{ "oom-14.25B": { "cass_state_dim": 1024, "ffn_dim": 16384, "hidden_size": 6144, "n_kv_heads": 8, "n_recurrent": 1, "num_attention_heads": 48, "num_logical_layers": 43, "num_physical_layers": 36, "params_B": 14.250608642, "status": "oom-single-A100-AdamW8bit", "vocab_size": 32768 }, "production-13.08B": { "cass_state_dim": 1024, "ffn_dim": 16384, "hidden_size": 6144, "n_kv_heads": 8, "n_recurrent": 1, "num_attention_heads": 48, "num_logical_layers": 41, "num_physical_layers": 33, "params_B": 13.080360962, "status": "default-long-runs", "vocab_size": 32768 }, "ref-9.96B": { "cass_state_dim": 1024, "ffn_dim": 16384, "hidden_size": 6144, "n_kv_heads": 8, "n_recurrent": 1, "num_attention_heads": 48, "num_logical_layers": 30, "num_physical_layers": 25, "params_B": 9.959700482, "status": "reference", "vocab_size": 32768 }, "safe-13.47B": { "cass_state_dim": 1024, "ffn_dim": 16384, "hidden_size": 6144, "n_kv_heads": 8, "n_recurrent": 1, "num_attention_heads": 48, "num_logical_layers": 42, "num_physical_layers": 34, "params_B": 13.470443522, "status": "squeeze-with-safety", "vocab_size": 32768 } }