| { | |
| "architectures": [ | |
| "TRMMoEForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_trm_moe.TRMMoEConfig", | |
| "AutoModelForCausalLM": "modeling_trm_moe.TRMMoEForCausalLM" | |
| }, | |
| "model_type": "trm_moe", | |
| "vocab_size": 50259, | |
| "dim": 768, | |
| "n_heads": 12, | |
| "head_dim": 64, | |
| "mlp_ratio": 2.6666666667, | |
| "mlp_hidden_size": null, | |
| "num_hidden_layers": 1, | |
| "recurrence_steps": 4, | |
| "residual_scale": 0.5, | |
| "gate_style": "stable", | |
| "gate_init": -1.5, | |
| "dropout": 0.0, | |
| "max_seq_len": 512, | |
| "bos_token_id": 50256, | |
| "eos_token_id": 50256, | |
| "pad_token_id": 50256, | |
| "num_experts": 32, | |
| "num_experts_per_token": 2, | |
| "moe_aux_loss_coeff": 0.01, | |
| "moe_z_loss_coeff": 0.001, | |
| "tie_word_embeddings": true, | |
| "transformers_version": "4.40.0" | |
| } |