Weight listing: Qwen GPTQ-2 W2A16
Source reports/results/QWEN_GPTQ2_20261116/weight_listing_20261117/qwen_gptq2_W2A16.tsv · source md5 888bd1f74178a6cf4eac5037f493d8fb · EAGLE4 FPGA archive, 2026-11-18
efficient_lm_head.g_idx I32 [128] 128 512
efficient_lm_head.qweight I32 [8, 248320] 1986560 7946240
efficient_lm_head.qzeros I32 [1, 15520] 15520 62080
efficient_lm_head.scales BF16 [1, 248320] 248320 496640
efficient_lm_head_down_proj.weight BF16 [128, 5120] 655360 1310720
embed_tokens.weight BF16 [248320, 5120] 1271398400 2542796800
fc.g_idx I32 [15360] 15360 61440
fc.qweight I32 [960, 5120] 4915200 19660800
fc.qzeros I32 [120, 320] 38400 153600
fc.scales BF16 [120, 5120] 614400 1228800
fc_norm.0.weight BF16 [5120] 5120 10240
fc_norm.1.weight BF16 [5120] 5120 10240
fc_norm.2.weight BF16 [5120] 5120 10240
layers.0.hidden_norm.weight BF16 [5120] 5120 10240
layers.0.input_layernorm.weight BF16 [5120] 5120 10240
layers.0.mlp.down_proj.g_idx I32 [16384] 16384 65536
layers.0.mlp.down_proj.qweight I32 [1024, 5120] 5242880 20971520
layers.0.mlp.down_proj.qzeros I32 [128, 320] 40960 163840
layers.0.mlp.down_proj.scales BF16 [128, 5120] 655360 1310720
layers.0.mlp.gate_proj.g_idx I32 [5120] 5120 20480
layers.0.mlp.gate_proj.qweight I32 [320, 16384] 5242880 20971520
layers.0.mlp.gate_proj.qzeros I32 [40, 1024] 40960 163840
layers.0.mlp.gate_proj.scales BF16 [40, 16384] 655360 1310720
layers.0.mlp.up_proj.g_idx I32 [5120] 5120 20480
layers.0.mlp.up_proj.qweight I32 [320, 16384] 5242880 20971520
layers.0.mlp.up_proj.qzeros I32 [40, 1024] 40960 163840
layers.0.mlp.up_proj.scales BF16 [40, 16384] 655360 1310720
layers.0.post_attention_layernorm.weight BF16 [5120] 5120 10240
layers.0.self_attn.k_proj.g_idx I32 [10240] 10240 40960
layers.0.self_attn.k_proj.qweight I32 [640, 5120] 3276800 13107200
layers.0.self_attn.k_proj.qzeros I32 [80, 320] 25600 102400
layers.0.self_attn.k_proj.scales BF16 [80, 5120] 409600 819200
layers.0.self_attn.o_proj.g_idx I32 [5120] 5120 20480
layers.0.self_attn.o_proj.qweight I32 [320, 5120] 1638400 6553600
layers.0.self_attn.o_proj.qzeros I32 [40, 320] 12800 51200
layers.0.self_attn.o_proj.scales BF16 [40, 5120] 204800 409600
layers.0.self_attn.q_proj.g_idx I32 [10240] 10240 40960
layers.0.self_attn.q_proj.qweight I32 [640, 5120] 3276800 13107200
layers.0.self_attn.q_proj.qzeros I32 [80, 320] 25600 102400
layers.0.self_attn.q_proj.scales BF16 [80, 5120] 409600 819200
layers.0.self_attn.v_proj.g_idx I32 [10240] 10240 40960
layers.0.self_attn.v_proj.qweight I32 [640, 5120] 3276800 13107200
layers.0.self_attn.v_proj.qzeros I32 [80, 320] 25600 102400
layers.0.self_attn.v_proj.scales BF16 [80, 5120] 409600 819200
lm_head.g_idx I32 [5120] 5120 20480
lm_head.qweight I32 [320, 248320] 79462400 317849600
lm_head.qzeros I32 [40, 15520] 620800 2483200
lm_head.scales BF16 [40, 248320] 9932800 19865600
norm.weight BF16 [5120] 5120 10240
TOTAL_BYTES 3030697088