| { | |
| "architectures": [ | |
| "RecurrentGemmaForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "attention_window_size": 2048, | |
| "block_types": [ | |
| "recurrent", | |
| "recurrent", | |
| "attention" | |
| ], | |
| "bos_token_id": 2, | |
| "conv1d_width": 4, | |
| "eos_token_id": 1, | |
| "final_w_init_variance_scale": 0.05263157894736842, | |
| "head_dim": 256, | |
| "hidden_activation": "gelu_pytorch_tanh", | |
| "hidden_size": 4096, | |
| "intermediate_size": 24576, | |
| "logits_soft_cap": 30.0, | |
| "lru_width": 4096, | |
| "model_type": "recurrent_gemma", | |
| "num_attention_heads": 16, | |
| "num_hidden_layers": 38, | |
| "num_key_value_heads": 1, | |
| "pad_token_id": 0, | |
| "partial_rotary_factor": 0.5, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 10000.0, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.42.0.dev0", | |
| "use_cache": true, | |
| "vocab_size": 256000, | |
| "w_init_variance_scale": 0.01 | |
| } | |