| { | |
| "architectures": [ | |
| "VanFastForCausalLM" | |
| ], | |
| "block_size": 1024, | |
| "bos_token_id": 50256, | |
| "d_ff": 4096, | |
| "d_model": 1024, | |
| "dropout": 0.0, | |
| "dtype": "float32", | |
| "eos_token_id": 50256, | |
| "initializer_range": 0.02, | |
| "is_decoder": true, | |
| "model_type": "van_fast_transformer", | |
| "n_head": 16, | |
| "n_kv_head": 4, | |
| "n_layer": 18, | |
| "pad_token_id": 50256, | |
| "tie_word_embeddings": false, | |
| "transformers_version": "5.0.0", | |
| "use_qk_norm": true, | |
| "vocab_size": 50257, | |
| "auto_map": { | |
| "AutoConfig": "modeling_van_fast.VanFastConfig", | |
| "AutoModelForCausalLM": "modeling_van_fast.VanFastForCausalLM" | |
| }, | |
| "torch_dtype": "bfloat16", | |
| "use_cache": true | |
| } |