{ | |
"d_model": 2048, | |
"ssm_cfg": { | |
"expand": 1 | |
}, | |
"rms_norm_eps": 1e-05, | |
"vocab_size": null, | |
"d_inner": null, | |
"d_xb": 512, | |
"intermediate_size": 8192, | |
"hidden_act": "silu", | |
"n_layer": 16, | |
"attn_layers": [ | |
3, | |
7, | |
11, | |
15 | |
] | |
} |
{ | |
"d_model": 2048, | |
"ssm_cfg": { | |
"expand": 1 | |
}, | |
"rms_norm_eps": 1e-05, | |
"vocab_size": null, | |
"d_inner": null, | |
"d_xb": 512, | |
"intermediate_size": 8192, | |
"hidden_act": "silu", | |
"n_layer": 16, | |
"attn_layers": [ | |
3, | |
7, | |
11, | |
15 | |
] | |
} |