{ "humaneval": { "pass@1": [ [ 0, 1.0 ], [ 1, 0.4999999999999999 ], [ 2, 1.0 ], [ 3, 1.0 ], [ 4, 1.0 ], [ 5, 1.0 ], [ 6, 1.0 ], [ 7, 1.0 ], [ 8, 1.0 ], [ 9, 0.0 ], [ 10, 1.0 ], [ 11, 1.0 ], [ 12, 1.0 ], [ 13, 1.0 ], [ 14, 1.0 ], [ 15, 1.0 ], [ 16, 0.95 ], [ 17, 1.0 ], [ 18, 1.0 ], [ 19, 0.25 ], [ 20, 1.0 ], [ 21, 1.0 ], [ 22, 1.0 ], [ 23, 1.0 ], [ 24, 1.0 ], [ 25, 1.0 ], [ 26, 0.0 ], [ 27, 1.0 ], [ 28, 1.0 ], [ 29, 1.0 ], [ 30, 1.0 ], [ 31, 1.0 ], [ 32, 0.050000000000000044 ], [ 33, 1.0 ], [ 34, 1.0 ], [ 35, 1.0 ], [ 36, 1.0 ], [ 37, 1.0 ], [ 38, 1.0 ], [ 39, 1.0 ], [ 40, 1.0 ], [ 41, 1.0 ], [ 42, 1.0 ], [ 43, 0.09999999999999998 ], [ 44, 1.0 ], [ 45, 1.0 ], [ 46, 0.65 ], [ 47, 1.0 ], [ 48, 1.0 ], [ 49, 1.0 ], [ 50, 1.0 ], [ 51, 1.0 ], [ 52, 1.0 ], [ 53, 1.0 ], [ 54, 0.0 ], [ 55, 1.0 ], [ 56, 1.0 ], [ 57, 1.0 ], [ 58, 1.0 ], [ 59, 0.95 ], [ 60, 1.0 ], [ 61, 1.0 ], [ 62, 0.0 ], [ 63, 0.95 ], [ 64, 0.20000000000000007 ], [ 65, 0.0 ], [ 66, 1.0 ], [ 67, 0.09999999999999998 ], [ 68, 0.7 ], [ 69, 1.0 ], [ 70, 1.0 ], [ 71, 1.0 ], [ 72, 1.0 ], [ 73, 1.0 ], [ 74, 0.7999999999999999 ], [ 75, 0.30000000000000004 ], [ 76, 0.15000000000000002 ], [ 77, 0.20000000000000007 ], [ 78, 0.85 ], [ 79, 1.0 ], [ 80, 1.0 ], [ 81, 0.0 ], [ 82, 1.0 ], [ 83, 0.0 ], [ 84, 0.0 ], [ 85, 1.0 ], [ 86, 0.95 ], [ 87, 1.0 ], [ 88, 0.7999999999999999 ], [ 89, 0.7 ], [ 90, 0.15000000000000002 ], [ 91, 0.0 ], [ 92, 0.75 ], [ 93, 0.0 ], [ 94, 1.0 ], [ 95, 0.0 ], [ 96, 1.0 ], [ 97, 1.0 ], [ 98, 1.0 ], [ 99, 0.4 ], [ 100, 1.0 ], [ 101, 1.0 ], [ 102, 1.0 ], [ 103, 0.8999999999999999 ], [ 104, 1.0 ], [ 105, 0.4 ], [ 106, 1.0 ], [ 107, 0.7999999999999999 ], [ 108, 0.0 ], [ 109, 0.09999999999999998 ], [ 110, 1.0 ], [ 111, 0.09999999999999998 ], [ 112, 1.0 ], [ 113, 0.44999999999999984 ], [ 114, 0.95 ], [ 115, 0.0 ], [ 116, 1.0 ], [ 117, 1.0 ], [ 118, 1.0 ], [ 119, 0.25 ], [ 120, 0.0 ], [ 121, 1.0 ], [ 122, 1.0 ], [ 123, 0.8999999999999999 ], [ 124, 0.44999999999999984 ], [ 125, 0.8999999999999999 ], [ 126, 0.0 ], [ 127, 0.0 ], [ 128, 0.15000000000000002 ], [ 129, 0.0 ], [ 130, 0.0 ], [ 131, 0.0 ], [ 132, 0.0 ], [ 133, 1.0 ], [ 134, 0.09999999999999998 ], [ 135, 0.5499999999999998 ], [ 136, 1.0 ], [ 137, 0.0 ], [ 138, 0.0 ], [ 139, 1.0 ], [ 140, 0.050000000000000044 ], [ 141, 0.3500000000000001 ], [ 142, 1.0 ], [ 143, 1.0 ], [ 144, 0.15000000000000002 ], [ 145, 0.0 ], [ 146, 1.0 ], [ 147, 0.95 ], [ 148, 1.0 ], [ 149, 0.7 ], [ 150, 1.0 ], [ 151, 0.0 ], [ 152, 1.0 ], [ 153, 1.0 ], [ 154, 0.8999999999999999 ], [ 155, 0.85 ], [ 156, 0.7 ], [ 157, 1.0 ], [ 158, 1.0 ], [ 159, 1.0 ], [ 160, 0.30000000000000004 ], [ 161, 1.0 ], [ 162, 1.0 ], [ 163, 0.0 ] ], "pass@10": [ [ 0, 1.0 ], [ 1, 0.9999945874558878 ], [ 2, 1.0 ], [ 3, 1.0 ], [ 4, 1.0 ], [ 5, 1.0 ], [ 6, 1.0 ], [ 7, 1.0 ], [ 8, 1.0 ], [ 9, 0.0 ], [ 10, 1.0 ], [ 11, 1.0 ], [ 12, 1.0 ], [ 13, 1.0 ], [ 14, 1.0 ], [ 15, 1.0 ], [ 16, 1.0 ], [ 17, 1.0 ], [ 18, 1.0 ], [ 19, 0.9837461300309598 ], [ 20, 1.0 ], [ 21, 1.0 ], [ 22, 1.0 ], [ 23, 1.0 ], [ 24, 1.0 ], [ 25, 1.0 ], [ 26, 0.0 ], [ 27, 1.0 ], [ 28, 1.0 ], [ 29, 1.0 ], [ 30, 1.0 ], [ 31, 1.0 ], [ 32, 0.5 ], [ 33, 1.0 ], [ 34, 1.0 ], [ 35, 1.0 ], [ 36, 1.0 ], [ 37, 1.0 ], [ 38, 1.0 ], [ 39, 1.0 ], [ 40, 1.0 ], [ 41, 1.0 ], [ 42, 1.0 ], [ 43, 0.763157894736842 ], [ 44, 1.0 ], [ 45, 1.0 ], [ 46, 1.0 ], [ 47, 1.0 ], [ 48, 1.0 ], [ 49, 1.0 ], [ 50, 1.0 ], [ 51, 1.0 ], [ 52, 1.0 ], [ 53, 1.0 ], [ 54, 0.0 ], [ 55, 1.0 ], [ 56, 1.0 ], [ 57, 1.0 ], [ 58, 1.0 ], [ 59, 1.0 ], [ 60, 1.0 ], [ 61, 1.0 ], [ 62, 0.0 ], [ 63, 1.0 ], [ 64, 0.956656346749226 ], [ 65, 0.0 ], [ 66, 1.0 ], [ 67, 0.763157894736842 ], [ 68, 1.0 ], [ 69, 1.0 ], [ 70, 1.0 ], [ 71, 1.0 ], [ 72, 1.0 ], [ 73, 1.0 ], [ 74, 1.0 ], [ 75, 0.9945820433436533 ], [ 76, 0.8947368421052632 ], [ 77, 0.956656346749226 ], [ 78, 1.0 ], [ 79, 1.0 ], [ 80, 1.0 ], [ 81, 0.0 ], [ 82, 1.0 ], [ 83, 0.0 ], [ 84, 0.0 ], [ 85, 1.0 ], [ 86, 1.0 ], [ 87, 1.0 ], [ 88, 1.0 ], [ 89, 1.0 ], [ 90, 0.8947368421052632 ], [ 91, 0.0 ], [ 92, 1.0 ], [ 93, 0.0 ], [ 94, 1.0 ], [ 95, 0.0 ], [ 96, 1.0 ], [ 97, 1.0 ], [ 98, 1.0 ], [ 99, 0.9996427720885925 ], [ 100, 1.0 ], [ 101, 1.0 ], [ 102, 1.0 ], [ 103, 1.0 ], [ 104, 1.0 ], [ 105, 0.9996427720885925 ], [ 106, 1.0 ], [ 107, 1.0 ], [ 108, 0.0 ], [ 109, 0.763157894736842 ], [ 110, 1.0 ], [ 111, 0.763157894736842 ], [ 112, 1.0 ], [ 113, 0.9999404620147654 ], [ 114, 1.0 ], [ 115, 0.0 ], [ 116, 1.0 ], [ 117, 1.0 ], [ 118, 1.0 ], [ 119, 0.9837461300309598 ], [ 120, 0.0 ], [ 121, 1.0 ], [ 122, 1.0 ], [ 123, 1.0 ], [ 124, 0.9999404620147654 ], [ 125, 1.0 ], [ 126, 0.0 ], [ 127, 0.0 ], [ 128, 0.8947368421052632 ], [ 129, 0.0 ], [ 130, 0.0 ], [ 131, 0.0 ], [ 132, 0.0 ], [ 133, 1.0 ], [ 134, 0.763157894736842 ], [ 135, 1.0 ], [ 136, 1.0 ], [ 137, 0.0 ], [ 138, 0.0 ], [ 139, 1.0 ], [ 140, 0.5 ], [ 141, 0.9984520123839009 ], [ 142, 1.0 ], [ 143, 1.0 ], [ 144, 0.8947368421052632 ], [ 145, 0.0 ], [ 146, 1.0 ], [ 147, 1.0 ], [ 148, 1.0 ], [ 149, 1.0 ], [ 150, 1.0 ], [ 151, 0.0 ], [ 152, 1.0 ], [ 153, 1.0 ], [ 154, 1.0 ], [ 155, 1.0 ], [ 156, 1.0 ], [ 157, 1.0 ], [ 158, 1.0 ], [ 159, 1.0 ], [ 160, 0.9945820433436533 ], [ 161, 1.0 ], [ 162, 1.0 ], [ 163, 0.0 ] ] }, "config": { "prefix": "", "do_sample": true, "temperature": 0.2, "top_k": 0, "top_p": 0.95, "n_samples": 20, "eos": "<|endoftext|>", "seed": 0, "model": "deepseek-coder-6.7b-instruct", "modeltype": "causal", "peft_model": null, "revision": null, "use_auth_token": false, "trust_remote_code": false, "tasks": "humaneval", "instruction_tokens": null, "batch_size": 1, "max_length_generation": 512, "precision": "fp32", "load_in_8bit": false, "load_in_4bit": false, "left_padding": false, "limit": null, "limit_start": 0, "save_every_k_tasks": -1, "postprocess": true, "allow_code_execution": true, "generation_only": false, "load_generations_path": "generations_humaneval_deepseek-coder-6.7b-instruct.json", "load_data_path": null, "metric_output_path": "evaluation_results.json", "save_generations": false, "load_generations_intermediate_paths": null, "save_generations_path": "generations.json", "save_references": false, "save_references_path": "references.json", "prompt": "prompt", "max_memory_per_gpu": null, "check_references": false } }