| { | |
| "cfg": { | |
| "d_in": 768, | |
| "d_sae": 12288, | |
| "k": 32, | |
| "k_aux": 256, | |
| "dead_after": 2000000, | |
| "aux_coef": 0.03125 | |
| }, | |
| "layer": 8, | |
| "scale": 0.10184143483638763, | |
| "model_run": "/data/runs/500M_ctx8k/model.pt" | |
| } |
| { | |
| "cfg": { | |
| "d_in": 768, | |
| "d_sae": 12288, | |
| "k": 32, | |
| "k_aux": 256, | |
| "dead_after": 2000000, | |
| "aux_coef": 0.03125 | |
| }, | |
| "layer": 8, | |
| "scale": 0.10184143483638763, | |
| "model_run": "/data/runs/500M_ctx8k/model.pt" | |
| } |