| { | |
| "_attn_implementation_autoset": true, | |
| "_block_types": [ | |
| "recurrent", | |
| "recurrent", | |
| "attention" | |
| ], | |
| "_name_or_path": "/tmp/models/tmp4ukbr2br/tmpc0zwa1g1", | |
| "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, | |
| "embeddings_scale_by_sqrt_dim": true, | |
| "eos_token_id": 1, | |
| "final_w_init_variance_scale": 0.07692307692307693, | |
| "head_dim": 256, | |
| "hidden_activation": "gelu_pytorch_tanh", | |
| "hidden_size": 2560, | |
| "intermediate_size": 15360, | |
| "logits_soft_cap": 30.0, | |
| "lru_width": 2560, | |
| "model_type": "recurrent_gemma", | |
| "num_attention_heads": 10, | |
| "num_hidden_layers": 26, | |
| "num_key_value_heads": 1, | |
| "pad_token_id": 0, | |
| "partial_rotary_factor": 0.5, | |
| "quantization_config": { | |
| "quant_config": { | |
| "offload_meta": false, | |
| "scale_quant_params": null, | |
| "weight_quant_params": { | |
| "axis": 1, | |
| "channel_wise": true, | |
| "group_size": 64, | |
| "nbits": 8, | |
| "optimize": true, | |
| "round_zero": false, | |
| "view_as_float": false | |
| }, | |
| "zero_quant_params": null | |
| }, | |
| "quant_method": "hqq", | |
| "skip_modules": [ | |
| "lm_head" | |
| ] | |
| }, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 10000.0, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.48.2", | |
| "use_cache": true, | |
| "vocab_size": 256000, | |
| "w_init_variance_scale": 0.01 | |
| } | |