我的主页My Profile
加载中…Loading…
请先登录Please log in
登录后才能查看你的模型与提交记录Log in to view your models and submissions.
加载中…Loading…
登录后才能查看你的模型与提交记录Log in to view your models and submissions.
填写模型基本信息,GitHub 与 Hugging Face 链接均为选填。提交后仅为一条申请,会以「待确认」状态出现在「我的模型」中,需管理员后台确认后方可用于提测。 Fill in the model details. The GitHub / Hugging Face links are both optional. Submitting only files a request: it appears under “My Models” as Pending Confirmation and becomes usable for testing once an admin confirms it.
为已登记的模型发起评测,分两种模式:
① 计分多任务——覆盖全部 32 个任务、计入排行榜;
② 不计分单任务调试——只测 1 个任务、不计入排行榜,仅供了解模型大概能力。
可预约的时段以下方排期表为准(灰色为当前类型不可选)。
Schedule an evaluation for a registered model. Two modes:
① Scored multi-task — covers all 32 tasks and counts toward the leaderboard.
② Unscored single-task debug — one task only, not counted toward the leaderboard (just a rough capability check).
Bookable slots are shown in the schedule below (greyed cells aren't available for the selected type).