| { | |
| "activation_type": "gelu", | |
| "architectures": [ | |
| "AxonForCausalLM" | |
| ], | |
| "attention_type": "standard", | |
| "batch_size": 15, | |
| "bias": true, | |
| "block_size": 1024, | |
| "block_type": "standard", | |
| "dropout": 0.2, | |
| "dropout_attention": 0.2, | |
| "dropout_resid": 0.2, | |
| "eval_interval": 25, | |
| "eval_iters": 200, | |
| "flash_attention": false, | |
| "grad_clip": 1.0, | |
| "gradient_accumulation_steps": 40, | |
| "layer_norm_type": "default", | |
| "log_interval": 10, | |
| "max_iters": 600000, | |
| "mlp_ratio": 4, | |
| "model_type": "gpt2", | |
| "n_embd": 256, | |
| "n_head": 4, | |
| "n_layer": 8, | |
| "out_path": "models/axon-test/", | |
| "rotary_embeddings": true, | |
| "transformers_version": "4.41.2", | |
| "use_cache": true, | |
| "vocab_size": 50304 | |
| } | |