| { | |
| "architectures": [ | |
| "NSAForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_nsa.NSAConfig", | |
| "AutoModelForCausalLM": "modeling_nsa.NSAForCausalLM" | |
| }, | |
| "d_k": 64, | |
| "d_v": 64, | |
| "dtype": "float32", | |
| "eos_token_id": 0, | |
| "hidden_size": 768, | |
| "max_position_embeddings": 2048, | |
| "model_type": "nsa", | |
| "n_kv_groups": 2, | |
| "nsa": { | |
| "block": 32, | |
| "branches": [ | |
| "cmp", | |
| "sel", | |
| "win" | |
| ], | |
| "gqa_groups": 2, | |
| "sel_block": 64, | |
| "sel_top_n": 16, | |
| "stride": 16, | |
| "window": 512 | |
| }, | |
| "num_attention_heads": 12, | |
| "num_hidden_layers": 12, | |
| "pad_token_id": 0, | |
| "rope_theta": 10000, | |
| "transformers_version": "4.56.0", | |
| "vocab_size": 256 | |
| } | |