clean up puncts; train longer

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "facebook/wav2vec2-large-xlsr-53",
+  "_name_or_path": "/workspace/checkpoints/rw/wav2vec2-large-xlsr-rw/checkpoint-15360",
   "activation_dropout": 0.05,
   "apply_spec_augment": true,
   "architectures": [
@@ -62,7 +62,7 @@
   "mask_time_length": 10,
   "mask_time_min_space": 1,
   "mask_time_other": 0.0,
-  "mask_time_prob": 0.05,
+  "mask_time_prob": 0.1,
   "mask_time_selection": "static",
   "model_type": "wav2vec2",
   "num_attention_heads": 16,
@@ -70,7 +70,7 @@
   "num_conv_pos_embeddings": 128,
   "num_feat_extract_layers": 7,
   "num_hidden_layers": 24,
-  "pad_token_id": 62,
+  "pad_token_id": 29,
   "transformers_version": "4.5.0.dev0",
-  "vocab_size": 63
+  "vocab_size": 32
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:91c5d0f15305b61780b33114710f3a1f57b704f35927d16f80e9900a148d51bb
-size 1262192151
+oid sha256:48b68303a54e7bc0086817f76c8202880355390ce8cddccd7a41baab2443ed70
+size 1262065047

vocab.json

@@ -1 +1 @@
-{"q": 0, "ê": 1, "d": 2, "ç": 3, "е": 4, "o": 5, "ö": 6, "©": 7, "u": 8, "&": 9, "y": 10, "ā": 11, "í": 12, "ô": 13, "l": 14, "с": 15, "k": 16, "â": 17, "f": 18, "a": 19, "z": 20, "½": 21, "s": 22, "è": 23, "p": 24, "á": 25, "w": 26, "c": 27, "t": 28, "i": 29, "™": 30, "а": 31, "і": 32, "x": 33, "à": 34, "r": 35, "e": 36, "å": 37, "€": 38, "b": 39, "$": 40, "ñ": 41, "v": 42, "ã": 43, "é": 44, "ü": 45, "ï": 46, "в": 47, "h": 48, "g": 50, "j": 51, "ú": 52, "m": 53, "ë": 54, "¼": 55, "ğ": 56, "ō": 57, "+": 58, "°": 59, "n": 60, "|": 49, "[UNK]": 61, "[PAD]": 62}
+{"x": 0, "h": 1, "c": 2, "f": 3, "r": 4, "d": 5, "e": 6, "t": 7, "'": 8, "q": 9, "k": 10, "w": 11, "u": 12, "m": 13, "i": 14, "j": 15, "y": 16, "l": 17, "v": 18, "g": 19, "b": 20, "a": 21, "z": 23, "p": 24, "n": 25, "s": 26, "o": 27, "|": 22, "[UNK]": 28, "[PAD]": 29}