| { | |
| "d_model": 3072, | |
| "ssm_cfg": { | |
| "expand": 1 | |
| }, | |
| "rms_norm_eps": 1e-05, | |
| "vocab_size": null, | |
| "d_inner": null, | |
| "d_xb": 1024, | |
| "intermediate_size": 8192, | |
| "hidden_act": "silu", | |
| "n_layer": 28, | |
| "attn_layers": [ | |
| 3, | |
| 8, | |
| 13, | |
| 18, | |
| 23, | |
| 27 | |
| ] | |
| } |
| { | |
| "d_model": 3072, | |
| "ssm_cfg": { | |
| "expand": 1 | |
| }, | |
| "rms_norm_eps": 1e-05, | |
| "vocab_size": null, | |
| "d_inner": null, | |
| "d_xb": 1024, | |
| "intermediate_size": 8192, | |
| "hidden_act": "silu", | |
| "n_layer": 28, | |
| "attn_layers": [ | |
| 3, | |
| 8, | |
| 13, | |
| 18, | |
| 23, | |
| 27 | |
| ] | |
| } |