Upload Minisun Trained using model.fit on lss dataset[400],lr 2e-4, 2 epoch,batch size 2,cosine with restart
b90c018
verified
| { | |
| "vocab_size": 50272, | |
| "max_position_embeddings": 2048, | |
| "hidden_size": 768, | |
| "num_attention_heads": 12, | |
| "intermediate_size": 3072, | |
| "num_hidden_layers": 12, | |
| "dropout_rate": 0.1, | |
| "weight_decay": 0.01, | |
| "learning_rate": 0.0002, | |
| "restart_period": 66, | |
| "total_steps": 200, | |
| "warmup_ratio": 0.2 | |
| } |