diff --git a/config.json b/config.json index d29c483..0330b86 100644 --- a/config.json +++ b/config.json @@ -2,7 +2,7 @@ "activation_dropout": 0.0, "activation_function": "gelu", "add_bias_logits": false, - "add_final_layer_norm": false, + "add_final_layer_norm": true, "architectures": [ "BlenderbotForConditionalGeneration" ], @@ -42,13 +42,13 @@ "model_type": "blenderbot", "no_repeat_ngram_size": 3, "normalize_before": true, - "normalize_embedding": true, + "normalize_embedding": false, "num_beams": 10, "num_hidden_layers": 2, "pad_token_id": 0, "scale_embedding": true, "static_position_embeddings": false, "unk_token_id": 3, - "variant": "prelayernorm", + "layernorm_variant": "prelayernorm", "vocab_size": 8008 }