| { | |
| "model_type": "metanova_1", | |
| "model_name": "metanova-1", | |
| "architectures": [ | |
| "TinyGPTForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_tiny_gpt.TinyGPTConfig", | |
| "AutoModel": "modeling_tiny_gpt.TinyGPTModel", | |
| "AutoModelForCausalLM": "modeling_tiny_gpt.TinyGPTForCausalLM" | |
| }, | |
| "vocab_size": 32000, | |
| "ctx_len": 3072, | |
| "max_position_embeddings": 3072, | |
| "n_layer": 12, | |
| "num_hidden_layers": 12, | |
| "n_head": 8, | |
| "num_attention_heads": 8, | |
| "n_embd": 512, | |
| "hidden_size": 512, | |
| "intermediate_size": 1408, | |
| "mlp_ratio": 2.75, | |
| "dropout": 0.0, | |
| "attention_backend": "torch", | |
| "available_attention_backends": [ | |
| "sage", | |
| "torch", | |
| "flash2", | |
| "flash3" | |
| ], | |
| "trained_attention_backend": "flash3", | |
| "torch_fallback": true, | |
| "positional_encoding": "rope", | |
| "trained_positional_encoding": "rope", | |
| "rope_base": 10000.0, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "custom", | |
| "pad_token_id": 0, | |
| "sep_token_id": 2, | |
| "unk_token_id": 1, | |
| "bos_token_id": 5, | |
| "eos_token_id": 6 | |
| } | |