Models
Docs
Pricing
Sign in
Download
Models
Download
Docs
Pricing
Sign in
free02
/
glm5.1
:part2
149
Downloads
Updated
1 month ago
Cancel
glm5.1:part2
...
/
model
cdab9f1a0484 · 50GB
Metadata
split.count
6
6
split.no
1
1
split.tensors.count
1809
1809
Tensor
Name
Type
Shape
token_embd.weight
Q5_K
Q5_K
[6144, 154880]
blk.0
blk.0.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.0.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.0.attn_kv_a_norm.weight
F32
F32
[512]
blk.0.attn_norm.weight
F32
F32
[6144]
blk.0.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.0.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.0.attn_q_a_norm.weight
F32
F32
[2048]
blk.0.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.0.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.0.ffn_down.weight
Q6_K
Q6_K
[12288, 6144]
blk.0.ffn_gate.weight
Q5_K
Q5_K
[6144, 12288]
blk.0.ffn_norm.weight
F32
F32
[6144]
blk.0.ffn_up.weight
Q5_K
Q5_K
[6144, 12288]
blk.0.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.0.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.0.indexer.k_norm.bias
F32
F32
[128]
blk.0.indexer.k_norm.weight
F32
F32
[128]
blk.0.indexer.proj.weight
F32
F32
[6144, 32]
blk.1
blk.1.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.1.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.1.attn_kv_a_norm.weight
F32
F32
[512]
blk.1.attn_norm.weight
F32
F32
[6144]
blk.1.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.1.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.1.attn_q_a_norm.weight
F32
F32
[2048]
blk.1.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.1.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.1.ffn_down.weight
Q6_K
Q6_K
[12288, 6144]
blk.1.ffn_gate.weight
Q5_K
Q5_K
[6144, 12288]
blk.1.ffn_norm.weight
F32
F32
[6144]
blk.1.ffn_up.weight
Q5_K
Q5_K
[6144, 12288]
blk.1.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.1.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.1.indexer.k_norm.bias
F32
F32
[128]
blk.1.indexer.k_norm.weight
F32
F32
[128]
blk.1.indexer.proj.weight
F32
F32
[6144, 32]
blk.2
blk.2.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.2.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.2.attn_kv_a_norm.weight
F32
F32
[512]
blk.2.attn_norm.weight
F32
F32
[6144]
blk.2.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.2.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.2.attn_q_a_norm.weight
F32
F32
[2048]
blk.2.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.2.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.2.ffn_down.weight
Q6_K
Q6_K
[12288, 6144]
blk.2.ffn_gate.weight
Q5_K
Q5_K
[6144, 12288]
blk.2.ffn_norm.weight
F32
F32
[6144]
blk.2.ffn_up.weight
Q5_K
Q5_K
[6144, 12288]
blk.2.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.2.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.2.indexer.k_norm.bias
F32
F32
[128]
blk.2.indexer.k_norm.weight
F32
F32
[128]
blk.2.indexer.proj.weight
F32
F32
[6144, 32]
blk.3
blk.3.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.3.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.3.attn_kv_a_norm.weight
F32
F32
[512]
blk.3.attn_norm.weight
F32
F32
[6144]
blk.3.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.3.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.3.attn_q_a_norm.weight
F32
F32
[2048]
blk.3.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.3.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.3.exp_probs_b.bias
F32
F32
[256]
blk.3.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.3.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.3.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.3.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.3.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.3.ffn_norm.weight
F32
F32
[6144]
blk.3.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.3.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.3.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.3.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.3.indexer.k_norm.bias
F32
F32
[128]
blk.3.indexer.k_norm.weight
F32
F32
[128]
blk.3.indexer.proj.weight
F32
F32
[6144, 32]
blk.4
blk.4.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.4.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.4.attn_kv_a_norm.weight
F32
F32
[512]
blk.4.attn_norm.weight
F32
F32
[6144]
blk.4.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.4.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.4.attn_q_a_norm.weight
F32
F32
[2048]
blk.4.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.4.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.4.exp_probs_b.bias
F32
F32
[256]
blk.4.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.4.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.4.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.4.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.4.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.4.ffn_norm.weight
F32
F32
[6144]
blk.4.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.4.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.4.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.4.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.4.indexer.k_norm.bias
F32
F32
[128]
blk.4.indexer.k_norm.weight
F32
F32
[128]
blk.4.indexer.proj.weight
F32
F32
[6144, 32]
blk.5
blk.5.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.5.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.5.attn_kv_a_norm.weight
F32
F32
[512]
blk.5.attn_norm.weight
F32
F32
[6144]
blk.5.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.5.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.5.attn_q_a_norm.weight
F32
F32
[2048]
blk.5.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.5.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.5.exp_probs_b.bias
F32
F32
[256]
blk.5.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.5.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.5.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.5.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.5.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.5.ffn_norm.weight
F32
F32
[6144]
blk.5.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.5.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.5.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.5.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.5.indexer.k_norm.bias
F32
F32
[128]
blk.5.indexer.k_norm.weight
F32
F32
[128]
blk.5.indexer.proj.weight
F32
F32
[6144, 32]
blk.6
blk.6.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.6.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.6.attn_kv_a_norm.weight
F32
F32
[512]
blk.6.attn_norm.weight
F32
F32
[6144]
blk.6.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.6.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.6.attn_q_a_norm.weight
F32
F32
[2048]
blk.6.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.6.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.6.exp_probs_b.bias
F32
F32
[256]
blk.6.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.6.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.6.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.6.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.6.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.6.ffn_norm.weight
F32
F32
[6144]
blk.6.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.6.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.6.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.6.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.6.indexer.k_norm.bias
F32
F32
[128]
blk.6.indexer.k_norm.weight
F32
F32
[128]
blk.6.indexer.proj.weight
F32
F32
[6144, 32]
blk.7
blk.7.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.7.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.7.attn_kv_a_norm.weight
F32
F32
[512]
blk.7.attn_norm.weight
F32
F32
[6144]
blk.7.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.7.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.7.attn_q_a_norm.weight
F32
F32
[2048]
blk.7.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.7.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.7.exp_probs_b.bias
F32
F32
[256]
blk.7.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.7.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.7.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.7.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.7.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.7.ffn_norm.weight
F32
F32
[6144]
blk.7.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.7.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.7.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.7.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.7.indexer.k_norm.bias
F32
F32
[128]
blk.7.indexer.k_norm.weight
F32
F32
[128]
blk.7.indexer.proj.weight
F32
F32
[6144, 32]
blk.8
blk.8.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.8.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.8.attn_kv_a_norm.weight
F32
F32
[512]
blk.8.attn_norm.weight
F32
F32
[6144]
blk.8.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.8.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.8.attn_q_a_norm.weight
F32
F32
[2048]
blk.8.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.8.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.8.exp_probs_b.bias
F32
F32
[256]
blk.8.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.8.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.8.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.8.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.8.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.8.ffn_norm.weight
F32
F32
[6144]
blk.8.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.8.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.8.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.8.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.8.indexer.k_norm.bias
F32
F32
[128]
blk.8.indexer.k_norm.weight
F32
F32
[128]
blk.8.indexer.proj.weight
F32
F32
[6144, 32]
blk.9
blk.9.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.9.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.9.attn_kv_a_norm.weight
F32
F32
[512]
blk.9.attn_norm.weight
F32
F32
[6144]
blk.9.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.9.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.9.attn_q_a_norm.weight
F32
F32
[2048]
blk.9.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.9.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.9.exp_probs_b.bias
F32
F32
[256]
blk.9.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.9.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.9.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.9.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.9.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.9.ffn_norm.weight
F32
F32
[6144]
blk.9.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.9.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.9.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.9.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.9.indexer.k_norm.bias
F32
F32
[128]
blk.9.indexer.k_norm.weight
F32
F32
[128]
blk.9.indexer.proj.weight
F32
F32
[6144, 32]
blk.10
blk.10.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.10.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.10.attn_kv_a_norm.weight
F32
F32
[512]
blk.10.attn_norm.weight
F32
F32
[6144]
blk.10.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.10.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.10.attn_q_a_norm.weight
F32
F32
[2048]
blk.10.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.10.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.10.exp_probs_b.bias
F32
F32
[256]
blk.10.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.10.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.10.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.10.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.10.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.10.ffn_norm.weight
F32
F32
[6144]
blk.10.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.10.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.10.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.10.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.10.indexer.k_norm.bias
F32
F32
[128]
blk.10.indexer.k_norm.weight
F32
F32
[128]
blk.10.indexer.proj.weight
F32
F32
[6144, 32]
blk.11
blk.11.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.11.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.11.attn_kv_a_norm.weight
F32
F32
[512]
blk.11.attn_norm.weight
F32
F32
[6144]
blk.11.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.11.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.11.attn_q_a_norm.weight
F32
F32
[2048]
blk.11.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.11.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.11.exp_probs_b.bias
F32
F32
[256]
blk.11.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.11.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.11.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.11.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.11.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.11.ffn_norm.weight
F32
F32
[6144]
blk.11.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.11.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.11.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.11.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.11.indexer.k_norm.bias
F32
F32
[128]
blk.11.indexer.k_norm.weight
F32
F32
[128]
blk.11.indexer.proj.weight
F32
F32
[6144, 32]
blk.12
blk.12.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.12.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.12.attn_kv_a_norm.weight
F32
F32
[512]
blk.12.attn_norm.weight
F32
F32
[6144]
blk.12.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.12.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.12.attn_q_a_norm.weight
F32
F32
[2048]
blk.12.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.12.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.12.exp_probs_b.bias
F32
F32
[256]
blk.12.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.12.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.12.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.12.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.12.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.12.ffn_norm.weight
F32
F32
[6144]
blk.12.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.12.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.12.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.12.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.12.indexer.k_norm.bias
F32
F32
[128]
blk.12.indexer.k_norm.weight
F32
F32
[128]
blk.12.indexer.proj.weight
F32
F32
[6144, 32]
blk.13
blk.13.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.13.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.13.attn_kv_a_norm.weight
F32
F32
[512]
blk.13.attn_norm.weight
F32
F32
[6144]
blk.13.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.13.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.13.attn_q_a_norm.weight
F32
F32
[2048]
blk.13.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.13.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.13.exp_probs_b.bias
F32
F32
[256]
blk.13.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.13.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.13.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.13.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.13.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.13.ffn_norm.weight
F32
F32
[6144]
blk.13.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.13.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.13.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.13.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.13.indexer.k_norm.bias
F32
F32
[128]
blk.13.indexer.k_norm.weight
F32
F32
[128]
blk.13.indexer.proj.weight
F32
F32
[6144, 32]
blk.14
blk.14.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.14.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.14.attn_kv_a_norm.weight
F32
F32
[512]
blk.14.attn_norm.weight
F32
F32
[6144]
blk.14.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.14.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.14.attn_q_a_norm.weight
F32
F32
[2048]
blk.14.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.14.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.14.exp_probs_b.bias
F32
F32
[256]
blk.14.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.14.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.14.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.14.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.14.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.14.ffn_norm.weight
F32
F32
[6144]
blk.14.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.14.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.14.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.14.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.14.indexer.k_norm.bias
F32
F32
[128]
blk.14.indexer.k_norm.weight
F32
F32
[128]
blk.14.indexer.proj.weight
F32
F32
[6144, 32]
blk.15
blk.15.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.15.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.15.attn_kv_a_norm.weight
F32
F32
[512]
blk.15.attn_norm.weight
F32
F32
[6144]
blk.15.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.15.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.15.attn_q_a_norm.weight
F32
F32
[2048]
blk.15.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.15.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.15.exp_probs_b.bias
F32
F32
[256]
blk.15.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.15.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.15.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.15.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.15.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.15.ffn_norm.weight
F32
F32
[6144]
blk.15.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.15.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.15.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.15.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.15.indexer.k_norm.bias
F32
F32
[128]
blk.15.indexer.k_norm.weight
F32
F32
[128]
blk.15.indexer.proj.weight
F32
F32
[6144, 32]
blk.16
blk.16.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.16.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.16.attn_kv_a_norm.weight
F32
F32
[512]
blk.16.attn_norm.weight
F32
F32
[6144]
blk.16.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.16.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.16.attn_q_a_norm.weight
F32
F32
[2048]
blk.16.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.16.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.16.exp_probs_b.bias
F32
F32
[256]
blk.16.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.16.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.16.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.16.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.16.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.16.ffn_norm.weight
F32
F32
[6144]
blk.16.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.16.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.16.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.16.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.16.indexer.k_norm.bias
F32
F32
[128]
blk.16.indexer.k_norm.weight
F32
F32
[128]
blk.16.indexer.proj.weight
F32
F32
[6144, 32]
blk.17
blk.17.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.17.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.17.attn_kv_a_norm.weight
F32
F32
[512]
blk.17.attn_norm.weight
F32
F32
[6144]
blk.17.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.17.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.17.attn_q_a_norm.weight
F32
F32
[2048]
blk.17.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.17.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.17.exp_probs_b.bias
F32
F32
[256]
blk.17.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.17.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
blk.17.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.17.ffn_gate_inp.weight
F32
F32
[6144, 256]
blk.17.ffn_gate_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.17.ffn_norm.weight
F32
F32
[6144]
blk.17.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[6144, 2048, 256]
blk.17.ffn_up_shexp.weight
Q5_K
Q5_K
[6144, 2048]
blk.17.indexer.attn_k.weight
Q8_0
Q8_0
[6144, 128]
blk.17.indexer.attn_q_b.weight
Q8_0
Q8_0
[2048, 4096]
blk.17.indexer.k_norm.bias
F32
F32
[128]
blk.17.indexer.k_norm.weight
F32
F32
[128]
blk.17.indexer.proj.weight
F32
F32
[6144, 32]
blk.18
blk.18.attn_k_b.weight
Q8_0
Q8_0
[192, 512, 64]
blk.18.attn_kv_a_mqa.weight
Q8_0
Q8_0
[6144, 576]
blk.18.attn_kv_a_norm.weight
F32
F32
[512]
blk.18.attn_norm.weight
F32
F32
[6144]
blk.18.attn_output.weight
Q5_K
Q5_K
[16384, 6144]
blk.18.attn_q_a.weight
Q5_K
Q5_K
[6144, 2048]
blk.18.attn_q_a_norm.weight
F32
F32
[2048]
blk.18.attn_q_b.weight
Q8_0
Q8_0
[2048, 16384]
blk.18.attn_v_b.weight
Q8_0
Q8_0
[512, 256, 64]
blk.18.exp_probs_b.bias
F32
F32
[256]
blk.18.ffn_down_exps.weight
IQ3_XXS
IQ3_XXS
[2048, 6144, 256]
blk.18.ffn_down_shexp.weight
Q6_K
Q6_K
[2048, 6144]
output.weight
Q4_K
Q4_K
[6144, 154880]
output_norm.weight
F32
F32
[6144]