# Samantha Thought-Loop SFT

- Base model: `google/t5gemma-l-l-prefixlm-it`
- Cleaned dataset repo: `BRlkl/timed_natural_conversations_5min4`
- Raw dataset repo: `BRlkl/timed_natural_conversations_5min4`
- Latent slots: `256`
- Fixed tick seconds: `1.0`
- Chunk ticks: `300`
- Max horizon ticks: `300`
- Explicit time features: `False`
- Gate loss mode: `bce`
- Hazard soft target window: `16`
- Hazard soft target tau ticks: `4.0`
- Hazard censor weight: `0.25`
- TBPTT by bucket: `{}`

## Best Validation Metrics

- `active_ticks`: `1888.000000`
- `decoder_token_count`: `2494.000000`
- `gate_accuracy`: `0.715042`
- `gate_f1`: `0.154088`
- `gate_positive_weight`: `16.000000`
- `gate_precision`: `0.091932`
- `gate_pred_positive_count`: `533.000000`
- `gate_recall`: `0.475728`
- `gate_target_positive_count`: `103.000000`
- `hazard_censor_count`: `0.000000`
- `hazard_event_count`: `0.000000`
- `loss_decoder`: `3.564287`
- `loss_gate`: `1.241260`
- `loss_gate_ranking`: `0.000000`
- `loss_state_delta`: `0.258536`
- `loss_total`: `4.521086`
- `soft_target_mass`: `0.000000`
- `soft_target_positive_count`: `0.000000`