use crate::burn_backend::onnx::{Source::*, Transform::*, Weight};
pub(crate) const WEIGHTS: &[Weight] = &[
Weight { path: "submodule1.constant1", shape: Some(&[16384, 1024]), source: Param(".model.decoder.embed_tokens.weight", Copy) },
Weight { path: "submodule1.constant2", shape: Some(&[1024, 1024]), source: Param("model.decoder.pos_emb.weight", Copy) },
Weight { path: "submodule1.constant237", shape: Some(&[1]), source: Param("/model/constants/FLOAT/1.0", Copy) },
Weight { path: "submodule1.constant238", shape: Some(&[1]), source: Param("/model/constants/FLOAT/-3.4028234663852886e+38", Copy) },
Weight { path: "submodule1.constant42", shape: Some(&[1024]), source: Param("model.layers.0.attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule1.constant43", shape: Some(&[1024]), source: Param("model.layers.0.attn.k_proj.Add.bias", Copy) },
Weight { path: "submodule1.constant44", shape: Some(&[1024]), source: Param("model.layers.0.attn.v_proj.Add.bias", Copy) },
Weight { path: "submodule1.layernormalization1.beta", shape: Some(&[1024]), source: Param("model.layers.9.embedding_layernorm_layernorm.bias", Copy) },
Weight { path: "submodule1.layernormalization1.gamma", shape: Some(&[1024]), source: Param("model.layers.9.embedding_layernorm_layernorm.weight", Copy) },
Weight { path: "submodule1.layernormalization2.beta", shape: Some(&[1024]), source: Param("model.layers.0.input_ln_layernorm.bias", Copy) },
Weight { path: "submodule1.layernormalization2.gamma", shape: Some(&[1024]), source: Param("model.layers.0.input_ln_layernorm.weight", Copy) },
Weight { path: "submodule1.linear1.bias", shape: Some(&[1024]), source: Param("model.layers.0.encoder_attn.k_proj.bias", Copy) },
Weight { path: "submodule1.linear1.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.0.encoder_attn.k_proj.weight", Copy) },
Weight { path: "submodule1.linear10.bias", shape: Some(&[1024]), source: Param("model.layers.4.encoder_attn.v_proj.bias", Copy) },
Weight { path: "submodule1.linear10.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.4.encoder_attn.v_proj.weight", Copy) },
Weight { path: "submodule1.linear11.bias", shape: Some(&[1024]), source: Param("model.layers.5.encoder_attn.k_proj.bias", Copy) },
Weight { path: "submodule1.linear11.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.5.encoder_attn.k_proj.weight", Copy) },
Weight { path: "submodule1.linear12.bias", shape: Some(&[1024]), source: Param("model.layers.5.encoder_attn.v_proj.bias", Copy) },
Weight { path: "submodule1.linear12.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.5.encoder_attn.v_proj.weight", Copy) },
Weight { path: "submodule1.linear13.bias", shape: Some(&[1024]), source: Param("model.layers.6.encoder_attn.k_proj.bias", Copy) },
Weight { path: "submodule1.linear13.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.6.encoder_attn.k_proj.weight", Copy) },
Weight { path: "submodule1.linear14.bias", shape: Some(&[1024]), source: Param("model.layers.6.encoder_attn.v_proj.bias", Copy) },
Weight { path: "submodule1.linear14.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.6.encoder_attn.v_proj.weight", Copy) },
Weight { path: "submodule1.linear15.bias", shape: Some(&[1024]), source: Param("model.layers.7.encoder_attn.k_proj.bias", Copy) },
Weight { path: "submodule1.linear15.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.7.encoder_attn.k_proj.weight", Copy) },
Weight { path: "submodule1.linear16.bias", shape: Some(&[1024]), source: Param("model.layers.7.encoder_attn.v_proj.bias", Copy) },
Weight { path: "submodule1.linear16.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.7.encoder_attn.v_proj.weight", Copy) },
Weight { path: "submodule1.linear17.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.0.attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule1.linear18.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.0.attn.k_proj.MatMul.weight", Copy) },
Weight { path: "submodule1.linear19.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.0.attn.v_proj.MatMul.weight", Copy) },
Weight { path: "submodule1.linear2.bias", shape: Some(&[1024]), source: Param("model.layers.0.encoder_attn.v_proj.bias", Copy) },
Weight { path: "submodule1.linear2.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.0.encoder_attn.v_proj.weight", Copy) },
Weight { path: "submodule1.linear20.bias", shape: Some(&[1024]), source: Param("model.layers.0.attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule1.linear20.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.0.attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule1.linear3.bias", shape: Some(&[1024]), source: Param("model.layers.1.encoder_attn.k_proj.bias", Copy) },
Weight { path: "submodule1.linear3.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.1.encoder_attn.k_proj.weight", Copy) },
Weight { path: "submodule1.linear4.bias", shape: Some(&[1024]), source: Param("model.layers.1.encoder_attn.v_proj.bias", Copy) },
Weight { path: "submodule1.linear4.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.1.encoder_attn.v_proj.weight", Copy) },
Weight { path: "submodule1.linear5.bias", shape: Some(&[1024]), source: Param("model.layers.2.encoder_attn.k_proj.bias", Copy) },
Weight { path: "submodule1.linear5.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.2.encoder_attn.k_proj.weight", Copy) },
Weight { path: "submodule1.linear6.bias", shape: Some(&[1024]), source: Param("model.layers.2.encoder_attn.v_proj.bias", Copy) },
Weight { path: "submodule1.linear6.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.2.encoder_attn.v_proj.weight", Copy) },
Weight { path: "submodule1.linear7.bias", shape: Some(&[1024]), source: Param("model.layers.3.encoder_attn.k_proj.bias", Copy) },
Weight { path: "submodule1.linear7.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.3.encoder_attn.k_proj.weight", Copy) },
Weight { path: "submodule1.linear8.bias", shape: Some(&[1024]), source: Param("model.layers.3.encoder_attn.v_proj.bias", Copy) },
Weight { path: "submodule1.linear8.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.3.encoder_attn.v_proj.weight", Copy) },
Weight { path: "submodule1.linear9.bias", shape: Some(&[1024]), source: Param("model.layers.4.encoder_attn.k_proj.bias", Copy) },
Weight { path: "submodule1.linear9.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.4.encoder_attn.k_proj.weight", Copy) },
Weight { path: "submodule2.constant64", shape: Some(&[1024]), source: Param("model.layers.1.attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule2.constant65", shape: Some(&[1024]), source: Param("model.layers.1.attn.k_proj.Add.bias", Copy) },
Weight { path: "submodule2.constant66", shape: Some(&[1024]), source: Param("model.layers.1.attn.v_proj.Add.bias", Copy) },
Weight { path: "submodule2.constant86", shape: Some(&[1024]), source: Param("model.layers.2.attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule2.constant87", shape: Some(&[1024]), source: Param("model.layers.2.attn.k_proj.Add.bias", Copy) },
Weight { path: "submodule2.constant88", shape: Some(&[1024]), source: Param("model.layers.2.attn.v_proj.Add.bias", Copy) },
Weight { path: "submodule2.layernormalization3.beta", shape: Some(&[1024]), source: Param("model.layers.0.post_attn_ln_layernorm.bias", Copy) },
Weight { path: "submodule2.layernormalization3.gamma", shape: Some(&[1024]), source: Param("model.layers.0.post_attn_ln_layernorm.weight", Copy) },
Weight { path: "submodule2.layernormalization4.beta", shape: Some(&[1024]), source: Param("model.layers.0.final_ln_layernorm.bias", Copy) },
Weight { path: "submodule2.layernormalization4.gamma", shape: Some(&[1024]), source: Param("model.layers.0.final_ln_layernorm.weight", Copy) },
Weight { path: "submodule2.layernormalization5.beta", shape: Some(&[1024]), source: Param("model.layers.1.input_ln_layernorm.bias", Copy) },
Weight { path: "submodule2.layernormalization5.gamma", shape: Some(&[1024]), source: Param("model.layers.1.input_ln_layernorm.weight", Copy) },
Weight { path: "submodule2.layernormalization6.beta", shape: Some(&[1024]), source: Param("model.layers.1.post_attn_ln_layernorm.bias", Copy) },
Weight { path: "submodule2.layernormalization6.gamma", shape: Some(&[1024]), source: Param("model.layers.1.post_attn_ln_layernorm.weight", Copy) },
Weight { path: "submodule2.layernormalization7.beta", shape: Some(&[1024]), source: Param("model.layers.1.final_ln_layernorm.bias", Copy) },
Weight { path: "submodule2.layernormalization7.gamma", shape: Some(&[1024]), source: Param("model.layers.1.final_ln_layernorm.weight", Copy) },
Weight { path: "submodule2.layernormalization8.beta", shape: Some(&[1024]), source: Param("model.layers.2.input_ln_layernorm.bias", Copy) },
Weight { path: "submodule2.layernormalization8.gamma", shape: Some(&[1024]), source: Param("model.layers.2.input_ln_layernorm.weight", Copy) },
Weight { path: "submodule2.linear21.bias", shape: Some(&[1024]), source: Param("model.layers.0.encoder_attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule2.linear21.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.0.encoder_attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule2.linear22.bias", shape: Some(&[1024]), source: Param("model.layers.0.encoder_attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule2.linear22.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.0.encoder_attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule2.linear23.bias", shape: Some(&[4096]), source: Param("model.layers.0.mlp.fc1.Add.bias", Copy) },
Weight { path: "submodule2.linear23.weight", shape: Some(&[1024, 4096]), source: Param("model.layers.0.mlp.fc1.MatMul.weight", Copy) },
Weight { path: "submodule2.linear24.bias", shape: Some(&[1024]), source: Param("model.layers.0.mlp.fc2.Add.bias", Copy) },
Weight { path: "submodule2.linear24.weight", shape: Some(&[4096, 1024]), source: Param("model.layers.0.mlp.fc2.MatMul.weight", Copy) },
Weight { path: "submodule2.linear25.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.1.attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule2.linear26.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.1.attn.k_proj.MatMul.weight", Copy) },
Weight { path: "submodule2.linear27.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.1.attn.v_proj.MatMul.weight", Copy) },
Weight { path: "submodule2.linear28.bias", shape: Some(&[1024]), source: Param("model.layers.1.attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule2.linear28.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.1.attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule2.linear29.bias", shape: Some(&[1024]), source: Param("model.layers.1.encoder_attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule2.linear29.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.1.encoder_attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule2.linear30.bias", shape: Some(&[1024]), source: Param("model.layers.1.encoder_attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule2.linear30.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.1.encoder_attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule2.linear31.bias", shape: Some(&[4096]), source: Param("model.layers.1.mlp.fc1.Add.bias", Copy) },
Weight { path: "submodule2.linear31.weight", shape: Some(&[1024, 4096]), source: Param("model.layers.1.mlp.fc1.MatMul.weight", Copy) },
Weight { path: "submodule2.linear32.bias", shape: Some(&[1024]), source: Param("model.layers.1.mlp.fc2.Add.bias", Copy) },
Weight { path: "submodule2.linear32.weight", shape: Some(&[4096, 1024]), source: Param("model.layers.1.mlp.fc2.MatMul.weight", Copy) },
Weight { path: "submodule2.linear33.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.2.attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule2.linear34.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.2.attn.k_proj.MatMul.weight", Copy) },
Weight { path: "submodule2.linear35.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.2.attn.v_proj.MatMul.weight", Copy) },
Weight { path: "submodule2.linear36.bias", shape: Some(&[1024]), source: Param("model.layers.2.attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule2.linear36.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.2.attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule3.constant108", shape: Some(&[1024]), source: Param("model.layers.3.attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule3.constant109", shape: Some(&[1024]), source: Param("model.layers.3.attn.k_proj.Add.bias", Copy) },
Weight { path: "submodule3.constant110", shape: Some(&[1024]), source: Param("model.layers.3.attn.v_proj.Add.bias", Copy) },
Weight { path: "submodule3.constant130", shape: Some(&[1024]), source: Param("model.layers.4.attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule3.constant131", shape: Some(&[1024]), source: Param("model.layers.4.attn.k_proj.Add.bias", Copy) },
Weight { path: "submodule3.constant132", shape: Some(&[1024]), source: Param("model.layers.4.attn.v_proj.Add.bias", Copy) },
Weight { path: "submodule3.layernormalization10.beta", shape: Some(&[1024]), source: Param("model.layers.2.final_ln_layernorm.bias", Copy) },
Weight { path: "submodule3.layernormalization10.gamma", shape: Some(&[1024]), source: Param("model.layers.2.final_ln_layernorm.weight", Copy) },
Weight { path: "submodule3.layernormalization11.beta", shape: Some(&[1024]), source: Param("model.layers.3.input_ln_layernorm.bias", Copy) },
Weight { path: "submodule3.layernormalization11.gamma", shape: Some(&[1024]), source: Param("model.layers.3.input_ln_layernorm.weight", Copy) },
Weight { path: "submodule3.layernormalization12.beta", shape: Some(&[1024]), source: Param("model.layers.3.post_attn_ln_layernorm.bias", Copy) },
Weight { path: "submodule3.layernormalization12.gamma", shape: Some(&[1024]), source: Param("model.layers.3.post_attn_ln_layernorm.weight", Copy) },
Weight { path: "submodule3.layernormalization13.beta", shape: Some(&[1024]), source: Param("model.layers.3.final_ln_layernorm.bias", Copy) },
Weight { path: "submodule3.layernormalization13.gamma", shape: Some(&[1024]), source: Param("model.layers.3.final_ln_layernorm.weight", Copy) },
Weight { path: "submodule3.layernormalization14.beta", shape: Some(&[1024]), source: Param("model.layers.4.input_ln_layernorm.bias", Copy) },
Weight { path: "submodule3.layernormalization14.gamma", shape: Some(&[1024]), source: Param("model.layers.4.input_ln_layernorm.weight", Copy) },
Weight { path: "submodule3.layernormalization9.beta", shape: Some(&[1024]), source: Param("model.layers.2.post_attn_ln_layernorm.bias", Copy) },
Weight { path: "submodule3.layernormalization9.gamma", shape: Some(&[1024]), source: Param("model.layers.2.post_attn_ln_layernorm.weight", Copy) },
Weight { path: "submodule3.linear37.bias", shape: Some(&[1024]), source: Param("model.layers.2.encoder_attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule3.linear37.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.2.encoder_attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule3.linear38.bias", shape: Some(&[1024]), source: Param("model.layers.2.encoder_attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule3.linear38.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.2.encoder_attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule3.linear39.bias", shape: Some(&[4096]), source: Param("model.layers.2.mlp.fc1.Add.bias", Copy) },
Weight { path: "submodule3.linear39.weight", shape: Some(&[1024, 4096]), source: Param("model.layers.2.mlp.fc1.MatMul.weight", Copy) },
Weight { path: "submodule3.linear40.bias", shape: Some(&[1024]), source: Param("model.layers.2.mlp.fc2.Add.bias", Copy) },
Weight { path: "submodule3.linear40.weight", shape: Some(&[4096, 1024]), source: Param("model.layers.2.mlp.fc2.MatMul.weight", Copy) },
Weight { path: "submodule3.linear41.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.3.attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule3.linear42.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.3.attn.k_proj.MatMul.weight", Copy) },
Weight { path: "submodule3.linear43.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.3.attn.v_proj.MatMul.weight", Copy) },
Weight { path: "submodule3.linear44.bias", shape: Some(&[1024]), source: Param("model.layers.3.attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule3.linear44.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.3.attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule3.linear45.bias", shape: Some(&[1024]), source: Param("model.layers.3.encoder_attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule3.linear45.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.3.encoder_attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule3.linear46.bias", shape: Some(&[1024]), source: Param("model.layers.3.encoder_attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule3.linear46.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.3.encoder_attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule3.linear47.bias", shape: Some(&[4096]), source: Param("model.layers.3.mlp.fc1.Add.bias", Copy) },
Weight { path: "submodule3.linear47.weight", shape: Some(&[1024, 4096]), source: Param("model.layers.3.mlp.fc1.MatMul.weight", Copy) },
Weight { path: "submodule3.linear48.bias", shape: Some(&[1024]), source: Param("model.layers.3.mlp.fc2.Add.bias", Copy) },
Weight { path: "submodule3.linear48.weight", shape: Some(&[4096, 1024]), source: Param("model.layers.3.mlp.fc2.MatMul.weight", Copy) },
Weight { path: "submodule3.linear49.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.4.attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule3.linear50.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.4.attn.k_proj.MatMul.weight", Copy) },
Weight { path: "submodule3.linear51.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.4.attn.v_proj.MatMul.weight", Copy) },
Weight { path: "submodule3.linear52.bias", shape: Some(&[1024]), source: Param("model.layers.4.attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule3.linear52.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.4.attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.constant152", shape: Some(&[1024]), source: Param("model.layers.5.attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule4.constant153", shape: Some(&[1024]), source: Param("model.layers.5.attn.k_proj.Add.bias", Copy) },
Weight { path: "submodule4.constant154", shape: Some(&[1024]), source: Param("model.layers.5.attn.v_proj.Add.bias", Copy) },
Weight { path: "submodule4.constant174", shape: Some(&[1024]), source: Param("model.layers.6.attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule4.constant175", shape: Some(&[1024]), source: Param("model.layers.6.attn.k_proj.Add.bias", Copy) },
Weight { path: "submodule4.constant176", shape: Some(&[1024]), source: Param("model.layers.6.attn.v_proj.Add.bias", Copy) },
Weight { path: "submodule4.constant196", shape: Some(&[1024]), source: Param("model.layers.7.attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule4.constant197", shape: Some(&[1024]), source: Param("model.layers.7.attn.k_proj.Add.bias", Copy) },
Weight { path: "submodule4.constant198", shape: Some(&[1024]), source: Param("model.layers.7.attn.v_proj.Add.bias", Copy) },
Weight { path: "submodule4.layernormalization15.beta", shape: Some(&[1024]), source: Param("model.layers.4.post_attn_ln_layernorm.bias", Copy) },
Weight { path: "submodule4.layernormalization15.gamma", shape: Some(&[1024]), source: Param("model.layers.4.post_attn_ln_layernorm.weight", Copy) },
Weight { path: "submodule4.layernormalization16.beta", shape: Some(&[1024]), source: Param("model.layers.4.final_ln_layernorm.bias", Copy) },
Weight { path: "submodule4.layernormalization16.gamma", shape: Some(&[1024]), source: Param("model.layers.4.final_ln_layernorm.weight", Copy) },
Weight { path: "submodule4.layernormalization17.beta", shape: Some(&[1024]), source: Param("model.layers.5.input_ln_layernorm.bias", Copy) },
Weight { path: "submodule4.layernormalization17.gamma", shape: Some(&[1024]), source: Param("model.layers.5.input_ln_layernorm.weight", Copy) },
Weight { path: "submodule4.layernormalization18.beta", shape: Some(&[1024]), source: Param("model.layers.5.post_attn_ln_layernorm.bias", Copy) },
Weight { path: "submodule4.layernormalization18.gamma", shape: Some(&[1024]), source: Param("model.layers.5.post_attn_ln_layernorm.weight", Copy) },
Weight { path: "submodule4.layernormalization19.beta", shape: Some(&[1024]), source: Param("model.layers.5.final_ln_layernorm.bias", Copy) },
Weight { path: "submodule4.layernormalization19.gamma", shape: Some(&[1024]), source: Param("model.layers.5.final_ln_layernorm.weight", Copy) },
Weight { path: "submodule4.layernormalization20.beta", shape: Some(&[1024]), source: Param("model.layers.6.input_ln_layernorm.bias", Copy) },
Weight { path: "submodule4.layernormalization20.gamma", shape: Some(&[1024]), source: Param("model.layers.6.input_ln_layernorm.weight", Copy) },
Weight { path: "submodule4.layernormalization21.beta", shape: Some(&[1024]), source: Param("model.layers.6.post_attn_ln_layernorm.bias", Copy) },
Weight { path: "submodule4.layernormalization21.gamma", shape: Some(&[1024]), source: Param("model.layers.6.post_attn_ln_layernorm.weight", Copy) },
Weight { path: "submodule4.layernormalization22.beta", shape: Some(&[1024]), source: Param("model.layers.6.final_ln_layernorm.bias", Copy) },
Weight { path: "submodule4.layernormalization22.gamma", shape: Some(&[1024]), source: Param("model.layers.6.final_ln_layernorm.weight", Copy) },
Weight { path: "submodule4.layernormalization23.beta", shape: Some(&[1024]), source: Param("model.layers.7.input_ln_layernorm.bias", Copy) },
Weight { path: "submodule4.layernormalization23.gamma", shape: Some(&[1024]), source: Param("model.layers.7.input_ln_layernorm.weight", Copy) },
Weight { path: "submodule4.layernormalization24.beta", shape: Some(&[1024]), source: Param("model.layers.7.post_attn_ln_layernorm.bias", Copy) },
Weight { path: "submodule4.layernormalization24.gamma", shape: Some(&[1024]), source: Param("model.layers.7.post_attn_ln_layernorm.weight", Copy) },
Weight { path: "submodule4.layernormalization25.beta", shape: Some(&[1024]), source: Param("model.layers.7.final_ln_layernorm.bias", Copy) },
Weight { path: "submodule4.layernormalization25.gamma", shape: Some(&[1024]), source: Param("model.layers.7.final_ln_layernorm.weight", Copy) },
Weight { path: "submodule4.layernormalization26.beta", shape: Some(&[1024]), source: Param("model.layers.8.final_norm_layernorm.bias", Copy) },
Weight { path: "submodule4.layernormalization26.gamma", shape: Some(&[1024]), source: Param("model.layers.8.final_norm_layernorm.weight", Copy) },
Weight { path: "submodule4.linear53.bias", shape: Some(&[1024]), source: Param("model.layers.4.encoder_attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule4.linear53.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.4.encoder_attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear54.bias", shape: Some(&[1024]), source: Param("model.layers.4.encoder_attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule4.linear54.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.4.encoder_attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear55.bias", shape: Some(&[4096]), source: Param("model.layers.4.mlp.fc1.Add.bias", Copy) },
Weight { path: "submodule4.linear55.weight", shape: Some(&[1024, 4096]), source: Param("model.layers.4.mlp.fc1.MatMul.weight", Copy) },
Weight { path: "submodule4.linear56.bias", shape: Some(&[1024]), source: Param("model.layers.4.mlp.fc2.Add.bias", Copy) },
Weight { path: "submodule4.linear56.weight", shape: Some(&[4096, 1024]), source: Param("model.layers.4.mlp.fc2.MatMul.weight", Copy) },
Weight { path: "submodule4.linear57.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.5.attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear58.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.5.attn.k_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear59.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.5.attn.v_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear60.bias", shape: Some(&[1024]), source: Param("model.layers.5.attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule4.linear60.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.5.attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear61.bias", shape: Some(&[1024]), source: Param("model.layers.5.encoder_attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule4.linear61.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.5.encoder_attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear62.bias", shape: Some(&[1024]), source: Param("model.layers.5.encoder_attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule4.linear62.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.5.encoder_attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear63.bias", shape: Some(&[4096]), source: Param("model.layers.5.mlp.fc1.Add.bias", Copy) },
Weight { path: "submodule4.linear63.weight", shape: Some(&[1024, 4096]), source: Param("model.layers.5.mlp.fc1.MatMul.weight", Copy) },
Weight { path: "submodule4.linear64.bias", shape: Some(&[1024]), source: Param("model.layers.5.mlp.fc2.Add.bias", Copy) },
Weight { path: "submodule4.linear64.weight", shape: Some(&[4096, 1024]), source: Param("model.layers.5.mlp.fc2.MatMul.weight", Copy) },
Weight { path: "submodule4.linear65.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.6.attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear66.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.6.attn.k_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear67.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.6.attn.v_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear68.bias", shape: Some(&[1024]), source: Param("model.layers.6.attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule4.linear68.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.6.attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear69.bias", shape: Some(&[1024]), source: Param("model.layers.6.encoder_attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule4.linear69.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.6.encoder_attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear70.bias", shape: Some(&[1024]), source: Param("model.layers.6.encoder_attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule4.linear70.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.6.encoder_attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear71.bias", shape: Some(&[4096]), source: Param("model.layers.6.mlp.fc1.Add.bias", Copy) },
Weight { path: "submodule4.linear71.weight", shape: Some(&[1024, 4096]), source: Param("model.layers.6.mlp.fc1.MatMul.weight", Copy) },
Weight { path: "submodule4.linear72.bias", shape: Some(&[1024]), source: Param("model.layers.6.mlp.fc2.Add.bias", Copy) },
Weight { path: "submodule4.linear72.weight", shape: Some(&[4096, 1024]), source: Param("model.layers.6.mlp.fc2.MatMul.weight", Copy) },
Weight { path: "submodule4.linear73.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.7.attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear74.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.7.attn.k_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear75.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.7.attn.v_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear76.bias", shape: Some(&[1024]), source: Param("model.layers.7.attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule4.linear76.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.7.attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear77.bias", shape: Some(&[1024]), source: Param("model.layers.7.encoder_attn.q_proj.Add.bias", Copy) },
Weight { path: "submodule4.linear77.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.7.encoder_attn.q_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear78.bias", shape: Some(&[1024]), source: Param("model.layers.7.encoder_attn.o_proj.Add.bias", Copy) },
Weight { path: "submodule4.linear78.weight", shape: Some(&[1024, 1024]), source: Param("model.layers.7.encoder_attn.o_proj.MatMul.weight", Copy) },
Weight { path: "submodule4.linear79.bias", shape: Some(&[4096]), source: Param("model.layers.7.mlp.fc1.Add.bias", Copy) },
Weight { path: "submodule4.linear79.weight", shape: Some(&[1024, 4096]), source: Param("model.layers.7.mlp.fc1.MatMul.weight", Copy) },
Weight { path: "submodule4.linear80.bias", shape: Some(&[1024]), source: Param("model.layers.7.mlp.fc2.Add.bias", Copy) },
Weight { path: "submodule4.linear80.weight", shape: Some(&[4096, 1024]), source: Param("model.layers.7.mlp.fc2.MatMul.weight", Copy) },
Weight { path: "submodule4.linear81.bias", shape: Some(&[16384]), source: Param("lm_head.Add.bias", Copy) },
Weight { path: "submodule4.linear81.weight", shape: Some(&[1024, 16384]), source: Param("lm_head.MatMul.weight", Copy) },
];