| { |
| "architectures": [ |
| "MomoForCausalLM" |
| ], |
| "auto_map": { |
| "AutoConfig": "configuration_momo.MomoConfig", |
| "AutoModelForCausalLM": "modeling_momo.MomoForCausalLM" |
| }, |
| "model_type": "momo", |
| "model_name": "Momo-336M", |
| "model_version": "1.0", |
| "vocab_size": 32000, |
| "hidden_size": 1024, |
| "intermediate_size": 2752, |
| "num_hidden_layers": 24, |
| "num_attention_heads": 16, |
| "num_key_value_heads": 8, |
| "max_position_embeddings": 512, |
| "rope_theta": 10000.0, |
| "rms_norm_eps": 1e-05, |
| "use_gradient_checkpointing": false, |
| "hidden_act": "silu", |
| "torch_dtype": "float16", |
| "transformers_version": "4.40.0", |
| "bos_token_id": 1, |
| "eos_token_id": 0, |
| "pad_token_id": 3 |
| } |