| { |
| "model_type": "gtransformer", |
| "architectures": ["GTransformerForCausalLM"], |
| "vocab_size": 65536, |
| "hidden_size": 8192, |
| "intermediate_size": 22016, |
| "num_hidden_layers": 48, |
| "num_attention_heads": 64, |
| "hidden_act": "swiglu", |
| "max_position_embeddings": 65536, |
| "initializer_range": 0.02, |
| "layer_norm_epsilon": 1e-5, |
| "attention_dropout": 0.05, |
| "hidden_dropout_prob": 0.05, |
| "rotary_emb_base": 10000, |
| "use_flash_attention": true, |
| "use_low_rank_ffn": true, |
| "use_entropy_gate": true, |
| "fp8_precision": true, |
| "dvfs_enabled": true, |
| "informational_constant_kI": 2.612e-20, |
| "energy_per_token_target_J": 0.07, |
| "training_objective": "cross_entropy + lambda*(I_total - I_useful)", |
| "optimizer": "adamw_torch", |
| "gradient_checkpointing": true, |
| "torch_dtype": "float16", |
| "transformers_version": "4.44.2", |
| "tokenizer_class": "GTransformerTokenizer", |
| "bos_token_id": 1, |
| "eos_token_id": 2, |
| "pad_token_id": 0, |
| "tie_word_embeddings": false, |
| "use_cache": true, |
| "model_revision": "main", |
| "git_informational_mode": { |
| "enable": true, |
| "delta_I_gate": 0.75, |
| "thermal_threshold_T": 2.3, |
| "energy_scaling_mode": "adaptive" |
| } |
| } |
|
|
