fareedkhan-dev--train-llm-from-scratch
10 行
156 B
JSON
10 行
156 B
JSON
{
|
|
"iterations": 2,
|
|
"prompts_per_iter": 4,
|
|
"rollout_len": 32,
|
|
"ppo_epochs": 2,
|
|
"minibatch_size": 2,
|
|
"eval_every": 10000,
|
|
"save_every": 10000
|
|
}
|