| { | |
| "architectures": [ | |
| "TwinkelLLMForCausalLM" | |
| ], | |
| "model_type": "twinkel_llm", | |
| "auto_map": { | |
| "AutoConfig": "configuration_twinkel_llm.TwinkelLLMConfig", | |
| "AutoModelForCausalLM": "modeling_twinkel_llm.TwinkelLLMForCausalLM" | |
| }, | |
| "vocab_size": 128256, | |
| "hidden_size": 448, | |
| "intermediate_size": 1344, | |
| "num_hidden_layers": 6, | |
| "num_attention_heads": 8, | |
| "num_key_value_heads": 4, | |
| "max_position_embeddings": 512, | |
| "dropout": 0.1, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 10000.0, | |
| "pad_token_id": 128012, | |
| "bos_token_id": 128012, | |
| "eos_token_id": 128012, | |
| "tie_word_embeddings": true, | |
| "torch_dtype": "float32" | |
| } |