82 lines
2.0 KiB
JSON
82 lines
2.0 KiB
JSON
|
|
{
|
||
|
|
"architectures": [
|
||
|
|
"Kanana2TinyForCausalLM"
|
||
|
|
],
|
||
|
|
"attention_bias": false,
|
||
|
|
"attention_dropout": 0.0,
|
||
|
|
"auto_map": {
|
||
|
|
"AutoConfig": "configuration_kanana2_tiny.Kanana2TinyConfig",
|
||
|
|
"AutoModel": "modeling_kanana2_tiny.Kanana2TinyModel",
|
||
|
|
"AutoModelForCausalLM": "modeling_kanana2_tiny.Kanana2TinyForCausalLM"
|
||
|
|
},
|
||
|
|
"bos_token_id": 128000,
|
||
|
|
"dtype": "bfloat16",
|
||
|
|
"eos_token_id": 128010,
|
||
|
|
"head_dim": 128,
|
||
|
|
"hidden_act": "silu",
|
||
|
|
"hidden_size": 1280,
|
||
|
|
"initializer_range": 0.02,
|
||
|
|
"intermediate_size": 5760,
|
||
|
|
"layer_types": [
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"full_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"full_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"full_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"full_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"full_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"full_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"full_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"sliding_attention",
|
||
|
|
"full_attention"
|
||
|
|
],
|
||
|
|
"max_position_embeddings": 32768,
|
||
|
|
"max_window_layers": 32,
|
||
|
|
"model_type": "kanana2_tiny",
|
||
|
|
"num_attention_heads": 32,
|
||
|
|
"num_hidden_layers": 32,
|
||
|
|
"num_key_value_heads": 8,
|
||
|
|
"pad_token_id": 128001,
|
||
|
|
"rms_norm_eps": 1e-06,
|
||
|
|
"rope_parameters": {
|
||
|
|
"full_attention": {
|
||
|
|
"factor": 40.0,
|
||
|
|
"original_max_position_embeddings": 4096,
|
||
|
|
"rope_theta": 10000,
|
||
|
|
"rope_type": "yarn"
|
||
|
|
},
|
||
|
|
"sliding_attention": {
|
||
|
|
"rope_theta": 10000.0,
|
||
|
|
"rope_type": "default"
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"rope_scaling": null,
|
||
|
|
"rope_theta": 10000,
|
||
|
|
"sliding_window": 1024,
|
||
|
|
"transformers_version": "4.57.1",
|
||
|
|
"use_cache": true,
|
||
|
|
"use_sliding_window": true,
|
||
|
|
"vocab_size": 128256
|
||
|
|
}
|