| { | |
| "model_type": "aresmath1.0-it", | |
| "architectures": [ | |
| "AresMath1.0ForCausalLM" | |
| ], | |
| "vocab_size": 21, | |
| "max_position_embeddings": 32, | |
| "hidden_size": 32, | |
| "num_attention_heads": 4, | |
| "num_hidden_layers": 1, | |
| "num_loops": 50, | |
| "shared_loop_weights": true, | |
| "intermediate_size": 64, | |
| "dropout": 0.0, | |
| "torch_dtype": "bfloat16", | |
| "tokenizer_class": "AresTokenizer", | |
| "rope_scaling": "linear", | |
| "mlp_type": "swiglu", | |
| "norm_type": "rmsnorm", | |
| "rl_temperature": 0.7 | |
| } |