Upload model from models/nemotron-speech-streaming-en-0.6b-mlx-4bit

README.md

@@ -1,77 +1,82 @@
 ---
-license: cc-by-4.0
+license: other
 track_downloads: true
 language:
 - en
-- es
-- fr
-- de
-- bg
-- hr
-- cs
-- da
-- nl
-- et
-- fi
-- el
-- hu
-- it
-- lv
-- lt
-- mt
-- pl
-- pt
-- ro
-- sk
-- sl
-- sv
-- ru
-- uk
-
 pipeline_tag: automatic-speech-recognition
 library_name: mlx
 datasets:
 - nvidia/Granary
-- nemo/asr-set-3.0
+- YTC
+- Yodas2
+- LibriLight
+- librispeech_asr
+- fisher_corpus
+- Switchboard-1
+- WSJ-0
+- WSJ-1
+- National-Singapore-Corpus-Part-1
+- National-Singapore-Corpus-Part-6
+- vctk
+- voxpopuli
+- europarl
+- multilingual_librispeech
+- fleurs
+- mozilla-foundation/common_voice_8_0
+- MLCommons/peoples_speech
+- google/speech_commands
 tags:
 - quantized
+- speech-recognition
+- cache-aware ASR
 - automatic-speech-recognition
+- streaming-asr
 - speech
 - audio
-- Transducer
-- TDT
 - FastConformer
-- Conformer
-- pytorch
+- RNNT
 - Parakeet
-base_model: nvidia/parakeet-tdt-0.6b-v3
+- ASR
+- pytorch
+- NeMo
+- mlx
+base_model: nvidia/nemotron-speech-streaming-en-0.6b
+base_model_relation: quantized
 ---
 
-# **animaslabs/parakeet-tdt-0.6b-v3-mlx**
+# **animaslabs/nemotron-speech-streaming-en-0.6b-mlx-4bit**
 
-This model was converted to MLX format from [nvidia/parakeet-tdt-0.6b-v3](https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3) using the scripts in this [github repo](https://github.com/animaslabs/mlx-models). Please refer to [original model card](https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3) for more details on the model.
+This model was converted to MLX format, 4-bit quantized from [nvidia/nemotron-speech-streaming-en-0.6b](https://huggingface.co/nvidia/nemotron-speech-streaming-en-0.6b) using the scripts in this [github repo](https://github.com/animaslabs/mlx-models). Please refer to [original model card](https://huggingface.co/nvidia/nemotron-speech-streaming-en-0.6b) for more details on the model.
 
-## Use with mlx
+## Usage
 
-NOTE: parakeet-mlx and mlx-audio do not support quantized models.
+Quantized models require calling `mlx.nn.quantize()` before loading weights.
 
-### parakeet-mlx
+```python
+import json
+import mlx.nn as nn
+from huggingface_hub import hf_hub_download
+from parakeet_mlx.utils import from_config
 
-```bash
-pip install -U parakeet-mlx
-```
+# Download and load config
+config_path = hf_hub_download("animaslabs/nemotron-speech-streaming-en-0.6b-mlx-4bit", "config.json")
+with open(config_path) as f:
+    config = json.load(f)
 
-```bash
-parakeet-mlx audio.wav --model mlx-community/parakeet-tdt-0.6b-v3
-```
-
-### mlx-audio
+# Build model and apply quantization structure
+model = from_config(config)
+nn.quantize(
+    model,
+    bits=config["quantization"]["bits"],
+    group_size=config["quantization"]["group_size"],
+)
 
-```bash
-pip install -U mlx-audio
-```
+# Load quantized weights
+weights_path = hf_hub_download("animaslabs/nemotron-speech-streaming-en-0.6b-mlx-4bit", "model.safetensors")
+model.load_weights(weights_path)
 
-```bash
-python -m mlx_audio.stt.generate --model mlx-community/parakeet-tdt-0.6b-v3 --audio audio.wav --output somewhere
+# Transcribe
+result = model.transcribe("audio.wav")
+print(result.text)
 ```
 

config.json

@@ -1,6 +1,2265 @@
 {
-  "model_type": "parakeet-tdt",
-  "framework": "mlx",
+  "sample_rate": 16000,
+  "compute_eval_loss": false,
+  "log_prediction": true,
+  "rnnt_reduction": "mean_volume",
+  "skip_nan_grad": false,
+  "model_defaults": {
+    "enc_hidden": 1024,
+    "pred_hidden": 640,
+    "joint_hidden": 640
+  },
+  "train_ds": {
+    "use_lhotse": true,
+    "skip_missing_manifest_entries": true,
+    "input_cfg": null,
+    "tarred_audio_filepaths": null,
+    "manifest_filepath": null,
+    "sample_rate": 16000,
+    "shuffle": true,
+    "num_workers": 2,
+    "pin_memory": true,
+    "max_duration": 40.0,
+    "min_duration": 0.1,
+    "text_field": "answer",
+    "batch_duration": null,
+    "use_bucketing": true,
+    "max_tps": null,
+    "bucket_duration_bins": null,
+    "bucket_batch_size": null,
+    "num_buckets": null,
+    "bucket_buffer_size": null,
+    "shuffle_buffer_size": null,
+    "augmentor": null
+  },
+  "validation_ds": {
+    "use_lhotse": true,
+    "manifest_filepath": "/data/ASR/en/librispeech/test-other.json",
+    "sample_rate": 16000,
+    "batch_size": 32,
+    "shuffle": false,
+    "max_duration": 40.0,
+    "min_duration": 0.1,
+    "num_workers": 2,
+    "pin_memory": true,
+    "text_field": "answer",
+    "tarred_audio_filepaths": null
+  },
+  "tokenizer": {
+    "dir": null,
+    "type": "bpe",
+    "model_path": "nemo:tokenizer.model",
+    "vocab_path": "nemo:vocab.txt",
+    "spe_tokenizer_vocab": "nemo:tokenizer.vocab"
+  },
+  "preprocessor": {
+    "_target_": "nemo.collections.asr.modules.AudioToMelSpectrogramPreprocessor",
+    "sample_rate": 16000,
+    "normalize": "NA",
+    "window_size": 0.025,
+    "window_stride": 0.01,
+    "window": "hann",
+    "features": 128,
+    "n_fft": 512,
+    "log": true,
+    "frame_splicing": 1,
+    "dither": 1e-05,
+    "pad_to": 0,
+    "pad_value": 0.0
+  },
+  "spec_augment": {
+    "_target_": "nemo.collections.asr.modules.SpectrogramAugmentation",
+    "freq_masks": 2,
+    "time_masks": 10,
+    "freq_width": 27,
+    "time_width": 0.05
+  },
+  "encoder": {
+    "_target_": "nemo.collections.asr.modules.ConformerEncoder",
+    "feat_in": 128,
+    "feat_out": -1,
+    "n_layers": 24,
+    "d_model": 1024,
+    "use_bias": false,
+    "subsampling": "dw_striding",
+    "subsampling_factor": 8,
+    "subsampling_conv_channels": 256,
+    "causal_downsampling": true,
+    "reduction": null,
+    "reduction_position": null,
+    "reduction_factor": 1,
+    "ff_expansion_factor": 4,
+    "self_attention_model": "rel_pos",
+    "n_heads": 8,
+    "att_context_size": [
+      [
+        70,
+        13
+      ],
+      [
+        70,
+        6
+      ],
+      [
+        70,
+        1
+      ],
+      [
+        70,
+        0
+      ]
+    ],
+    "att_context_style": "chunked_limited",
+    "xscaling": false,
+    "untie_biases": true,
+    "pos_emb_max_len": 5000,
+    "conv_kernel_size": 9,
+    "conv_norm_type": "layer_norm",
+    "conv_context_size": "causal",
+    "dropout": 0.1,
+    "dropout_pre_encoder": 0.1,
+    "dropout_emb": 0.0,
+    "dropout_att": 0.1,
+    "stochastic_depth_drop_prob": 0.0,
+    "stochastic_depth_mode": "linear",
+    "stochastic_depth_start_layer": 1
+  },
+  "decoder": {
+    "_target_": "nemo.collections.asr.modules.RNNTDecoder",
+    "normalization_mode": null,
+    "random_state_sampling": false,
+    "blank_as_pad": true,
+    "prednet": {
+      "pred_hidden": 640,
+      "pred_rnn_layers": 2,
+      "t_max": null,
+      "dropout": 0.2
+    },
+    "vocab_size": 1024
+  },
+  "joint": {
+    "_target_": "nemo.collections.asr.modules.RNNTJoint",
+    "log_softmax": null,
+    "preserve_memory": false,
+    "fuse_loss_wer": true,
+    "fused_batch_size": 4,
+    "jointnet": {
+      "joint_hidden": 640,
+      "activation": "relu",
+      "dropout": 0.2,
+      "encoder_hidden": 1024,
+      "pred_hidden": 640
+    },
+    "num_classes": 1024,
+    "vocabulary": [
+      "<unk>",
+      "\u2581t",
+      "\u2581th",
+      "\u2581a",
+      "in",
+      "\u2581the",
+      "re",
+      "\u2581w",
+      "\u2581o",
+      "\u2581s",
+      "er",
+      "at",
+      "ou",
+      "nd",
+      "it",
+      "is",
+      "\u2581h",
+      "\u2581b",
+      "on",
+      "\u2581c",
+      "ing",
+      "en",
+      "\u2581to",
+      "\u2581m",
+      "\u2581f",
+      "\u2581p",
+      "or",
+      "an",
+      "es",
+      "\u2581of",
+      "\u2581d",
+      "ed",
+      "ll",
+      "\u2581and",
+      "\u2581I",
+      "\u2581in",
+      "\u2581l",
+      "ar",
+      "\u2581y",
+      "\u2581g",
+      "as",
+      "\u2581you",
+      "om",
+      "\u2581n",
+      "ic",
+      "ve",
+      "al",
+      "ion",
+      "us",
+      "\u2581be",
+      "ow",
+      "le",
+      "\u2581wh",
+      "\u2581e",
+      "ot",
+      "ut",
+      "\u2581it",
+      "\u2581is",
+      "\u2581we",
+      "\u2581T",
+      "\u2581re",
+      "et",
+      "\u2581A",
+      "ent",
+      "\u2581on",
+      "\u2581ha",
+      "ay",
+      "\u2581S",
+      "ct",
+      "\u2581Th",
+      "ver",
+      "id",
+      "ig",
+      "im",
+      "ro",
+      "\u2581for",
+      "ly",
+      "\u2581he",
+      "ke",
+      "ld",
+      "se",
+      "st",
+      "ch",
+      "\u2581st",
+      "all",
+      "ce",
+      "ur",
+      "ith",
+      "am",
+      "if",
+      "ir",
+      "\u2581go",
+      "\u2581u",
+      "\u2581as",
+      "\u2581was",
+      "ad",
+      "\u2581W",
+      "\u2581k",
+      "\u2581an",
+      "ht",
+      "th",
+      "\u2581r",
+      "\u2581are",
+      "ere",
+      "\u2581se",
+      "\u2581do",
+      "\u2581B",
+      "\u2581so",
+      "\u2581sh",
+      "\u2581not",
+      "\u2581li",
+      "od",
+      "\u2581C",
+      "ust",
+      "ill",
+      "ight",
+      "ally",
+      "\u2581And",
+      "ter",
+      "\u2581or",
+      "\u2581me",
+      "\u2581M",
+      "ome",
+      "op",
+      "\u2581at",
+      "il",
+      "\u2581The",
+      "ould",
+      "\u2581j",
+      "ant",
+      "\u2581So",
+      "\u2581H",
+      "ol",
+      "ain",
+      "\u2581can",
+      "\u2581de",
+      "\u2581ne",
+      "ore",
+      "\u2581con",
+      "\u2581kn",
+      "ck",
+      "ul",
+      "\u2581fr",
+      "\u2581ab",
+      "ers",
+      "ess",
+      "ge",
+      "\u2581pro",
+      "pe",
+      "ate",
+      "\u2581su",
+      "\u2581com",
+      "\u2581but",
+      "\u2581all",
+      "est",
+      "qu",
+      "\u2581ex",
+      "\u2581al",
+      "ra",
+      "\u2581O",
+      "out",
+      "use",
+      "very",
+      "pp",
+      "\u2581Y",
+      "\u2581ch",
+      "ri",
+      "ist",
+      "\u2581v",
+      "\u2581lo",
+      "ment",
+      "art",
+      "\u2581P",
+      "nt",
+      "ab",
+      "\u2581one",
+      "\u2581N",
+      "ive",
+      "\u2581wor",
+      "ions",
+      "ort",
+      "\u2581L",
+      "\u2581by",
+      "ich",
+      "\u2581my",
+      "ity",
+      "ok",
+      "\u2581G",
+      "res",
+      "\u2581up",
+      "un",
+      "um",
+      "ea",
+      "ind",
+      "and",
+      "ink",
+      "el",
+      "\u2581D",
+      "em",
+      "\u2581E",
+      "os",
+      "oug",
+      "\u2581if",
+      "ca",
+      "\u2581out",
+      "\u2581int",
+      "ie",
+      "\u2581F",
+      "\u2581It",
+      "\u2581his",
+      "ard",
+      "\u2581had",
+      "\u2581tr",
+      "her",
+      "our",
+      "ies",
+      "ake",
+      "\u2581R",
+      "\u2581We",
+      "\u2581get",
+      "\u2581don",
+      "\u2581us",
+      "ak",
+      "\u2581pl",
+      "ect",
+      "ure",
+      "ame",
+      "ast",
+      "\u2581who",
+      "ack",
+      "\u2581le",
+      "\u2581sa",
+      "iv",
+      "ci",
+      "ide",
+      "\u2581tim",
+      "\u2581our",
+      "ound",
+      "ous",
+      "\u2581co",
+      "\u2581pe",
+      "ose",
+      "ud",
+      "\u2581see",
+      "ough",
+      "\u2581man",
+      "\u2581qu",
+      "\u2581You",
+      "so",
+      "ople",
+      "\u2581Wh",
+      "ong",
+      "ap",
+      "ther",
+      "\u2581J",
+      "are",
+      "ine",
+      "\u2581say",
+      "\u2581im",
+      "\u2581But",
+      "ings",
+      "\u2581has",
+      "\u2581ag",
+      "ff",
+      "\u2581her",
+      "itt",
+      "one",
+      "\u2581en",
+      "\u2581ar",
+      "\u2581fe",
+      "ven",
+      "\u2581any",
+      "\u2581mo",
+      "reat",
+      "ag",
+      "\u2581how",
+      "\u2581cl",
+      "pt",
+      "\u2581now",
+      "own",
+      "ber",
+      "\u2581him",
+      "\u2581act",
+      "hing",
+      "ice",
+      "\u2581no",
+      "ans",
+      "iz",
+      "\u2581fa",
+      "per",
+      "pl",
+      "\u2581te",
+      "\u2581ad",
+      "age",
+      "ree",
+      "\u2581tw",
+      "ank",
+      "\u2581He",
+      "ple",
+      "ite",
+      "ry",
+      "\u2581U",
+      "ish",
+      "ire",
+      "ue",
+      "\u2581In",
+      "\u2581she",
+      "ble",
+      "cc",
+      "nder",
+      "\u2581way",
+      "\u2581pr",
+      "ear",
+      "\u2581did",
+      "\u2581po",
+      "eah",
+      "\u2581un",
+      "omet",
+      "ence",
+      "ep",
+      "uch",
+      "\u2581sp",
+      "ach",
+      "og",
+      "ance",
+      "able",
+      "iff",
+      "sel",
+      "\u2581got",
+      "way",
+      "\u2581gr",
+      "alk",
+      "\u2581res",
+      "ated",
+      "irst",
+      "ick",
+      "ass",
+      "\u2581two",
+      "\u2581dis",
+      "ord",
+      "\u2581pre",
+      "ount",
+      "ase",
+      "ip",
+      "ult",
+      "ical",
+      "orm",
+      "ary",
+      "ace",
+      "\u2581spe",
+      "\u2581Ch",
+      "\u2581thr",
+      "\u2581imp",
+      "int",
+      "\u2581am",
+      "\u2581off",
+      "act",
+      "ia",
+      "\u2581ro",
+      "ress",
+      "\u2581per",
+      "\u2581fo",
+      "\u2581br",
+      "\u2581K",
+      "vel",
+      "\u2581gu",
+      "\u2581bo",
+      "ang",
+      "kay",
+      "ub",
+      "ign",
+      "\u2581may",
+      "ving",
+      "ces",
+      "ens",
+      "cl",
+      "\u2581lot",
+      "ru",
+      "ade",
+      "\u2581bet",
+      "\u2581bl",
+      "\u2581let",
+      "fore",
+      "co",
+      "ild",
+      "ning",
+      "xt",
+      "ile",
+      "ark",
+      "self",
+      "\u2581app",
+      "ory",
+      "du",
+      "\u2581day",
+      "\u2581St",
+      "ater",
+      "\u2581use",
+      "ys",
+      "fter",
+      "\u2581new",
+      "ious",
+      "ial",
+      "he",
+      "wn",
+      "ved",
+      "red",
+      "\u2581fl",
+      "iss",
+      "ody",
+      "form",
+      "ian",
+      "tain",
+      "\u2581bu",
+      "\u2581V",
+      "\u2581rec",
+      "ty",
+      "be",
+      "\u2581sc",
+      "ors",
+      "vers",
+      "\u2581put",
+      "ife",
+      "\u2581If",
+      "we",
+      "te",
+      "ject",
+      "ath",
+      "ting",
+      "\u2581rem",
+      "\u2581acc",
+      "ull",
+      "ons",
+      "\u2581ind",
+      "\u2581ser",
+      "\u2581ke",
+      "ates",
+      "ves",
+      "na",
+      "lic",
+      "\u2581des",
+      "\u2581its",
+      "ful",
+      "ents",
+      "erm",
+      "ac",
+      "ered",
+      "ise",
+      "\u2581sy",
+      "urn",
+      "\u2581em",
+      "oth",
+      "ual",
+      "ne",
+      "ward",
+      "ib",
+      "\u2581try",
+      "\u2581pos",
+      "nds",
+      "ft",
+      "get",
+      "ph",
+      "\u2581ob",
+      "ady",
+      "igh",
+      "ood",
+      "\u2581rel",
+      "\u2581wr",
+      "ug",
+      "ears",
+      "ail",
+      "\u2581Now",
+      "\u2581bit",
+      "ng",
+      "\u2581Oh",
+      "\u2581hel",
+      "ange",
+      "\u2581reg",
+      "\u2581rep",
+      "\u2581bel",
+      "\u2581sm",
+      "ost",
+      "tern",
+      "gr",
+      "\u2581own",
+      "\u2581end",
+      "pect",
+      "ily",
+      "day",
+      "ied",
+      "ific",
+      "ower",
+      "\u2581add",
+      "cess",
+      "ict",
+      "ible",
+      "\u2581bas",
+      "\u2581i",
+      "\u2581op",
+      "cial",
+      "ular",
+      "\u2581Be",
+      "ced",
+      "\u2581too",
+      "ks",
+      "ew",
+      "mer",
+      "\u2581ph",
+      "ob",
+      "==",
+      "\u2581la",
+      "\u2581set",
+      "\u2581min",
+      "\u2581sub",
+      "\u2581gen",
+      "atch",
+      "..",
+      "\u2581inv",
+      "\u2581As",
+      "\u2581nat",
+      "\u2581sl",
+      "\u2581num",
+      "av",
+      "ways",
+      "\u2581God",
+      "stem",
+      "\u2581ac",
+      "\u2581att",
+      "\u2581ev",
+      "\u2581def",
+      "llow",
+      "\u2581str",
+      "lect",
+      "ars",
+      "\u2581cr",
+      "\u2581Is",
+      "olog",
+      "les",
+      "oy",
+      "\u2581ask",
+      "\u2581inc",
+      "body",
+      "\u2581ent",
+      "\u2581pol",
+      "ness",
+      "ix",
+      "\u2581why",
+      "onna",
+      "\u2581ear",
+      "\u2581tak",
+      "\u2581Un",
+      "ited",
+      "mun",
+      "li",
+      "ute",
+      "ract",
+      "\u2581dec",
+      "uro",
+      "\u2581mak",
+      "\u2581fin",
+      "ween",
+      "\u2581No",
+      "arch",
+      "\u2581bec",
+      "gan",
+      "old",
+      "cy",
+      "\u2581big",
+      "\u2581For",
+      "ren",
+      "als",
+      "und",
+      "\u2581Al",
+      "\u2581All",
+      "ss",
+      "ows",
+      "\u2581mod",
+      "ock",
+      "\u2581id",
+      "ism",
+      "cus",
+      "\u2581gl",
+      "ably",
+      "\u2581ass",
+      "\u2581car",
+      "ata",
+      "ppen",
+      "led",
+      "\u2581sim",
+      "\u2581mon",
+      "ics",
+      "\u2581giv",
+      "cept",
+      "\u2581Mr",
+      "pan",
+      "\u2581pub",
+      "\u2581eff",
+      "\u2581How",
+      "ps",
+      "vern",
+      "end",
+      "hip",
+      "iew",
+      "ope",
+      "\u2581An",
+      "\u2581She",
+      "\u2581Com",
+      "ee",
+      "ures",
+      "ell",
+      "ouse",
+      "cond",
+      "king",
+      "oc",
+      "ues",
+      "ever",
+      "\u2581To",
+      "clud",
+      "\u2581ins",
+      "\u2581exp",
+      "\u2581old",
+      "\u2581mem",
+      "\u2581ref",
+      "\u2581tra",
+      "\u2581far",
+      "ave",
+      "rat",
+      "\u2581sur",
+      "ruct",
+      "rib",
+      "duct",
+      "uff",
+      "\u2581met",
+      "\u2581sch",
+      "ince",
+      "\u2581run",
+      "ense",
+      "\u2581cle",
+      "\u2581==",
+      "mon",
+      "ize",
+      "\u2581ord",
+      "blem",
+      "tin",
+      "\u2581Let",
+      "ner",
+      "ond",
+      "its",
+      "\u2581cor",
+      "land",
+      "\u2581cur",
+      "\u2581Re",
+      "\u2581bus",
+      "\u2581uh",
+      "air",
+      "ote",
+      "ants",
+      "ason",
+      "ric",
+      "\u2581el",
+      "\u2581cer",
+      "nce",
+      "\u2581fam",
+      "\u2581cap",
+      "uck",
+      "ool",
+      "ried",
+      "\u2581cou",
+      "\u2581fun",
+      "\u2581wom",
+      "\u2581hum",
+      "\u2581ty",
+      "\u2581ap",
+      "ike",
+      "\u2581few",
+      "oney",
+      "\u2581inf",
+      "ont",
+      "ese",
+      "ook",
+      "gy",
+      "uth",
+      "ulat",
+      "ieve",
+      "ized",
+      "ross",
+      "\u2581ple",
+      "\u2581um",
+      "\u2581val",
+      "\u2581equ",
+      "\u2581lea",
+      "\u2581lar",
+      "ah",
+      "eral",
+      "\u2581ed",
+      "ared",
+      "lish",
+      "arn",
+      "ds",
+      "esn",
+      "\u2581iss",
+      "\u2581ca",
+      "ted",
+      "ices",
+      "\u2581wee",
+      "ash",
+      "\u2581top",
+      "ten",
+      "up",
+      "ts",
+      "gin",
+      "con",
+      "ari",
+      "\u2581opp",
+      "osed",
+      "\u2581eas",
+      "\u2581ext",
+      "gg",
+      "az",
+      "\u2581Fr",
+      "ideo",
+      "izat",
+      "\u2581men",
+      "\u2581mom",
+      "\u2581ret",
+      "tty",
+      "rist",
+      "\u2581gra",
+      "alth",
+      "ef",
+      "\u2581det",
+      "ax",
+      "\u2581mat",
+      "chn",
+      "ern",
+      "peri",
+      "\u2581bre",
+      "\u2581Sh",
+      "sw",
+      "erat",
+      "\u2581sit",
+      "ters",
+      "ale",
+      "man",
+      "\u2581sol",
+      "ork",
+      "\u2581adv",
+      "ety",
+      "\u2581vis",
+      "\u2581med",
+      "uc",
+      "less",
+      "\u2581unt",
+      "gram",
+      "ets",
+      "ists",
+      "\u2581ey",
+      "\u2581col",
+      "imes",
+      "\u2581law",
+      "\u2581pri",
+      "sid",
+      "\u2581On",
+      "\u2581mot",
+      "ield",
+      "\u2581Do",
+      "\u2581At",
+      "ages",
+      "amp",
+      "\u2581art",
+      "miss",
+      "\u2581sk",
+      "alf",
+      "pr",
+      "ier",
+      "\u2581beh",
+      "\u2581Yes",
+      "ural",
+      "ime",
+      "\u2581wa",
+      "oks",
+      "bers",
+      "ger",
+      "ient",
+      "ries",
+      "...",
+      "\u2581che",
+      "\u2581Br",
+      "ird",
+      "\u2581Ar",
+      "\u2581war",
+      "inat",
+      "\u2581My",
+      "ital",
+      "wh",
+      "med",
+      "\u2581pur",
+      "ully",
+      "\u2581One",
+      "\u2581rat",
+      "ines",
+      "\u2581Of",
+      "io",
+      "\u2581loc",
+      "ret",
+      "ctor",
+      "\u2581leg",
+      "stit",
+      "ined",
+      "ught",
+      "\u2581dur",
+      "\u2581es",
+      "vent",
+      "aj",
+      "\u2581bro",
+      "\u2581saw",
+      "\u2581sec",
+      "ream",
+      "\u2581pop",
+      "reen",
+      "\u2581Ind",
+      "els",
+      "\u2581yet",
+      "ired",
+      "\u2581sw",
+      "tro",
+      "oup",
+      "most",
+      "pean",
+      "eds",
+      "ush",
+      "oh",
+      "\u2581Se",
+      "\u2581tea",
+      "ann",
+      "ilit",
+      "err",
+      "pend",
+      "ton",
+      "ased",
+      "\u2581aff",
+      "\u2581mor",
+      "\u2581dra",
+      "put",
+      "\u2581dr",
+      "ins",
+      "uat",
+      "nect",
+      "cri",
+      "outh",
+      "\u2581ra",
+      "\u2581pay",
+      "ms",
+      "\u2581av",
+      "bs",
+      "ling",
+      "\u2581De",
+      "\u2581Or",
+      "ove",
+      "\u2581Can",
+      "\u2581eng",
+      "ames",
+      "ided",
+      "\u2581Go",
+      "mitt",
+      "ode",
+      "\u2581cre",
+      "par",
+      "ides",
+      "pos",
+      "\u2581fav",
+      "\u2581air",
+      "\u2581New",
+      "\u2581bad",
+      "\u2581six",
+      "vat",
+      "\u2581pat",
+      "not",
+      "\u2581di",
+      "rop",
+      "ral",
+      "orn",
+      "\u2581par",
+      "cing",
+      "\u2581aw",
+      "orts",
+      "ox",
+      "\u2581yes",
+      "cuss",
+      "eng",
+      "ives",
+      "erms",
+      "\u2581job",
+      "mand",
+      "ying",
+      "\u2581occ",
+      "aps",
+      "ases",
+      "\u2581Not",
+      "rent",
+      "ency",
+      "att",
+      "ised",
+      "vice",
+      "\u2581Eng",
+      "\u2581est",
+      "oked",
+      "\u2581Q",
+      "iron",
+      "idd",
+      "me",
+      "unch",
+      "ane",
+      "\u2581z",
+      "br",
+      "arts",
+      "\u2581fat",
+      "ery",
+      "anks",
+      "\u2581jo",
+      "\u2581mar",
+      "aw",
+      "ott",
+      "ards",
+      "\u2581oh",
+      "ians",
+      "\u2581sci",
+      "row",
+      "unt",
+      "ury",
+      "\u2581abs",
+      "ergy",
+      "\u2581Z",
+      "ump",
+      "\u2581Am",
+      "ened",
+      "angu",
+      "\u2581Pro",
+      "icat",
+      "itch",
+      "\u2581dri",
+      "iat",
+      "\u2581",
+      "e",
+      "t",
+      "o",
+      "a",
+      "n",
+      "i",
+      "s",
+      "r",
+      "h",
+      "l",
+      "d",
+      "u",
+      "c",
+      "m",
+      "y",
+      "g",
+      "w",
+      "f",
+      "p",
+      ",",
+      ".",
+      "b",
+      "v",
+      "k",
+      "'",
+      "I",
+      "T",
+      "A",
+      "S",
+      "x",
+      "W",
+      "j",
+      "C",
+      "B",
+      "M",
+      "?",
+      "H",
+      "O",
+      "0",
+      "P",
+      "q",
+      "Y",
+      "N",
+      "L",
+      "D",
+      "1",
+      "E",
+      "G",
+      "z",
+      "F",
+      "R",
+      "-",
+      "2",
+      "J",
+      "U",
+      "9",
+      "K",
+      "5",
+      "3",
+      "V",
+      "=",
+      "4",
+      "8",
+      "6",
+      "7",
+      "!",
+      "%",
+      ":",
+      "Q",
+      "Z",
+      "$",
+      "X",
+      "\"",
+      "&",
+      "*",
+      "/",
+      "\u00a3",
+      "+",
+      "\u20ac",
+      "_",
+      "^",
+      "\u00a5"
+    ]
+  },
+  "decoding": {
+    "strategy": "greedy_batch",
+    "greedy": {
+      "max_symbols": 10
+    },
+    "beam": {
+      "beam_size": 2,
+      "return_best_hypothesis": false,
+      "score_norm": true,
+      "tsd_max_sym_exp": 50,
+      "alsd_max_target_len": 2.0
+    }
+  },
+  "aux_ctc": {
+    "ctc_loss_weight": 0.3,
+    "use_cer": false,
+    "ctc_reduction": "mean_batch",
+    "decoder": {
+      "_target_": "nemo.collections.asr.modules.ConvASRDecoder",
+      "feat_in": null,
+      "num_classes": -1,
+      "vocabulary": []
+    },
+    "decoding": {
+      "strategy": "greedy"
+    }
+  },
+  "interctc": {
+    "loss_weights": [],
+    "apply_at_layers": []
+  },
+  "loss": {
+    "loss_name": "default",
+    "warprnnt_numba_kwargs": {
+      "fastemit_lambda": 0.005,
+      "clamp": -1.0
+    }
+  },
+  "optim": {
+    "name": "adamw",
+    "lr": 3.0,
+    "betas": [
+      0.9,
+      0.98
+    ],
+    "weight_decay": 0.001,
+    "sched": {
+      "name": "NoamAnnealing",
+      "d_model": 1024,
+      "warmup_steps": 8000,
+      "warmup_ratio": null,
+      "min_lr": 1e-06
+    }
+  },
+  "labels": [
+    "<unk>",
+    "\u2581t",
+    "\u2581th",
+    "\u2581a",
+    "in",
+    "\u2581the",
+    "re",
+    "\u2581w",
+    "\u2581o",
+    "\u2581s",
+    "er",
+    "at",
+    "ou",
+    "nd",
+    "it",
+    "is",
+    "\u2581h",
+    "\u2581b",
+    "on",
+    "\u2581c",
+    "ing",
+    "en",
+    "\u2581to",
+    "\u2581m",
+    "\u2581f",
+    "\u2581p",
+    "or",
+    "an",
+    "es",
+    "\u2581of",
+    "\u2581d",
+    "ed",
+    "ll",
+    "\u2581and",
+    "\u2581I",
+    "\u2581in",
+    "\u2581l",
+    "ar",
+    "\u2581y",
+    "\u2581g",
+    "as",
+    "\u2581you",
+    "om",
+    "\u2581n",
+    "ic",
+    "ve",
+    "al",
+    "ion",
+    "us",
+    "\u2581be",
+    "ow",
+    "le",
+    "\u2581wh",
+    "\u2581e",
+    "ot",
+    "ut",
+    "\u2581it",
+    "\u2581is",
+    "\u2581we",
+    "\u2581T",
+    "\u2581re",
+    "et",
+    "\u2581A",
+    "ent",
+    "\u2581on",
+    "\u2581ha",
+    "ay",
+    "\u2581S",
+    "ct",
+    "\u2581Th",
+    "ver",
+    "id",
+    "ig",
+    "im",
+    "ro",
+    "\u2581for",
+    "ly",
+    "\u2581he",
+    "ke",
+    "ld",
+    "se",
+    "st",
+    "ch",
+    "\u2581st",
+    "all",
+    "ce",
+    "ur",
+    "ith",
+    "am",
+    "if",
+    "ir",
+    "\u2581go",
+    "\u2581u",
+    "\u2581as",
+    "\u2581was",
+    "ad",
+    "\u2581W",
+    "\u2581k",
+    "\u2581an",
+    "ht",
+    "th",
+    "\u2581r",
+    "\u2581are",
+    "ere",
+    "\u2581se",
+    "\u2581do",
+    "\u2581B",
+    "\u2581so",
+    "\u2581sh",
+    "\u2581not",
+    "\u2581li",
+    "od",
+    "\u2581C",
+    "ust",
+    "ill",
+    "ight",
+    "ally",
+    "\u2581And",
+    "ter",
+    "\u2581or",
+    "\u2581me",
+    "\u2581M",
+    "ome",
+    "op",
+    "\u2581at",
+    "il",
+    "\u2581The",
+    "ould",
+    "\u2581j",
+    "ant",
+    "\u2581So",
+    "\u2581H",
+    "ol",
+    "ain",
+    "\u2581can",
+    "\u2581de",
+    "\u2581ne",
+    "ore",
+    "\u2581con",
+    "\u2581kn",
+    "ck",
+    "ul",
+    "\u2581fr",
+    "\u2581ab",
+    "ers",
+    "ess",
+    "ge",
+    "\u2581pro",
+    "pe",
+    "ate",
+    "\u2581su",
+    "\u2581com",
+    "\u2581but",
+    "\u2581all",
+    "est",
+    "qu",
+    "\u2581ex",
+    "\u2581al",
+    "ra",
+    "\u2581O",
+    "out",
+    "use",
+    "very",
+    "pp",
+    "\u2581Y",
+    "\u2581ch",
+    "ri",
+    "ist",
+    "\u2581v",
+    "\u2581lo",
+    "ment",
+    "art",
+    "\u2581P",
+    "nt",
+    "ab",
+    "\u2581one",
+    "\u2581N",
+    "ive",
+    "\u2581wor",
+    "ions",
+    "ort",
+    "\u2581L",
+    "\u2581by",
+    "ich",
+    "\u2581my",
+    "ity",
+    "ok",
+    "\u2581G",
+    "res",
+    "\u2581up",
+    "un",
+    "um",
+    "ea",
+    "ind",
+    "and",
+    "ink",
+    "el",
+    "\u2581D",
+    "em",
+    "\u2581E",
+    "os",
+    "oug",
+    "\u2581if",
+    "ca",
+    "\u2581out",
+    "\u2581int",
+    "ie",
+    "\u2581F",
+    "\u2581It",
+    "\u2581his",
+    "ard",
+    "\u2581had",
+    "\u2581tr",
+    "her",
+    "our",
+    "ies",
+    "ake",
+    "\u2581R",
+    "\u2581We",
+    "\u2581get",
+    "\u2581don",
+    "\u2581us",
+    "ak",
+    "\u2581pl",
+    "ect",
+    "ure",
+    "ame",
+    "ast",
+    "\u2581who",
+    "ack",
+    "\u2581le",
+    "\u2581sa",
+    "iv",
+    "ci",
+    "ide",
+    "\u2581tim",
+    "\u2581our",
+    "ound",
+    "ous",
+    "\u2581co",
+    "\u2581pe",
+    "ose",
+    "ud",
+    "\u2581see",
+    "ough",
+    "\u2581man",
+    "\u2581qu",
+    "\u2581You",
+    "so",
+    "ople",
+    "\u2581Wh",
+    "ong",
+    "ap",
+    "ther",
+    "\u2581J",
+    "are",
+    "ine",
+    "\u2581say",
+    "\u2581im",
+    "\u2581But",
+    "ings",
+    "\u2581has",
+    "\u2581ag",
+    "ff",
+    "\u2581her",
+    "itt",
+    "one",
+    "\u2581en",
+    "\u2581ar",
+    "\u2581fe",
+    "ven",
+    "\u2581any",
+    "\u2581mo",
+    "reat",
+    "ag",
+    "\u2581how",
+    "\u2581cl",
+    "pt",
+    "\u2581now",
+    "own",
+    "ber",
+    "\u2581him",
+    "\u2581act",
+    "hing",
+    "ice",
+    "\u2581no",
+    "ans",
+    "iz",
+    "\u2581fa",
+    "per",
+    "pl",
+    "\u2581te",
+    "\u2581ad",
+    "age",
+    "ree",
+    "\u2581tw",
+    "ank",
+    "\u2581He",
+    "ple",
+    "ite",
+    "ry",
+    "\u2581U",
+    "ish",
+    "ire",
+    "ue",
+    "\u2581In",
+    "\u2581she",
+    "ble",
+    "cc",
+    "nder",
+    "\u2581way",
+    "\u2581pr",
+    "ear",
+    "\u2581did",
+    "\u2581po",
+    "eah",
+    "\u2581un",
+    "omet",
+    "ence",
+    "ep",
+    "uch",
+    "\u2581sp",
+    "ach",
+    "og",
+    "ance",
+    "able",
+    "iff",
+    "sel",
+    "\u2581got",
+    "way",
+    "\u2581gr",
+    "alk",
+    "\u2581res",
+    "ated",
+    "irst",
+    "ick",
+    "ass",
+    "\u2581two",
+    "\u2581dis",
+    "ord",
+    "\u2581pre",
+    "ount",
+    "ase",
+    "ip",
+    "ult",
+    "ical",
+    "orm",
+    "ary",
+    "ace",
+    "\u2581spe",
+    "\u2581Ch",
+    "\u2581thr",
+    "\u2581imp",
+    "int",
+    "\u2581am",
+    "\u2581off",
+    "act",
+    "ia",
+    "\u2581ro",
+    "ress",
+    "\u2581per",
+    "\u2581fo",
+    "\u2581br",
+    "\u2581K",
+    "vel",
+    "\u2581gu",
+    "\u2581bo",
+    "ang",
+    "kay",
+    "ub",
+    "ign",
+    "\u2581may",
+    "ving",
+    "ces",
+    "ens",
+    "cl",
+    "\u2581lot",
+    "ru",
+    "ade",
+    "\u2581bet",
+    "\u2581bl",
+    "\u2581let",
+    "fore",
+    "co",
+    "ild",
+    "ning",
+    "xt",
+    "ile",
+    "ark",
+    "self",
+    "\u2581app",
+    "ory",
+    "du",
+    "\u2581day",
+    "\u2581St",
+    "ater",
+    "\u2581use",
+    "ys",
+    "fter",
+    "\u2581new",
+    "ious",
+    "ial",
+    "he",
+    "wn",
+    "ved",
+    "red",
+    "\u2581fl",
+    "iss",
+    "ody",
+    "form",
+    "ian",
+    "tain",
+    "\u2581bu",
+    "\u2581V",
+    "\u2581rec",
+    "ty",
+    "be",
+    "\u2581sc",
+    "ors",
+    "vers",
+    "\u2581put",
+    "ife",
+    "\u2581If",
+    "we",
+    "te",
+    "ject",
+    "ath",
+    "ting",
+    "\u2581rem",
+    "\u2581acc",
+    "ull",
+    "ons",
+    "\u2581ind",
+    "\u2581ser",
+    "\u2581ke",
+    "ates",
+    "ves",
+    "na",
+    "lic",
+    "\u2581des",
+    "\u2581its",
+    "ful",
+    "ents",
+    "erm",
+    "ac",
+    "ered",
+    "ise",
+    "\u2581sy",
+    "urn",
+    "\u2581em",
+    "oth",
+    "ual",
+    "ne",
+    "ward",
+    "ib",
+    "\u2581try",
+    "\u2581pos",
+    "nds",
+    "ft",
+    "get",
+    "ph",
+    "\u2581ob",
+    "ady",
+    "igh",
+    "ood",
+    "\u2581rel",
+    "\u2581wr",
+    "ug",
+    "ears",
+    "ail",
+    "\u2581Now",
+    "\u2581bit",
+    "ng",
+    "\u2581Oh",
+    "\u2581hel",
+    "ange",
+    "\u2581reg",
+    "\u2581rep",
+    "\u2581bel",
+    "\u2581sm",
+    "ost",
+    "tern",
+    "gr",
+    "\u2581own",
+    "\u2581end",
+    "pect",
+    "ily",
+    "day",
+    "ied",
+    "ific",
+    "ower",
+    "\u2581add",
+    "cess",
+    "ict",
+    "ible",
+    "\u2581bas",
+    "\u2581i",
+    "\u2581op",
+    "cial",
+    "ular",
+    "\u2581Be",
+    "ced",
+    "\u2581too",
+    "ks",
+    "ew",
+    "mer",
+    "\u2581ph",
+    "ob",
+    "==",
+    "\u2581la",
+    "\u2581set",
+    "\u2581min",
+    "\u2581sub",
+    "\u2581gen",
+    "atch",
+    "..",
+    "\u2581inv",
+    "\u2581As",
+    "\u2581nat",
+    "\u2581sl",
+    "\u2581num",
+    "av",
+    "ways",
+    "\u2581God",
+    "stem",
+    "\u2581ac",
+    "\u2581att",
+    "\u2581ev",
+    "\u2581def",
+    "llow",
+    "\u2581str",
+    "lect",
+    "ars",
+    "\u2581cr",
+    "\u2581Is",
+    "olog",
+    "les",
+    "oy",
+    "\u2581ask",
+    "\u2581inc",
+    "body",
+    "\u2581ent",
+    "\u2581pol",
+    "ness",
+    "ix",
+    "\u2581why",
+    "onna",
+    "\u2581ear",
+    "\u2581tak",
+    "\u2581Un",
+    "ited",
+    "mun",
+    "li",
+    "ute",
+    "ract",
+    "\u2581dec",
+    "uro",
+    "\u2581mak",
+    "\u2581fin",
+    "ween",
+    "\u2581No",
+    "arch",
+    "\u2581bec",
+    "gan",
+    "old",
+    "cy",
+    "\u2581big",
+    "\u2581For",
+    "ren",
+    "als",
+    "und",
+    "\u2581Al",
+    "\u2581All",
+    "ss",
+    "ows",
+    "\u2581mod",
+    "ock",
+    "\u2581id",
+    "ism",
+    "cus",
+    "\u2581gl",
+    "ably",
+    "\u2581ass",
+    "\u2581car",
+    "ata",
+    "ppen",
+    "led",
+    "\u2581sim",
+    "\u2581mon",
+    "ics",
+    "\u2581giv",
+    "cept",
+    "\u2581Mr",
+    "pan",
+    "\u2581pub",
+    "\u2581eff",
+    "\u2581How",
+    "ps",
+    "vern",
+    "end",
+    "hip",
+    "iew",
+    "ope",
+    "\u2581An",
+    "\u2581She",
+    "\u2581Com",
+    "ee",
+    "ures",
+    "ell",
+    "ouse",
+    "cond",
+    "king",
+    "oc",
+    "ues",
+    "ever",
+    "\u2581To",
+    "clud",
+    "\u2581ins",
+    "\u2581exp",
+    "\u2581old",
+    "\u2581mem",
+    "\u2581ref",
+    "\u2581tra",
+    "\u2581far",
+    "ave",
+    "rat",
+    "\u2581sur",
+    "ruct",
+    "rib",
+    "duct",
+    "uff",
+    "\u2581met",
+    "\u2581sch",
+    "ince",
+    "\u2581run",
+    "ense",
+    "\u2581cle",
+    "\u2581==",
+    "mon",
+    "ize",
+    "\u2581ord",
+    "blem",
+    "tin",
+    "\u2581Let",
+    "ner",
+    "ond",
+    "its",
+    "\u2581cor",
+    "land",
+    "\u2581cur",
+    "\u2581Re",
+    "\u2581bus",
+    "\u2581uh",
+    "air",
+    "ote",
+    "ants",
+    "ason",
+    "ric",
+    "\u2581el",
+    "\u2581cer",
+    "nce",
+    "\u2581fam",
+    "\u2581cap",
+    "uck",
+    "ool",
+    "ried",
+    "\u2581cou",
+    "\u2581fun",
+    "\u2581wom",
+    "\u2581hum",
+    "\u2581ty",
+    "\u2581ap",
+    "ike",
+    "\u2581few",
+    "oney",
+    "\u2581inf",
+    "ont",
+    "ese",
+    "ook",
+    "gy",
+    "uth",
+    "ulat",
+    "ieve",
+    "ized",
+    "ross",
+    "\u2581ple",
+    "\u2581um",
+    "\u2581val",
+    "\u2581equ",
+    "\u2581lea",
+    "\u2581lar",
+    "ah",
+    "eral",
+    "\u2581ed",
+    "ared",
+    "lish",
+    "arn",
+    "ds",
+    "esn",
+    "\u2581iss",
+    "\u2581ca",
+    "ted",
+    "ices",
+    "\u2581wee",
+    "ash",
+    "\u2581top",
+    "ten",
+    "up",
+    "ts",
+    "gin",
+    "con",
+    "ari",
+    "\u2581opp",
+    "osed",
+    "\u2581eas",
+    "\u2581ext",
+    "gg",
+    "az",
+    "\u2581Fr",
+    "ideo",
+    "izat",
+    "\u2581men",
+    "\u2581mom",
+    "\u2581ret",
+    "tty",
+    "rist",
+    "\u2581gra",
+    "alth",
+    "ef",
+    "\u2581det",
+    "ax",
+    "\u2581mat",
+    "chn",
+    "ern",
+    "peri",
+    "\u2581bre",
+    "\u2581Sh",
+    "sw",
+    "erat",
+    "\u2581sit",
+    "ters",
+    "ale",
+    "man",
+    "\u2581sol",
+    "ork",
+    "\u2581adv",
+    "ety",
+    "\u2581vis",
+    "\u2581med",
+    "uc",
+    "less",
+    "\u2581unt",
+    "gram",
+    "ets",
+    "ists",
+    "\u2581ey",
+    "\u2581col",
+    "imes",
+    "\u2581law",
+    "\u2581pri",
+    "sid",
+    "\u2581On",
+    "\u2581mot",
+    "ield",
+    "\u2581Do",
+    "\u2581At",
+    "ages",
+    "amp",
+    "\u2581art",
+    "miss",
+    "\u2581sk",
+    "alf",
+    "pr",
+    "ier",
+    "\u2581beh",
+    "\u2581Yes",
+    "ural",
+    "ime",
+    "\u2581wa",
+    "oks",
+    "bers",
+    "ger",
+    "ient",
+    "ries",
+    "...",
+    "\u2581che",
+    "\u2581Br",
+    "ird",
+    "\u2581Ar",
+    "\u2581war",
+    "inat",
+    "\u2581My",
+    "ital",
+    "wh",
+    "med",
+    "\u2581pur",
+    "ully",
+    "\u2581One",
+    "\u2581rat",
+    "ines",
+    "\u2581Of",
+    "io",
+    "\u2581loc",
+    "ret",
+    "ctor",
+    "\u2581leg",
+    "stit",
+    "ined",
+    "ught",
+    "\u2581dur",
+    "\u2581es",
+    "vent",
+    "aj",
+    "\u2581bro",
+    "\u2581saw",
+    "\u2581sec",
+    "ream",
+    "\u2581pop",
+    "reen",
+    "\u2581Ind",
+    "els",
+    "\u2581yet",
+    "ired",
+    "\u2581sw",
+    "tro",
+    "oup",
+    "most",
+    "pean",
+    "eds",
+    "ush",
+    "oh",
+    "\u2581Se",
+    "\u2581tea",
+    "ann",
+    "ilit",
+    "err",
+    "pend",
+    "ton",
+    "ased",
+    "\u2581aff",
+    "\u2581mor",
+    "\u2581dra",
+    "put",
+    "\u2581dr",
+    "ins",
+    "uat",
+    "nect",
+    "cri",
+    "outh",
+    "\u2581ra",
+    "\u2581pay",
+    "ms",
+    "\u2581av",
+    "bs",
+    "ling",
+    "\u2581De",
+    "\u2581Or",
+    "ove",
+    "\u2581Can",
+    "\u2581eng",
+    "ames",
+    "ided",
+    "\u2581Go",
+    "mitt",
+    "ode",
+    "\u2581cre",
+    "par",
+    "ides",
+    "pos",
+    "\u2581fav",
+    "\u2581air",
+    "\u2581New",
+    "\u2581bad",
+    "\u2581six",
+    "vat",
+    "\u2581pat",
+    "not",
+    "\u2581di",
+    "rop",
+    "ral",
+    "orn",
+    "\u2581par",
+    "cing",
+    "\u2581aw",
+    "orts",
+    "ox",
+    "\u2581yes",
+    "cuss",
+    "eng",
+    "ives",
+    "erms",
+    "\u2581job",
+    "mand",
+    "ying",
+    "\u2581occ",
+    "aps",
+    "ases",
+    "\u2581Not",
+    "rent",
+    "ency",
+    "att",
+    "ised",
+    "vice",
+    "\u2581Eng",
+    "\u2581est",
+    "oked",
+    "\u2581Q",
+    "iron",
+    "idd",
+    "me",
+    "unch",
+    "ane",
+    "\u2581z",
+    "br",
+    "arts",
+    "\u2581fat",
+    "ery",
+    "anks",
+    "\u2581jo",
+    "\u2581mar",
+    "aw",
+    "ott",
+    "ards",
+    "\u2581oh",
+    "ians",
+    "\u2581sci",
+    "row",
+    "unt",
+    "ury",
+    "\u2581abs",
+    "ergy",
+    "\u2581Z",
+    "ump",
+    "\u2581Am",
+    "ened",
+    "angu",
+    "\u2581Pro",
+    "icat",
+    "itch",
+    "\u2581dri",
+    "iat",
+    "\u2581",
+    "e",
+    "t",
+    "o",
+    "a",
+    "n",
+    "i",
+    "s",
+    "r",
+    "h",
+    "l",
+    "d",
+    "u",
+    "c",
+    "m",
+    "y",
+    "g",
+    "w",
+    "f",
+    "p",
+    ",",
+    ".",
+    "b",
+    "v",
+    "k",
+    "'",
+    "I",
+    "T",
+    "A",
+    "S",
+    "x",
+    "W",
+    "j",
+    "C",
+    "B",
+    "M",
+    "?",
+    "H",
+    "O",
+    "0",
+    "P",
+    "q",
+    "Y",
+    "N",
+    "L",
+    "D",
+    "1",
+    "E",
+    "G",
+    "z",
+    "F",
+    "R",
+    "-",
+    "2",
+    "J",
+    "U",
+    "9",
+    "K",
+    "5",
+    "3",
+    "V",
+    "=",
+    "4",
+    "8",
+    "6",
+    "7",
+    "!",
+    "%",
+    ":",
+    "Q",
+    "Z",
+    "$",
+    "X",
+    "\"",
+    "&",
+    "*",
+    "/",
+    "\u00a3",
+    "+",
+    "\u20ac",
+    "_",
+    "^",
+    "\u00a5"
+  ],
+  "target": "nemo.collections.asr.models.rnnt_bpe_models.EncDecRNNTBPEModel",
+  "nemo_version": "2.4.0rc0",
   "quantization": {
     "bits": 4,
     "group_size": 64

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ab5abc14c63e82c9baed231bcdd7061a463da3da30b94df02b87ac0dcfa652ce
-size 351359434
+oid sha256:e9309913fadc9c72a98cf791c9f7066cd012e7e9287dfd1aa893c8a57d66b005
+size 631143202