config.json

{
  "architectures": [
    "DiffusionLMForMaskedLM"
  ],
  "attention_dropout": 0.0,
  "bos_token_id": 1,
  "dtype": "float32",
  "eos_token_id": 2,
  "head_dim": 64,
  "hidden_size": 512,
  "intermediate_size": 1536,
  "mask_token_id": 32768,
  "max_position_embeddings": 2048,
  "model_type": "diffusion_lm",
  "num_attention_heads": 8,
  "num_diffusion_steps": 64,
  "num_hidden_layers": 10,
  "pad_token_id": 0,
  "rms_norm_eps": 1e-05,
  "rope_theta": 10000.0,
  "tie_word_embeddings": true,
  "time_conditioning": "none",
  "time_conditioning_scale": 0.02,
  "training_objective": "absorbing-linear-1overT-v2",
  "transformers_version": "5.17.0",
  "vocab_size": 32768,
  "auto_map": {
    "AutoConfig": "configuration_diffusion_lm.DiffusionLMConfig",
    "AutoModelForMaskedLM": "modeling_diffusion_lm.DiffusionLMForMaskedLM"
  }
}