| { | |
| "activation_function": "gelu", | |
| "architectures": [ | |
| "RwkvForCausalLM" | |
| ], | |
| "attention_hidden_size": 32, | |
| "attn_pdrop": 0.1, | |
| "bos_token_id": 0, | |
| "context_length": 1024, | |
| "eos_token_id": 0, | |
| "gradient_checkpointing": false, | |
| "hidden_size": 32, | |
| "intermediate_size": 37, | |
| "is_decoder": true, | |
| "layer_norm_epsilon": 1e-05, | |
| "model_type": "rwkv", | |
| "n_positions": 512, | |
| "num_hidden_layers": 5, | |
| "pad_token_id": 1023, | |
| "reorder_and_upcast_attn": false, | |
| "rescale_every": 6, | |
| "resid_pdrop": 0.1, | |
| "scale_attn_by_inverse_layer_idx": false, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.30.0.dev0", | |
| "type_vocab_size": 16, | |
| "use_cache": true, | |
| "vocab_size": 1024 | |
| } | |