| { | |
| "architectures": ["KimiForCausalLM"], | |
| "auto_map": { | |
| "AutoConfig": "configuration_kimi.KimiConfig", | |
| "AutoModelForCausalLM": "modeling_kimi.KimiForCausalLM", | |
| "AutoTokenizer": "tokenization_kimi.TikTokenTokenizer" | |
| }, | |
| "bos_token": "[BOS]", | |
| "eos_token": "[EOS]", | |
| "hidden_size": 4096, | |
| "intermediate_size": 11008, | |
| "max_position_embeddings": 32768, | |
| "model_type": "kimi", | |
| "num_attention_heads": 32, | |
| "num_hidden_layers": 32, | |
| "num_key_value_heads": 8, | |
| "pad_token": "[PAD]", | |
| "rope_scaling": { | |
| "factor": 2.0, | |
| "type": "linear" | |
| }, | |
| "rope_theta": 10000.0, | |
| "rms_norm_eps": 1e-06, | |
| "sliding_window": 4096, | |
| "tie_word_embeddings": false, | |
| "tokenizer_class": "TikTokenTokenizer", | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "5.0.0", | |
| "use_cache": true, | |
| "vocab_size": 32000 | |
| } | |