arpanpathak's picture
Add StaleyNET training checkpoints, configs, metrics and logs
b5948a7 verified
Raw History Blame Contribute Delete
637 Bytes
{
"model": {
"width": 256,
"heads": 8,
"layers": 6,
"feed_forward_multiplier": 4,
"context_length": 768,
"prompt_bytes": 126,
"codebooks": 4,
"codebook_size": 1024
},
"optimizer": {
"peak_learning_rate": 0.0006,
"final_learning_rate": 0.00006,
"warmup_steps": 500,
"decay_steps": 30000,
"beta1": 0.9,
"beta2": 0.95,
"epsilon": 1e-8,
"weight_decay": 0.1,
"max_gradient_norm": 1.0
},
"batch_size": 8,
"sequence_length": 768,
"steps": 30000,
"log_every": 25,
"validate_every": 500,
"validation_batches": 16,
"checkpoint_every": 500,
"seed": 1234
}