{"model_id": "nvidia/nemotron-3.5-asr-streaming-0.6b", "name": "nemotron-3.5-asr-streaming-0.6b", "owner": "nvidia", "source_url": "https://huggingface.co/nvidia/nemotron-3.5-asr-streaming-0.6b", "task": "automatic-speech-recognition", "languages": ["en", "es", "de", "fr", "it", "ar", "ja", "ko", "pt", "ru", "hi", "zh", "vi", "he", "nl", "cs", "da", "pl", "no", "sv", "th", "tr", "bg", "el", "et", "fi", "hr", "hu", "lt", "lv"], "tags": ["nemo", "safetensors", "gguf", "nemotron3_5_asr", "feature-extraction", "transformers", "speech-recognition", "cache-aware ASR", "automatic-speech-recognition", "streaming-asr", "multilingual", "speech", "audio", "FastConformer", "RNNT", "Parakeet", "ASR", "pytorch", "NeMo", "en", "es", "de", "fr", "it", "ar", "ja", "ko", "pt", "ru", "hi", "zh", "vi", "he", "nl", "cs", "da", "pl", "no", "sv", "th", "tr", "bg", "el", "et", "fi", "hr", "hu", "lt", "lv", "ro", "sk", "uk", "mt", "sl", "dataset:nvidia/Granary", "dataset:multilingual_librispeech", "dataset:fleurs", "dataset:mozilla-foundation/common_voice_8_0", "dataset:voxpopuli", "dataset:europarl", "arxiv:2312.17279", "arxiv:2305.05084", "license:other", "model-index", "eval-results", "deploy:sagemaker", "region:us"], "license_id": "other", "license_status": "a_verifier", "license_label": "Autre licence déclarée", "commercial_use": "a_verifier", "gated": false, "private": false, "downloads": 1289845, "likes": 1177, "updated_at": "2026-09-10T16:49:07.000Z", "description": "", "files": [{"name": ".gitattributes", "size": 1647, "format": "autre", "precision": "inconnue"}, {"name": "README.md", "size": 54239, "format": "autre", "precision": "inconnue"}, {"name": "arch_slide10.png", "size": 87843, "format": "autre", "precision": "inconnue"}, {"name": "avg_wer_summary.png", "size": 65515, "format": "autre", "precision": "inconnue"}, {"name": "bias.md", "size": 2104, "format": "autre", "precision": "inconnue"}, {"name": "config.json", "size": 1376, "format": "autre", "precision": "inconnue"}, {"name": "explainability.md", "size": 2388, "format": "autre", "precision": "inconnue"}, {"name": "fleurs_langid_vs_auto.png", "size": 83849, "format": "autre", "precision": "inconnue"}, {"name": "fleurs_wer_vs_chunk_size.png", "size": 92234, "format": "autre", "precision": "inconnue"}, {"name": "generation_config.json", "size": 193, "format": "autre", "precision": "inconnue"}, {"name": "latency_vs_parallel.png", "size": 139167, "format": "autre", "precision": "inconnue"}, {"name": "model.safetensors", "size": 2552062944, "format": "safetensors", "precision": "inconnue"}, {"name": "model_architecture.png", "size": 150586, "format": "autre", "precision": "inconnue"}, {"name": "model_overview.png", "size": 114032, "format": "autre", "precision": "inconnue"}, {"name": "nemotron-3.5-asr-streaming-0.6b.nemo", "size": 2368284501, "format": "autre", "precision": "inconnue"}, {"name": "nemotron-3.5-asr-streaming-0.6b.q8_0.gguf", "size": 742090464, "format": "gguf", "precision": "8-bit"}, {"name": "privacy.md", "size": 2097, "format": "autre", "precision": "inconnue"}, {"name": "processor_config.json", "size": 2519, "format": "autre", "precision": "inconnue"}, {"name": "safety.md", "size": 737, "format": "autre", "precision": "inconnue"}, {"name": "throughput_vs_chunk.png", "size": 69454, "format": "autre", "precision": "inconnue"}, {"name": "tokenizer.json", "size": 752051, "format": "autre", "precision": "inconnue"}, {"name": "tokenizer_config.json", "size": 881, "format": "autre", "precision": "inconnue"}], "parameter_count": 637997088, "library_name": "nemo", "pipeline_tag": "automatic-speech-recognition", "datasets": ["nvidia/Granary", "multilingual_librispeech", "fleurs", "mozilla-foundation/common_voice_8_0", "voxpopuli", "europarl"], "metrics": [{"name": "WER (1.12s frame size, LangID)", "value": 7.91, "task": "Automatic Speech Recognition", "dataset": "FLEURS (English)"}, {"name": "WER (1.12s frame size, LangID)", "value": 4.11, "task": "Automatic Speech Recognition", "dataset": "FLEURS (Spanish)"}, {"name": "WER (1.12s frame size, LangID)", "value": 9.03, "task": "Automatic Speech Recognition", "dataset": "FLEURS (French)"}, {"name": "WER (1.12s frame size, LangID)", "value": 4.25, "task": "Automatic Speech Recognition", "dataset": "FLEURS (Italian)"}, {"name": "WER (1.12s frame size, LangID)", "value": 5.48, "task": "Automatic Speech Recognition", "dataset": "FLEURS (Portuguese)"}, {"name": "WER (1.12s frame size, LangID)", "value": 8.31, "task": "Automatic Speech Recognition", "dataset": "FLEURS (German)"}, {"name": "WER (1.12s frame size, LangID)", "value": 6.81, "task": "Automatic Speech Recognition", "dataset": "FLEURS (Hindi)"}, {"name": "WER (1.12s frame size, LangID)", "value": 7.12, "task": "Automatic Speech Recognition", "dataset": "FLEURS (Korean)"}], "github_url": "", "github_metadata": {}, "paper_url": "https://arxiv.org/abs/2312.17279", "paper_doi": "", "paper_metadata": {}, "card_complete": true, "safetensors": true, "provenance": [{"source": "Hugging Face Hub API", "url": "https://huggingface.co/nvidia/nemotron-3.5-asr-streaming-0.6b", "retrieved_at": "2026-10-08T10:20:52.151415+00:00"}], "retrieved_at": "2026-10-08T10:20:52.151415+00:00", "warnings": ["L’usage commercial doit être confirmé dans le texte intégral de la licence du modèle."], "rank_score": 0.0, "rank_reasons": [], "compatibility": {"state": "compatible", "label": "Compatible selon l’estimation", "summary": "La mémoire indiquée dépasse l’estimation prudente.", "factors": ["Mémoire estimée : 3.1 Go, marge d’exécution de 30% incluse.", "Espace disque minimal indicatif : 2.4 Go.", "Une exécution CPU peut être possible mais sa vitesse ne peut pas être déduite de ces métadonnées."], "estimate": {"known": true, "precision": "float32", "source": "taille réelle déclarée des fichiers de poids", "base_bytes": 2552062944, "estimated_bytes": 3317681827, "overhead": 0.3}}, "quantization": {"available": true, "variants": [{"label": "Q8_0", "filename": "nemotron-3.5-asr-streaming-0.6b.q8_0.gguf", "size": 742090464, "size_human": "708 Mo", "memory_bytes": 964717603, "memory_human": "920 Mo", "note": "Pratiquement sans perte perceptible.", "rank": 30, "memory_gb": 0.9, "state": "tient", "verdict": "Tient dans la mémoire déclarée"}], "recommended": {"label": "Q8_0", "filename": "nemotron-3.5-asr-streaming-0.6b.q8_0.gguf", "size": 742090464, "size_human": "708 Mo", "memory_bytes": 964717603, "memory_human": "920 Mo", "note": "Pratiquement sans perte perceptible.", "rank": 30, "memory_gb": 0.9, "state": "tient", "verdict": "Tient dans la mémoire déclarée"}, "lightest": {"label": "Q8_0", "filename": "nemotron-3.5-asr-streaming-0.6b.q8_0.gguf", "size": 742090464, "size_human": "708 Mo", "memory_bytes": 964717603, "memory_human": "920 Mo", "note": "Pratiquement sans perte perceptible.", "rank": 30, "memory_gb": 0.9, "state": "tient", "verdict": "Tient dans la mémoire déclarée"}, "memory_gb": 8.0, "disk_gb": 20.0, "fitting_count": 1, "overhead": 0.3, "note": "Tailles issues des fichiers déclarés par le producteur, majorées de 30% pour l’exécution. Une variante plus compressée consomme moins de mémoire et dégrade les réponses : cette dégradation n’est pas mesurée ici."}, "editorial_quality_score": null, "editorial_quality_blockers": []}