| { | |
| "architectures": [ | |
| "ZambaForCausalLM" | |
| ], | |
| "attention_dropout": 0.0, | |
| "attn_layer_offset": 4, | |
| "attn_layer_period": 6, | |
| "bos_token_id": 1, | |
| "eos_token_id": 2, | |
| "hidden_act": "gelu", | |
| "hidden_mamba_act": "silu", | |
| "hidden_size": 3712, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 14848, | |
| "mamba_conv_bias": true, | |
| "mamba_d_conv": 4, | |
| "mamba_d_state": 16, | |
| "mamba_dt_rank": 232, | |
| "mamba_expand": 2, | |
| "mamba_proj_bias": false, | |
| "max_position_embeddings": 4096, | |
| "model_type": "zamba", | |
| "n_mamba_heads": 2, | |
| "num_attention_heads": 16, | |
| "num_hidden_layers": 76, | |
| "num_key_value_heads": 16, | |
| "num_logits_to_keep": 1, | |
| "rms_norm_eps": 1e-05, | |
| "rope_theta": 10000, | |
| "sliding_window": null, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.42.0.dev0", | |
| "unk_token_id": 0, | |
| "use_cache": true, | |
| "use_mamba_kernels": true, | |
| "vocab_size": 32000 | |
| } | |