vistec-en-th-pytorch / config.json
dewdue's picture
Upload FSMTForConditionalGeneration
a71be04
{
"_name_or_path": "/Users/dew/Desktop/OpenThai/language-image-pretraining/outputs/fsmt_scb/SCB_1M+TBASE_en-th_moses-spm_130000-16000_v1.0_pytorch",
"activation_dropout": 0.0,
"activation_function": "relu",
"architectures": [
"FSMTForConditionalGeneration"
],
"attention_dropout": 0.0,
"bos_token_id": 0,
"d_model": 512,
"decoder": {
"bos_token_id": 2,
"model_type": "fsmt_decoder",
"vocab_size": 16000
},
"decoder_attention_heads": 8,
"decoder_ffn_dim": 2048,
"decoder_layerdrop": 0,
"decoder_layers": 6,
"decoder_start_token_id": 2,
"dropout": 0.3,
"encoder_attention_heads": 8,
"encoder_ffn_dim": 2048,
"encoder_layerdrop": 0,
"encoder_layers": 6,
"eos_token_id": 2,
"forced_eos_token_id": 2,
"init_std": 0.02,
"is_encoder_decoder": true,
"langs": [
"en",
"th"
],
"max_length": 200,
"max_position_embeddings": 1024,
"model_type": "fsmt",
"num_beams": 5,
"num_hidden_layers": 6,
"pad_token_id": 1,
"scale_embedding": true,
"src_vocab_size": 130000,
"tgt_vocab_size": 16000,
"tie_word_embeddings": false,
"torch_dtype": "float32",
"transformers_version": "4.32.0",
"use_cache": true
}