Training in progress, epoch 0

config.json

@@ -7,19 +7,19 @@
   "d_kv": 64,
   "d_model": 1024,
   "decoder_start_token_id": 0,
-  "dense_act_fn": "relu",
+  "dense_act_fn": "gelu_new",
   "dropout_rate": 0.1,
   "dtype": "float32",
   "eos_token_id": 1,
-  "feed_forward_proj": "relu",
+  "feed_forward_proj": "gated-gelu_new",
   "initializer_factor": 1.0,
   "is_encoder_decoder": true,
-  "is_gated_act": false,
+  "is_gated_act": true,
   "layer_norm_epsilon": 1e-06,
   "lookahead_size": 2,
   "lookahead_type": "la",
   "model_type": "t5la",
-  "num_decoder_layers": 6,
+  "num_decoder_layers": 24,
   "num_heads": 16,
   "num_layers": 24,
   "pad_token_id": 0,