| { |
| "status": "all_target_full_attention_layers_accepted", |
| "architecture": "functiongemma-memory-fusion-sequential-v1", |
| "base_model": "vtava/functiongemma-270m-it-simple-tool-calling", |
| "target_layers": [ |
| 5, |
| 11, |
| 17 |
| ], |
| "accepted_layers": [ |
| 5, |
| 11, |
| 17 |
| ], |
| "feature_dim": 32, |
| "memory_rank": 64, |
| "teacher_probe_nll": 8.525523066520691, |
| "final_probe_nll": 6.230323672294617, |
| "final_probe_delta_nll": -2.295199394226074, |
| "structure": { |
| "memory_fusion_layers": [ |
| 5, |
| 11, |
| 17 |
| ], |
| "remaining_full_attention_layers": [], |
| "sliding_attention_layers": [ |
| 0, |
| 1, |
| 2, |
| 3, |
| 4, |
| 6, |
| 7, |
| 8, |
| 9, |
| 10, |
| 12, |
| 13, |
| 14, |
| 15, |
| 16 |
| ] |
| }, |
| "layer_reports": [ |
| { |
| "layer": 5, |
| "accepted": false, |
| "steps": 50, |
| "nmse": 0.39167162775993347, |
| "cosine": 0.77734375, |
| "probe_nll": 6.578028440475464, |
| "incremental_delta_nll": -1.947494626045227, |
| "cumulative_delta_nll": -1.947494626045227, |
| "round": 1 |
| }, |
| { |
| "layer": 5, |
| "accepted": false, |
| "steps": 175, |
| "nmse": 0.24159154295921326, |
| "cosine": 0.8691575527191162, |
| "probe_nll": 6.600935220718384, |
| "incremental_delta_nll": -1.9245878458023071, |
| "cumulative_delta_nll": -1.9245878458023071, |
| "round": 2 |
| }, |
| { |
| "layer": 5, |
| "accepted": true, |
| "steps": 225, |
| "nmse": 0.15754102170467377, |
| "cosine": 0.922616183757782, |
| "probe_nll": 6.462509989738464, |
| "incremental_delta_nll": -2.0630130767822266, |
| "cumulative_delta_nll": -2.0630130767822266, |
| "round": 3 |
| }, |
| { |
| "layer": 11, |
| "accepted": true, |
| "steps": 150, |
| "nmse": 0.16000066697597504, |
| "cosine": 0.9214716553688049, |
| "probe_nll": 6.223196983337402, |
| "incremental_delta_nll": -0.239313006401062, |
| "cumulative_delta_nll": -2.3023260831832886, |
| "round": 1 |
| }, |
| { |
| "layer": 17, |
| "accepted": false, |
| "steps": 300, |
| "nmse": 0.15326528251171112, |
| "cosine": 0.9228067398071289, |
| "probe_nll": 6.2936941385269165, |
| "incremental_delta_nll": 0.07049715518951416, |
| "cumulative_delta_nll": -2.2318289279937744, |
| "round": 1 |
| }, |
| { |
| "layer": 17, |
| "accepted": false, |
| "steps": 225, |
| "nmse": 0.11938813328742981, |
| "cosine": 0.9484158754348755, |
| "probe_nll": 6.2485164403915405, |
| "incremental_delta_nll": 0.025319457054138184, |
| "cumulative_delta_nll": -2.2770066261291504, |
| "round": 2 |
| }, |
| { |
| "layer": 17, |
| "accepted": true, |
| "steps": 225, |
| "nmse": 0.13086512684822083, |
| "cosine": 0.9407033920288086, |
| "probe_nll": 6.230323672294617, |
| "incremental_delta_nll": 0.0071266889572143555, |
| "cumulative_delta_nll": -2.295199394226074, |
| "round": 3 |
| } |
| ], |
| "elapsed_minutes": 24.377013247816663 |
| } |
