| { |
| "T": 0, |
| "architectures": [ |
| "E2D2" |
| ], |
| "attn_backend": "sdpa", |
| "auto_map": { |
| "AutoConfig": "diffusion.E2D2Config", |
| "AutoModel": "diffusion.E2D2", |
| "AutoModelForMaskedLM": "diffusion.E2D2" |
| }, |
| "backbone_config": { |
| "_target_": "backbone_encoder_decoder.LLMasEncoderDecoder", |
| "attn_backend": "sdpa", |
| "freeze_encoder": false, |
| "hidden_size": 768, |
| "intermediate_size": 3072, |
| "keep_top_decoder_layers": false, |
| "keep_top_encoder_layers": false, |
| "max_length": 1024, |
| "num_decoder_layers": 2, |
| "num_encoder_layers": 10, |
| "pretrained_model_name_or_path": "Qwen/Qwen3-0.6B-Base", |
| "reinit_decoder": true, |
| "reinit_encoder": true, |
| "tie_encoder_decoder_weights": false, |
| "use_encoder_causal_mask": false, |
| "use_gradient_checkpointing": false, |
| "vocab_size": 50258 |
| }, |
| "block_size": 4, |
| "bos_token_id": 50256, |
| "diffusion_type": "absorbing", |
| "eos_token_id": 50256, |
| "eval_block_size": 4, |
| "keep_clean_bos": true, |
| "length": 1024, |
| "mask_token_id": 50257, |
| "model_type": "e2d2", |
| "noise_config": { |
| "_target_": "noise_schedule_noise_schedules.LinearNoise" |
| }, |
| "pad_token_id": 50256, |
| "pad_vocab_size_multiple": 1, |
| "shift_logits": false, |
| "time_conditioned_backbone": false, |
| "tokenization_config": { |
| "bos_token_id": 50256, |
| "eos_token_id": 50256, |
| "mask_token_id": 50257, |
| "pad_token_id": 50256, |
| "pad_vocab_size_multiple": 1, |
| "vocab_size": 50258 |
| }, |
| "tokenizer_name": "gpt2", |
| "torch_dtype": "float32", |
| "train_on_context": false, |
| "transformers_version": "4.52.4", |
| "vocab_size": 50258 |
| } |
|
|
