config.json
846 B · 35 lines · json Raw
1 {
2 "activation_dropout": 0.0,
3 "activation_function": "relu",
4 "architectures": [
5 "M2M100ForConditionalGeneration"
6 ],
7 "attention_dropout": 0.1,
8 "bos_token_id": 0,
9 "d_model": 1024,
10 "decoder_attention_heads": 16,
11 "decoder_ffn_dim": 4096,
12 "decoder_layerdrop": 0,
13 "decoder_layers": 12,
14 "decoder_start_token_id": 2,
15 "dropout": 0.1,
16 "encoder_attention_heads": 16,
17 "encoder_ffn_dim": 4096,
18 "encoder_layerdrop": 0,
19 "encoder_layers": 12,
20 "eos_token_id": 2,
21 "init_std": 0.02,
22 "is_encoder_decoder": true,
23 "max_position_embeddings": 1024,
24 "model_type": "m2m_100",
25 "num_hidden_layers": 12,
26 "pad_token_id": 1,
27 "scale_embedding": true,
28 "torch_dtype": "float32",
29 "transformers_version": "4.21.0.dev0",
30 "use_cache": true,
31 "vocab_size": 256206,
32 "tokenizer_class": "NllbTokenizer",
33 "max_length": 200
34 }
35