__gemma-2-2b-it/samples_gsm_plus_2026-05-22T02-31-51.763404.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results/vllm/qwen3-1.7B\#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/Qwen__Qwen3-1.7B/samples_mmlu_pro_math_2026-05-19T16-04-10.333630.jsonl filter=lfs diff=lfs merge=lfs -text
+eval_results_2/vllm/gemma-2-2b-it/nnm0.1_K128_L4_epoch2_lr1e-4_kdr0.9_ckpt2492/google__gemma-2-2b-it/samples_gsm_plus_2026-05-22T03-04-17.698928.jsonl filter=lfs diff=lfs merge=lfs -text

eval_results/vllm/gemma-2-2b-it/nnm0.1_K128_L4_epoch2_lr1e-4_kdr1.0/google__gemma-2-2b-it/samples_gsm_plus_2026-05-20T03-58-28.135424.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13d98302f2dd81464e51f67b88d5b99f95e431f2d9f6f49e439014512fa3c02c
+size 133305231

eval_results/vllm/gemma-2-2b-it/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/google__gemma-2-2b-it/samples_gsm_plus_2026-05-20T05-58-40.592353.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35ea12d5db9a140dcb4b87de1cb16be03a8ae2b4afe135872d92525dd936c62f
+size 130569938

eval_results/vllm/gemma-2-2b-it/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/google__gemma-2-2b-it/samples_gsm_plus_2026-05-20T11-11-49.737097.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:04398e2b52a3bf8555ce97e5ffd78df755f307f3163c3447e408b96fbc6a5544
+size 130467045

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/erank0.2_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__erank0.2_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-20T10-27-27.647641.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:11aed7e2004f70c1e9d37273ecc49b8319395b325e3fe813a9c0a7d6a9308a00
+size 13751094

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/erank0.2_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__erank0.2_K128_L4_epoch2_lr1e-4_kdr1.0/samples_mmlu_pro_math_2026-05-20T10-44-29.653753.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:851c454b6328382ccc28c804de824fac18813973e920e5eb2c07cb52e7fc43ab
+size 11364001

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-19T15-47-08.528666.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:085f8548456ef1b470bba2bc16f351de49588d2ad977bff9e3556c68fc733a9f
+size 14111501

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-20T04-02-44.931627.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8ac64407c921ba0584cb3937abeb14bfd2541b2d9012b6a8bf8d03d11b1b2f4
+size 13878674

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm_plus_2026-05-19T16-02-01.567296.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58eb6c7658a77e20d2d684df36fcbc8ab752bcaac979b2ad4d2366c2fb174c26
+size 139042463

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__nnm0.2_K128_L4_epoch2_lr1e-4_kdr1.0/samples_mmlu_pro_math_2026-05-19T16-03-33.329973.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22430f28759e5c4ee32800f509e68244b9e7afe7d22b348ed9c30e6ee8950352
+size 10587255

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.3_K128_L4_epoch2_lr1e-4_kdr1.0/Qwen__Qwen3-1.7B/samples_gsm8k_2026-05-19T14-42-25.634092.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9895bd962f91ccafa1394e7bf8972b14ff4e6ff9d66e182701719d8e478d5f2
+size 19791473

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.3_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__nnm0.3_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-19T05-03-19.631875.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88c1caa51233ff737adcac1e55d85d66b07e2e97e79d201589cdc0884ab3cd6e
+size 14313588

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-19T04-46-37.564280.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce54bdb239a644bd20f7d53f100e55357813760bb29f43928e7fae43a3b0031b
+size 14007107

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__nnm0.5_K128_L4_epoch2_lr1e-4_kdr1.0/samples_mbpp_2026-05-19T04-59-33.294652.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25b5d33397b0f8f24ce445bbf5c04b657bebd4abc27bb4bd4ccfbe9a1ad6de76
+size 19539640

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-20T04-30-26.192040.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f08e0ac8abcded019fba54fa4ae534f4ae8ac7d93896d3c72f634d800b28d701
+size 13876573

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-20T04-45-38.942117.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb2e0df8d65337b47fe0c6b2a7264fd092bfebc98d6fd01753a005c92f27144a
+size 13876591

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-20T04-59-22.053427.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f08e0ac8abcded019fba54fa4ae534f4ae8ac7d93896d3c72f634d800b28d701
+size 13876573

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm8k_2026-05-20T09-37-48.075763.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a36b9b3c49b0da29da7e50b1cbfbd573375ca2b75dbde775dc2364aaa716e358
+size 13875351

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lora__nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/samples_gsm_plus_2026-05-20T05-14-37.474356.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22cc73a1294ba67ffc932ed04740568e0980a3088f4d464c129bb3a7550177e7
+size 132622224

eval_results/vllm/qwen2.5-1.5B-Instruct#sfkl_nnm_lora/nnm0.7_K128_L4_epoch2_lr1e-4_kdr1.0/results__qwen2.5-1.5B-Instruct#sfkl_nnm_lw__nnm0.2_K128_L4_epoch1_lr1e-4_kdr1.0/samples_gsm_plus_2026-05-20T10-06-20.055408.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:287765468003dda2ac14033569d29a7f9b42d2a96178cac07bad5a4075c4a6ca
+size 131870241

eval_results/vllm/qwen2.5-1.5B-it-amid/results__qwen2.5-1.5B-Instruct#amid__ab_pr_0.5_0.5_4_1e-4/results_2026-05-17T08-18-54.582733.json

@@ -0,0 +1,135 @@
+{
+  "results": {
+    "mbpp": {
+      "name": "mbpp",
+      "alias": "mbpp",
+      "sample_len": 500,
+      "pass_at_1,none": 0.436,
+      "pass_at_1_stderr,none": 0.0221989546414768
+    }
+  },
+  "group_subtasks": {},
+  "configs": {
+    "mbpp": {
+      "task": "mbpp",
+      "dataset_path": "google-research-datasets/mbpp",
+      "dataset_name": "full",
+      "test_split": "test",
+      "doc_to_text": "You are an expert Python programmer, and here is your task: {{text}} Your code should pass these tests:\n\n{{test_list[0]}}\n{{test_list[1]}}\n{{test_list[2]}}\n[BEGIN]\n",
+      "doc_to_target": "{% if is_fewshot is defined %}{{code}}\n[DONE]{% else %}{{test_list[0]}}\n{{test_list[1]}}\n{{test_list[2]}}{% endif %}",
+      "unsafe_code": true,
+      "description": "",
+      "target_delimiter": "",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "first_n",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": "<function list_fewshot_samples at 0x7742e379bb50>",
+        "doc_to_text": "You are an expert Python programmer, and here is your task: {{text}} Your code should pass these tests:\n\n{{test_list[0]}}\n{{test_list[1]}}\n{{test_list[2]}}\n[BEGIN]\n",
+        "doc_to_choice": null,
+        "doc_to_target": "{% if is_fewshot is defined %}{{code}}\n[DONE]{% else %}{{test_list[0]}}\n{{test_list[1]}}\n{{test_list[2]}}{% endif %}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": ""
+      },
+      "num_fewshot": 3,
+      "metric_list": [
+        {
+          "metric": "def pass_at_1(\n    references: Union[str, list[str]], predictions: Union[str, list[list[str]]]\n) -> float:\n    if isinstance(references, str):\n        references = [references]\n    if isinstance(predictions[0], str):\n        predictions = [[p] for p in predictions]\n    return pass_at_k.compute(\n        references=references,\n        predictions=predictions,\n        k=[1],\n    )[0][\"pass@1\"]\n",
+          "aggregation": "mean",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": [
+          "[DONE]"
+        ],
+        "do_sample": false
+      },
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "results/qwen2.5-1.5B-Instruct#amid/ab_pr_0.5_0.5_4_1e-4",
+        "tensor_parallel_size": 2,
+        "dtype": "float16",
+        "gpu_memory_utilization": 0.75,
+        "trust_remote_code": true,
+        "config_source": "/home/ubuntu/AMiD/.venv/lib/python3.10/site-packages/lm_eval/tasks/mbpp/mbpp.yaml"
+      }
+    }
+  },
+  "versions": {
+    "mbpp": 1.0
+  },
+  "n-shot": {
+    "mbpp": 3
+  },
+  "higher_is_better": {
+    "mbpp": {
+      "pass_at_1": true
+    }
+  },
+  "n-samples": {
+    "mbpp": {
+      "original": 500,
+      "effective": 500
+    }
+  },
+  "config": {
+    "model": "vllm",
+    "model_args": {
+      "pretrained": "results/qwen2.5-1.5B-Instruct#amid/ab_pr_0.5_0.5_4_1e-4",
+      "tensor_parallel_size": 2,
+      "dtype": "float16",
+      "gpu_memory_utilization": 0.75,
+      "trust_remote_code": true
+    },
+    "batch_size": "auto",
+    "batch_sizes": [],
+    "device": "cuda:0",
+    "use_cache": null,
+    "limit": null,
+    "bootstrap_iters": 100000,
+    "gen_kwargs": {},
+    "random_seed": 0,
+    "numpy_seed": 1234,
+    "torch_seed": 1234,
+    "fewshot_seed": 1234
+  },
+  "git_hash": "3e47bef",
+  "date": 1779005739.493435,
+  "pretty_env_info": "PyTorch version: 2.9.1+cu128\nIs debug build: False\nCUDA used to build PyTorch: 12.8\nROCM used to build PyTorch: N/A\n\nOS: Ubuntu 24.04.4 LTS (x86_64)\nGCC version: (Ubuntu 13.3.0-6ubuntu2~24.04.1) 13.3.0\nClang version: Could not collect\nCMake version: version 3.28.3\nLibc version: glibc-2.39\n\nPython version: 3.10.20 (main, May 10 2026, 19:27:29) [Clang 22.1.3 ] (64-bit runtime)\nPython platform: Linux-6.17.0-1013-aws-x86_64-with-glibc2.39\nIs CUDA available: True\nCUDA runtime version: Could not collect\nCUDA_MODULE_LOADING set to: \nGPU models and configuration: \nGPU 0: NVIDIA H200\nGPU 1: NVIDIA H200\nGPU 2: NVIDIA H200\nGPU 3: NVIDIA H200\nGPU 4: NVIDIA H200\nGPU 5: NVIDIA H200\nGPU 6: NVIDIA H200\nGPU 7: NVIDIA H200\n\nNvidia driver version: 580.126.16\ncuDNN version: Probably one of the following:\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_adv.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_cnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_precompiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_runtime_compiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_graph.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_heuristic.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_ops.so.9\nIs XPU available: False\nHIP runtime version: N/A\nMIOpen runtime version: N/A\nIs XNNPACK available: True\n\nCPU:\nArchitecture:                            x86_64\nCPU op-mode(s):                          32-bit, 64-bit\nAddress sizes:                           48 bits physical, 48 bits virtual\nByte Order:                              Little Endian\nCPU(s):                                  192\nOn-line CPU(s) list:                     0-191\nVendor ID:                               AuthenticAMD\nModel name:                              AMD EPYC 7R13 Processor\nCPU family:                              25\nModel:                                   1\nThread(s) per core:                      2\nCore(s) per socket:                      48\nSocket(s):                               2\nStepping:                                1\nBogoMIPS:                                5299.99\nFlags:                                   fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology nonstop_tsc cpuid extd_apicid aperfmperf tsc_known_freq pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy cr8_legacy abm sse4a misalignsse 3dnowprefetch topoext perfctr_core ssbd ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 invpcid rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 clzero xsaveerptr rdpru wbnoinvd arat npt nrip_save vaes vpclmulqdq rdpid\nHypervisor vendor:                       KVM\nVirtualization type:                     full\nL1d cache:                               3 MiB (96 instances)\nL1i cache:                               3 MiB (96 instances)\nL2 cache:                                48 MiB (96 instances)\nL3 cache:                                384 MiB (12 instances)\nNUMA node(s):                            4\nNUMA node0 CPU(s):                       0-23,96-119\nNUMA node1 CPU(s):                       24-47,120-143\nNUMA node2 CPU(s):                       48-71,144-167\nNUMA node3 CPU(s):                       72-95,168-191\nVulnerability Gather data sampling:      Not affected\nVulnerability Ghostwrite:                Not affected\nVulnerability Indirect target selection: Not affected\nVulnerability Itlb multihit:             Not affected\nVulnerability L1tf:                      Not affected\nVulnerability Mds:                       Not affected\nVulnerability Meltdown:                  Not affected\nVulnerability Mmio stale data:           Not affected\nVulnerability Old microcode:             Not affected\nVulnerability Reg file data sampling:    Not affected\nVulnerability Retbleed:                  Not affected\nVulnerability Spec rstack overflow:      Mitigation; Safe RET\nVulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl\nVulnerability Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization\nVulnerability Spectre v2:                Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected\nVulnerability Srbds:                     Not affected\nVulnerability Tsa:                       Mitigation; Clear CPU buffers\nVulnerability Tsx async abort:           Not affected\nVulnerability Vmscape:                   Not affected\n\nVersions of relevant libraries:\n[pip3] Could not collect\n[conda] numpy                       2.4.5            pypi_0              pypi\n[conda] nvidia-cublas-cu12          12.4.5.8         pypi_0              pypi\n[conda] nvidia-cuda-cupti-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-nvrtc-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-runtime-cu12    12.4.127         pypi_0              pypi\n[conda] nvidia-cudnn-cu12           9.1.0.70         pypi_0              pypi\n[conda] nvidia-cufft-cu12           11.2.1.3         pypi_0              pypi\n[conda] nvidia-curand-cu12          10.3.5.147       pypi_0              pypi\n[conda] nvidia-cusolver-cu12        11.6.1.9         pypi_0              pypi\n[conda] nvidia-cusparse-cu12        12.3.1.170       pypi_0              pypi\n[conda] nvidia-nccl-cu12            2.21.5           pypi_0              pypi\n[conda] nvidia-nvjitlink-cu12       12.4.127         pypi_0              pypi\n[conda] nvidia-nvtx-cu12            12.4.127         pypi_0              pypi\n[conda] torch                       2.5.1+cu124      pypi_0              pypi",
+  "transformers_version": "4.57.3",
+  "lm_eval_version": "0.4.12",
+  "upper_git_hash": null,
+  "tokenizer_pad_token": [
+    "<|endoftext|>",
+    "151643"
+  ],
+  "tokenizer_eos_token": [
+    "<|im_end|>",
+    "151645"
+  ],
+  "tokenizer_bos_token": [
+    null,
+    "None"
+  ],
+  "eot_token_id": 151645,
+  "max_length": 32768,
+  "task_hashes": {
+    "mbpp": "62356f643ed511c895814c093ddb55451941156ba9febb31dc87c6e00d9723e3"
+  },
+  "model_source": "vllm",
+  "model_name": "results/qwen2.5-1.5B-Instruct#amid/ab_pr_0.5_0.5_4_1e-4",
+  "model_name_sanitized": "results__qwen2.5-1.5B-Instruct#amid__ab_pr_0.5_0.5_4_1e-4",
+  "system_instruction": null,
+  "system_instruction_sha": null,
+  "fewshot_as_multiturn": null,
+  "chat_template": null,
+  "chat_template_sha": null,
+  "total_evaluation_time_seconds": "203.03449699201155"
+}

eval_results/vllm/qwen2.5-1.5B-it-amid/results__qwen2.5-1.5B-Instruct#amid__ab_pr_0.5_0.5_4_1e-4/results_2026-05-17T08-22-15.616148.json

@@ -0,0 +1,176 @@
+{
+  "results": {
+    "gsm_plus": {
+      "name": "gsm_plus",
+      "alias": "gsm_plus",
+      "sample_len": 10552,
+      "exact_match,strict-match": 0.20612206216830933,
+      "exact_match_stderr,strict-match": 0.003938150366088475,
+      "exact_match,flexible-extract": 0.42124715693707354,
+      "exact_match_stderr,flexible-extract": 0.004806934989070513
+    }
+  },
+  "group_subtasks": {},
+  "configs": {
+    "gsm_plus": {
+      "task": "gsm_plus",
+      "dataset_path": "qintongli/GSM-Plus",
+      "training_split": "test",
+      "test_split": "test",
+      "fewshot_split": "test",
+      "doc_to_text": "Question: {{question}}\nAnswer:",
+      "doc_to_target": "{{solution}}",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": "test",
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Question: {{question}}\nAnswer:",
+        "doc_to_choice": null,
+        "doc_to_target": "{{solution}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 5,
+      "metric_list": [
+        {
+          "metric": "exact_match",
+          "aggregation": "mean",
+          "higher_is_better": true,
+          "ignore_case": true,
+          "ignore_punctuation": false,
+          "regexes_to_ignore": [
+            ",",
+            "\\$",
+            "(?s).*#### ",
+            "\\.$"
+          ]
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": [
+          "Question:",
+          "</s>",
+          "<|im_end|>"
+        ],
+        "do_sample": false,
+        "temperature": 0.0
+      },
+      "repeats": 1,
+      "filter_list": [
+        {
+          "name": "strict-match",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "#### (\\-?[0-9\\.\\,]+)"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        },
+        {
+          "name": "flexible-extract",
+          "filter": [
+            {
+              "function": "regex",
+              "group_select": -1,
+              "regex_pattern": "(-?[$0-9.,]{2,})|(-?[0-9]+)"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "results/qwen2.5-1.5B-Instruct#amid/ab_pr_0.5_0.5_4_1e-4",
+        "tensor_parallel_size": 2,
+        "dtype": "float16",
+        "gpu_memory_utilization": 0.75,
+        "trust_remote_code": true,
+        "config_source": "/home/ubuntu/AMiD/.venv/lib/python3.10/site-packages/lm_eval/tasks/gsm_plus/gsm_plus.yaml"
+      }
+    }
+  },
+  "versions": {
+    "gsm_plus": 1.0
+  },
+  "n-shot": {
+    "gsm_plus": 5
+  },
+  "higher_is_better": {
+    "gsm_plus": {
+      "exact_match": true
+    }
+  },
+  "n-samples": {
+    "gsm_plus": {
+      "original": 10552,
+      "effective": 10552
+    }
+  },
+  "config": {
+    "model": "vllm",
+    "model_args": {
+      "pretrained": "results/qwen2.5-1.5B-Instruct#amid/ab_pr_0.5_0.5_4_1e-4",
+      "tensor_parallel_size": 2,
+      "dtype": "float16",
+      "gpu_memory_utilization": 0.75,
+      "trust_remote_code": true
+    },
+    "batch_size": "auto",
+    "batch_sizes": [],
+    "device": "cuda:0",
+    "use_cache": null,
+    "limit": null,
+    "bootstrap_iters": 100000,
+    "gen_kwargs": {},
+    "random_seed": 0,
+    "numpy_seed": 1234,
+    "torch_seed": 1234,
+    "fewshot_seed": 1234
+  },
+  "git_hash": "3e47bef",
+  "date": 1779005945.558593,
+  "pretty_env_info": "PyTorch version: 2.9.1+cu128\nIs debug build: False\nCUDA used to build PyTorch: 12.8\nROCM used to build PyTorch: N/A\n\nOS: Ubuntu 24.04.4 LTS (x86_64)\nGCC version: (Ubuntu 13.3.0-6ubuntu2~24.04.1) 13.3.0\nClang version: Could not collect\nCMake version: version 3.28.3\nLibc version: glibc-2.39\n\nPython version: 3.10.20 (main, May 10 2026, 19:27:29) [Clang 22.1.3 ] (64-bit runtime)\nPython platform: Linux-6.17.0-1013-aws-x86_64-with-glibc2.39\nIs CUDA available: True\nCUDA runtime version: Could not collect\nCUDA_MODULE_LOADING set to: \nGPU models and configuration: \nGPU 0: NVIDIA H200\nGPU 1: NVIDIA H200\nGPU 2: NVIDIA H200\nGPU 3: NVIDIA H200\nGPU 4: NVIDIA H200\nGPU 5: NVIDIA H200\nGPU 6: NVIDIA H200\nGPU 7: NVIDIA H200\n\nNvidia driver version: 580.126.16\ncuDNN version: Probably one of the following:\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_adv.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_cnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_precompiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_runtime_compiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_graph.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_heuristic.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_ops.so.9\nIs XPU available: False\nHIP runtime version: N/A\nMIOpen runtime version: N/A\nIs XNNPACK available: True\n\nCPU:\nArchitecture:                            x86_64\nCPU op-mode(s):                          32-bit, 64-bit\nAddress sizes:                           48 bits physical, 48 bits virtual\nByte Order:                              Little Endian\nCPU(s):                                  192\nOn-line CPU(s) list:                     0-191\nVendor ID:                               AuthenticAMD\nModel name:                              AMD EPYC 7R13 Processor\nCPU family:                              25\nModel:                                   1\nThread(s) per core:                      2\nCore(s) per socket:                      48\nSocket(s):                               2\nStepping:                                1\nBogoMIPS:                                5299.99\nFlags:                                   fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology nonstop_tsc cpuid extd_apicid aperfmperf tsc_known_freq pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy cr8_legacy abm sse4a misalignsse 3dnowprefetch topoext perfctr_core ssbd ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 invpcid rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 clzero xsaveerptr rdpru wbnoinvd arat npt nrip_save vaes vpclmulqdq rdpid\nHypervisor vendor:                       KVM\nVirtualization type:                     full\nL1d cache:                               3 MiB (96 instances)\nL1i cache:                               3 MiB (96 instances)\nL2 cache:                                48 MiB (96 instances)\nL3 cache:                                384 MiB (12 instances)\nNUMA node(s):                            4\nNUMA node0 CPU(s):                       0-23,96-119\nNUMA node1 CPU(s):                       24-47,120-143\nNUMA node2 CPU(s):                       48-71,144-167\nNUMA node3 CPU(s):                       72-95,168-191\nVulnerability Gather data sampling:      Not affected\nVulnerability Ghostwrite:                Not affected\nVulnerability Indirect target selection: Not affected\nVulnerability Itlb multihit:             Not affected\nVulnerability L1tf:                      Not affected\nVulnerability Mds:                       Not affected\nVulnerability Meltdown:                  Not affected\nVulnerability Mmio stale data:           Not affected\nVulnerability Old microcode:             Not affected\nVulnerability Reg file data sampling:    Not affected\nVulnerability Retbleed:                  Not affected\nVulnerability Spec rstack overflow:      Mitigation; Safe RET\nVulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl\nVulnerability Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization\nVulnerability Spectre v2:                Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected\nVulnerability Srbds:                     Not affected\nVulnerability Tsa:                       Mitigation; Clear CPU buffers\nVulnerability Tsx async abort:           Not affected\nVulnerability Vmscape:                   Not affected\n\nVersions of relevant libraries:\n[pip3] Could not collect\n[conda] numpy                       2.4.5            pypi_0              pypi\n[conda] nvidia-cublas-cu12          12.4.5.8         pypi_0              pypi\n[conda] nvidia-cuda-cupti-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-nvrtc-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-runtime-cu12    12.4.127         pypi_0              pypi\n[conda] nvidia-cudnn-cu12           9.1.0.70         pypi_0              pypi\n[conda] nvidia-cufft-cu12           11.2.1.3         pypi_0              pypi\n[conda] nvidia-curand-cu12          10.3.5.147       pypi_0              pypi\n[conda] nvidia-cusolver-cu12        11.6.1.9         pypi_0              pypi\n[conda] nvidia-cusparse-cu12        12.3.1.170       pypi_0              pypi\n[conda] nvidia-nccl-cu12            2.21.5           pypi_0              pypi\n[conda] nvidia-nvjitlink-cu12       12.4.127         pypi_0              pypi\n[conda] nvidia-nvtx-cu12            12.4.127         pypi_0              pypi\n[conda] torch                       2.5.1+cu124      pypi_0              pypi",
+  "transformers_version": "4.57.3",
+  "lm_eval_version": "0.4.12",
+  "upper_git_hash": null,
+  "tokenizer_pad_token": [
+    "<|endoftext|>",
+    "151643"
+  ],
+  "tokenizer_eos_token": [
+    "<|im_end|>",
+    "151645"
+  ],
+  "tokenizer_bos_token": [
+    null,
+    "None"
+  ],
+  "eot_token_id": 151645,
+  "max_length": 32768,
+  "task_hashes": {
+    "gsm_plus": "9fc5cf479857a1eb1a143016f6217e0f6f7a6a3251c77540c70b68178e0daa41"
+  },
+  "model_source": "vllm",
+  "model_name": "results/qwen2.5-1.5B-Instruct#amid/ab_pr_0.5_0.5_4_1e-4",
+  "model_name_sanitized": "results__qwen2.5-1.5B-Instruct#amid__ab_pr_0.5_0.5_4_1e-4",
+  "system_instruction": null,
+  "system_instruction_sha": null,
+  "fewshot_as_multiturn": true,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "chat_template_sha": "cd8e9439f0570856fd70470bf8889ebd8b5d1107207f67a5efb46e342330527f",
+  "total_evaluation_time_seconds": "197.96630131197162"
+}

eval_results/vllm/qwen2.5-1.5B-it-amid/results__qwen2.5-1.5B-Instruct#amid__ab_pr_0.5_0.5_4_1e-4/samples_gsm8k_2026-05-17T08-02-58.299918.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09e7dcb06cbabf80fe3fdbc1d70142a844a5ac115487cf66a12fcbdb7a7806ae
+size 14205616

eval_results/vllm/qwen2.5-1.5B-it-amid/results__qwen2.5-1.5B-Instruct#csd__ab_pr_0.5_0.5_8_1e-4/results_2026-05-18T01-45-38.483394.json

@@ -0,0 +1,177 @@
+{
+  "results": {
+    "gsm8k": {
+      "name": "gsm8k",
+      "alias": "gsm8k",
+      "sample_len": 1319,
+      "exact_match,strict-match": 0.39878695981804396,
+      "exact_match_stderr,strict-match": 0.013487360477060839,
+      "exact_match,flexible-extract": 0.6148597422289613,
+      "exact_match_stderr,flexible-extract": 0.013404165536474303
+    }
+  },
+  "group_subtasks": {},
+  "configs": {
+    "gsm8k": {
+      "task": "gsm8k",
+      "dataset_path": "openai/gsm8k",
+      "dataset_name": "main",
+      "training_split": "train",
+      "test_split": "test",
+      "fewshot_split": "train",
+      "doc_to_text": "Question: {{question}}\nAnswer:",
+      "doc_to_target": "{{answer}}",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": "train",
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Question: {{question}}\nAnswer:",
+        "doc_to_choice": null,
+        "doc_to_target": "{{answer}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 5,
+      "metric_list": [
+        {
+          "metric": "exact_match",
+          "aggregation": "mean",
+          "higher_is_better": true,
+          "ignore_case": true,
+          "ignore_punctuation": false,
+          "regexes_to_ignore": [
+            ",",
+            "\\$",
+            "(?s).*#### ",
+            "\\.$"
+          ]
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": [
+          "Question:",
+          "</s>",
+          "<|im_end|>"
+        ],
+        "do_sample": false,
+        "temperature": 0.0
+      },
+      "repeats": 1,
+      "filter_list": [
+        {
+          "name": "strict-match",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "#### (\\-?[0-9\\.\\,]+)"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        },
+        {
+          "name": "flexible-extract",
+          "filter": [
+            {
+              "function": "regex",
+              "group_select": -1,
+              "regex_pattern": "(-?[$0-9.,]{2,})|(-?[0-9]+)"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 3.0,
+        "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+        "tensor_parallel_size": 2,
+        "dtype": "float16",
+        "gpu_memory_utilization": 0.75,
+        "trust_remote_code": true,
+        "config_source": "/home/ubuntu/AMiD/.venv/lib/python3.10/site-packages/lm_eval/tasks/gsm8k/gsm8k.yaml"
+      }
+    }
+  },
+  "versions": {
+    "gsm8k": 3.0
+  },
+  "n-shot": {
+    "gsm8k": 5
+  },
+  "higher_is_better": {
+    "gsm8k": {
+      "exact_match": true
+    }
+  },
+  "n-samples": {
+    "gsm8k": {
+      "original": 1319,
+      "effective": 1319
+    }
+  },
+  "config": {
+    "model": "vllm",
+    "model_args": {
+      "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+      "tensor_parallel_size": 2,
+      "dtype": "float16",
+      "gpu_memory_utilization": 0.75,
+      "trust_remote_code": true
+    },
+    "batch_size": "auto",
+    "batch_sizes": [],
+    "device": "cuda:0",
+    "use_cache": null,
+    "limit": null,
+    "bootstrap_iters": 100000,
+    "gen_kwargs": {},
+    "random_seed": 0,
+    "numpy_seed": 1234,
+    "torch_seed": 1234,
+    "fewshot_seed": 1234
+  },
+  "git_hash": "e49ad15",
+  "date": 1779068671.219799,
+  "pretty_env_info": "PyTorch version: 2.9.1+cu128\nIs debug build: False\nCUDA used to build PyTorch: 12.8\nROCM used to build PyTorch: N/A\n\nOS: Ubuntu 24.04.4 LTS (x86_64)\nGCC version: (Ubuntu 13.3.0-6ubuntu2~24.04.1) 13.3.0\nClang version: Could not collect\nCMake version: version 3.28.3\nLibc version: glibc-2.39\n\nPython version: 3.10.20 (main, May 10 2026, 19:27:29) [Clang 22.1.3 ] (64-bit runtime)\nPython platform: Linux-6.17.0-1013-aws-x86_64-with-glibc2.39\nIs CUDA available: True\nCUDA runtime version: Could not collect\nCUDA_MODULE_LOADING set to: \nGPU models and configuration: \nGPU 0: NVIDIA H200\nGPU 1: NVIDIA H200\nGPU 2: NVIDIA H200\nGPU 3: NVIDIA H200\nGPU 4: NVIDIA H200\nGPU 5: NVIDIA H200\nGPU 6: NVIDIA H200\nGPU 7: NVIDIA H200\n\nNvidia driver version: 580.126.16\ncuDNN version: Probably one of the following:\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_adv.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_cnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_precompiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_runtime_compiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_graph.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_heuristic.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_ops.so.9\nIs XPU available: False\nHIP runtime version: N/A\nMIOpen runtime version: N/A\nIs XNNPACK available: True\n\nCPU:\nArchitecture:                            x86_64\nCPU op-mode(s):                          32-bit, 64-bit\nAddress sizes:                           48 bits physical, 48 bits virtual\nByte Order:                              Little Endian\nCPU(s):                                  192\nOn-line CPU(s) list:                     0-191\nVendor ID:                               AuthenticAMD\nModel name:                              AMD EPYC 7R13 Processor\nCPU family:                              25\nModel:                                   1\nThread(s) per core:                      2\nCore(s) per socket:                      48\nSocket(s):                               2\nStepping:                                1\nBogoMIPS:                                5299.99\nFlags:                                   fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology nonstop_tsc cpuid extd_apicid aperfmperf tsc_known_freq pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy cr8_legacy abm sse4a misalignsse 3dnowprefetch topoext perfctr_core ssbd ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 invpcid rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 clzero xsaveerptr rdpru wbnoinvd arat npt nrip_save vaes vpclmulqdq rdpid\nHypervisor vendor:                       KVM\nVirtualization type:                     full\nL1d cache:                               3 MiB (96 instances)\nL1i cache:                               3 MiB (96 instances)\nL2 cache:                                48 MiB (96 instances)\nL3 cache:                                384 MiB (12 instances)\nNUMA node(s):                            4\nNUMA node0 CPU(s):                       0-23,96-119\nNUMA node1 CPU(s):                       24-47,120-143\nNUMA node2 CPU(s):                       48-71,144-167\nNUMA node3 CPU(s):                       72-95,168-191\nVulnerability Gather data sampling:      Not affected\nVulnerability Ghostwrite:                Not affected\nVulnerability Indirect target selection: Not affected\nVulnerability Itlb multihit:             Not affected\nVulnerability L1tf:                      Not affected\nVulnerability Mds:                       Not affected\nVulnerability Meltdown:                  Not affected\nVulnerability Mmio stale data:           Not affected\nVulnerability Old microcode:             Not affected\nVulnerability Reg file data sampling:    Not affected\nVulnerability Retbleed:                  Not affected\nVulnerability Spec rstack overflow:      Mitigation; Safe RET\nVulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl\nVulnerability Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization\nVulnerability Spectre v2:                Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected\nVulnerability Srbds:                     Not affected\nVulnerability Tsa:                       Mitigation; Clear CPU buffers\nVulnerability Tsx async abort:           Not affected\nVulnerability Vmscape:                   Not affected\n\nVersions of relevant libraries:\n[pip3] Could not collect\n[conda] numpy                       2.4.5            pypi_0              pypi\n[conda] nvidia-cublas-cu12          12.4.5.8         pypi_0              pypi\n[conda] nvidia-cuda-cupti-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-nvrtc-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-runtime-cu12    12.4.127         pypi_0              pypi\n[conda] nvidia-cudnn-cu12           9.1.0.70         pypi_0              pypi\n[conda] nvidia-cufft-cu12           11.2.1.3         pypi_0              pypi\n[conda] nvidia-curand-cu12          10.3.5.147       pypi_0              pypi\n[conda] nvidia-cusolver-cu12        11.6.1.9         pypi_0              pypi\n[conda] nvidia-cusparse-cu12        12.3.1.170       pypi_0              pypi\n[conda] nvidia-nccl-cu12            2.21.5           pypi_0              pypi\n[conda] nvidia-nvjitlink-cu12       12.4.127         pypi_0              pypi\n[conda] nvidia-nvtx-cu12            12.4.127         pypi_0              pypi\n[conda] torch                       2.5.1+cu124      pypi_0              pypi",
+  "transformers_version": "4.57.3",
+  "lm_eval_version": "0.4.12",
+  "upper_git_hash": null,
+  "tokenizer_pad_token": [
+    "<|endoftext|>",
+    "151643"
+  ],
+  "tokenizer_eos_token": [
+    "<|im_end|>",
+    "151645"
+  ],
+  "tokenizer_bos_token": [
+    null,
+    "None"
+  ],
+  "eot_token_id": 151645,
+  "max_length": 32768,
+  "task_hashes": {
+    "gsm8k": "1876b0d29670c836f6ef54bca0a415304e7755fdf9babadb7a5d844de9d52cf2"
+  },
+  "model_source": "vllm",
+  "model_name": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+  "model_name_sanitized": "results__qwen2.5-1.5B-Instruct#csd__ab_pr_0.5_0.5_8_1e-4",
+  "system_instruction": null,
+  "system_instruction_sha": null,
+  "fewshot_as_multiturn": true,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "chat_template_sha": "cd8e9439f0570856fd70470bf8889ebd8b5d1107207f67a5efb46e342330527f",
+  "total_evaluation_time_seconds": "75.01892117102398"
+}

eval_results/vllm/qwen2.5-1.5B-it-amid/results__qwen2.5-1.5B-Instruct#csd__ab_pr_0.5_0.5_8_1e-4/results_2026-05-18T01-55-41.590782.json

@@ -0,0 +1,140 @@
+{
+  "results": {
+    "sciq": {
+      "name": "sciq",
+      "alias": "sciq",
+      "sample_len": 1000,
+      "acc,none": 0.819,
+      "acc_stderr,none": 0.012181436179177916,
+      "acc_norm,none": 0.584,
+      "acc_norm_stderr,none": 0.0155944601441406
+    }
+  },
+  "group_subtasks": {},
+  "configs": {
+    "sciq": {
+      "task": "sciq",
+      "dataset_path": "allenai/sciq",
+      "training_split": "train",
+      "validation_split": "validation",
+      "test_split": "test",
+      "doc_to_text": "{{support.lstrip()}}\nQuestion: {{question}}\nAnswer:",
+      "doc_to_target": 3,
+      "unsafe_code": false,
+      "doc_to_choice": "{{[distractor1, distractor2, distractor3, correct_answer]}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "{{support.lstrip()}}\nQuestion: {{question}}\nAnswer:",
+        "doc_to_choice": "{{[distractor1, distractor2, distractor3, correct_answer]}}",
+        "doc_to_target": 3,
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "aggregation": "mean",
+          "higher_is_better": true
+        },
+        {
+          "metric": "acc_norm",
+          "aggregation": "mean",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": true,
+      "doc_to_decontamination_query": "{{support}} {{question}}",
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+        "tensor_parallel_size": 2,
+        "dtype": "float16",
+        "gpu_memory_utilization": 0.75,
+        "trust_remote_code": true,
+        "config_source": "/home/ubuntu/AMiD/.venv/lib/python3.10/site-packages/lm_eval/tasks/sciq/sciq.yaml"
+      }
+    }
+  },
+  "versions": {
+    "sciq": 1.0
+  },
+  "n-shot": {
+    "sciq": 0
+  },
+  "higher_is_better": {
+    "sciq": {
+      "acc": true,
+      "acc_norm": true
+    }
+  },
+  "n-samples": {
+    "sciq": {
+      "original": 1000,
+      "effective": 1000
+    }
+  },
+  "config": {
+    "model": "vllm",
+    "model_args": {
+      "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+      "tensor_parallel_size": 2,
+      "dtype": "float16",
+      "gpu_memory_utilization": 0.75,
+      "trust_remote_code": true
+    },
+    "batch_size": "auto",
+    "batch_sizes": [],
+    "device": "cuda:0",
+    "use_cache": null,
+    "limit": null,
+    "bootstrap_iters": 100000,
+    "gen_kwargs": {},
+    "random_seed": 0,
+    "numpy_seed": 1234,
+    "torch_seed": 1234,
+    "fewshot_seed": 1234
+  },
+  "git_hash": "e49ad15",
+  "date": 1779069282.5380046,
+  "pretty_env_info": "PyTorch version: 2.9.1+cu128\nIs debug build: False\nCUDA used to build PyTorch: 12.8\nROCM used to build PyTorch: N/A\n\nOS: Ubuntu 24.04.4 LTS (x86_64)\nGCC version: (Ubuntu 13.3.0-6ubuntu2~24.04.1) 13.3.0\nClang version: Could not collect\nCMake version: version 3.28.3\nLibc version: glibc-2.39\n\nPython version: 3.10.20 (main, May 10 2026, 19:27:29) [Clang 22.1.3 ] (64-bit runtime)\nPython platform: Linux-6.17.0-1013-aws-x86_64-with-glibc2.39\nIs CUDA available: True\nCUDA runtime version: Could not collect\nCUDA_MODULE_LOADING set to: \nGPU models and configuration: \nGPU 0: NVIDIA H200\nGPU 1: NVIDIA H200\nGPU 2: NVIDIA H200\nGPU 3: NVIDIA H200\nGPU 4: NVIDIA H200\nGPU 5: NVIDIA H200\nGPU 6: NVIDIA H200\nGPU 7: NVIDIA H200\n\nNvidia driver version: 580.126.16\ncuDNN version: Probably one of the following:\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_adv.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_cnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_precompiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_runtime_compiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_graph.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_heuristic.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_ops.so.9\nIs XPU available: False\nHIP runtime version: N/A\nMIOpen runtime version: N/A\nIs XNNPACK available: True\n\nCPU:\nArchitecture:                            x86_64\nCPU op-mode(s):                          32-bit, 64-bit\nAddress sizes:                           48 bits physical, 48 bits virtual\nByte Order:                              Little Endian\nCPU(s):                                  192\nOn-line CPU(s) list:                     0-191\nVendor ID:                               AuthenticAMD\nModel name:                              AMD EPYC 7R13 Processor\nCPU family:                              25\nModel:                                   1\nThread(s) per core:                      2\nCore(s) per socket:                      48\nSocket(s):                               2\nStepping:                                1\nBogoMIPS:                                5299.99\nFlags:                                   fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology nonstop_tsc cpuid extd_apicid aperfmperf tsc_known_freq pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy cr8_legacy abm sse4a misalignsse 3dnowprefetch topoext perfctr_core ssbd ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 invpcid rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 clzero xsaveerptr rdpru wbnoinvd arat npt nrip_save vaes vpclmulqdq rdpid\nHypervisor vendor:                       KVM\nVirtualization type:                     full\nL1d cache:                               3 MiB (96 instances)\nL1i cache:                               3 MiB (96 instances)\nL2 cache:                                48 MiB (96 instances)\nL3 cache:                                384 MiB (12 instances)\nNUMA node(s):                            4\nNUMA node0 CPU(s):                       0-23,96-119\nNUMA node1 CPU(s):                       24-47,120-143\nNUMA node2 CPU(s):                       48-71,144-167\nNUMA node3 CPU(s):                       72-95,168-191\nVulnerability Gather data sampling:      Not affected\nVulnerability Ghostwrite:                Not affected\nVulnerability Indirect target selection: Not affected\nVulnerability Itlb multihit:             Not affected\nVulnerability L1tf:                      Not affected\nVulnerability Mds:                       Not affected\nVulnerability Meltdown:                  Not affected\nVulnerability Mmio stale data:           Not affected\nVulnerability Old microcode:             Not affected\nVulnerability Reg file data sampling:    Not affected\nVulnerability Retbleed:                  Not affected\nVulnerability Spec rstack overflow:      Mitigation; Safe RET\nVulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl\nVulnerability Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization\nVulnerability Spectre v2:                Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected\nVulnerability Srbds:                     Not affected\nVulnerability Tsa:                       Mitigation; Clear CPU buffers\nVulnerability Tsx async abort:           Not affected\nVulnerability Vmscape:                   Not affected\n\nVersions of relevant libraries:\n[pip3] Could not collect\n[conda] numpy                       2.4.5            pypi_0              pypi\n[conda] nvidia-cublas-cu12          12.4.5.8         pypi_0              pypi\n[conda] nvidia-cuda-cupti-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-nvrtc-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-runtime-cu12    12.4.127         pypi_0              pypi\n[conda] nvidia-cudnn-cu12           9.1.0.70         pypi_0              pypi\n[conda] nvidia-cufft-cu12           11.2.1.3         pypi_0              pypi\n[conda] nvidia-curand-cu12          10.3.5.147       pypi_0              pypi\n[conda] nvidia-cusolver-cu12        11.6.1.9         pypi_0              pypi\n[conda] nvidia-cusparse-cu12        12.3.1.170       pypi_0              pypi\n[conda] nvidia-nccl-cu12            2.21.5           pypi_0              pypi\n[conda] nvidia-nvjitlink-cu12       12.4.127         pypi_0              pypi\n[conda] nvidia-nvtx-cu12            12.4.127         pypi_0              pypi\n[conda] torch                       2.5.1+cu124      pypi_0              pypi",
+  "transformers_version": "4.57.3",
+  "lm_eval_version": "0.4.12",
+  "upper_git_hash": null,
+  "tokenizer_pad_token": [
+    "<|endoftext|>",
+    "151643"
+  ],
+  "tokenizer_eos_token": [
+    "<|im_end|>",
+    "151645"
+  ],
+  "tokenizer_bos_token": [
+    null,
+    "None"
+  ],
+  "eot_token_id": 151645,
+  "max_length": 32768,
+  "task_hashes": {
+    "sciq": "94d7c305efca6804164ebf166179faaee89d73aba3a1539fb62b358258a4b95f"
+  },
+  "model_source": "vllm",
+  "model_name": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+  "model_name_sanitized": "results__qwen2.5-1.5B-Instruct#csd__ab_pr_0.5_0.5_8_1e-4",
+  "system_instruction": null,
+  "system_instruction_sha": null,
+  "fewshot_as_multiturn": true,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "chat_template_sha": "cd8e9439f0570856fd70470bf8889ebd8b5d1107207f67a5efb46e342330527f",
+  "total_evaluation_time_seconds": "67.09884223603876"
+}

eval_results/vllm/qwen2.5-1.5B-it-amid/results__qwen2.5-1.5B-Instruct#csd__ab_pr_0.5_0.5_8_1e-4/results_2026-05-18T02-02-06.353674.json

@@ -0,0 +1,176 @@
+{
+  "results": {
+    "gsm_plus": {
+      "name": "gsm_plus",
+      "alias": "gsm_plus",
+      "sample_len": 10552,
+      "exact_match,strict-match": 0.20754359363153904,
+      "exact_match_stderr,strict-match": 0.0039481672795388615,
+      "exact_match,flexible-extract": 0.41537149355572406,
+      "exact_match_stderr,flexible-extract": 0.0047974618117091255
+    }
+  },
+  "group_subtasks": {},
+  "configs": {
+    "gsm_plus": {
+      "task": "gsm_plus",
+      "dataset_path": "qintongli/GSM-Plus",
+      "training_split": "test",
+      "test_split": "test",
+      "fewshot_split": "test",
+      "doc_to_text": "Question: {{question}}\nAnswer:",
+      "doc_to_target": "{{solution}}",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": "test",
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Question: {{question}}\nAnswer:",
+        "doc_to_choice": null,
+        "doc_to_target": "{{solution}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 5,
+      "metric_list": [
+        {
+          "metric": "exact_match",
+          "aggregation": "mean",
+          "higher_is_better": true,
+          "ignore_case": true,
+          "ignore_punctuation": false,
+          "regexes_to_ignore": [
+            ",",
+            "\\$",
+            "(?s).*#### ",
+            "\\.$"
+          ]
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": [
+          "Question:",
+          "</s>",
+          "<|im_end|>"
+        ],
+        "do_sample": false,
+        "temperature": 0.0
+      },
+      "repeats": 1,
+      "filter_list": [
+        {
+          "name": "strict-match",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "#### (\\-?[0-9\\.\\,]+)"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        },
+        {
+          "name": "flexible-extract",
+          "filter": [
+            {
+              "function": "regex",
+              "group_select": -1,
+              "regex_pattern": "(-?[$0-9.,]{2,})|(-?[0-9]+)"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+        "tensor_parallel_size": 2,
+        "dtype": "float16",
+        "gpu_memory_utilization": 0.75,
+        "trust_remote_code": true,
+        "config_source": "/home/ubuntu/AMiD/.venv/lib/python3.10/site-packages/lm_eval/tasks/gsm_plus/gsm_plus.yaml"
+      }
+    }
+  },
+  "versions": {
+    "gsm_plus": 1.0
+  },
+  "n-shot": {
+    "gsm_plus": 5
+  },
+  "higher_is_better": {
+    "gsm_plus": {
+      "exact_match": true
+    }
+  },
+  "n-samples": {
+    "gsm_plus": {
+      "original": 10552,
+      "effective": 10552
+    }
+  },
+  "config": {
+    "model": "vllm",
+    "model_args": {
+      "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+      "tensor_parallel_size": 2,
+      "dtype": "float16",
+      "gpu_memory_utilization": 0.75,
+      "trust_remote_code": true
+    },
+    "batch_size": "auto",
+    "batch_sizes": [],
+    "device": "cuda:0",
+    "use_cache": null,
+    "limit": null,
+    "bootstrap_iters": 100000,
+    "gen_kwargs": {},
+    "random_seed": 0,
+    "numpy_seed": 1234,
+    "torch_seed": 1234,
+    "fewshot_seed": 1234
+  },
+  "git_hash": "e49ad15",
+  "date": 1779069544.020459,
+  "pretty_env_info": "PyTorch version: 2.9.1+cu128\nIs debug build: False\nCUDA used to build PyTorch: 12.8\nROCM used to build PyTorch: N/A\n\nOS: Ubuntu 24.04.4 LTS (x86_64)\nGCC version: (Ubuntu 13.3.0-6ubuntu2~24.04.1) 13.3.0\nClang version: Could not collect\nCMake version: version 3.28.3\nLibc version: glibc-2.39\n\nPython version: 3.10.20 (main, May 10 2026, 19:27:29) [Clang 22.1.3 ] (64-bit runtime)\nPython platform: Linux-6.17.0-1013-aws-x86_64-with-glibc2.39\nIs CUDA available: True\nCUDA runtime version: Could not collect\nCUDA_MODULE_LOADING set to: \nGPU models and configuration: \nGPU 0: NVIDIA H200\nGPU 1: NVIDIA H200\nGPU 2: NVIDIA H200\nGPU 3: NVIDIA H200\nGPU 4: NVIDIA H200\nGPU 5: NVIDIA H200\nGPU 6: NVIDIA H200\nGPU 7: NVIDIA H200\n\nNvidia driver version: 580.126.16\ncuDNN version: Probably one of the following:\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_adv.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_cnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_precompiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_runtime_compiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_graph.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_heuristic.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_ops.so.9\nIs XPU available: False\nHIP runtime version: N/A\nMIOpen runtime version: N/A\nIs XNNPACK available: True\n\nCPU:\nArchitecture:                            x86_64\nCPU op-mode(s):                          32-bit, 64-bit\nAddress sizes:                           48 bits physical, 48 bits virtual\nByte Order:                              Little Endian\nCPU(s):                                  192\nOn-line CPU(s) list:                     0-191\nVendor ID:                               AuthenticAMD\nModel name:                              AMD EPYC 7R13 Processor\nCPU family:                              25\nModel:                                   1\nThread(s) per core:                      2\nCore(s) per socket:                      48\nSocket(s):                               2\nStepping:                                1\nBogoMIPS:                                5299.99\nFlags:                                   fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology nonstop_tsc cpuid extd_apicid aperfmperf tsc_known_freq pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy cr8_legacy abm sse4a misalignsse 3dnowprefetch topoext perfctr_core ssbd ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 invpcid rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 clzero xsaveerptr rdpru wbnoinvd arat npt nrip_save vaes vpclmulqdq rdpid\nHypervisor vendor:                       KVM\nVirtualization type:                     full\nL1d cache:                               3 MiB (96 instances)\nL1i cache:                               3 MiB (96 instances)\nL2 cache:                                48 MiB (96 instances)\nL3 cache:                                384 MiB (12 instances)\nNUMA node(s):                            4\nNUMA node0 CPU(s):                       0-23,96-119\nNUMA node1 CPU(s):                       24-47,120-143\nNUMA node2 CPU(s):                       48-71,144-167\nNUMA node3 CPU(s):                       72-95,168-191\nVulnerability Gather data sampling:      Not affected\nVulnerability Ghostwrite:                Not affected\nVulnerability Indirect target selection: Not affected\nVulnerability Itlb multihit:             Not affected\nVulnerability L1tf:                      Not affected\nVulnerability Mds:                       Not affected\nVulnerability Meltdown:                  Not affected\nVulnerability Mmio stale data:           Not affected\nVulnerability Old microcode:             Not affected\nVulnerability Reg file data sampling:    Not affected\nVulnerability Retbleed:                  Not affected\nVulnerability Spec rstack overflow:      Mitigation; Safe RET\nVulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl\nVulnerability Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization\nVulnerability Spectre v2:                Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected\nVulnerability Srbds:                     Not affected\nVulnerability Tsa:                       Mitigation; Clear CPU buffers\nVulnerability Tsx async abort:           Not affected\nVulnerability Vmscape:                   Not affected\n\nVersions of relevant libraries:\n[pip3] Could not collect\n[conda] numpy                       2.4.5            pypi_0              pypi\n[conda] nvidia-cublas-cu12          12.4.5.8         pypi_0              pypi\n[conda] nvidia-cuda-cupti-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-nvrtc-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-runtime-cu12    12.4.127         pypi_0              pypi\n[conda] nvidia-cudnn-cu12           9.1.0.70         pypi_0              pypi\n[conda] nvidia-cufft-cu12           11.2.1.3         pypi_0              pypi\n[conda] nvidia-curand-cu12          10.3.5.147       pypi_0              pypi\n[conda] nvidia-cusolver-cu12        11.6.1.9         pypi_0              pypi\n[conda] nvidia-cusparse-cu12        12.3.1.170       pypi_0              pypi\n[conda] nvidia-nccl-cu12            2.21.5           pypi_0              pypi\n[conda] nvidia-nvjitlink-cu12       12.4.127         pypi_0              pypi\n[conda] nvidia-nvtx-cu12            12.4.127         pypi_0              pypi\n[conda] torch                       2.5.1+cu124      pypi_0              pypi",
+  "transformers_version": "4.57.3",
+  "lm_eval_version": "0.4.12",
+  "upper_git_hash": null,
+  "tokenizer_pad_token": [
+    "<|endoftext|>",
+    "151643"
+  ],
+  "tokenizer_eos_token": [
+    "<|im_end|>",
+    "151645"
+  ],
+  "tokenizer_bos_token": [
+    null,
+    "None"
+  ],
+  "eot_token_id": 151645,
+  "max_length": 32768,
+  "task_hashes": {
+    "gsm_plus": "9fc5cf479857a1eb1a143016f6217e0f6f7a6a3251c77540c70b68178e0daa41"
+  },
+  "model_source": "vllm",
+  "model_name": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+  "model_name_sanitized": "results__qwen2.5-1.5B-Instruct#csd__ab_pr_0.5_0.5_8_1e-4",
+  "system_instruction": null,
+  "system_instruction_sha": null,
+  "fewshot_as_multiturn": true,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "chat_template_sha": "cd8e9439f0570856fd70470bf8889ebd8b5d1107207f67a5efb46e342330527f",
+  "total_evaluation_time_seconds": "190.207013798994"
+}

eval_results/vllm/qwen2.5-1.5B-it-amid/results__qwen2.5-1.5B-Instruct#csd__ab_pr_0.5_0.5_8_1e-4/results_2026-05-18T02-03-29.883190.json

@@ -0,0 +1,155 @@
+{
+  "results": {
+    "mmlu_pro_math": {
+      "name": "mmlu_pro_math",
+      "alias": "math",
+      "sample_len": 1351,
+      "exact_match,custom-extract": 0.44411547002220575,
+      "exact_match_stderr,custom-extract": 0.013523009727159878
+    }
+  },
+  "group_subtasks": {},
+  "configs": {
+    "mmlu_pro_math": {
+      "task": "mmlu_pro_math",
+      "task_alias": "math",
+      "dataset_path": "TIGER-Lab/MMLU-Pro",
+      "test_split": "test",
+      "fewshot_split": "validation",
+      "process_docs": "functools.partial(<function process_docs at 0x7b58eab1fd90>, subject='math')",
+      "doc_to_text": "functools.partial(<function format_cot_example at 0x7b58eab1feb0>, including_answer=False)",
+      "doc_to_target": "answer",
+      "unsafe_code": false,
+      "description": "The following are multiple choice questions (with answers) about math. Think step by step and then finish your answer with \"the answer is (X)\" where X is the correct letter choice.\n",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "first_n",
+        "split": "validation",
+        "process_docs": "functools.partial(<function process_docs at 0x7b58eab1fd90>, subject='math')",
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "functools.partial(<function format_cot_example at 0x7b58eab1feb0>, including_answer=False)",
+        "doc_to_choice": null,
+        "doc_to_target": "functools.partial(<function format_cot_target at 0x7b58eab1f910>, including_answer=True)",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 5,
+      "metric_list": [
+        {
+          "metric": "exact_match",
+          "aggregation": "mean",
+          "higher_is_better": true,
+          "ignore_case": true,
+          "ignore_punctuation": true
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": [
+          "Question:"
+        ],
+        "max_gen_toks": 2048,
+        "do_sample": false,
+        "temperature": 0.0
+      },
+      "repeats": 1,
+      "filter_list": [
+        {
+          "name": "custom-extract",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "answer is \\(?([ABCDEFGHIJ])\\)?"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 3.1,
+        "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+        "tensor_parallel_size": 2,
+        "dtype": "float16",
+        "gpu_memory_utilization": 0.75,
+        "trust_remote_code": true,
+        "config_source": "/home/ubuntu/AMiD/.venv/lib/python3.10/site-packages/lm_eval/tasks/mmlu_pro/mmlu_pro_math.yaml"
+      }
+    }
+  },
+  "versions": {
+    "mmlu_pro_math": 3.1
+  },
+  "n-shot": {
+    "mmlu_pro_math": 5
+  },
+  "higher_is_better": {
+    "mmlu_pro_math": {
+      "exact_match": true
+    }
+  },
+  "n-samples": {
+    "mmlu_pro_math": {
+      "original": 1351,
+      "effective": 1351
+    }
+  },
+  "config": {
+    "model": "vllm",
+    "model_args": {
+      "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+      "tensor_parallel_size": 2,
+      "dtype": "float16",
+      "gpu_memory_utilization": 0.75,
+      "trust_remote_code": true
+    },
+    "batch_size": "auto",
+    "batch_sizes": [],
+    "device": "cuda:0",
+    "use_cache": null,
+    "limit": null,
+    "bootstrap_iters": 100000,
+    "gen_kwargs": {},
+    "random_seed": 0,
+    "numpy_seed": 1234,
+    "torch_seed": 1234,
+    "fewshot_seed": 1234
+  },
+  "git_hash": "e49ad15",
+  "date": 1779069738.4071019,
+  "pretty_env_info": "PyTorch version: 2.9.1+cu128\nIs debug build: False\nCUDA used to build PyTorch: 12.8\nROCM used to build PyTorch: N/A\n\nOS: Ubuntu 24.04.4 LTS (x86_64)\nGCC version: (Ubuntu 13.3.0-6ubuntu2~24.04.1) 13.3.0\nClang version: Could not collect\nCMake version: version 3.28.3\nLibc version: glibc-2.39\n\nPython version: 3.10.20 (main, May 10 2026, 19:27:29) [Clang 22.1.3 ] (64-bit runtime)\nPython platform: Linux-6.17.0-1013-aws-x86_64-with-glibc2.39\nIs CUDA available: True\nCUDA runtime version: Could not collect\nCUDA_MODULE_LOADING set to: \nGPU models and configuration: \nGPU 0: NVIDIA H200\nGPU 1: NVIDIA H200\nGPU 2: NVIDIA H200\nGPU 3: NVIDIA H200\nGPU 4: NVIDIA H200\nGPU 5: NVIDIA H200\nGPU 6: NVIDIA H200\nGPU 7: NVIDIA H200\n\nNvidia driver version: 580.126.16\ncuDNN version: Probably one of the following:\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_adv.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_cnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_precompiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_runtime_compiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_graph.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_heuristic.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_ops.so.9\nIs XPU available: False\nHIP runtime version: N/A\nMIOpen runtime version: N/A\nIs XNNPACK available: True\n\nCPU:\nArchitecture:                            x86_64\nCPU op-mode(s):                          32-bit, 64-bit\nAddress sizes:                           48 bits physical, 48 bits virtual\nByte Order:                              Little Endian\nCPU(s):                                  192\nOn-line CPU(s) list:                     0-191\nVendor ID:                               AuthenticAMD\nModel name:                              AMD EPYC 7R13 Processor\nCPU family:                              25\nModel:                                   1\nThread(s) per core:                      2\nCore(s) per socket:                      48\nSocket(s):                               2\nStepping:                                1\nBogoMIPS:                                5299.99\nFlags:                                   fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology nonstop_tsc cpuid extd_apicid aperfmperf tsc_known_freq pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy cr8_legacy abm sse4a misalignsse 3dnowprefetch topoext perfctr_core ssbd ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 invpcid rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 clzero xsaveerptr rdpru wbnoinvd arat npt nrip_save vaes vpclmulqdq rdpid\nHypervisor vendor:                       KVM\nVirtualization type:                     full\nL1d cache:                               3 MiB (96 instances)\nL1i cache:                               3 MiB (96 instances)\nL2 cache:                                48 MiB (96 instances)\nL3 cache:                                384 MiB (12 instances)\nNUMA node(s):                            4\nNUMA node0 CPU(s):                       0-23,96-119\nNUMA node1 CPU(s):                       24-47,120-143\nNUMA node2 CPU(s):                       48-71,144-167\nNUMA node3 CPU(s):                       72-95,168-191\nVulnerability Gather data sampling:      Not affected\nVulnerability Ghostwrite:                Not affected\nVulnerability Indirect target selection: Not affected\nVulnerability Itlb multihit:             Not affected\nVulnerability L1tf:                      Not affected\nVulnerability Mds:                       Not affected\nVulnerability Meltdown:                  Not affected\nVulnerability Mmio stale data:           Not affected\nVulnerability Old microcode:             Not affected\nVulnerability Reg file data sampling:    Not affected\nVulnerability Retbleed:                  Not affected\nVulnerability Spec rstack overflow:      Mitigation; Safe RET\nVulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl\nVulnerability Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization\nVulnerability Spectre v2:                Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected\nVulnerability Srbds:                     Not affected\nVulnerability Tsa:                       Mitigation; Clear CPU buffers\nVulnerability Tsx async abort:           Not affected\nVulnerability Vmscape:                   Not affected\n\nVersions of relevant libraries:\n[pip3] Could not collect\n[conda] numpy                       2.4.5            pypi_0              pypi\n[conda] nvidia-cublas-cu12          12.4.5.8         pypi_0              pypi\n[conda] nvidia-cuda-cupti-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-nvrtc-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-runtime-cu12    12.4.127         pypi_0              pypi\n[conda] nvidia-cudnn-cu12           9.1.0.70         pypi_0              pypi\n[conda] nvidia-cufft-cu12           11.2.1.3         pypi_0              pypi\n[conda] nvidia-curand-cu12          10.3.5.147       pypi_0              pypi\n[conda] nvidia-cusolver-cu12        11.6.1.9         pypi_0              pypi\n[conda] nvidia-cusparse-cu12        12.3.1.170       pypi_0              pypi\n[conda] nvidia-nccl-cu12            2.21.5           pypi_0              pypi\n[conda] nvidia-nvjitlink-cu12       12.4.127         pypi_0              pypi\n[conda] nvidia-nvtx-cu12            12.4.127         pypi_0              pypi\n[conda] torch                       2.5.1+cu124      pypi_0              pypi",
+  "transformers_version": "4.57.3",
+  "lm_eval_version": "0.4.12",
+  "upper_git_hash": null,
+  "tokenizer_pad_token": [
+    "<|endoftext|>",
+    "151643"
+  ],
+  "tokenizer_eos_token": [
+    "<|im_end|>",
+    "151645"
+  ],
+  "tokenizer_bos_token": [
+    null,
+    "None"
+  ],
+  "eot_token_id": 151645,
+  "max_length": 32768,
+  "task_hashes": {
+    "mmlu_pro_math": "2ddae5317c3ae5b3b96f4724d31f9877fdaf30fb4e532897de01a2e284b82493"
+  },
+  "model_source": "vllm",
+  "model_name": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+  "model_name_sanitized": "results__qwen2.5-1.5B-Instruct#csd__ab_pr_0.5_0.5_8_1e-4",
+  "system_instruction": null,
+  "system_instruction_sha": null,
+  "fewshot_as_multiturn": true,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "chat_template_sha": "cd8e9439f0570856fd70470bf8889ebd8b5d1107207f67a5efb46e342330527f",
+  "total_evaluation_time_seconds": "79.42120214103488"
+}

eval_results/vllm/qwen2.5-1.5B-it-amid/results__qwen2.5-1.5B-Instruct#csd__ab_pr_0.5_0.5_8_1e-4/results_2026-05-18T02-07-43.167479.json

@@ -0,0 +1,286 @@
+{
+  "results": {
+    "leaderboard_musr_murder_mysteries": {
+      "name": "leaderboard_musr_murder_mysteries",
+      "alias": "leaderboard_musr_murder_mysteries",
+      "sample_len": 250,
+      "acc_norm,none": 0.504,
+      "acc_norm_stderr,none": 0.031685198551199154
+    },
+    "leaderboard_musr_object_placements": {
+      "name": "leaderboard_musr_object_placements",
+      "alias": "leaderboard_musr_object_placements",
+      "sample_len": 256,
+      "acc_norm,none": 0.2265625,
+      "acc_norm_stderr,none": 0.026214195644894838
+    },
+    "leaderboard_musr_team_allocation": {
+      "name": "leaderboard_musr_team_allocation",
+      "alias": "leaderboard_musr_team_allocation",
+      "sample_len": 250,
+      "acc_norm,none": 0.284,
+      "acc_norm_stderr,none": 0.02857695873043741
+    },
+    "leaderboard_musr": {
+      "alias": "leaderboard_musr",
+      "name": "leaderboard_musr",
+      "sample_len": 756,
+      "acc_norm,none": 0.3373015873015873,
+      "acc_norm_stderr,none": 0.01666990076971401,
+      "sample_count": {
+        "acc_norm,none": 756
+      }
+    }
+  },
+  "groups": {
+    "leaderboard_musr": {
+      "alias": "leaderboard_musr",
+      "name": "leaderboard_musr",
+      "sample_len": 756,
+      "acc_norm,none": 0.3373015873015873,
+      "acc_norm_stderr,none": 0.01666990076971401,
+      "sample_count": {
+        "acc_norm,none": 756
+      }
+    }
+  },
+  "group_subtasks": {
+    "leaderboard_musr": [
+      "leaderboard_musr_murder_mysteries",
+      "leaderboard_musr_object_placements",
+      "leaderboard_musr_team_allocation"
+    ]
+  },
+  "configs": {
+    "leaderboard_musr_murder_mysteries": {
+      "task": "leaderboard_musr_murder_mysteries",
+      "dataset_path": "TAUR-Lab/MuSR",
+      "test_split": "murder_mysteries",
+      "doc_to_text": "def doc_to_text(doc):\n    \"\"\"\n    Convert a doc to text.\n    \"\"\"\n    choices = \"\"\n    for i, choice in enumerate(ast.literal_eval(doc[\"choices\"])):\n        choices += f\"{i + 1} - {choice}\\n\"\n\n    text = DOC_TO_TEXT.format(\n        narrative=doc[\"narrative\"], question=doc[\"question\"], choices=choices\n    )\n\n    return text\n",
+      "doc_to_target": "{{answer_choice}}",
+      "unsafe_code": false,
+      "doc_to_choice": "{{choices}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "<function doc_to_text at 0x7b719ce40040>",
+        "doc_to_choice": "{{choices}}",
+        "doc_to_target": "{{answer_choice}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc_norm",
+          "aggregation": "mean",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+        "tensor_parallel_size": 2,
+        "dtype": "float16",
+        "gpu_memory_utilization": 0.75,
+        "trust_remote_code": true,
+        "config_source": "/home/ubuntu/AMiD/.venv/lib/python3.10/site-packages/lm_eval/tasks/leaderboard/musr/musr_murder_mysteries.yaml"
+      }
+    },
+    "leaderboard_musr_object_placements": {
+      "task": "leaderboard_musr_object_placements",
+      "dataset_path": "TAUR-Lab/MuSR",
+      "test_split": "object_placements",
+      "doc_to_text": "def doc_to_text(doc):\n    \"\"\"\n    Convert a doc to text.\n    \"\"\"\n    choices = \"\"\n    for i, choice in enumerate(ast.literal_eval(doc[\"choices\"])):\n        choices += f\"{i + 1} - {choice}\\n\"\n\n    text = DOC_TO_TEXT.format(\n        narrative=doc[\"narrative\"], question=doc[\"question\"], choices=choices\n    )\n\n    return text\n",
+      "doc_to_target": "{{answer_choice}}",
+      "unsafe_code": false,
+      "doc_to_choice": "{{choices}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "<function doc_to_text at 0x7b719ce40040>",
+        "doc_to_choice": "{{choices}}",
+        "doc_to_target": "{{answer_choice}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc_norm",
+          "aggregation": "mean",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+        "tensor_parallel_size": 2,
+        "dtype": "float16",
+        "gpu_memory_utilization": 0.75,
+        "trust_remote_code": true,
+        "config_source": "/home/ubuntu/AMiD/.venv/lib/python3.10/site-packages/lm_eval/tasks/leaderboard/musr/musr_object_placements.yaml"
+      }
+    },
+    "leaderboard_musr_team_allocation": {
+      "task": "leaderboard_musr_team_allocation",
+      "dataset_path": "TAUR-Lab/MuSR",
+      "test_split": "team_allocation",
+      "doc_to_text": "def doc_to_text(doc):\n    \"\"\"\n    Convert a doc to text.\n    \"\"\"\n    choices = \"\"\n    for i, choice in enumerate(ast.literal_eval(doc[\"choices\"])):\n        choices += f\"{i + 1} - {choice}\\n\"\n\n    text = DOC_TO_TEXT.format(\n        narrative=doc[\"narrative\"], question=doc[\"question\"], choices=choices\n    )\n\n    return text\n",
+      "doc_to_target": "{{answer_choice}}",
+      "unsafe_code": false,
+      "doc_to_choice": "{{choices}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "<function doc_to_text at 0x7b719ce40040>",
+        "doc_to_choice": "{{choices}}",
+        "doc_to_target": "{{answer_choice}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc_norm",
+          "aggregation": "mean",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+        "tensor_parallel_size": 2,
+        "dtype": "float16",
+        "gpu_memory_utilization": 0.75,
+        "trust_remote_code": true,
+        "config_source": "/home/ubuntu/AMiD/.venv/lib/python3.10/site-packages/lm_eval/tasks/leaderboard/musr/musr_team_allocation.yaml"
+      }
+    }
+  },
+  "versions": {
+    "leaderboard_musr": "N/A",
+    "leaderboard_musr_murder_mysteries": 1.0,
+    "leaderboard_musr_object_placements": 1.0,
+    "leaderboard_musr_team_allocation": 1.0
+  },
+  "n-shot": {
+    "leaderboard_musr": 0,
+    "leaderboard_musr_murder_mysteries": 0,
+    "leaderboard_musr_object_placements": 0,
+    "leaderboard_musr_team_allocation": 0
+  },
+  "higher_is_better": {
+    "leaderboard_musr": {
+      "acc_norm": true
+    },
+    "leaderboard_musr_murder_mysteries": {
+      "acc_norm": true
+    },
+    "leaderboard_musr_object_placements": {
+      "acc_norm": true
+    },
+    "leaderboard_musr_team_allocation": {
+      "acc_norm": true
+    }
+  },
+  "n-samples": {
+    "leaderboard_musr_murder_mysteries": {
+      "original": 250,
+      "effective": 250
+    },
+    "leaderboard_musr_object_placements": {
+      "original": 256,
+      "effective": 256
+    },
+    "leaderboard_musr_team_allocation": {
+      "original": 250,
+      "effective": 250
+    }
+  },
+  "config": {
+    "model": "vllm",
+    "model_args": {
+      "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+      "tensor_parallel_size": 2,
+      "dtype": "float16",
+      "gpu_memory_utilization": 0.75,
+      "trust_remote_code": true
+    },
+    "batch_size": "auto",
+    "batch_sizes": [],
+    "device": "cuda:0",
+    "use_cache": null,
+    "limit": null,
+    "bootstrap_iters": 100000,
+    "gen_kwargs": {},
+    "random_seed": 0,
+    "numpy_seed": 1234,
+    "torch_seed": 1234,
+    "fewshot_seed": 1234
+  },
+  "git_hash": "e49ad15",
+  "date": 1779069987.8120446,
+  "pretty_env_info": "PyTorch version: 2.9.1+cu128\nIs debug build: False\nCUDA used to build PyTorch: 12.8\nROCM used to build PyTorch: N/A\n\nOS: Ubuntu 24.04.4 LTS (x86_64)\nGCC version: (Ubuntu 13.3.0-6ubuntu2~24.04.1) 13.3.0\nClang version: Could not collect\nCMake version: version 3.28.3\nLibc version: glibc-2.39\n\nPython version: 3.10.20 (main, May 10 2026, 19:27:29) [Clang 22.1.3 ] (64-bit runtime)\nPython platform: Linux-6.17.0-1013-aws-x86_64-with-glibc2.39\nIs CUDA available: True\nCUDA runtime version: Could not collect\nCUDA_MODULE_LOADING set to: \nGPU models and configuration: \nGPU 0: NVIDIA H200\nGPU 1: NVIDIA H200\nGPU 2: NVIDIA H200\nGPU 3: NVIDIA H200\nGPU 4: NVIDIA H200\nGPU 5: NVIDIA H200\nGPU 6: NVIDIA H200\nGPU 7: NVIDIA H200\n\nNvidia driver version: 580.126.16\ncuDNN version: Probably one of the following:\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_adv.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_cnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_precompiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_runtime_compiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_graph.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_heuristic.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_ops.so.9\nIs XPU available: False\nHIP runtime version: N/A\nMIOpen runtime version: N/A\nIs XNNPACK available: True\n\nCPU:\nArchitecture:                            x86_64\nCPU op-mode(s):                          32-bit, 64-bit\nAddress sizes:                           48 bits physical, 48 bits virtual\nByte Order:                              Little Endian\nCPU(s):                                  192\nOn-line CPU(s) list:                     0-191\nVendor ID:                               AuthenticAMD\nModel name:                              AMD EPYC 7R13 Processor\nCPU family:                              25\nModel:                                   1\nThread(s) per core:                      2\nCore(s) per socket:                      48\nSocket(s):                               2\nStepping:                                1\nBogoMIPS:                                5299.99\nFlags:                                   fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology nonstop_tsc cpuid extd_apicid aperfmperf tsc_known_freq pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy cr8_legacy abm sse4a misalignsse 3dnowprefetch topoext perfctr_core ssbd ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 invpcid rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 clzero xsaveerptr rdpru wbnoinvd arat npt nrip_save vaes vpclmulqdq rdpid\nHypervisor vendor:                       KVM\nVirtualization type:                     full\nL1d cache:                               3 MiB (96 instances)\nL1i cache:                               3 MiB (96 instances)\nL2 cache:                                48 MiB (96 instances)\nL3 cache:                                384 MiB (12 instances)\nNUMA node(s):                            4\nNUMA node0 CPU(s):                       0-23,96-119\nNUMA node1 CPU(s):                       24-47,120-143\nNUMA node2 CPU(s):                       48-71,144-167\nNUMA node3 CPU(s):                       72-95,168-191\nVulnerability Gather data sampling:      Not affected\nVulnerability Ghostwrite:                Not affected\nVulnerability Indirect target selection: Not affected\nVulnerability Itlb multihit:             Not affected\nVulnerability L1tf:                      Not affected\nVulnerability Mds:                       Not affected\nVulnerability Meltdown:                  Not affected\nVulnerability Mmio stale data:           Not affected\nVulnerability Old microcode:             Not affected\nVulnerability Reg file data sampling:    Not affected\nVulnerability Retbleed:                  Not affected\nVulnerability Spec rstack overflow:      Mitigation; Safe RET\nVulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl\nVulnerability Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization\nVulnerability Spectre v2:                Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected\nVulnerability Srbds:                     Not affected\nVulnerability Tsa:                       Mitigation; Clear CPU buffers\nVulnerability Tsx async abort:           Not affected\nVulnerability Vmscape:                   Not affected\n\nVersions of relevant libraries:\n[pip3] Could not collect\n[conda] numpy                       2.4.5            pypi_0              pypi\n[conda] nvidia-cublas-cu12          12.4.5.8         pypi_0              pypi\n[conda] nvidia-cuda-cupti-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-nvrtc-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-runtime-cu12    12.4.127         pypi_0              pypi\n[conda] nvidia-cudnn-cu12           9.1.0.70         pypi_0              pypi\n[conda] nvidia-cufft-cu12           11.2.1.3         pypi_0              pypi\n[conda] nvidia-curand-cu12          10.3.5.147       pypi_0              pypi\n[conda] nvidia-cusolver-cu12        11.6.1.9         pypi_0              pypi\n[conda] nvidia-cusparse-cu12        12.3.1.170       pypi_0              pypi\n[conda] nvidia-nccl-cu12            2.21.5           pypi_0              pypi\n[conda] nvidia-nvjitlink-cu12       12.4.127         pypi_0              pypi\n[conda] nvidia-nvtx-cu12            12.4.127         pypi_0              pypi\n[conda] torch                       2.5.1+cu124      pypi_0              pypi",
+  "transformers_version": "4.57.3",
+  "lm_eval_version": "0.4.12",
+  "upper_git_hash": null,
+  "tokenizer_pad_token": [
+    "<|endoftext|>",
+    "151643"
+  ],
+  "tokenizer_eos_token": [
+    "<|im_end|>",
+    "151645"
+  ],
+  "tokenizer_bos_token": [
+    null,
+    "None"
+  ],
+  "eot_token_id": 151645,
+  "max_length": 32768,
+  "task_hashes": {
+    "leaderboard_musr_murder_mysteries": "10b0f39591b5077ca738884f3a41fe6dee70cf50a346410ea0523f6c3a9d4597",
+    "leaderboard_musr_object_placements": "d9fd900d3ebf1b133577f77ce3e75afc425aad0f4a76ad4d00b2225b01ebfecd",
+    "leaderboard_musr_team_allocation": "2cc8896ce61ba9f272a2e2de62fc7b29c0999fc91b81e79187e0a256f3d3d539"
+  },
+  "model_source": "vllm",
+  "model_name": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+  "model_name_sanitized": "results__qwen2.5-1.5B-Instruct#csd__ab_pr_0.5_0.5_8_1e-4",
+  "system_instruction": null,
+  "system_instruction_sha": null,
+  "fewshot_as_multiturn": true,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "chat_template_sha": "cd8e9439f0570856fd70470bf8889ebd8b5d1107207f67a5efb46e342330527f",
+  "total_evaluation_time_seconds": "83.26320017699618"
+}

eval_results/vllm/qwen2.5-1.5B-it-amid/results__qwen2.5-1.5B-Instruct#csd__ab_pr_0.5_0.5_8_1e-4/results_2026-05-18T02-08-46.469693.json

@@ -0,0 +1,159 @@
+{
+  "results": {
+    "leaderboard_ifeval": {
+      "name": "leaderboard_ifeval",
+      "alias": "leaderboard_ifeval",
+      "sample_len": 541,
+      "prompt_level_strict_acc,none": 0.3456561922365989,
+      "prompt_level_strict_acc_stderr,none": 0.02046577943268255,
+      "inst_level_strict_acc,none": 0.44244604316546765,
+      "inst_level_strict_acc_stderr,none": "N/A",
+      "prompt_level_loose_acc,none": 0.3844731977818854,
+      "prompt_level_loose_acc_stderr,none": 0.02093435763458466,
+      "inst_level_loose_acc,none": 0.4832134292565947,
+      "inst_level_loose_acc_stderr,none": "N/A"
+    }
+  },
+  "group_subtasks": {},
+  "configs": {
+    "leaderboard_ifeval": {
+      "task": "leaderboard_ifeval",
+      "dataset_path": "wis-k/instruction-following-eval",
+      "test_split": "train",
+      "doc_to_text": "prompt",
+      "doc_to_target": 0,
+      "unsafe_code": false,
+      "process_results": "def process_results(doc, results):\n    inp = InputExample(\n        key=doc[\"key\"],\n        instruction_id_list=doc[\"instruction_id_list\"],\n        prompt=doc[\"prompt\"],\n        kwargs=doc[\"kwargs\"],\n    )\n    response = results[0]\n\n    out_strict = test_instruction_following_strict(inp, response)\n    out_loose = test_instruction_following_loose(inp, response)\n\n    return {\n        \"prompt_level_strict_acc\": out_strict.follow_all_instructions,\n        \"inst_level_strict_acc\": out_strict.follow_instruction_list,\n        \"prompt_level_loose_acc\": out_loose.follow_all_instructions,\n        \"inst_level_loose_acc\": out_loose.follow_instruction_list,\n    }\n",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "first_n",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "prompt",
+        "doc_to_choice": null,
+        "doc_to_target": 0,
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "prompt_level_strict_acc",
+          "aggregation": "mean",
+          "higher_is_better": true
+        },
+        {
+          "metric": "inst_level_strict_acc",
+          "aggregation": "def agg_inst_level_acc(items):\n    flat_items = [item for sublist in items for item in sublist]\n    inst_level_acc = sum(flat_items) / len(flat_items)\n    return inst_level_acc\n",
+          "higher_is_better": true
+        },
+        {
+          "metric": "prompt_level_loose_acc",
+          "aggregation": "mean",
+          "higher_is_better": true
+        },
+        {
+          "metric": "inst_level_loose_acc",
+          "aggregation": "def agg_inst_level_acc(items):\n    flat_items = [item for sublist in items for item in sublist]\n    inst_level_acc = sum(flat_items) / len(flat_items)\n    return inst_level_acc\n",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": [],
+        "do_sample": false,
+        "temperature": 0.0,
+        "max_gen_toks": 1280
+      },
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 3.0,
+        "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+        "tensor_parallel_size": 2,
+        "dtype": "float16",
+        "gpu_memory_utilization": 0.75,
+        "trust_remote_code": true,
+        "config_source": "/home/ubuntu/AMiD/.venv/lib/python3.10/site-packages/lm_eval/tasks/leaderboard/ifeval/ifeval.yaml"
+      }
+    }
+  },
+  "versions": {
+    "leaderboard_ifeval": 3.0
+  },
+  "n-shot": {
+    "leaderboard_ifeval": 0
+  },
+  "higher_is_better": {
+    "leaderboard_ifeval": {
+      "prompt_level_strict_acc": true,
+      "inst_level_strict_acc": true,
+      "prompt_level_loose_acc": true,
+      "inst_level_loose_acc": true
+    }
+  },
+  "n-samples": {
+    "leaderboard_ifeval": {
+      "original": 541,
+      "effective": 541
+    }
+  },
+  "config": {
+    "model": "vllm",
+    "model_args": {
+      "pretrained": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+      "tensor_parallel_size": 2,
+      "dtype": "float16",
+      "gpu_memory_utilization": 0.75,
+      "trust_remote_code": true
+    },
+    "batch_size": "auto",
+    "batch_sizes": [],
+    "device": "cuda:0",
+    "use_cache": null,
+    "limit": null,
+    "bootstrap_iters": 100000,
+    "gen_kwargs": {},
+    "random_seed": 0,
+    "numpy_seed": 1234,
+    "torch_seed": 1234,
+    "fewshot_seed": 1234
+  },
+  "git_hash": "e49ad15",
+  "date": 1779070073.9039183,
+  "pretty_env_info": "PyTorch version: 2.9.1+cu128\nIs debug build: False\nCUDA used to build PyTorch: 12.8\nROCM used to build PyTorch: N/A\n\nOS: Ubuntu 24.04.4 LTS (x86_64)\nGCC version: (Ubuntu 13.3.0-6ubuntu2~24.04.1) 13.3.0\nClang version: Could not collect\nCMake version: version 3.28.3\nLibc version: glibc-2.39\n\nPython version: 3.10.20 (main, May 10 2026, 19:27:29) [Clang 22.1.3 ] (64-bit runtime)\nPython platform: Linux-6.17.0-1013-aws-x86_64-with-glibc2.39\nIs CUDA available: True\nCUDA runtime version: Could not collect\nCUDA_MODULE_LOADING set to: \nGPU models and configuration: \nGPU 0: NVIDIA H200\nGPU 1: NVIDIA H200\nGPU 2: NVIDIA H200\nGPU 3: NVIDIA H200\nGPU 4: NVIDIA H200\nGPU 5: NVIDIA H200\nGPU 6: NVIDIA H200\nGPU 7: NVIDIA H200\n\nNvidia driver version: 580.126.16\ncuDNN version: Probably one of the following:\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_adv.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_cnn.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_precompiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_engines_runtime_compiled.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_graph.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_heuristic.so.9\n/opt/pytorch/lib/python3.13/site-packages/nvidia/cu13/lib/libcudnn_ops.so.9\nIs XPU available: False\nHIP runtime version: N/A\nMIOpen runtime version: N/A\nIs XNNPACK available: True\n\nCPU:\nArchitecture:                            x86_64\nCPU op-mode(s):                          32-bit, 64-bit\nAddress sizes:                           48 bits physical, 48 bits virtual\nByte Order:                              Little Endian\nCPU(s):                                  192\nOn-line CPU(s) list:                     0-191\nVendor ID:                               AuthenticAMD\nModel name:                              AMD EPYC 7R13 Processor\nCPU family:                              25\nModel:                                   1\nThread(s) per core:                      2\nCore(s) per socket:                      48\nSocket(s):                               2\nStepping:                                1\nBogoMIPS:                                5299.99\nFlags:                                   fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology nonstop_tsc cpuid extd_apicid aperfmperf tsc_known_freq pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy cr8_legacy abm sse4a misalignsse 3dnowprefetch topoext perfctr_core ssbd ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 invpcid rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 clzero xsaveerptr rdpru wbnoinvd arat npt nrip_save vaes vpclmulqdq rdpid\nHypervisor vendor:                       KVM\nVirtualization type:                     full\nL1d cache:                               3 MiB (96 instances)\nL1i cache:                               3 MiB (96 instances)\nL2 cache:                                48 MiB (96 instances)\nL3 cache:                                384 MiB (12 instances)\nNUMA node(s):                            4\nNUMA node0 CPU(s):                       0-23,96-119\nNUMA node1 CPU(s):                       24-47,120-143\nNUMA node2 CPU(s):                       48-71,144-167\nNUMA node3 CPU(s):                       72-95,168-191\nVulnerability Gather data sampling:      Not affected\nVulnerability Ghostwrite:                Not affected\nVulnerability Indirect target selection: Not affected\nVulnerability Itlb multihit:             Not affected\nVulnerability L1tf:                      Not affected\nVulnerability Mds:                       Not affected\nVulnerability Meltdown:                  Not affected\nVulnerability Mmio stale data:           Not affected\nVulnerability Old microcode:             Not affected\nVulnerability Reg file data sampling:    Not affected\nVulnerability Retbleed:                  Not affected\nVulnerability Spec rstack overflow:      Mitigation; Safe RET\nVulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl\nVulnerability Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization\nVulnerability Spectre v2:                Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected\nVulnerability Srbds:                     Not affected\nVulnerability Tsa:                       Mitigation; Clear CPU buffers\nVulnerability Tsx async abort:           Not affected\nVulnerability Vmscape:                   Not affected\n\nVersions of relevant libraries:\n[pip3] Could not collect\n[conda] numpy                       2.4.5            pypi_0              pypi\n[conda] nvidia-cublas-cu12          12.4.5.8         pypi_0              pypi\n[conda] nvidia-cuda-cupti-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-nvrtc-cu12      12.4.127         pypi_0              pypi\n[conda] nvidia-cuda-runtime-cu12    12.4.127         pypi_0              pypi\n[conda] nvidia-cudnn-cu12           9.1.0.70         pypi_0              pypi\n[conda] nvidia-cufft-cu12           11.2.1.3         pypi_0              pypi\n[conda] nvidia-curand-cu12          10.3.5.147       pypi_0              pypi\n[conda] nvidia-cusolver-cu12        11.6.1.9         pypi_0              pypi\n[conda] nvidia-cusparse-cu12        12.3.1.170       pypi_0              pypi\n[conda] nvidia-nccl-cu12            2.21.5           pypi_0              pypi\n[conda] nvidia-nvjitlink-cu12       12.4.127         pypi_0              pypi\n[conda] nvidia-nvtx-cu12            12.4.127         pypi_0              pypi\n[conda] torch                       2.5.1+cu124      pypi_0              pypi",
+  "transformers_version": "4.57.3",
+  "lm_eval_version": "0.4.12",
+  "upper_git_hash": null,
+  "tokenizer_pad_token": [
+    "<|endoftext|>",
+    "151643"
+  ],
+  "tokenizer_eos_token": [
+    "<|im_end|>",
+    "151645"
+  ],
+  "tokenizer_bos_token": [
+    null,
+    "None"
+  ],
+  "eot_token_id": 151645,
+  "max_length": 32768,
+  "task_hashes": {
+    "leaderboard_ifeval": "279a13ef55eb3ed0f9c83e4e0ce1bab5e125f34e0cef9998d6364d6e53c7a584"
+  },
+  "model_source": "vllm",
+  "model_name": "results/qwen2.5-1.5B-Instruct#csd/ab_pr_0.5_0.5_8_1e-4",
+  "model_name_sanitized": "results__qwen2.5-1.5B-Instruct#csd__ab_pr_0.5_0.5_8_1e-4",
+  "system_instruction": null,
+  "system_instruction_sha": null,
+  "fewshot_as_multiturn": true,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "chat_template_sha": "cd8e9439f0570856fd70470bf8889ebd8b5d1107207f67a5efb46e342330527f",
+  "total_evaluation_time_seconds": "60.6238280350226"
+}