update to pretraining

config.json

@@ -2,10 +2,12 @@
   "activation_dropout": 0.0,
   "apply_spec_augment": true,
   "architectures": [
-    "Wav2Vec2Model"
+    "Wav2Vec2ForPreTraining"
   ],
   "attention_dropout": 0.1,
   "bos_token_id": 1,
+  "codevector_dim": 768,
+  "contrastive_logits_temperature": 0.1,
   "conv_bias": true,
   "conv_dim": [
     512,
@@ -36,12 +38,14 @@
   ],
   "ctc_loss_reduction": "sum",
   "ctc_zero_infinity": false,
+  "diversity_loss_weight": 0.1,
   "do_stable_layer_norm": true,
   "eos_token_id": 2,
   "feat_extract_activation": "gelu",
   "feat_extract_dropout": 0.0,
   "feat_extract_norm": "layer",
   "feat_proj_dropout": 0.1,
+  "feat_quantizer_dropout": 0.0,
   "final_dropout": 0.0,
   "gradient_checkpointing": false,
   "hidden_act": "gelu",
@@ -65,11 +69,15 @@
   "mask_time_selection": "static",
   "model_type": "wav2vec2",
   "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": 24,
+  "num_negatives": 100,
   "pad_token_id": 0,
-  "transformers_version": "4.6.0.dev0",
+  "proj_codevector_dim": 768,
+  "transformers_version": "4.7.0.dev0",
   "vocab_size": 32
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1e00a85284dd3fb3adc5b29e5d1652c74f77e903929caa3e1ddbe0124d31d059
-size 1261920069
+oid sha256:49aeba6f083455b382d4ec01e6c592e33612419b37b68b7b9fa26c3e4bf1729f
+size 1269737156