| { | |
| "architectures": [ | |
| "RandyGPTForCausalLM" | |
| ], | |
| "model_type": "randygpt", | |
| "auto_map": { | |
| "AutoConfig": "modeling_randygpt.RandyGPTConfig", | |
| "AutoModelForCausalLM": "modeling_randygpt.RandyGPTForCausalLM" | |
| }, | |
| "vocab_size": 2000, | |
| "n_embd": 128, | |
| "n_head": 4, | |
| "n_layer": 12, | |
| "block_size": 256, | |
| "n_experts": 4, | |
| "expert_dim": 256, | |
| "moe_top_k": 2, | |
| "bos_token_id": 0, | |
| "eos_token_id": 1, | |
| "trained_iters": 12850, | |
| "best_val_loss": 3.5011932849884033 | |
| } |