Training in progress, step 400

added_tokens.json

@@ -1,4 +1,4 @@
 {
-  "</s>": 44,
-  "<s>": 43
+  "</s>": 40,
+  "<s>": 39
 }

config.json

@@ -1,12 +1,13 @@
 {
   "_name_or_path": "ylacombe/w2v-bert-2.0",
   "activation_dropout": 0.0,
+  "adapter_act": "relu",
   "adapter_kernel_size": 3,
   "adapter_stride": 2,
   "add_adapter": true,
   "apply_spec_augment": false,
   "architectures": [
-    "Wav2Vec2BERTForCTC"
+    "Wav2Vec2BertForCTC"
   ],
   "attention_dropout": 0.0,
   "bos_token_id": 1,
@@ -16,7 +17,7 @@
   "contrastive_logits_temperature": 0.1,
   "conv_depthwise_kernel_size": 31,
   "ctc_loss_reduction": "mean",
-  "ctc_zero_infinity": false,
+  "ctc_zero_infinity": true,
   "diversity_loss_weight": 0.1,
   "eos_token_id": 2,
   "feat_proj_dropout": 0.0,
@@ -46,7 +47,7 @@
   "num_hidden_layers": 24,
   "num_negatives": 100,
   "output_hidden_size": 1024,
-  "pad_token_id": 42,
+  "pad_token_id": 38,
   "position_embeddings_type": "relative_key",
   "proj_codevector_dim": 768,
   "right_max_position_embeddings": 8,
@@ -76,6 +77,6 @@
   "transformers_version": "4.37.0.dev0",
   "use_intermediate_ffn_before_adapter": false,
   "use_weighted_layer_sum": false,
-  "vocab_size": 45,
+  "vocab_size": 41,
   "xvector_output_dim": 512
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5600e15d85cd8ffffdb0ff8dfd191738b49b52b5f867bca8154cfb7d9aa8120d
-size 2422999060
+oid sha256:402995db86f95097cfa076e02367acc469a241885744074124832242aeba68a1
+size 2422982660

preprocessor_config.json

@@ -1,11 +1,10 @@
 {
-  "do_normalize": true,
-  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_extractor_type": "SeamlessM4TFeatureExtractor",
   "feature_size": 80,
   "num_mel_bins": 80,
   "padding_side": "right",
   "padding_value": 1,
-  "processor_class": "Wav2Vec2Processor",
+  "processor_class": "Wav2Vec2BertProcessor",
   "return_attention_mask": true,
   "sampling_rate": 16000,
   "stride": 2

runs/Jan15_14-05-01_vorace/events.out.tfevents.1705327546.vorace.343655.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:615ab56e90cda16e37513cc0336e578c4532a54c4f92e57e90956b280bb65a90
+size 4136

runs/Jan15_14-09-37_vorace/events.out.tfevents.1705327823.vorace.344138.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac92050bdb781c16e946dee95f11fca3cc883ef6dbf8572bc4be8398460e5c60
+size 4136

runs/Jan15_14-13-12_vorace/events.out.tfevents.1705328027.vorace.344504.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a25cb677c5b089e5660c6ae93531eceaf6cdbf571303dd2b09d370afd72aeb0f
+size 6648

runs/Jan15_14-28-43_vorace/events.out.tfevents.1705329032.vorace.347376.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ec06b9bacd698152fbbda2fe9742b62bede44317d954d3a07a3062ae603aa42
+size 4136

runs/Jan15_15-02-19_vorace/events.out.tfevents.1705331015.vorace.348037.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:322330b6e54dfdccf83e665265ee90e434f1c7b65075462dfa0dbd7be67e282f
+size 6912

tokenizer_config.json

@@ -1,6 +1,6 @@
 {
   "added_tokens_decoder": {
-    "41": {
+    "37": {
       "content": "[UNK]",
       "lstrip": true,
       "normalized": false,
@@ -8,7 +8,7 @@
       "single_word": false,
       "special": false
     },
-    "42": {
+    "38": {
       "content": "[PAD]",
       "lstrip": true,
       "normalized": false,
@@ -16,7 +16,7 @@
       "single_word": false,
       "special": false
     },
-    "43": {
+    "39": {
       "content": "<s>",
       "lstrip": false,
       "normalized": false,
@@ -24,7 +24,7 @@
       "single_word": false,
       "special": true
     },
-    "44": {
+    "40": {
       "content": "</s>",
       "lstrip": false,
       "normalized": false,
@@ -39,7 +39,7 @@
   "eos_token": "</s>",
   "model_max_length": 1000000000000000019884624838656,
   "pad_token": "[PAD]",
-  "processor_class": "Wav2Vec2Processor",
+  "processor_class": "Wav2Vec2BertProcessor",
   "replace_word_delimiter_char": " ",
   "target_lang": null,
   "tokenizer_class": "Wav2Vec2CTCTokenizer",

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c59160a475baad18bee490cd0c48f5e0db7fcbb10dbd0a3bf1ab3076b53beb12
+oid sha256:55faed694c10f073dceb228b3f6f82f891efab95cf41709bee3e5c14337451b2
 size 4728

vocab.json

@@ -1,6 +1,6 @@
 {
-  "[PAD]": 42,
-  "[UNK]": 41,
+  "[PAD]": 38,
+  "[UNK]": 37,
   "a": 1,
   "b": 2,
   "c": 3,
@@ -30,16 +30,12 @@
   "|": 0,
   "â": 27,
   "ç": 28,
-  "é": 29,
-  "ë": 30,
-  "î": 31,
-  "ö": 32,
-  "û": 33,
-  "ü": 34,
-  "ğ": 35,
-  "ı": 36,
-  "ş": 37,
-  "̇": 38,
-  "’": 39,
-  "…": 40
+  "ë": 29,
+  "î": 30,
+  "ö": 31,
+  "ü": 32,
+  "ğ": 33,
+  "ı": 34,
+  "ş": 35,
+  "̇": 36
 }