| { | |
| "model_type": "mirage", | |
| "architecture": { | |
| "text": 8192, | |
| "dim": 512, | |
| "kv_heads": 2, | |
| "qk_dim": 48, | |
| "v_dim": 64, | |
| "dec_layers": 8, | |
| "taps": 3, | |
| "lanes": 4, | |
| "sites": [ | |
| 3, | |
| 7 | |
| ], | |
| "orders": [ | |
| 2, | |
| 3 | |
| ], | |
| "slots": 18432, | |
| "hashes": 2, | |
| "rope": 100000.0, | |
| "canvas": 512, | |
| "stem": [ | |
| 32, | |
| 64, | |
| 128, | |
| 256, | |
| 512 | |
| ], | |
| "body_layers": 16, | |
| "hconv_kernel": 65, | |
| "mix_kernel": 3, | |
| "attn_layers": 12, | |
| "attn_heads": 4, | |
| "attn_kv_heads": 1, | |
| "attn_qk_dim": 32, | |
| "attn_v_dim": 64, | |
| "enc_rope": 10000.0, | |
| "registers": 4, | |
| "pooled_slots": true, | |
| "grn": true, | |
| "neck": true, | |
| "level_memory": true, | |
| "conv_norm": true, | |
| "max_tokens": 1536, | |
| "buckets": [ | |
| 512, | |
| 1024, | |
| 1536 | |
| ], | |
| "form_share": 0.1, | |
| "trace_share": 0.2, | |
| "name_loss": 0.5, | |
| "seg_loss": 0.0, | |
| "detections": 64, | |
| "regions": 32, | |
| "reads": 8, | |
| "heat_alpha": 0.54, | |
| "head_grad": 0.1, | |
| "stride": 8, | |
| "dense_dim": 128, | |
| "reader_dim": 384, | |
| "reader_blocks": 4, | |
| "read_width": 320, | |
| "pad": 0, | |
| "eos": 1, | |
| "bos": 2, | |
| "dtype": "float32" | |
| }, | |
| "training_step": 1100000, | |
| "source_checkpoint_sha256": "dc8db0c4f1a9f516a9080224514de7b6b85193a8c6648a35b0f0486c5bc25224", | |
| "cact_bits": 2, | |
| "cact_bits_map": { | |
| "embedding": 4, | |
| "mhc": 4, | |
| "default": 2 | |
| } | |
| } | |