↓ 1 callersFunctionparam_count(vocab_size, d_model, n_layers, n_heads, ffn_expansion=2.67, block_size=128)
phases/07-transformers-deep-dive/14-build-a-transformer-capstone/code/main.py:58
↓ 1 callersFunctionppo_training(
policy_model,
reference_model,
reward_model,
prompts,
num_episodes=20,
lr=1.5e-5,
phases/10-llms-from-scratch/07-rlhf/code/main.py:220
↓ 1 callersFunctionppo_update(theta, w, buffer, advantages, returns, lr_a=0.05, lr_v=0.1, eps=0.2, epochs=4, batch=32, rng=None)
phases/09-reinforcement-learning/08-ppo/code/main.py:112