| { |
| "architectures": [ |
| "H3ForCausalLM" |
| ], |
| "attn_layer_idx": [ |
| 8, |
| 16 |
| ], |
| "bos_token_id": 50256, |
| "embedding_dropout": 0.1, |
| "eos_token_id": 50256, |
| "glu_act": false, |
| "hidden_size": 2048, |
| "initializer_range": 0.02, |
| "layer_norm_epsilon": 1e-05, |
| "model_type": "h3", |
| "n_inner": null, |
| "num_attention_heads": 16, |
| "num_hidden_layers": 24, |
| "rescale_prenorm_residual": true, |
| "residual_dropout": 0.0, |
| "residual_in_fp32": false, |
| "ssm_measure": "diag-lin", |
| "ssm_mode": "diag", |
| "torch_dtype": "float32", |
| "transformers_version": "4.28.0.dev0", |
| "vocab_size": 50264 |
| } |
|
|