tiny-random-HYV3Model / config.json
hf-transformers-bot's picture
Update tiny models for HYV3Model
e10de14 verified
Raw
History Blame Contribute Delete
971 Bytes
{
"architectures": [
"HYV3Model"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": 120000,
"dtype": "float32",
"enable_moe_fp32_combine": true,
"eos_token_id": 120025,
"head_dim": 16,
"hidden_act": "gelu",
"hidden_size": 32,
"initializer_range": 0.02,
"intermediate_size": 32,
"max_position_embeddings": 512,
"mlp_bias": false,
"mlp_layer_types": [
"dense",
"sparse"
],
"model_type": "hy_v3",
"moe_intermediate_size": 16,
"num_attention_heads": 2,
"num_experts": 8,
"num_experts_per_tok": 2,
"num_hidden_layers": 2,
"num_key_value_heads": 2,
"num_shared_experts": 1,
"output_router_logits": false,
"pad_token_id": 120002,
"rms_norm_eps": 1e-05,
"rope_parameters": {
"rope_theta": 11158840.0,
"rope_type": "default"
},
"router_scaling_factor": 2.826,
"tie_word_embeddings": false,
"transformers_version": "5.16.0.dev0",
"use_cache": true,
"vocab_size": 120818
}