add files

config.json

@@ -0,0 +1,48 @@
+{
+  "activation_dropout": 0.1,
+  "activation_function": "relu",
+  "architectures": [
+    "Speech2TextTransformerForConditionalGeneration"
+  ],
+  "attention_dropout": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": 0.0,
+  "conv_channels": 1024,
+  "conv_kernel_sizes": [
+    5,
+    5
+  ],
+  "d_model": 256,
+  "decoder_attention_heads": 4,
+  "decoder_ffn_dim": 2048,
+  "decoder_layerdrop": 0.0,
+  "decoder_layers": 6,
+  "decoder_start_token_id": 2,
+  "dropout": 0.1,
+  "early_stopping": true,
+  "encoder_attention_heads": 4,
+  "encoder_ffn_dim": 2048,
+  "encoder_layerdrop": 0.0,
+  "encoder_layers": 12,
+  "eos_token_id": 2,
+  "gradient_checkpointing": false,
+  "init_std": 0.02,
+  "input_channels": 1,
+  "input_feat_per_channel": 80,
+  "is_encoder_decoder": true,
+  "max_length": 200,
+  "max_position_embeddings": 1024,
+  "max_source_positions": 6000,
+  "max_target_positions": 1024,
+  "model_type": "speech_to_text_transformer",
+  "num_beams": 5,
+  "num_conv_layers": 2,
+  "num_hidden_layers": 12,
+  "pad_token_id": 1,
+  "scale_embedding": true,
+  "share_embeds": false,
+  "tie_word_embeddings": false,
+  "transformers_version": "4.4.0.dev0",
+  "use_cache": true,
+  "vocab_size": 181
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50642832df4ac676d148bd3e5881c4447a1bf7ce8571e6b9a358eec28ab55ba5
+size 108412503

sentencepiece.bpe.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4880a4424d0a477e623a5d48e4f0d9e51d6c9794d2d3c76630248fa89f88d990
+size 239648

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}

vocab.json

@@ -0,0 +1,183 @@
+{
+  "<s>": 0,
+  "<pad>": 1,
+  "</s>": 2,
+  "<unk>": 3,
+  "\u2581": 4,
+  "e": 5,
+  "n": 6,
+  "i": 7,
+  "r": 8,
+  "t": 9,
+  "s": 10,
+  "a": 11,
+  "d": 12,
+  "h": 13,
+  "u": 14,
+  "l": 15,
+  "o": 16,
+  "c": 17,
+  "g": 18,
+  "m": 19,
+  ".": 20,
+  "b": 21,
+  "f": 22,
+  "w": 23,
+  "k": 24,
+  "z": 25,
+  "S": 26,
+  "v": 27,
+  "p": 28,
+  ",": 29,
+  "D": 30,
+  "\u00fc": 31,
+  "\u00e4": 32,
+  "E": 33,
+  "A": 34,
+  "B": 35,
+  "M": 36,
+  "G": 37,
+  "F": 38,
+  "K": 39,
+  "P": 40,
+  "W": 41,
+  "T": 42,
+  "y": 43,
+  "H": 44,
+  "\u00f6": 45,
+  "I": 46,
+  "R": 47,
+  "L": 48,
+  "-": 49,
+  "C": 50,
+  "V": 51,
+  "N": 52,
+  "\u00df": 53,
+  "\u201e": 54,
+  "Z": 55,
+  "\u201c": 56,
+  "J": 57,
+  "U": 58,
+  "j": 59,
+  "O": 60,
+  "x": 61,
+  "?": 62,
+  "\"": 63,
+  "!": 64,
+  "q": 65,
+  "Y": 66,
+  "\u00dc": 67,
+  ":": 68,
+  "Q": 69,
+  "'": 70,
+  "\u00c4": 71,
+  "\u2018": 72,
+  "\u00d6": 73,
+  ";": 74,
+  "(": 75,
+  ")": 76,
+  "\u2013": 77,
+  "X": 78,
+  "\u201a": 79,
+  "\u201d": 80,
+  "0": 81,
+  "1": 82,
+  "\u00e9": 83,
+  "2": 84,
+  "\u2019": 85,
+  "&": 86,
+  "3": 87,
+  "5": 88,
+  "4": 89,
+  "7": 90,
+  "\u2014": 91,
+  "9": 92,
+  "8": 93,
+  "6": 94,
+  "/": 95,
+  "\u0301": 96,
+  "[": 97,
+  "]": 98,
+  "\u00e1": 99,
+  "\u014d": 100,
+  "\u00f3": 101,
+  "\u00f1": 102,
+  "\u00fa": 103,
+  "\u00ed": 104,
+  "\u0101": 105,
+  "\u00e8": 106,
+  "*": 107,
+  "\u0107": 108,
+  "\u00e0": 109,
+  "\u00ea": 110,
+  "\u00eb": 111,
+  "\u00a1": 112,
+  "\u00e7": 113,
+  "\u00f0": 114,
+  "\u00e3": 115,
+  "\u010d": 116,
+  "\u016b": 117,
+  "%": 118,
+  "\u00c9": 119,
+  "\u00e2": 120,
+  "\u00f8": 121,
+  "\u0161": 122,
+  "`": 123,
+  "\u00ad": 124,
+  "\u00e5": 125,
+  "\u00f4": 126,
+  "\u0142": 127,
+  "\u0153": 128,
+  "\u015f": 129,
+  "\u0160": 130,
+  "_": 131,
+  "\u00ce": 132,
+  "\u00d3": 133,
+  "\u00e6": 134,
+  "\u00ef": 135,
+  "\u0103": 136,
+  "\u011b": 137,
+  "\u012b": 138,
+  "\u0131": 139,
+  "\u02bb": 140,
+  "\u02bf": 141,
+  "\u03c0": 142,
+  "\u0438": 143,
+  "\u043a": 144,
+  "=": 145,
+  "\u00ab": 146,
+  "\u00bb": 147,
+  "\u00c3": 148,
+  "\u00d8": 149,
+  "\u00ee": 150,
+  "\u00fb": 151,
+  "\u00fe": 152,
+  "\u010b": 153,
+  "\u010c": 154,
+  "\u0119": 155,
+  "\u011f": 156,
+  "\u0144": 157,
+  "\u014c": 158,
+  "\u0151": 159,
+  "\u0159": 160,
+  "\u017e": 161,
+  "\u01ce": 162,
+  "\u03b1": 163,
+  "\u0412": 164,
+  "\u0435": 165,
+  "\u0437": 166,
+  "\u0439": 167,
+  "\u043b": 168,
+  "\u043d": 169,
+  "\u044c": 170,
+  "\u044f": 171,
+  "\u1e43": 172,
+  "\u1ea1": 173,
+  "\u1ee5": 174,
+  "\u2192": 175,
+  "\u2261": 176,
+  "\u4eac": 177,
+  "\u5927": 178,
+  "\u90fd": 179,
+  "\u962a": 180
+}