nvidia/Efficient-DLM-4B
28489
1{2 "adaptive_mask_rate": false,3 "architectures": [4 "EfficientDLM"5 ],6 "attention_bias": false,7 "attention_dropout": 0.0,8 "attn_implementation": "sdpa",9 "auto_map": {10 "AutoConfig": "configuration_edlm.EfficientDLMConfig",11 "AutoModel": "modeling_edlm.EfficientDLM"12 },13 "block_size": 32,14 "diff_loss_weight": 1,15 "disable_qk_norm": false,16 "dlm_arch": "encoder",17 "dlm_paradigm": "bidirectional",18 "dlm_type": "llada",19 "enforce_mask": false,20 "head_dim": 128,21 "hidden_act": "silu",22 "hidden_size": 2560,23 "initializer_range": 0.02,24 "intermediate_size": 9728,25 "intl_mask": false,26 "mask_token_id": 151662,27 "max_position_embeddings": 32768,28 "max_window_layers": 28,29 "model_type": "qwen3",30 "multi_sampling": null,31 "num_ar_layers": 0,32 "num_attention_heads": 32,33 "num_diffusion_layers": 0,34 "num_hidden_layers": 36,35 "num_key_value_heads": 8,36 "prefix_ratio": 0.8,37 "random_length_prob": 0,38 "rms_norm_eps": 1e-06,39 "rope_scaling": null,40 "rope_theta": 1000000,41 "sliding_window": null,42 "tie_word_embeddings": false,43 "tok_mask_half_life_ratio": null,44 "torch_dtype": "bfloat16",45 "transformers_version": "4.52.2",46 "use_cache": false,47 "use_sliding_window": false,48 "vocab_size": 15193649}50 