24 lines
537 B
JSON
24 lines
537 B
JSON
|
|
{
|
||
|
|
"vocab_size": 128000,
|
||
|
|
"hidden_size": 2048,
|
||
|
|
"intermediate_size": 4096,
|
||
|
|
"num_hidden_layers": 28,
|
||
|
|
"num_attention_heads": 32,
|
||
|
|
"num_key_value_heads": 8,
|
||
|
|
"max_position_embeddings": 4096,
|
||
|
|
"rope_theta": 500000.0,
|
||
|
|
"rms_norm_eps": 1e-05,
|
||
|
|
"initializer_range": 0.02,
|
||
|
|
"tie_word_embeddings": true,
|
||
|
|
"attention_dropout": 0.0,
|
||
|
|
"residual_dropout": 0.0,
|
||
|
|
"gradient_checkpointing": true,
|
||
|
|
"pad_token_id": 0,
|
||
|
|
"bos_token_id": 1,
|
||
|
|
"eos_token_id": 2,
|
||
|
|
"model_type": "jumini",
|
||
|
|
"architectures": [
|
||
|
|
"JuminiForCausalLM"
|
||
|
|
]
|
||
|
|
}
|