configs/micro_moe_llama_1b.yml

run-title: micro-moe-llama-1b

model: micro-moe-llama-1b

base-model: meta-llama/Llama-3.2-1B
tokenizer: meta-llama/Llama-3.2-1B-Instruct
num-experts: 4
top-k-experts: 1
jitter-noise: 0
router-aux-loss-coef: 0.000
use-load-balancing: False
use-router: True
mask-input: True
max-length: 8192

trainable:
  - model