{ "dataset": "VoxLingua107", "split": "dev", "repo_id": "espnet/geolid_vl107only_shared_trainable", "source_hf_revision": "82c40013919a914cb3a77c237df1b0e67ece2f15", "implementation_revision": "301372e4da311d2a3d9574a5a43b3037ae6e67c1", "model": "ESPnetLIDModel", "model_sha256": "d77542f79e39f876336b650a70dbb980240b30494a67d75ddf1d10096a09619e", "config_sha256": "021aa904d3d70ccfec413de20c22714fa8419b2234431a63fd5a72292969cfaf", "manifest_sha256": "6ad3c8fc6c74fab2720fe186525fc3bd20c4ab7b1be35d227c5c7f5456ba6c54", "utterances": 1609, "languages": 33, "metric": "utterance-level accuracy (%)", "correct": 1526, "accuracy_percent": 94.84151646985705, "paper_accuracy_percent": 94.9, "entrypoint": "espnet2.bin.lid_inference", "device": "cuda", "dtype": "float32", "batch_size": 1, "num_workers": 8, "use_preprocessor": false, "fix_duration": false, "input_order": "wav.scp sorted by utterance ID; standard ESPnet streaming iterator", "environment": { "python": "3.12.3", "torch": "2.11.0+cu128", "torchaudio": "2.11.0+cu128", "transformers": "5.14.1", "espnet-s3prl": "2026.5.14", "soundfile": "0.14.0" }, "finished_at": "2026-10-23T01:05:59.999050+00:00", "seconds": 151.15425174683332, "predictions_file": "evaluation_predictions.jsonl", "predictions_sha256": "c8488577ce62c588e7f8d3df48b84bf5f1e2a52df980ca296703f229ea5b8957", "comparison": { "baseline_implementation_revision": "ca777d59eb49519e3827e40cc34997f2123d6682", "baseline_correct": 1526, "matched_predictions": 1609, "changed_predictions": 0, "baseline_predictions_sha256": "c8488577ce62c588e7f8d3df48b84bf5f1e2a52df980ca296703f229ea5b8957" }, "per_language": { "ara": { "total": 100, "correct": 96 }, "aze": { "total": 34, "correct": 33 }, "cmn": { "total": 23, "correct": 23 }, "dan": { "total": 99, "correct": 94 }, "deu": { "total": 82, "correct": 78 }, "ell": { "total": 5, "correct": 4 }, "eng": { "total": 80, "correct": 75 }, "est": { "total": 85, "correct": 81 }, "fas": { "total": 100, "correct": 92 }, "fin": { "total": 93, "correct": 91 }, "fra": { "total": 100, "correct": 96 }, "hrv": { "total": 4, "correct": 2 }, "hun": { "total": 4, "correct": 4 }, "hye": { "total": 100, "correct": 97 }, "isl": { "total": 26, "correct": 25 }, "ita": { "total": 49, "correct": 48 }, "jpn": { "total": 42, "correct": 41 }, "lav": { "total": 95, "correct": 95 }, "lit": { "total": 13, "correct": 11 }, "mkd": { "total": 4, "correct": 4 }, "nld": { "total": 100, "correct": 97 }, "nno": { "total": 3, "correct": 2 }, "nor": { "total": 24, "correct": 19 }, "pol": { "total": 2, "correct": 2 }, "por": { "total": 1, "correct": 1 }, "rus": { "total": 29, "correct": 29 }, "slv": { "total": 9, "correct": 9 }, "spa": { "total": 55, "correct": 54 }, "srp": { "total": 7, "correct": 4 }, "swe": { "total": 100, "correct": 96 }, "tur": { "total": 72, "correct": 68 }, "ukr": { "total": 17, "correct": 17 }, "urd": { "total": 52, "correct": 38 } }, "batch_dependence": { "s3prl_frame_alignment": "S3PRL reports ceil(samples/320) frames and repeats the final frame only at the batch boundary. Shorter items can instead expose padded-query frames from Wav2Vec2.", "ecapa_tdnn": "Temporal convolutions and unmasked SE global means depend on the padded sequence length before the masked final pooling.", "mvn_and_final_pooling_mask_padding": true, "reported_measurement_batch_size": 1 }, "resolved_config_sha256": "d53f6021d2c5ce3ddae8d032a8b307c84631bf57875d6378d454fd63ff60d121", "languages_sha256": "1dbd458b4cb9148d9ab5cc83abc3fcf9c4835f02c7f2afb94589d2c7190194bc", "log_file": "evaluation.log", "log_sha256": "16813ed34bc75245b60c5f743b8ca52e8583d90c02813c9df773023f19e87e69", "prediction_differences_file": "evaluation_prediction_differences.json", "previous_evaluation_file": "evaluation_previous.json", "source_config_url": "https://huggingface.co/espnet/geolid_vl107only_shared_trainable/blob/82c40013919a914cb3a77c237df1b0e67ece2f15/config.yaml", "loader": "Unmodified standard ESPnet CLI loader; strict loading was not forced" }