Training in progress, step 500

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "nguyenvulebinh/wav2vec2-base-vi-vlsp2020",
+  "_name_or_path": "nguyenvulebinh/wav2vec2-large-vi",
   "activation_dropout": 0.0,
   "adapter_attn_dim": null,
   "adapter_kernel_size": 3,
@@ -9,12 +9,12 @@
   "architectures": [
     "Wav2Vec2ForCTC"
   ],
-  "attention_dropout": 0.0,
+  "attention_dropout": 0.1,
   "bos_token_id": 1,
   "classifier_proj_size": 256,
-  "codevector_dim": 256,
+  "codevector_dim": 768,
   "contrastive_logits_temperature": 0.1,
-  "conv_bias": false,
+  "conv_bias": true,
   "conv_dim": [
     512,
     512,
@@ -45,41 +45,48 @@
   "ctc_loss_reduction": "mean",
   "ctc_zero_infinity": false,
   "diversity_loss_weight": 0.1,
-  "do_stable_layer_norm": false,
+  "do_stable_layer_norm": true,
   "eos_token_id": 2,
   "feat_extract_activation": "gelu",
   "feat_extract_dropout": 0.0,
-  "feat_extract_norm": "group",
-  "feat_proj_dropout": 0.0,
+  "feat_extract_norm": "layer",
+  "feat_proj_dropout": 0.1,
   "feat_quantizer_dropout": 0.0,
   "final_dropout": 0.0,
   "hidden_act": "gelu",
-  "hidden_dropout": 0.3,
-  "hidden_dropout_prob": 0.1,
-  "hidden_size": 768,
+  "hidden_dropout": 0.1,
+  "hidden_size": 1024,
   "initializer_range": 0.02,
-  "intermediate_size": 3072,
+  "intermediate_size": 4096,
   "layer_norm_eps": 1e-05,
   "layerdrop": 0.1,
+  "mask_channel_length": 10,
+  "mask_channel_min_space": 1,
+  "mask_channel_other": 0.0,
+  "mask_channel_prob": 0.0,
+  "mask_channel_selection": "static",
   "mask_feature_length": 10,
   "mask_feature_min_masks": 0,
   "mask_feature_prob": 0.0,
   "mask_time_length": 10,
   "mask_time_min_masks": 2,
-  "mask_time_prob": 0.05,
+  "mask_time_min_space": 1,
+  "mask_time_other": 0.0,
+  "mask_time_prob": 0.075,
+  "mask_time_selection": "static",
   "model_type": "wav2vec2",
   "num_adapter_layers": 3,
-  "num_attention_heads": 12,
+  "num_attention_heads": 16,
   "num_codevector_groups": 2,
   "num_codevectors_per_group": 320,
   "num_conv_pos_embedding_groups": 16,
   "num_conv_pos_embeddings": 128,
   "num_feat_extract_layers": 7,
-  "num_hidden_layers": 12,
+  "num_hidden_layers": 24,
   "num_negatives": 100,
-  "output_hidden_size": 768,
+  "output_hidden_size": 1024,
   "pad_token_id": 109,
-  "proj_codevector_dim": 256,
+  "proj_codevector_dim": 768,
   "tdnn_dilation": [
     1,
     2,

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0d7b9b021632dcf71be459c18c1b8149ab55aa377db6c43f712219f0438c1991
-size 377904865
+oid sha256:e632d662029c4418a0a5df5566d40870c7e1d433e7c045df9c1bf3758f1cf1f1
+size 1262361069

runs/Sep09_06-03-17_6fc807377663/events.out.tfevents.1694240047.6fc807377663.28.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b47adb2082d23d37ce525e38489eb9e18e110f589886400027cf8dd04f472e4
+size 6193

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6c24fec51db3cf3308efe767fe4df0a28fe6c66dbf9f996428f7976aab7d2a11
+oid sha256:e0d933fa2df79d271b4d76ed50177f0582d75a0f0c33cff5039eee45626815b9
 size 3963