| { |
| "trainer": { |
| "trainer_class": "WTATrainer", |
| "dict_class": "WTASAE", |
| "lr": 0.000282842712474619, |
| "steps": 12207, |
| "auxk_alpha": 0.03125, |
| "seed": 42, |
| "activation_dim": 1024, |
| "dict_size": 8192, |
| "sparsity_rate": 5e-06, |
| "device": "cuda", |
| "layer": 3, |
| "lm_name": "EleutherAI/pythia-410m-deduped", |
| "wandb_name": "WTATrainer-EleutherAI/pythia-410m-deduped-resid_post_layer_3_trainer_0", |
| "submodule_name": "resid_post_layer_3", |
| "warmup_steps": 1000, |
| "decay_start": null, |
| "threshold_beta": 0.999, |
| "threshold_start_step": 1000 |
| }, |
| "buffer": { |
| "d_submodule": 1024, |
| "io": "out", |
| "n_ctxs": 1280, |
| "ctx_len": 128, |
| "refresh_batch_size": 512, |
| "out_batch_size": 8192, |
| "device": "cuda" |
| } |
| } |
