{ "sweep_A_max_model_len_32768_n16": [ { "bucket": "2k", "target_tokens": 2048, "n": 16, "acc_len": 5.276287886050829, "acc_len_median": 5.354166666666666, "proposal_acceptance": 0.6108982694358327, "n_acc_unmatched": 0, "retrieval_accuracy": 1.0, "mean_prompt_tokens": 2109, "mean_completion_tokens": 80, "truncated": 0, "elapsed_s": 13.5, "sampling": { "temperature": 1.0, "top_p": 0.95 }, "max_tokens": 1024, "concurrency": 1 }, { "bucket": "4k", "target_tokens": 4096, "n": 16, "acc_len": 5.680358749706176, "acc_len_median": 5.464285714285714, "proposal_acceptance": 0.6686226785294538, "n_acc_unmatched": 0, "retrieval_accuracy": 1.0, "mean_prompt_tokens": 4143, "mean_completion_tokens": 76, "truncated": 0, "elapsed_s": 16.2, "sampling": { "temperature": 1.0, "top_p": 0.95 }, "max_tokens": 1024, "concurrency": 1 }, { "bucket": "8k", "target_tokens": 8192, "n": 16, "acc_len": 5.214891198896133, "acc_len_median": 5.323809523809524, "proposal_acceptance": 0.602127314128019, "n_acc_unmatched": 0, "retrieval_accuracy": 1.0, "mean_prompt_tokens": 8203, "mean_completion_tokens": 82, "truncated": 0, "elapsed_s": 30.6, "sampling": { "temperature": 1.0, "top_p": 0.95 }, "max_tokens": 1024, "concurrency": 1 }, { "bucket": "16k", "target_tokens": 16384, "n": 16, "acc_len": 4.877537646373459, "acc_len_median": 4.839583333333334, "proposal_acceptance": 0.5539339494819228, "n_acc_unmatched": 0, "retrieval_accuracy": 1.0, "mean_prompt_tokens": 16336, "mean_completion_tokens": 84, "truncated": 0, "elapsed_s": 40.6, "sampling": { "temperature": 1.0, "top_p": 0.95 }, "max_tokens": 1024, "concurrency": 1 }, { "bucket": "28k", "target_tokens": 28672, "n": 16, "acc_len": 4.809826773995698, "acc_len_median": 4.933035714285714, "proposal_acceptance": 0.5442609677136712, "n_acc_unmatched": 0, "retrieval_accuracy": 1.0, "mean_prompt_tokens": 28535, "mean_completion_tokens": 81, "truncated": 0, "elapsed_s": 72.6, "sampling": { "temperature": 1.0, "top_p": 0.95 }, "max_tokens": 1024, "concurrency": 1 } ], "sweep_B_max_model_len_139264_n8": [ { "bucket": "8k", "target_tokens": 8192, "n": 8, "acc_len": 5.6663430867378235, "acc_len_median": 5.746153846153846, "proposal_acceptance": 0.6666204409625462, "n_acc_unmatched": 0, "retrieval_accuracy": 1.0, "mean_prompt_tokens": 8203, "mean_completion_tokens": 79, "truncated": 0, "elapsed_s": 16.1, "sampling": { "temperature": 1.0, "top_p": 0.95 }, "max_tokens": 1024, "concurrency": 1 }, { "bucket": "32k", "target_tokens": 32768, "n": 8, "acc_len": 4.972686982401656, "acc_len_median": 5.0, "proposal_acceptance": 0.5675267117716652, "n_acc_unmatched": 0, "retrieval_accuracy": 1.0, "mean_prompt_tokens": 32596, "mean_completion_tokens": 90, "truncated": 0, "elapsed_s": 42.0, "sampling": { "temperature": 1.0, "top_p": 0.95 }, "max_tokens": 1024, "concurrency": 1 }, { "bucket": "64k", "target_tokens": 65536, "n": 8, "acc_len": 4.805144610291669, "acc_len_median": 4.719047619047618, "proposal_acceptance": 0.5435920871845241, "n_acc_unmatched": 0, "retrieval_accuracy": 1.0, "mean_prompt_tokens": 65113, "mean_completion_tokens": 77, "truncated": 0, "elapsed_s": 81.8, "sampling": { "temperature": 1.0, "top_p": 0.95 }, "max_tokens": 1024, "concurrency": 1 }, { "bucket": "128k", "target_tokens": 131072, "n": 8, "acc_len": 4.854379881893468, "acc_len_median": 4.931818181818182, "proposal_acceptance": 0.5506256974133527, "n_acc_unmatched": 0, "retrieval_accuracy": 1.0, "mean_prompt_tokens": 130151, "mean_completion_tokens": 85, "truncated": 0, "elapsed_s": 201.9, "sampling": { "temperature": 1.0, "top_p": 0.95 }, "max_tokens": 1024, "concurrency": 1 } ] }