| { | |
| "_name_or_path": "Mxode/NanoTranslator-M2", | |
| "architectures": [ | |
| "Qwen2ForCausalLM" | |
| ], | |
| "attention_dropout": 0.0, | |
| "hidden_act": "gelu", | |
| "hidden_size": 432, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 2304, | |
| "max_position_embeddings": 1024, | |
| "max_window_layers": 24, | |
| "model_type": "qwen2", | |
| "num_attention_heads": 24, | |
| "num_hidden_layers": 6, | |
| "num_key_value_heads": 8, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 10000.0, | |
| "sliding_window": 4096, | |
| "tie_word_embeddings": true, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.42.4", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "vocab_size": 4000 | |
| } | |