sie_id: convaiinnovations/laya-multilingual description: "Multilingual typed-decision classifier (Laya), uncalibrated: answers noul, choice and score questions, one row per question. 1,024-token rows." release_date: "2026-09-23" hf_id: convaiinnovations/laya-multilingual hf_revision: 82d57fc4f2d1be3d2caac494045f2ec51d0842f3 # This checkpoint ships no calibration temperatures (rl_agent_config.json has # temperature [1, 1, 1] and no per-option buckets), so its probabilities and # confidences are the raw softmax of the decision head: uncalibrated. inputs: text: true image: false audio: false video: false tasks: encode: null score: null extract: {} max_sequence_length: 1024 profiles: default: max_batch_tokens: 16384 compute_precision: bfloat16 adapter_path: sie_server.adapters.laya.adapter:LayaAdapter adapter_options: loadtime: {} runtime: {} # One state with a few questions is a small forward that kernel launches, # not the GPU, bound. This profile replays its encoder as a CUDA graph # (request it as `MODEL:graphs`). It is not the default because a graph # moves probabilities by more than the 0.02 the server README allows a # default speed-up, though by less than batching already does: see # "Laya CUDA graphs" there. graphs: extends: default adapter_options: loadtime: cuda_graphs: bucketed runtime: {}