2026-07-18:14:17:11 WARNING [config.evaluate_config:281] --limit SHOULD ONLY BE USED FOR TESTING. REAL METRICS SHOULD NOT BE COMPUTED USING LIMIT. 2026-07-18:14:17:15 INFO [_cli.run:376] Selected Tasks: ['niah_single_1', 'niah_single_3', 'niah_multikey_2', 'niah_multiquery', 'ruler_vt', 'ruler_fwe', 'ruler_qa_hotpot'] 2026-07-18:14:17:16 INFO [evaluator:211] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2026-07-18:14:17:16 INFO [evaluator:236] Initializing hf model, with arguments: {'pretrained': '/workspace', 'trust_remote_code': True, 'dtype': 'bfloat16', 'max_length': 16384, 'attn_implementation': 'sdpa'} 2026-07-18:14:17:20 INFO [models.huggingface:161] Using device 'cuda:0' 2026-07-18:14:17:20 INFO [models.huggingface:423] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:0'} 2026-07-18:14:17:29 WARNING [api.task:856] niah_single_1: Custom kwargs can be passed to `--metadata` in console (as json string) or to the TaskManager. For example --metadata='{"max_seq_lengths":[4096, 8192]}'. For details see task Readme. 2026-07-18:14:17:29 INFO [tasks.ruler.common_utils:26] Using tokenizer /workspace for synthetic tasks. Generating synthetic samples: repeat | 8192: 0%| | 0/500 [00:00