| { | |
| "model_path": "trained_gpt2_clts/tied-per-target-skip/tied_per_target_skip_global_batchtopk_jumprelu", | |
| "config": { | |
| "num_features": 122880, | |
| "num_layers": 12, | |
| "d_model": 768, | |
| "model_name": null, | |
| "normalization_method": "mean_std", | |
| "activation_fn": "jumprelu", | |
| "jumprelu_threshold": 0.0, | |
| "batchtopk_k": null, | |
| "batchtopk_straight_through": false, | |
| "topk_k": null, | |
| "topk_straight_through": true, | |
| "topk_mode": "global", | |
| "two_stage_batchtopk": false, | |
| "two_stage_topk": false, | |
| "clt_dtype": null, | |
| "expected_input_dtype": null, | |
| "mlp_input_template": null, | |
| "mlp_output_template": null, | |
| "tl_input_template": null, | |
| "tl_output_template": null, | |
| "decoder_tying": "per_target", | |
| "enable_feature_offset": false, | |
| "enable_feature_scale": false, | |
| "skip_connection": true | |
| }, | |
| "layer_stats": { | |
| "0": { | |
| "avg_l0": 4.6845703125, | |
| "max_l0": 45.0 | |
| }, | |
| "1": { | |
| "avg_l0": 3.1630859375, | |
| "max_l0": 56.0 | |
| }, | |
| "2": { | |
| "avg_l0": 6.1865234375, | |
| "max_l0": 232.0 | |
| }, | |
| "3": { | |
| "avg_l0": 4.947265625, | |
| "max_l0": 26.0 | |
| }, | |
| "4": { | |
| "avg_l0": 5.6328125, | |
| "max_l0": 83.0 | |
| }, | |
| "5": { | |
| "avg_l0": 4.9423828125, | |
| "max_l0": 189.0 | |
| }, | |
| "6": { | |
| "avg_l0": 8.0615234375, | |
| "max_l0": 240.0 | |
| }, | |
| "7": { | |
| "avg_l0": 9.8701171875, | |
| "max_l0": 210.0 | |
| }, | |
| "8": { | |
| "avg_l0": 12.326171875, | |
| "max_l0": 332.0 | |
| }, | |
| "9": { | |
| "avg_l0": 21.96875, | |
| "max_l0": 624.0 | |
| }, | |
| "10": { | |
| "avg_l0": 32.90234375, | |
| "max_l0": 306.0 | |
| }, | |
| "11": { | |
| "avg_l0": 44.5126953125, | |
| "max_l0": 521.0 | |
| } | |
| } | |
| } |