{ "producer": { "name": "modelopt", "version": "0.47.0.dev393+ga4bc45b30.d20260828", "in_proj_layout": "split-qkv-bfg" }, "quantization": { "quant_algo": "MIXED_PRECISION", "kv_cache_quant_algo": "FP8", "group_size": 16, "exclude_modules": [ "model.language_model.embed_tokens", "model.language_model.layers.10.mlp.gate", "model.language_model.layers.10.mlp.shared_experts*", "model.language_model.layers.11.mlp.gate", "model.language_model.layers.11.mlp.shared_experts*", "model.language_model.layers.12.mlp.gate", "model.language_model.layers.12.mlp.shared_experts*", "model.language_model.layers.13.mlp.gate", "model.language_model.layers.13.mlp.shared_experts*", "model.language_model.layers.14.mlp.gate", "model.language_model.layers.14.mlp.shared_experts*", "model.language_model.layers.15.mlp.gate", "model.language_model.layers.15.mlp.shared_experts*", "model.language_model.layers.16.mlp.gate", "model.language_model.layers.16.mlp.shared_experts*", "model.language_model.layers.17.mlp.gate", "model.language_model.layers.17.mlp.shared_experts*", "model.language_model.layers.18.mlp.gate", "model.language_model.layers.18.mlp.shared_experts*", "model.language_model.layers.19.mlp.gate", "model.language_model.layers.19.mlp.shared_experts*", "model.language_model.layers.20.mlp.gate", "model.language_model.layers.20.mlp.shared_experts*", "model.language_model.layers.21.mlp.gate", "model.language_model.layers.21.mlp.shared_experts*", "model.language_model.layers.22.mlp.gate", "model.language_model.layers.22.mlp.shared_experts*", "model.language_model.layers.23.mlp.gate", "model.language_model.layers.23.mlp.shared_experts*", "model.language_model.layers.24.mlp.gate", "model.language_model.layers.24.mlp.shared_experts*", "model.language_model.layers.25.mlp.gate", "model.language_model.layers.25.mlp.shared_experts*", "model.language_model.layers.26.mlp.gate", "model.language_model.layers.26.mlp.shared_experts*", "model.language_model.layers.27.mlp.gate", "model.language_model.layers.27.mlp.shared_experts*", "model.language_model.layers.28.mlp.gate", "model.language_model.layers.28.mlp.shared_experts*", "model.language_model.layers.29.mlp.gate", "model.language_model.layers.29.mlp.shared_experts*", "model.language_model.layers.3.mlp.gate", "model.language_model.layers.3.mlp.shared_experts*", "model.language_model.layers.30.mlp.gate", "model.language_model.layers.30.mlp.shared_experts*", "model.language_model.layers.31.mlp.gate", "model.language_model.layers.31.mlp.shared_experts*", "model.language_model.layers.32.mlp.gate", "model.language_model.layers.32.mlp.shared_experts*", "model.language_model.layers.33.mlp.gate", "model.language_model.layers.33.mlp.shared_experts*", "model.language_model.layers.34.mlp.gate", "model.language_model.layers.34.mlp.shared_experts*", "model.language_model.layers.35.mlp.gate", "model.language_model.layers.35.mlp.shared_experts*", "model.language_model.layers.36.mlp.gate", "model.language_model.layers.36.mlp.shared_experts*", "model.language_model.layers.37.mlp.gate", "model.language_model.layers.37.mlp.shared_experts*", "model.language_model.layers.38.mlp.gate", "model.language_model.layers.38.mlp.shared_experts*", "model.language_model.layers.39.mlp.gate", "model.language_model.layers.39.mlp.shared_experts*", "model.language_model.layers.4.mlp.gate", "model.language_model.layers.4.mlp.shared_experts*", "model.language_model.layers.40.mlp.gate", "model.language_model.layers.40.mlp.shared_experts*", "model.language_model.layers.41.mlp.gate", "model.language_model.layers.41.mlp.shared_experts*", "model.language_model.layers.42.mlp.gate", "model.language_model.layers.42.mlp.shared_experts*", "model.language_model.layers.43.mlp.gate", "model.language_model.layers.43.mlp.shared_experts*", "model.language_model.layers.44.mlp.gate", "model.language_model.layers.44.mlp.shared_experts*", "model.language_model.layers.5.mlp.gate", "model.language_model.layers.5.mlp.shared_experts*", "model.language_model.layers.6.mlp.gate", "model.language_model.layers.6.mlp.shared_experts*", "model.language_model.layers.7.mlp.gate", "model.language_model.layers.7.mlp.shared_experts*", "model.language_model.layers.8.mlp.gate", "model.language_model.layers.8.mlp.shared_experts*", "model.language_model.layers.9.mlp.gate", "model.language_model.layers.9.mlp.shared_experts*", "model.visual*", "model.language_model.layers.45*" ], "quantized_layers": { "lm_head": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.0.mlp.down_proj": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.0.mlp.gate_proj": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.0.mlp.up_proj": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.0.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.0.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.0.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.0.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.0.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.0.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.0.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.1.mlp.down_proj": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.1.mlp.gate_proj": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.1.mlp.up_proj": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.1.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.1.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.1.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.1.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.1.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.1.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.1.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.10.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.10.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.10.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.10.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.10.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.10.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.10.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.10.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.11.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.11.self_attn.fused_qkv_a_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.11.self_attn.indexer.wq_b": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.11.self_attn.kv_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.11.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.11.self_attn.q_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.12.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.12.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.12.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.12.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.12.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.12.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.12.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.12.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.13.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.13.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.13.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.13.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.13.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.13.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.13.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.13.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.14.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.14.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.14.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.14.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.14.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.14.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.14.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.14.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.15.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.15.self_attn.fused_qkv_a_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.15.self_attn.indexer.wq_b": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.15.self_attn.kv_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.15.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.15.self_attn.q_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.16.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.16.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.16.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.16.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.16.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.16.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.16.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.16.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.17.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.17.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.17.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.17.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.17.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.17.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.17.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.17.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.18.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.18.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.18.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.18.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.18.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.18.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.18.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.18.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.19.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.19.self_attn.fused_qkv_a_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.19.self_attn.indexer.wq_b": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.19.self_attn.kv_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.19.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.19.self_attn.q_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.2.mlp.down_proj": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.2.mlp.gate_proj": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.2.mlp.up_proj": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.2.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.2.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.2.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.2.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.2.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.2.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.2.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.20.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.20.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.20.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.20.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.20.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.20.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.20.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.20.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.21.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.21.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.21.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.21.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.21.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.21.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.21.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.21.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.22.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.22.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.22.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.22.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.22.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.22.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.22.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.22.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.23.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.23.self_attn.fused_qkv_a_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.23.self_attn.indexer.wq_b": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.23.self_attn.kv_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.23.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.23.self_attn.q_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.24.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.24.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.24.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.24.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.24.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.24.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.24.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.24.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.25.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.25.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.25.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.25.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.25.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.25.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.25.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.25.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.26.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.26.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.26.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.26.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.26.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.26.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.26.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.26.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.27.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.27.self_attn.fused_qkv_a_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.27.self_attn.indexer.wq_b": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.27.self_attn.kv_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.27.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.27.self_attn.q_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.28.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.28.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.28.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.28.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.28.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.28.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.28.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.28.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.29.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.29.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.29.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.29.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.29.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.29.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.29.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.29.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.3.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.3.self_attn.fused_qkv_a_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.3.self_attn.indexer.wq_b": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.3.self_attn.kv_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.3.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.3.self_attn.q_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.30.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.30.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.30.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.30.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.30.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.30.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.30.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.30.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.31.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.31.self_attn.fused_qkv_a_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.31.self_attn.indexer.wq_b": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.31.self_attn.kv_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.31.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.31.self_attn.q_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.32.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.32.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.32.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.32.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.32.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.32.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.32.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.32.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.33.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.33.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.33.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.33.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.33.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.33.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.33.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.33.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.34.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.34.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.34.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.34.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.34.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.34.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.34.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.34.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.35.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.35.self_attn.fused_qkv_a_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.35.self_attn.indexer.wq_b": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.35.self_attn.kv_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.35.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.35.self_attn.q_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.36.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.36.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.36.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.36.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.36.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.36.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.36.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.36.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.37.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.37.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.37.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.37.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.37.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.37.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.37.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.37.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.38.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.38.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.38.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.38.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.38.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.38.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.38.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.38.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.39.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.39.self_attn.fused_qkv_a_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.39.self_attn.indexer.wq_b": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.39.self_attn.kv_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.39.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.39.self_attn.q_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.4.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.4.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.4.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.4.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.4.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.4.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.4.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.4.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.40.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.40.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.40.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.40.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.40.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.40.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.40.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.40.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.41.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.41.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.41.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.41.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.41.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.41.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.41.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.41.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.42.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.42.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.42.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.42.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.42.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.42.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.42.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.42.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.43.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.43.self_attn.fused_qkv_a_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.43.self_attn.indexer.wq_b": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.43.self_attn.kv_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.43.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.43.self_attn.q_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.44.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.44.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.44.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.44.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.44.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.44.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.44.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.44.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.5.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.5.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.5.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.5.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.5.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.5.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.5.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.5.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.6.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.6.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.6.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.6.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.6.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.6.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.6.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.6.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.7.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.7.self_attn.fused_qkv_a_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.7.self_attn.indexer.wq_b": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.7.self_attn.kv_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.7.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.7.self_attn.q_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.8.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.8.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.8.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.8.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.8.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.8.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.8.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.8.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.9.mlp.experts": { "quant_algo": "NVFP4", "group_size": 16 }, "model.language_model.layers.9.self_attn.f_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.9.self_attn.g_b_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.9.self_attn.q_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.9.self_attn.k_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.9.self_attn.v_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.9.self_attn.in_proj_bfg_a": { "quant_algo": "W4A16_NVFP4", "group_size": 16 }, "model.language_model.layers.9.self_attn.o_proj": { "quant_algo": "W4A16_NVFP4", "group_size": 16 } } } }