Invalid JSON:Unexpected token 'N', ..."al_loss": NaN,
"to"... is not valid JSON
| { | |
| "vocab_size": 50304, | |
| "d_model": 3072, | |
| "n_layers": 48, | |
| "n_heads": 24, | |
| "n_kv_heads": 8, | |
| "max_len": 2048, | |
| "rope_theta": 5000000.0, | |
| "ffn_hidden": 8192, | |
| "use_moe": true, | |
| "moe_experts": 64, | |
| "moe_top_k": 8, | |
| "moe_shared": 2, | |
| "moe_hidden": 1024, | |
| "moe_every": 3, | |
| "moe_bias_speed": 0.001, | |
| "hec_junior_frac": 0.5, | |
| "hec_senior_frac": 0.3, | |
| "hec_principal_frac": 0.2, | |
| "n_code_types": 8, | |
| "use_ter": true, | |
| "ter_confidence_threshold": 0.85, | |
| "ter_min_layers": 6, | |
| "ter_gate_hidden": 256, | |
| "use_sas": true, | |
| "sas_tier_breaks": [ | |
| 12, | |
| 30, | |
| 44 | |
| ], | |
| "sas_tier_freqs": [ | |
| 1, | |
| 3, | |
| 6, | |
| 12 | |
| ], | |
| "use_stm": true, | |
| "stm_every": 8, | |
| "stm_merge_threshold": 0.92, | |
| "stm_min_tokens": 64, | |
| "use_psm": true, | |
| "psm_slots": 256, | |
| "psm_dim": 512, | |
| "psm_read_heads": 4, | |
| "psm_write_heads": 2, | |
| "use_mrcv": true, | |
| "mrcv_micro_window": 1, | |
| "mrcv_meso_window": 16, | |
| "mrcv_macro_window": 128, | |
| "use_fda": true, | |
| "fda_head_frac": 0.25, | |
| "use_dsv": true, | |
| "dsv_verifier_layers": 4, | |
| "dsv_threshold": 0.6, | |
| "use_ccd": true, | |
| "ccd_high_thresh": 0.9, | |
| "ccd_low_thresh": 0.5, | |
| "ccd_n_candidates": 5, | |
| "use_bct": true, | |
| "bct_rate": 0.15, | |
| "bct_bug_types": 6, | |
| "use_gv": true, | |
| "gv_noise_std": 0.01, | |
| "gv_max_noise": 0.05, | |
| "gv_warmup_frac": 0.1, | |
| "use_egm": true, | |
| "egm_bank_size": 1024, | |
| "egm_similarity_thresh": 0.8, | |
| "egm_correction_mult": 3.0, | |
| "use_rgt": true, | |
| "rgt_rate": 0.2, | |
| "use_egl": true, | |
| "egl_keyword_weight": 2.0, | |
| "egl_whitespace_weight": 0.3, | |
| "use_mcp": true, | |
| "mcp_every_steps": 500, | |
| "mcp_replay_batches": 10, | |
| "use_qk_norm": true, | |
| "use_structural_bias": true, | |
| "n_struct_rel": 4, | |
| "use_fim": true, | |
| "fim_rate": 0.5, | |
| "fim_pre_id": 50269, | |
| "fim_suf_id": 50270, | |
| "fim_mid_id": 50271, | |
| "grad_checkpointing": true, | |
| "lr": 0.0003, | |
| "min_lr": 3e-05, | |
| "weight_decay": 0.1, | |
| "beta1": 0.9, | |
| "beta2": 0.95, | |
| "grad_clip": 1.0, | |
| "warmup_tokens": 500000, | |
| "target_total_tokens": 10000000000, | |
| "micro_batch": 4, | |
| "grad_accum": 8, | |
| "log_every": 10, | |
| "ckpt_every_steps": 200, | |
| "eval_every_steps": 200, | |
| "seed": 42, | |
| "max_session_hours": 11.5, | |
| "final_step": 0, | |
| "final_epoch": 0, | |
| "final_val_loss": NaN, | |
| "total_tokens": 0 | |
| } |