HuskyDoge's picture
Initial release
2cdd3ba
Raw History Blame Contribute Delete
2.39 kB
{
"model": {
"apply_attn_gate": false,
"apply_bias_term": true,
"apply_rmsnorm": true,
"arch": "huginn",
"attention_dropout": 0.0,
"attn_gate_func": "silu",
"causal_attn_backend": "flash",
"chunk_size": 1024,
"dense_prelude_diagonal_a_init": 1.0,
"dense_prelude_diagonal_dt_init": 1.0,
"dense_prelude_input_injection": "none",
"dense_prelude_input_layers": "",
"dense_prelude_source_layer": -1,
"dropout": 0.0,
"expert_inter_dim": 0,
"ffn_hidden_dim": 4352,
"fused_block": true,
"fused_output_layer": true,
"head_dim": 64,
"hidden_dropout": 0.0,
"huginn_coda_layers": 8,
"huginn_diagonal_a_init": 1.0,
"huginn_diagonal_dt_init": 1.0,
"huginn_hierarchical_h_cycles": 2,
"huginn_hierarchical_l_cycles": 3,
"huginn_hierarchical_state": "split_hl",
"huginn_input_injection": "diagonal",
"huginn_prelude_layers": 8,
"huginn_recurrent_layers": 12,
"huginn_state_init": "normal",
"init_embed_std": null,
"init_logits_std": null,
"init_mode": "gaussian",
"init_std": 0.02,
"layernorm_num_groups": 1,
"layerwise_ckpt": false,
"loop_diagonal_a_init": 1.0,
"loop_diagonal_dt_init": 1.0,
"loop_end_layers": 28,
"loop_input_injection": "none",
"loop_start_layer": 0,
"loop_times": 8,
"memory_efficient_norm": false,
"model_dim": 1536,
"moe_expert_backend": "sequential",
"moe_router_bias": false,
"moe_router_bias_update_rate": null,
"moe_router_scaling_factor": null,
"moe_router_score_func": "sigmoid",
"norm_affine": true,
"num_activated_experts": 0,
"num_activated_values": 0,
"num_dense_layers": null,
"num_experts": 0,
"num_heads": 24,
"num_kv_heads": 6,
"num_layers": 28,
"num_shared_experts": 0,
"num_values": 0,
"output_size": -1,
"qknorm": false,
"recompute_attention": true,
"recompute_fc1_out": true,
"recompute_fc3_out": true,
"recompute_logits": false,
"recompute_q": true,
"recompute_router": false,
"recompute_v": true,
"rmsnorm_eps": 1e-06,
"rope_base": 1000000.0,
"rope_head_dim": 64,
"scale_emb": false,
"swiglu": true,
"vocab_size": 250624
},
"tokenizer": {
"num_reserved_special_tokens": 0,
"path": "tokenizer",
"template_dict_json": null,
"type": "huggingface"
}
}