slm-125m-instruct / code_completion_training_metadata.json
tohio's picture
Export slm-125m-instruct (125.3M params)
d8ff2d9 verified
Raw
History Blame Contribute Delete
466 Bytes
{
"config": "finetune/configs/code_completion_125m.yaml",
"base_model": "results/runs/125m/sft_code/final",
"train_path": "/data/slm/data/runs/125m/code_completion/train.jsonl",
"val_path": "/data/slm/data/runs/125m/code_completion/val.jsonl",
"updates": 500,
"max_length": 768,
"micro_batch_size": 8,
"gradient_accumulation_steps": 4,
"learning_rate": 1e-05,
"weight_decay": 0.01,
"warmup_ratio": 0.05,
"best_val_loss": 0.8386935763888889
}