| { |
| "architectures": [ |
| "HymbaForCausalLM" |
| ], |
| "attention_window_size": 2048, |
| "auto_map": { |
| "AutoConfig": "configuration_hymba.HymbaConfig", |
| "AutoModelForCausalLM": "modeling_hymba.HymbaForCausalLM" |
| }, |
| "bos_token_id": 1, |
| "conv_kernel_size": 3, |
| "cpu_offload": false, |
| "dtype": null, |
| "eos_token_id": 2, |
| "global_layer_list": [ |
| 5, |
| 11, |
| 18, |
| 25, |
| 31 |
| ], |
| "hidden_size": 1536, |
| "intermediate_size": 3072, |
| "low_memory_mode": false, |
| "mamba_expand": 2, |
| "mlp_hidden_act": "silu", |
| "model_type": "hymba", |
| "modify_attention_mask": true, |
| "num_attention_heads": 12, |
| "num_key_value_heads": 2, |
| "num_layers": 32, |
| "num_meta_tokens": 256, |
| "pad_token_id": 0, |
| "rope_base": 10000, |
| "seq_length": 4096, |
| "ssm_state_size": 16, |
| "time_step_rank": 8, |
| "torch_dtype": "float32", |
| "transformers_version": "4.51.3", |
| "use_cache": true, |
| "use_gradient_checkpointing": false, |
| "use_positional_embedding": true, |
| "vocab_size": 151936 |
| } |
|
|
