| { | |
| "architectures": [ | |
| "Mamba3CausalLM" | |
| ], | |
| "model_type": "mamba3", | |
| "auto_map": { | |
| "AutoConfig": "modeling_mamba3.Mamba3Config", | |
| "AutoModelForCausalLM": "modeling_mamba3.Mamba3CausalLM" | |
| }, | |
| "d_model": 4096, | |
| "n_layer": 32, | |
| "vocab_size": 32768, | |
| "d_state": 128, | |
| "expand": 2, | |
| "headdim": 64, | |
| "ngroups": 1, | |
| "rope_fraction": 0.5, | |
| "is_safe_A": true, | |
| "tie_embeddings": true, | |
| "torch_dtype": "bfloat16", | |
| "source_model": "mistralai/Mistral-7B-Instruct-v0.3", | |
| "heist_type": "mistral_embedding_transplant" | |
| } |