{ | |
"architectures": [ | |
"DiffusionLLM" | |
], | |
"attention_probs_dropout_prob": 0.1, | |
"eos_token_id": 50256, | |
"hidden_dropout_prob": 0.1, | |
"hidden_size": 256, | |
"initializer_range": 0.02, | |
"intermediate_size": 3072, | |
"layer_norm_eps": 1e-12, | |
"mask_token_id": 50257, | |
"max_position_embeddings": 256, | |
"model_type": "diffusionLM", | |
"num_attention_heads": 4, | |
"num_hidden_layers": 4, | |
"num_timesteps": 100, | |
"pad_token_id": 50258, | |
"time_embed_dim": 128, | |
"torch_dtype": "float32", | |
"transformers_version": "4.48.3", | |
"vocab_size": 50259 | |
} | |