recipes/docker-images/glm52-nf3-hybrid/overlay/mxfp8_tier.json

{
"module_prefixes": [
"layers.1.mlp.down_proj",
"layers.1.mlp.gate_up_proj",
"layers.1.self_attn.kv_a_proj_with_mqa",
"layers.1.self_attn.kv_b_proj",
"layers.1.self_attn.o_proj",
"layers.1.self_attn.q_a_proj",
"layers.1.self_attn.q_b_proj",
"layers.10.mlp.shared_experts.down_proj",
"layers.10.mlp.shared_experts.gate_up_proj",
"layers.10.self_attn.kv_a_proj_with_mqa",
"layers.10.self_attn.kv_b_proj",
"layers.10.self_attn.o_proj",
"layers.10.self_attn.q_a_proj",
"layers.10.self_attn.q_b_proj",
"layers.11.mlp.shared_experts.down_proj",
"layers.11.mlp.shared_experts.gate_up_proj",
"layers.11.self_attn.kv_a_proj_with_mqa",
"layers.11.self_attn.kv_b_proj",
"layers.11.self_attn.o_proj",
"layers.11.self_attn.q_a_proj",
"layers.11.self_attn.q_b_proj",
"layers.12.mlp.shared_experts.down_proj",
"layers.12.mlp.shared_experts.gate_up_proj",
"layers.12.self_attn.kv_a_proj_with_mqa",
"layers.12.self_attn.kv_b_proj",
"layers.12.self_attn.o_proj",
"layers.12.self_attn.q_a_proj",
"layers.12.self_attn.q_b_proj",
"layers.13.mlp.shared_experts.down_proj",
"layers.13.mlp.shared_experts.gate_up_proj",
"layers.13.self_attn.kv_a_proj_with_mqa",
"layers.13.self_attn.kv_b_proj",
"layers.13.self_attn.o_proj",
"layers.13.self_attn.q_a_proj",
"layers.13.self_attn.q_b_proj",
"layers.14.mlp.shared_experts.down_proj",
"layers.14.mlp.shared_experts.gate_up_proj",
"layers.14.self_attn.kv_a_proj_with_mqa",
"layers.14.self_attn.kv_b_proj",
"layers.14.self_attn.o_proj",
"layers.14.self_attn.q_a_proj",
"layers.14.self_attn.q_b_proj",
"layers.15.mlp.shared_experts.down_proj",
"layers.15.mlp.shared_experts.gate_up_proj",
"layers.15.self_attn.kv_a_proj_with_mqa",
"layers.15.self_attn.kv_b_proj",
"layers.15.self_attn.o_proj",
"layers.15.self_attn.q_a_proj",
"layers.15.self_attn.q_b_proj",
"layers.16.mlp.shared_experts.down_proj",
"layers.16.mlp.shared_experts.gate_up_proj",
"layers.16.self_attn.kv_a_proj_with_mqa",
"layers.16.self_attn.kv_b_proj",
"layers.16.self_attn.o_proj",
"layers.16.self_attn.q_a_proj",
"layers.16.self_attn.q_b_proj",
"layers.17.mlp.shared_experts.down_proj",
"layers.17.mlp.shared_experts.gate_up_proj",
"layers.17.self_attn.kv_a_proj_with_mqa",
"layers.17.self_attn.kv_b_proj",
"layers.17.self_attn.o_proj",
"layers.17.self_attn.q_a_proj",
"layers.17.self_attn.q_b_proj",
"layers.18.mlp.shared_experts.down_proj",
"layers.18.mlp.shared_experts.gate_up_proj",
"layers.18.self_attn.kv_a_proj_with_mqa",
"layers.18.self_attn.kv_b_proj",
"layers.18.self_attn.o_proj",
"layers.18.self_attn.q_a_proj",
"layers.18.self_attn.q_b_proj",
"layers.19.mlp.shared_experts.down_proj",
"layers.19.mlp.shared_experts.gate_up_proj",
"layers.19.self_attn.kv_a_proj_with_mqa",
"layers.19.self_attn.kv_b_proj",
"layers.19.self_attn.o_proj",
"layers.19.self_attn.q_a_proj",
"layers.19.self_attn.q_b_proj",
"layers.2.mlp.down_proj",
"layers.2.mlp.gate_up_proj",
"layers.2.self_attn.kv_a_proj_with_mqa",
"layers.2.self_attn.kv_b_proj",
"layers.2.self_attn.o_proj",
"layers.2.self_attn.q_a_proj",
"layers.2.self_attn.q_b_proj",
"layers.20.mlp.shared_experts.down_proj",
"layers.20.mlp.shared_experts.gate_up_proj",
"layers.20.self_attn.kv_a_proj_with_mqa",
"layers.20.self_attn.kv_b_proj",
"layers.20.self_attn.o_proj",
"layers.20.self_attn.q_a_proj",
"layers.20.self_attn.q_b_proj",
"layers.21.mlp.shared_experts.down_proj",
"layers.21.mlp.shared_experts.gate_up_proj",
"layers.21.self_attn.kv_a_proj_with_mqa",
"layers.21.self_attn.kv_b_proj",
"layers.21.self_attn.o_proj",
"layers.21.self_attn.q_a_proj",
"layers.21.self_attn.q_b_proj",
"layers.22.mlp.shared_experts.down_proj",
"layers.22.mlp.shared_experts.gate_up_proj",
"layers.22.self_attn.kv_a_proj_with_mqa",
"layers.22.self_attn.kv_b_proj",
"layers.22.self_attn.o_proj",
"layers.22.self_attn.q_a_proj",
"layers.22.self_attn.q_b_proj",
"layers.23.mlp.shared_experts.down_proj",
"layers.23.mlp.shared_experts.gate_up_proj",
"layers.23.self_attn.kv_a_proj_with_mqa",
"layers.23.self_attn.kv_b_proj",
"layers.23.self_attn.o_proj",
"layers.23.self_attn.q_a_proj",
"layers.23.self_attn.q_b_proj",
"layers.24.mlp.shared_experts.down_proj",
"layers.24.mlp.shared_experts.gate_up_proj",
"layers.24.self_attn.kv_a_proj_with_mqa",
"layers.24.self_attn.kv_b_proj",
"layers.24.self_attn.o_proj",
"layers.24.self_attn.q_a_proj",
"layers.24.self_attn.q_b_proj",
"layers.25.mlp.shared_experts.down_proj",
"layers.25.mlp.shared_experts.gate_up_proj",
"layers.25.self_attn.kv_a_proj_with_mqa",
"layers.25.self_attn.kv_b_proj",
"layers.25.self_attn.o_proj",
"layers.25.self_attn.q_a_proj",
"layers.25.self_attn.q_b_proj",
"layers.26.mlp.shared_experts.down_proj",
"layers.26.mlp.shared_experts.gate_up_proj",
"layers.26.self_attn.kv_a_proj_with_mqa",
"layers.26.self_attn.kv_b_proj",
"layers.26.self_attn.o_proj",
"layers.26.self_attn.q_a_proj",
"layers.26.self_attn.q_b_proj",
"layers.27.mlp.shared_experts.down_proj",
"layers.27.mlp.shared_experts.gate_up_proj",
"layers.27.self_attn.kv_a_proj_with_mqa",
"layers.27.self_attn.kv_b_proj",
"layers.27.self_attn.o_proj",
"layers.27.self_attn.q_a_proj",
"layers.27.self_attn.q_b_proj",
"layers.28.mlp.shared_experts.down_proj",
"layers.28.mlp.shared_experts.gate_up_proj",
"layers.28.self_attn.kv_a_proj_with_mqa",
"layers.28.self_attn.kv_b_proj",
"layers.28.self_attn.o_proj",
"layers.28.self_attn.q_a_proj",
"layers.28.self_attn.q_b_proj",
"layers.29.mlp.shared_experts.down_proj",
"layers.29.mlp.shared_experts.gate_up_proj",
"layers.29.self_attn.kv_a_proj_with_mqa",
"layers.29.self_attn.kv_b_proj",
"layers.29.self_attn.o_proj",
"layers.29.self_attn.q_a_proj",
"layers.29.self_attn.q_b_proj",
"layers.3.mlp.shared_experts.down_proj",
"layers.3.mlp.shared_experts.gate_up_proj",
"layers.3.self_attn.kv_a_proj_with_mqa",
"layers.3.self_attn.kv_b_proj",
"layers.3.self_attn.o_proj",
"layers.3.self_attn.q_a_proj",
"layers.3.self_attn.q_b_proj",
"layers.30.mlp.shared_experts.down_proj",
"layers.30.mlp.shared_experts.gate_up_proj",
"layers.30.self_attn.kv_a_proj_with_mqa",
"layers.30.self_attn.kv_b_proj",
"layers.30.self_attn.o_proj",
"layers.30.self_attn.q_a_proj",
"layers.30.self_attn.q_b_proj",
"layers.31.mlp.shared_experts.down_proj",
"layers.31.mlp.shared_experts.gate_up_proj",
"layers.31.self_attn.kv_a_proj_with_mqa",
"layers.31.self_attn.kv_b_proj",
"layers.31.self_attn.o_proj",
"layers.31.self_attn.q_a_proj",
"layers.31.self_attn.q_b_proj",
"layers.32.mlp.shared_experts.down_proj",
"layers.32.mlp.shared_experts.gate_up_proj",
"layers.32.self_attn.kv_a_proj_with_mqa",
"layers.32.self_attn.kv_b_proj",
"layers.32.self_attn.o_proj",
"layers.32.self_attn.q_a_proj",
"layers.32.self_attn.q_b_proj",
"layers.33.mlp.shared_experts.down_proj",
"layers.33.mlp.shared_experts.gate_up_proj",
"layers.33.self_attn.kv_a_proj_with_mqa",
"layers.33.self_attn.kv_b_proj",
"layers.33.self_attn.o_proj",
"layers.33.self_attn.q_a_proj",
"layers.33.self_attn.q_b_proj",
"layers.34.mlp.shared_experts.down_proj",
"layers.34.mlp.shared_experts.gate_up_proj",
"layers.34.self_attn.kv_a_proj_with_mqa",
"layers.34.self_attn.kv_b_proj",
"layers.34.self_attn.o_proj",
"layers.34.self_attn.q_a_proj",
"layers.34.self_attn.q_b_proj",
"layers.35.mlp.shared_experts.down_proj",
"layers.35.mlp.shared_experts.gate_up_proj",
"layers.35.self_attn.kv_a_proj_with_mqa",
"layers.35.self_attn.kv_b_proj",
"layers.35.self_attn.o_proj",
"layers.35.self_attn.q_a_proj",
"layers.35.self_attn.q_b_proj",
"layers.36.mlp.shared_experts.down_proj",
"layers.36.mlp.shared_experts.gate_up_proj",
"layers.36.self_attn.kv_a_proj_with_mqa",
"layers.36.self_attn.kv_b_proj",
"layers.36.self_attn.o_proj",
"layers.36.self_attn.q_a_proj",
"layers.36.self_attn.q_b_proj",
"layers.37.mlp.shared_experts.down_proj",
"layers.37.mlp.shared_experts.gate_up_proj",
"layers.37.self_attn.kv_a_proj_with_mqa",
"layers.37.self_attn.kv_b_proj",
"layers.37.self_attn.o_proj",
"layers.37.self_attn.q_a_proj",
"layers.37.self_attn.q_b_proj",
"layers.38.mlp.shared_experts.down_proj",
"layers.38.mlp.shared_experts.gate_up_proj",
"layers.38.self_attn.kv_a_proj_with_mqa",
"layers.38.self_attn.kv_b_proj",
"layers.38.self_attn.o_proj",
"layers.38.self_attn.q_a_proj",
"layers.38.self_attn.q_b_proj",
"layers.39.mlp.shared_experts.down_proj",
"layers.39.mlp.shared_experts.gate_up_proj",
"layers.39.self_attn.kv_a_proj_with_mqa",
"layers.39.self_attn.kv_b_proj",
"layers.39.self_attn.o_proj",
"layers.39.self_attn.q_a_proj",
"layers.39.self_attn.q_b_proj",
"layers.4.mlp.shared_experts.down_proj",
"layers.4.mlp.shared_experts.gate_up_proj",
"layers.4.self_attn.kv_a_proj_with_mqa",
"layers.4.self_attn.kv_b_proj",
"layers.4.self_attn.o_proj",
"layers.4.self_attn.q_a_proj",
"layers.4.self_attn.q_b_proj",
"layers.40.mlp.shared_experts.down_proj",
"layers.40.mlp.shared_experts.gate_up_proj",
"layers.40.self_attn.kv_a_proj_with_mqa",
"layers.40.self_attn.kv_b_proj",
"layers.40.self_attn.o_proj",
"layers.40.self_attn.q_a_proj",
"layers.40.self_attn.q_b_proj",
"layers.41.mlp.shared_experts.down_proj",
"layers.41.mlp.shared_experts.gate_up_proj",
"layers.41.self_attn.kv_a_proj_with_mqa",
"layers.41.self_attn.kv_b_proj",
"layers.41.self_attn.o_proj",
"layers.41.self_attn.q_a_proj",
"layers.41.self_attn.q_b_proj",
"layers.42.mlp.shared_experts.down_proj",
"layers.42.mlp.shared_experts.gate_up_proj",
"layers.42.self_attn.kv_a_proj_with_mqa",
"layers.42.self_attn.kv_b_proj",
"layers.42.self_attn.o_proj",
"layers.42.self_attn.q_a_proj",
"layers.42.self_attn.q_b_proj",
"layers.43.mlp.shared_experts.down_proj",
"layers.43.mlp.shared_experts.gate_up_proj",
"layers.43.self_attn.kv_a_proj_with_mqa",
"layers.43.self_attn.kv_b_proj",
"layers.43.self_attn.o_proj",
"layers.43.self_attn.q_a_proj",
"layers.43.self_attn.q_b_proj",
"layers.44.mlp.shared_experts.down_proj",
"layers.44.mlp.shared_experts.gate_up_proj",
"layers.44.self_attn.kv_a_proj_with_mqa",
"layers.44.self_attn.kv_b_proj",
"layers.44.self_attn.o_proj",
"layers.44.self_attn.q_a_proj",
"layers.44.self_attn.q_b_proj",
"layers.45.mlp.shared_experts.down_proj",
"layers.45.mlp.shared_experts.gate_up_proj",
"layers.45.self_attn.kv_a_proj_with_mqa",
"layers.45.self_attn.kv_b_proj",
"layers.45.self_attn.o_proj",
"layers.45.self_attn.q_a_proj",
"layers.45.self_attn.q_b_proj",
"layers.46.mlp.shared_experts.down_proj",
"layers.46.mlp.shared_experts.gate_up_proj",
"layers.46.self_attn.kv_a_proj_with_mqa",
"layers.46.self_attn.kv_b_proj",
"layers.46.self_attn.o_proj",
"layers.46.self_attn.q_a_proj",
"layers.46.self_attn.q_b_proj",
"layers.47.mlp.shared_experts.down_proj",
"layers.47.mlp.shared_experts.gate_up_proj",
"layers.47.self_attn.kv_a_proj_with_mqa",
"layers.47.self_attn.kv_b_proj",
"layers.47.self_attn.o_proj",
"layers.47.self_attn.q_a_proj",
"layers.47.self_attn.q_b_proj",
"layers.48.mlp.shared_experts.down_proj",
"layers.48.mlp.shared_experts.gate_up_proj",
"layers.48.self_attn.kv_a_proj_with_mqa",
"layers.48.self_attn.kv_b_proj",
"layers.48.self_attn.o_proj",
"layers.48.self_attn.q_a_proj",
"layers.48.self_attn.q_b_proj",
"layers.49.mlp.shared_experts.down_proj",
"layers.49.mlp.shared_experts.gate_up_proj",
"layers.49.self_attn.kv_a_proj_with_mqa",
"layers.49.self_attn.kv_b_proj",
"layers.49.self_attn.o_proj",
"layers.49.self_attn.q_a_proj",
"layers.49.self_attn.q_b_proj",
"layers.5.mlp.shared_experts.down_proj",
"layers.5.mlp.shared_experts.gate_up_proj",
"layers.5.self_attn.kv_a_proj_with_mqa",
"layers.5.self_attn.kv_b_proj",
"layers.5.self_attn.o_proj",
"layers.5.self_attn.q_a_proj",
"layers.5.self_attn.q_b_proj",
"layers.50.mlp.shared_experts.down_proj",
"layers.50.mlp.shared_experts.gate_up_proj",
"layers.50.self_attn.kv_a_proj_with_mqa",
"layers.50.self_attn.kv_b_proj",
"layers.50.self_attn.o_proj",
"layers.50.self_attn.q_a_proj",
"layers.50.self_attn.q_b_proj",
"layers.51.mlp.shared_experts.down_proj",
"layers.51.mlp.shared_experts.gate_up_proj",
"layers.51.self_attn.kv_a_proj_with_mqa",
"layers.51.self_attn.kv_b_proj",
"layers.51.self_attn.o_proj",
"layers.51.self_attn.q_a_proj",
"layers.51.self_attn.q_b_proj",
"layers.52.mlp.shared_experts.down_proj",
"layers.52.mlp.shared_experts.gate_up_proj",
"layers.52.self_attn.kv_a_proj_with_mqa",
"layers.52.self_attn.kv_b_proj",
"layers.52.self_attn.o_proj",
"layers.52.self_attn.q_a_proj",
"layers.52.self_attn.q_b_proj",
"layers.53.mlp.shared_experts.down_proj",
"layers.53.mlp.shared_experts.gate_up_proj",
"layers.53.self_attn.kv_a_proj_with_mqa",
"layers.53.self_attn.kv_b_proj",
"layers.53.self_attn.o_proj",
"layers.53.self_attn.q_a_proj",
"layers.53.self_attn.q_b_proj",
"layers.54.mlp.shared_experts.down_proj",
"layers.54.mlp.shared_experts.gate_up_proj",
"layers.54.self_attn.kv_a_proj_with_mqa",
"layers.54.self_attn.kv_b_proj",
"layers.54.self_attn.o_proj",
"layers.54.self_attn.q_a_proj",
"layers.54.self_attn.q_b_proj",
"layers.55.mlp.shared_experts.down_proj",
"layers.55.mlp.shared_experts.gate_up_proj",
"layers.55.self_attn.kv_a_proj_with_mqa",
"layers.55.self_attn.kv_b_proj",
"layers.55.self_attn.o_proj",
"layers.55.self_attn.q_a_proj",
"layers.55.self_attn.q_b_proj",
"layers.56.mlp.shared_experts.down_proj",
"layers.56.mlp.shared_experts.gate_up_proj",
"layers.56.self_attn.kv_a_proj_with_mqa",
"layers.56.self_attn.kv_b_proj",
"layers.56.self_attn.o_proj",
"layers.56.self_attn.q_a_proj",
"layers.56.self_attn.q_b_proj",
"layers.57.mlp.shared_experts.down_proj",
"layers.57.mlp.shared_experts.gate_up_proj",
"layers.57.self_attn.kv_a_proj_with_mqa",
"layers.57.self_attn.kv_b_proj",
"layers.57.self_attn.o_proj",
"layers.57.self_attn.q_a_proj",
"layers.57.self_attn.q_b_proj",
"layers.58.mlp.shared_experts.down_proj",
"layers.58.mlp.shared_experts.gate_up_proj",
"layers.58.self_attn.kv_a_proj_with_mqa",
"layers.58.self_attn.kv_b_proj",
"layers.58.self_attn.o_proj",
"layers.58.self_attn.q_a_proj",
"layers.58.self_attn.q_b_proj",
"layers.59.mlp.shared_experts.down_proj",
"layers.59.mlp.shared_experts.gate_up_proj",
"layers.59.self_attn.kv_a_proj_with_mqa",
"layers.59.self_attn.kv_b_proj",
"layers.59.self_attn.o_proj",
"layers.59.self_attn.q_a_proj",
"layers.59.self_attn.q_b_proj",
"layers.6.mlp.shared_experts.down_proj",
"layers.6.mlp.shared_experts.gate_up_proj",
"layers.6.self_attn.kv_a_proj_with_mqa",
"layers.6.self_attn.kv_b_proj",
"layers.6.self_attn.o_proj",
"layers.6.self_attn.q_a_proj",
"layers.6.self_attn.q_b_proj",
"layers.60.mlp.shared_experts.down_proj",
"layers.60.mlp.shared_experts.gate_up_proj",
"layers.60.self_attn.kv_a_proj_with_mqa",
"layers.60.self_attn.kv_b_proj",
"layers.60.self_attn.o_proj",
"layers.60.self_attn.q_a_proj",
"layers.60.self_attn.q_b_proj",
"layers.61.mlp.shared_experts.down_proj",
"layers.61.mlp.shared_experts.gate_up_proj",
"layers.61.self_attn.kv_a_proj_with_mqa",
"layers.61.self_attn.kv_b_proj",
"layers.61.self_attn.o_proj",
"layers.61.self_attn.q_a_proj",
"layers.61.self_attn.q_b_proj",
"layers.62.mlp.shared_experts.down_proj",
"layers.62.mlp.shared_experts.gate_up_proj",
"layers.62.self_attn.kv_a_proj_with_mqa",
"layers.62.self_attn.kv_b_proj",
"layers.62.self_attn.o_proj",
"layers.62.self_attn.q_a_proj",
"layers.62.self_attn.q_b_proj",
"layers.63.mlp.shared_experts.down_proj",
"layers.63.mlp.shared_experts.gate_up_proj",
"layers.63.self_attn.kv_a_proj_with_mqa",
"layers.63.self_attn.kv_b_proj",
"layers.63.self_attn.o_proj",
"layers.63.self_attn.q_a_proj",
"layers.63.self_attn.q_b_proj",
"layers.64.mlp.shared_experts.down_proj",
"layers.64.mlp.shared_experts.gate_up_proj",
"layers.64.self_attn.kv_a_proj_with_mqa",
"layers.64.self_attn.kv_b_proj",
"layers.64.self_attn.o_proj",
"layers.64.self_attn.q_a_proj",
"layers.64.self_attn.q_b_proj",
"layers.65.mlp.shared_experts.down_proj",
"layers.65.mlp.shared_experts.gate_up_proj",
"layers.65.self_attn.kv_a_proj_with_mqa",
"layers.65.self_attn.kv_b_proj",
"layers.65.self_attn.o_proj",
"layers.65.self_attn.q_a_proj",
"layers.65.self_attn.q_b_proj",
"layers.66.mlp.shared_experts.down_proj",
"layers.66.mlp.shared_experts.gate_up_proj",
"layers.66.self_attn.kv_a_proj_with_mqa",
"layers.66.self_attn.kv_b_proj",
"layers.66.self_attn.o_proj",
"layers.66.self_attn.q_a_proj",
"layers.66.self_attn.q_b_proj",
"layers.67.mlp.shared_experts.down_proj",
"layers.67.mlp.shared_experts.gate_up_proj",
"layers.67.self_attn.kv_a_proj_with_mqa",
"layers.67.self_attn.kv_b_proj",
"layers.67.self_attn.o_proj",
"layers.67.self_attn.q_a_proj",
"layers.67.self_attn.q_b_proj",
"layers.68.mlp.shared_experts.down_proj",
"layers.68.mlp.shared_experts.gate_up_proj",
"layers.68.self_attn.kv_a_proj_with_mqa",
"layers.68.self_attn.kv_b_proj",
"layers.68.self_attn.o_proj",
"layers.68.self_attn.q_a_proj",
"layers.68.self_attn.q_b_proj",
"layers.69.mlp.shared_experts.down_proj",
"layers.69.mlp.shared_experts.gate_up_proj",
"layers.69.self_attn.kv_a_proj_with_mqa",
"layers.69.self_attn.kv_b_proj",
"layers.69.self_attn.o_proj",
"layers.69.self_attn.q_a_proj",
"layers.69.self_attn.q_b_proj",
"layers.7.mlp.shared_experts.down_proj",
"layers.7.mlp.shared_experts.gate_up_proj",
"layers.7.self_attn.kv_a_proj_with_mqa",
"layers.7.self_attn.kv_b_proj",
"layers.7.self_attn.o_proj",
"layers.7.self_attn.q_a_proj",
"layers.7.self_attn.q_b_proj",
"layers.70.mlp.shared_experts.down_proj",
"layers.70.mlp.shared_experts.gate_up_proj",
"layers.70.self_attn.kv_a_proj_with_mqa",
"layers.70.self_attn.kv_b_proj",
"layers.70.self_attn.o_proj",
"layers.70.self_attn.q_a_proj",
"layers.70.self_attn.q_b_proj",
"layers.71.mlp.shared_experts.down_proj",
"layers.71.mlp.shared_experts.gate_up_proj",
"layers.71.self_attn.kv_a_proj_with_mqa",
"layers.71.self_attn.kv_b_proj",
"layers.71.self_attn.o_proj",
"layers.71.self_attn.q_a_proj",
"layers.71.self_attn.q_b_proj",
"layers.72.mlp.shared_experts.down_proj",
"layers.72.mlp.shared_experts.gate_up_proj",
"layers.72.self_attn.kv_a_proj_with_mqa",
"layers.72.self_attn.kv_b_proj",
"layers.72.self_attn.o_proj",
"layers.72.self_attn.q_a_proj",
"layers.72.self_attn.q_b_proj",
"layers.73.mlp.shared_experts.down_proj",
"layers.73.mlp.shared_experts.gate_up_proj",
"layers.73.self_attn.kv_a_proj_with_mqa",
"layers.73.self_attn.kv_b_proj",
"layers.73.self_attn.o_proj",
"layers.73.self_attn.q_a_proj",
"layers.73.self_attn.q_b_proj",
"layers.74.mlp.shared_experts.down_proj",
"layers.74.mlp.shared_experts.gate_up_proj",
"layers.74.self_attn.kv_a_proj_with_mqa",
"layers.74.self_attn.kv_b_proj",
"layers.74.self_attn.o_proj",
"layers.74.self_attn.q_a_proj",
"layers.74.self_attn.q_b_proj",
"layers.75.mlp.shared_experts.down_proj",
"layers.75.mlp.shared_experts.gate_up_proj",
"layers.75.self_attn.kv_a_proj_with_mqa",
"layers.75.self_attn.kv_b_proj",
"layers.75.self_attn.o_proj",
"layers.75.self_attn.q_a_proj",
"layers.75.self_attn.q_b_proj",
"layers.76.mlp.shared_experts.down_proj",
"layers.76.mlp.shared_experts.gate_up_proj",
"layers.76.self_attn.kv_a_proj_with_mqa",
"layers.76.self_attn.kv_b_proj",
"layers.76.self_attn.o_proj",
"layers.76.self_attn.q_a_proj",
"layers.76.self_attn.q_b_proj",
"layers.77.mlp.shared_experts.down_proj",
"layers.77.mlp.shared_experts.gate_up_proj",
"layers.77.self_attn.kv_a_proj_with_mqa",
"layers.77.self_attn.kv_b_proj",
"layers.77.self_attn.o_proj",
"layers.77.self_attn.q_a_proj",
"layers.77.self_attn.q_b_proj",
"layers.78.mlp.shared_experts.down_proj",
"layers.78.mlp.shared_experts.gate_up_proj",
"layers.78.self_attn.kv_a_proj_with_mqa",
"layers.78.self_attn.kv_b_proj",
"layers.78.self_attn.o_proj",
"layers.78.self_attn.q_a_proj",
"layers.78.self_attn.q_b_proj",
"layers.8.mlp.shared_experts.down_proj",
"layers.8.mlp.shared_experts.gate_up_proj",
"layers.8.self_attn.kv_a_proj_with_mqa",
"layers.8.self_attn.kv_b_proj",
"layers.8.self_attn.o_proj",
"layers.8.self_attn.q_a_proj",
"layers.8.self_attn.q_b_proj",
"layers.9.mlp.shared_experts.down_proj",
"layers.9.mlp.shared_experts.gate_up_proj",
"layers.9.self_attn.kv_a_proj_with_mqa",
"layers.9.self_attn.kv_b_proj",
"layers.9.self_attn.o_proj",
"layers.9.self_attn.q_a_proj",
"layers.9.self_attn.q_b_proj"
]
}