TRM-text-v4 / config.json
summerMC's picture
Upload fine-tuned model v4 (Stable check)
a26ead1 verified
Raw
History Blame Contribute Delete
788 Bytes
{
"architectures": [
"TRMMoEForCausalLM"
],
"auto_map": {
"AutoConfig": "configuration_trm_moe.TRMMoEConfig",
"AutoModelForCausalLM": "modeling_trm_moe.TRMMoEForCausalLM"
},
"bos_token_id": 50256,
"dim": 768,
"dropout": 0.0,
"dtype": "bfloat16",
"eos_token_id": 50256,
"gate_init": -1.5,
"gate_style": "stable",
"head_dim": 64,
"max_seq_len": 512,
"mlp_hidden_size": null,
"mlp_ratio": 2.6666666667,
"model_type": "trm_moe",
"moe_aux_loss_coeff": 0.01,
"moe_z_loss_coeff": 0.001,
"n_heads": 12,
"num_experts": 32,
"num_experts_per_token": 2,
"num_hidden_layers": 1,
"pad_token_id": 50256,
"recurrence_steps": 4,
"residual_scale": 0.5,
"tie_word_embeddings": true,
"transformers_version": "5.10.2",
"vocab_size": 50259
}