| { | |
| "model": { | |
| "baseline_d_ff": 4704, | |
| "chart_read_dim": 256, | |
| "constituent_categories": 64, | |
| "context_length": 8192, | |
| "d_ff": 4096, | |
| "d_model": 1024, | |
| "dependency_relations": 64, | |
| "dropout": 0.0, | |
| "events_per_segment": 2, | |
| "local_query_chunk": 256, | |
| "local_window": 1024, | |
| "max_event_role_memory_slots": 1024, | |
| "max_primitive_memory_slots": 256, | |
| "max_span_width": 12, | |
| "num_heads": 16, | |
| "num_layers": 24, | |
| "replacement_stage": "R0", | |
| "segment_size": 256, | |
| "spans_per_segment": 24, | |
| "vocab_size": 48029 | |
| }, | |
| "name": "strata-ph-core-v2-dense-base-8k", | |
| "optimizer": { | |
| "betas": [ | |
| 0.9, | |
| 0.95 | |
| ], | |
| "distributed": "pure_ddp", | |
| "grad_clip": 1.0, | |
| "learning_rate": 5e-05, | |
| "precision": "bf16", | |
| "warmup_fraction": 0.02, | |
| "weight_decay": 0.1 | |
| }, | |
| "seed": 20260718, | |
| "token_budget": 100000000, | |
| "tokenizer": "tokenizer/strata_sp48k.model", | |
| "training": { | |
| "gradient_checkpointing": true, | |
| "log_every": 10, | |
| "long_range_acquisition_fraction": 0.1, | |
| "save_every_tokens": 25000000 | |
| } | |
| } | |