| { | |
| "Transformer": { | |
| "encoder": { | |
| "sentence_embedding": { | |
| "embedding": { | |
| "num_embeddings": 71, | |
| "embedding_dim": 512 | |
| }, | |
| "position_encoder": {}, | |
| "dropout": 0.1 | |
| }, | |
| "layers": [ | |
| { | |
| "attention": { | |
| "qkv_layer": { | |
| "in_features": 512, | |
| "out_features": 1536, | |
| "bias": true | |
| }, | |
| "linear_layer": { | |
| "in_features": 512, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "norm1": {}, | |
| "dropout1": 0.1, | |
| "ffn": { | |
| "linear1": { | |
| "in_features": 512, | |
| "out_features": 2048, | |
| "bias": true | |
| }, | |
| "linear2": { | |
| "in_features": 2048, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "norm2": {}, | |
| "dropout2": 0.1 | |
| }, | |
| { | |
| "attention": { | |
| "qkv_layer": { | |
| "in_features": 512, | |
| "out_features": 1536, | |
| "bias": true | |
| }, | |
| "linear_layer": { | |
| "in_features": 512, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "norm1": {}, | |
| "dropout1": 0.1, | |
| "ffn": { | |
| "linear1": { | |
| "in_features": 512, | |
| "out_features": 2048, | |
| "bias": true | |
| }, | |
| "linear2": { | |
| "in_features": 2048, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "norm2": {}, | |
| "dropout2": 0.1 | |
| }, | |
| { | |
| "attention": { | |
| "qkv_layer": { | |
| "in_features": 512, | |
| "out_features": 1536, | |
| "bias": true | |
| }, | |
| "linear_layer": { | |
| "in_features": 512, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "norm1": {}, | |
| "dropout1": 0.1, | |
| "ffn": { | |
| "linear1": { | |
| "in_features": 512, | |
| "out_features": 2048, | |
| "bias": true | |
| }, | |
| "linear2": { | |
| "in_features": 2048, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "norm2": {}, | |
| "dropout2": 0.1 | |
| } | |
| ] | |
| }, | |
| "decoder": { | |
| "sentence_embedding": { | |
| "embedding": { | |
| "num_embeddings": 125, | |
| "embedding_dim": 512 | |
| }, | |
| "position_encoder": {}, | |
| "dropout": 0.1 | |
| }, | |
| "layers": [ | |
| { | |
| "self_attention": { | |
| "qkv_layer": { | |
| "in_features": 512, | |
| "out_features": 1536, | |
| "bias": true | |
| }, | |
| "linear_layer": { | |
| "in_features": 512, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "layer_norm1": {}, | |
| "dropout1": 0.1, | |
| "encoder_decoder_attention": { | |
| "kv_layer": { | |
| "in_features": 512, | |
| "out_features": 1024, | |
| "bias": true | |
| }, | |
| "q_layer": { | |
| "in_features": 512, | |
| "out_features": 512, | |
| "bias": true | |
| }, | |
| "linear_layer": { | |
| "in_features": 512, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "layer_norm2": {}, | |
| "dropout2": 0.1, | |
| "ffn": { | |
| "linear1": { | |
| "in_features": 512, | |
| "out_features": 2048, | |
| "bias": true | |
| }, | |
| "linear2": { | |
| "in_features": 2048, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "layer_norm3": {}, | |
| "dropout3": 0.1 | |
| }, | |
| { | |
| "self_attention": { | |
| "qkv_layer": { | |
| "in_features": 512, | |
| "out_features": 1536, | |
| "bias": true | |
| }, | |
| "linear_layer": { | |
| "in_features": 512, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "layer_norm1": {}, | |
| "dropout1": 0.1, | |
| "encoder_decoder_attention": { | |
| "kv_layer": { | |
| "in_features": 512, | |
| "out_features": 1024, | |
| "bias": true | |
| }, | |
| "q_layer": { | |
| "in_features": 512, | |
| "out_features": 512, | |
| "bias": true | |
| }, | |
| "linear_layer": { | |
| "in_features": 512, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "layer_norm2": {}, | |
| "dropout2": 0.1, | |
| "ffn": { | |
| "linear1": { | |
| "in_features": 512, | |
| "out_features": 2048, | |
| "bias": true | |
| }, | |
| "linear2": { | |
| "in_features": 2048, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "layer_norm3": {}, | |
| "dropout3": 0.1 | |
| }, | |
| { | |
| "self_attention": { | |
| "qkv_layer": { | |
| "in_features": 512, | |
| "out_features": 1536, | |
| "bias": true | |
| }, | |
| "linear_layer": { | |
| "in_features": 512, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "layer_norm1": {}, | |
| "dropout1": 0.1, | |
| "encoder_decoder_attention": { | |
| "kv_layer": { | |
| "in_features": 512, | |
| "out_features": 1024, | |
| "bias": true | |
| }, | |
| "q_layer": { | |
| "in_features": 512, | |
| "out_features": 512, | |
| "bias": true | |
| }, | |
| "linear_layer": { | |
| "in_features": 512, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "layer_norm2": {}, | |
| "dropout2": 0.1, | |
| "ffn": { | |
| "linear1": { | |
| "in_features": 512, | |
| "out_features": 2048, | |
| "bias": true | |
| }, | |
| "linear2": { | |
| "in_features": 2048, | |
| "out_features": 512, | |
| "bias": true | |
| } | |
| }, | |
| "layer_norm3": {}, | |
| "dropout3": 0.1 | |
| } | |
| ] | |
| }, | |
| "linear": { | |
| "in_features": 512, | |
| "out_features": 125, | |
| "bias": true | |
| } | |
| } | |
| } | |