lnp_ml/scripts/fixed_hparams.json

8 lines
331 B
JSON

{
"dropout": 0.33, "lr": 0.0005, "weight_decay": 0.0001, "backbone_lr_ratio": 0.3,
"moe_n_experts": 4, "moe_top_k": 2, "moe_expert_hidden_mult": 1,
"llm_lora_r": 8,
"d_model": 256, "num_heads": 8, "n_attn_layers": 4,
"fusion_strategy": "attention", "head_hidden_dim": 128,
"set_transformer_block": "sab"
}