{ "architecture": "Foundry4", "config": { "vocab_size": 65536, "n_specials": 16, "d_model": 1280, "n_layers": 24, "n_heads": 10, "head_dim": 128, "seq_len": 4096, "dropout": 0.0, "rope_theta": 10000.0, "qk_norm": true, "attn_every": 4, "gdn_heads": 10, "gdn_head_k": 128, "gdn_head_v": 128, "gdn_chunk": 32, "n_experts": 384, "expert_hidden": 128, "moe_topk": 8, "shared_hidden": 640, "lb_weight": 0.01, "kg_buckets": 262144, "kg_dim": 80, "kg_max_n": 4, "kg_probes": 2, "ponder_iters": 4, "ponder_hidden": 2560, "ponder_iter_weight_decay": 1.0, "judge_hidden": 512, "use_mtp": true, "mtp_weight": 0.05, "moe_ckpt": false }, "trained_steps": 2000, "dtype": "bfloat16" }