| { | |
| "architectures": [ | |
| "SMDMForCausalLM" | |
| ], | |
| "model_type": "smdm", | |
| "vocab_size": 32000, | |
| "n_embd": 2048, | |
| "n_layer": 22, | |
| "n_head": 32, | |
| "n_query_groups": 32, | |
| "intermediate_size": 5632, | |
| "hidden_dropout_prob": 0.0, | |
| "attention_probs_dropout_prob": 0.0, | |
| "block_size": 2048, | |
| "initializer_range": 0.02, | |
| "norm_eps": 1e-5, | |
| "use_cache": true, | |
| "rotary_percentage": 1.0, | |
| "condense_ratio": 1, | |
| "parallel_residual": true, | |
| "shared_attention_norm": false, | |
| "bias": true, | |
| "bos_token_id": 1, | |
| "eos_token_id": 2, | |
| "pad_token_id": 0, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "float16", | |
| "transformers_version": "4.31.0" | |
| } |