{ "adaptive": true, "attn_type": 0, "clamp_len": 1000, "cutoffs": [], "d_embed": 256, "d_head": 64, "d_inner": 2048, "d_model": 256, "div_val": 1, "dropatt": 0.0, "dropout": 0.1, "eos_token_id": 0, "init": "normal", "init_range": 0.01, "init_std": 0.02, "layer_norm_epsilon": 1e-05, "mem_len": 1600, "model_type": "transfo-xl", "n_head": 4, "n_layer": 6, "pre_lnorm": false, "proj_init_std": 0.01, "same_length": true, "sample_softmax": -1, "tie_projs": [ false ], "tie_word_embeddings": false, "transformers_version": "4.31.0.dev0", "untie_r": true, "vocab_size": 500 }