GitHub
185.8K
Download
Models
Docs
Pricing
Sign in
Download
Models
Docs
Pricing
Sign in
frob
/
deepseek-v4.1-flash
:552b-a8b-1017-q4_K_M
232
Downloads
Updated
6 days ago
1017
1017
Cancel
vision
552b
deepseek-v4.1-flash:552b-a8b-1017-q4_K_M
...
/
model
63962ffd461c ยท 5.2GB
Metadata
split.count
9
9
split.no
8
8
split.tensors.count
1043
1043
Tensor
Name
Type
Shape
token_embd.weight
Q4_K
Q4_K
[5120, 129280]
blk.0
blk.0.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.0.attn_kv_a_norm.weight
F32
F32
[512]
blk.0.attn_norm.weight
F32
F32
[5120]
blk.0.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.0.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.0.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.0.attn_q_a_norm.weight
F32
F32
[1280]
blk.0.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.0.attn_sinks.weight
F32
F32
[64]
blk.0.exp_probs_b.bias
F32
F32
[384]
blk.0.exp_probs_b_vl.bias
F32
F32
[384]
blk.0.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.0.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.0.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.0.ffn_norm.weight
F32
F32
[5120]
blk.0.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.0.hc_attn_base.weight
F32
F32
[24]
blk.0.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.0.hc_attn_scale.weight
F32
F32
[3]
blk.0.hc_ffn_base.weight
F32
F32
[24]
blk.0.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.0.hc_ffn_scale.weight
F32
F32
[3]
blk.1
blk.1.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.1.attn_kv_a_norm.weight
F32
F32
[512]
blk.1.attn_norm.weight
F32
F32
[5120]
blk.1.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.1.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.1.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.1.attn_q_a_norm.weight
F32
F32
[1280]
blk.1.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.1.attn_sinks.weight
F32
F32
[64]
blk.1.engram_gate.weight
F32
F32
[5120, 4]
blk.1.engram_kv.weight
Q4_K
Q4_K
[6144, 25600]
blk.1.exp_probs_b.bias
F32
F32
[384]
blk.1.exp_probs_b_vl.bias
F32
F32
[384]
blk.1.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.1.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.1.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.1.ffn_norm.weight
F32
F32
[5120]
blk.1.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.1.hc_attn_base.weight
F32
F32
[24]
blk.1.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.1.hc_attn_scale.weight
F32
F32
[3]
blk.1.hc_ffn_base.weight
F32
F32
[24]
blk.1.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.1.hc_ffn_scale.weight
F32
F32
[3]
blk.2
blk.2.attn_compressor_gate.weight
Q4_K
Q4_K
[5120, 512]
blk.2.attn_compressor_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.2.attn_compressor_norm.weight
F32
F32
[512]
blk.2.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.2.attn_kv_a_norm.weight
F32
F32
[512]
blk.2.attn_norm.weight
F32
F32
[5120]
blk.2.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.2.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.2.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.2.attn_q_a_norm.weight
F32
F32
[1280]
blk.2.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.2.attn_sinks.weight
F32
F32
[64]
blk.2.exp_probs_b.bias
F32
F32
[384]
blk.2.exp_probs_b_vl.bias
F32
F32
[384]
blk.2.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.2.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.2.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.2.ffn_norm.weight
F32
F32
[5120]
blk.2.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.2.hc_attn_base.weight
F32
F32
[24]
blk.2.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.2.hc_attn_scale.weight
F32
F32
[3]
blk.2.hc_ffn_base.weight
F32
F32
[24]
blk.2.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.2.hc_ffn_scale.weight
F32
F32
[3]
blk.2.indexer.attn_k.weight
Q4_K
Q4_K
[512, 128]
blk.2.indexer.attn_q_b.weight
Q4_K
Q4_K
[1280, 4096]
blk.2.indexer.k_norm.weight
F32
F32
[128]
blk.2.indexer.proj.weight
Q4_K
Q4_K
[5120, 32]
blk.3
blk.3.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.3.attn_kv_a_norm.weight
F32
F32
[512]
blk.3.attn_norm.weight
F32
F32
[5120]
blk.3.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.3.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.3.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.3.attn_q_a_norm.weight
F32
F32
[1280]
blk.3.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.3.attn_sinks.weight
F32
F32
[64]
blk.3.exp_probs_b.bias
F32
F32
[384]
blk.3.exp_probs_b_vl.bias
F32
F32
[384]
blk.3.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.3.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.3.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.3.ffn_norm.weight
F32
F32
[5120]
blk.3.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.3.hc_attn_base.weight
F32
F32
[24]
blk.3.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.3.hc_attn_scale.weight
F32
F32
[3]
blk.3.hc_ffn_base.weight
F32
F32
[24]
blk.3.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.3.hc_ffn_scale.weight
F32
F32
[3]
blk.4
blk.4.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.4.attn_kv_a_norm.weight
F32
F32
[512]
blk.4.attn_norm.weight
F32
F32
[5120]
blk.4.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.4.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.4.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.4.attn_q_a_norm.weight
F32
F32
[1280]
blk.4.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.4.attn_sinks.weight
F32
F32
[64]
blk.4.exp_probs_b.bias
F32
F32
[384]
blk.4.exp_probs_b_vl.bias
F32
F32
[384]
blk.4.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.4.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.4.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.4.ffn_norm.weight
F32
F32
[5120]
blk.4.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.4.hc_attn_base.weight
F32
F32
[24]
blk.4.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.4.hc_attn_scale.weight
F32
F32
[3]
blk.4.hc_ffn_base.weight
F32
F32
[24]
blk.4.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.4.hc_ffn_scale.weight
F32
F32
[3]
blk.5
blk.5.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.5.attn_kv_a_norm.weight
F32
F32
[512]
blk.5.attn_norm.weight
F32
F32
[5120]
blk.5.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.5.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.5.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.5.attn_q_a_norm.weight
F32
F32
[1280]
blk.5.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.5.attn_sinks.weight
F32
F32
[64]
blk.5.exp_probs_b.bias
F32
F32
[384]
blk.5.exp_probs_b_vl.bias
F32
F32
[384]
blk.5.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.5.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.5.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.5.ffn_norm.weight
F32
F32
[5120]
blk.5.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.5.hc_attn_base.weight
F32
F32
[24]
blk.5.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.5.hc_attn_scale.weight
F32
F32
[3]
blk.5.hc_ffn_base.weight
F32
F32
[24]
blk.5.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.5.hc_ffn_scale.weight
F32
F32
[3]
blk.6
blk.6.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.6.attn_kv_a_norm.weight
F32
F32
[512]
blk.6.attn_norm.weight
F32
F32
[5120]
blk.6.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.6.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.6.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.6.attn_q_a_norm.weight
F32
F32
[1280]
blk.6.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.6.attn_sinks.weight
F32
F32
[64]
blk.6.exp_probs_b.bias
F32
F32
[384]
blk.6.exp_probs_b_vl.bias
F32
F32
[384]
blk.6.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.6.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.6.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.6.ffn_norm.weight
F32
F32
[5120]
blk.6.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.6.hc_attn_base.weight
F32
F32
[24]
blk.6.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.6.hc_attn_scale.weight
F32
F32
[3]
blk.6.hc_ffn_base.weight
F32
F32
[24]
blk.6.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.6.hc_ffn_scale.weight
F32
F32
[3]
blk.7
blk.7.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.7.attn_kv_a_norm.weight
F32
F32
[512]
blk.7.attn_norm.weight
F32
F32
[5120]
blk.7.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.7.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.7.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.7.attn_q_a_norm.weight
F32
F32
[1280]
blk.7.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.7.attn_sinks.weight
F32
F32
[64]
blk.7.exp_probs_b.bias
F32
F32
[384]
blk.7.exp_probs_b_vl.bias
F32
F32
[384]
blk.7.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.7.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.7.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.7.ffn_norm.weight
F32
F32
[5120]
blk.7.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.7.hc_attn_base.weight
F32
F32
[24]
blk.7.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.7.hc_attn_scale.weight
F32
F32
[3]
blk.7.hc_ffn_base.weight
F32
F32
[24]
blk.7.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.7.hc_ffn_scale.weight
F32
F32
[3]
blk.8
blk.8.attn_compressor_gate.weight
Q4_K
Q4_K
[5120, 512]
blk.8.attn_compressor_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.8.attn_compressor_norm.weight
F32
F32
[512]
blk.8.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.8.attn_kv_a_norm.weight
F32
F32
[512]
blk.8.attn_norm.weight
F32
F32
[5120]
blk.8.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.8.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.8.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.8.attn_q_a_norm.weight
F32
F32
[1280]
blk.8.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.8.attn_sinks.weight
F32
F32
[64]
blk.8.exp_probs_b.bias
F32
F32
[384]
blk.8.exp_probs_b_vl.bias
F32
F32
[384]
blk.8.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.8.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.8.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.8.ffn_norm.weight
F32
F32
[5120]
blk.8.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.8.hc_attn_base.weight
F32
F32
[24]
blk.8.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.8.hc_attn_scale.weight
F32
F32
[3]
blk.8.hc_ffn_base.weight
F32
F32
[24]
blk.8.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.8.hc_ffn_scale.weight
F32
F32
[3]
blk.8.indexer.attn_k.weight
Q4_K
Q4_K
[512, 128]
blk.8.indexer.attn_q_b.weight
Q4_K
Q4_K
[1280, 4096]
blk.8.indexer.k_norm.weight
F32
F32
[128]
blk.8.indexer.proj.weight
Q4_K
Q4_K
[5120, 32]
blk.9
blk.9.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.9.attn_kv_a_norm.weight
F32
F32
[512]
blk.9.attn_norm.weight
F32
F32
[5120]
blk.9.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.9.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.9.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.9.attn_q_a_norm.weight
F32
F32
[1280]
blk.9.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.9.attn_sinks.weight
F32
F32
[64]
blk.9.exp_probs_b.bias
F32
F32
[384]
blk.9.exp_probs_b_vl.bias
F32
F32
[384]
blk.9.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.9.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.9.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.9.ffn_norm.weight
F32
F32
[5120]
blk.9.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.9.hc_attn_base.weight
F32
F32
[24]
blk.9.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.9.hc_attn_scale.weight
F32
F32
[3]
blk.9.hc_ffn_base.weight
F32
F32
[24]
blk.9.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.9.hc_ffn_scale.weight
F32
F32
[3]
blk.10
blk.10.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.10.attn_kv_a_norm.weight
F32
F32
[512]
blk.10.attn_norm.weight
F32
F32
[5120]
blk.10.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.10.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.10.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.10.attn_q_a_norm.weight
F32
F32
[1280]
blk.10.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.10.attn_sinks.weight
F32
F32
[64]
blk.10.exp_probs_b.bias
F32
F32
[384]
blk.10.exp_probs_b_vl.bias
F32
F32
[384]
blk.10.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.10.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.10.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.10.ffn_norm.weight
F32
F32
[5120]
blk.10.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.10.hc_attn_base.weight
F32
F32
[24]
blk.10.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.10.hc_attn_scale.weight
F32
F32
[3]
blk.10.hc_ffn_base.weight
F32
F32
[24]
blk.10.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.10.hc_ffn_scale.weight
F32
F32
[3]
blk.11
blk.11.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.11.attn_kv_a_norm.weight
F32
F32
[512]
blk.11.attn_norm.weight
F32
F32
[5120]
blk.11.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.11.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.11.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.11.attn_q_a_norm.weight
F32
F32
[1280]
blk.11.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.11.attn_sinks.weight
F32
F32
[64]
blk.11.exp_probs_b.bias
F32
F32
[384]
blk.11.exp_probs_b_vl.bias
F32
F32
[384]
blk.11.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.11.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.11.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.11.ffn_norm.weight
F32
F32
[5120]
blk.11.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.11.hc_attn_base.weight
F32
F32
[24]
blk.11.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.11.hc_attn_scale.weight
F32
F32
[3]
blk.11.hc_ffn_base.weight
F32
F32
[24]
blk.11.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.11.hc_ffn_scale.weight
F32
F32
[3]
blk.12
blk.12.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.12.attn_kv_a_norm.weight
F32
F32
[512]
blk.12.attn_norm.weight
F32
F32
[5120]
blk.12.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.12.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.12.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.12.attn_q_a_norm.weight
F32
F32
[1280]
blk.12.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.12.attn_sinks.weight
F32
F32
[64]
blk.12.exp_probs_b.bias
F32
F32
[384]
blk.12.exp_probs_b_vl.bias
F32
F32
[384]
blk.12.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.12.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.12.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.12.ffn_norm.weight
F32
F32
[5120]
blk.12.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.12.hc_attn_base.weight
F32
F32
[24]
blk.12.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.12.hc_attn_scale.weight
F32
F32
[3]
blk.12.hc_ffn_base.weight
F32
F32
[24]
blk.12.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.12.hc_ffn_scale.weight
F32
F32
[3]
blk.13
blk.13.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.13.attn_kv_a_norm.weight
F32
F32
[512]
blk.13.attn_norm.weight
F32
F32
[5120]
blk.13.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.13.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.13.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.13.attn_q_a_norm.weight
F32
F32
[1280]
blk.13.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.13.attn_sinks.weight
F32
F32
[64]
blk.13.exp_probs_b.bias
F32
F32
[384]
blk.13.exp_probs_b_vl.bias
F32
F32
[384]
blk.13.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.13.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.13.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.13.ffn_norm.weight
F32
F32
[5120]
blk.13.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.13.hc_attn_base.weight
F32
F32
[24]
blk.13.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.13.hc_attn_scale.weight
F32
F32
[3]
blk.13.hc_ffn_base.weight
F32
F32
[24]
blk.13.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.13.hc_ffn_scale.weight
F32
F32
[3]
blk.14
blk.14.attn_compressor_gate.weight
Q4_K
Q4_K
[5120, 512]
blk.14.attn_compressor_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.14.attn_compressor_norm.weight
F32
F32
[512]
blk.14.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.14.attn_kv_a_norm.weight
F32
F32
[512]
blk.14.attn_norm.weight
F32
F32
[5120]
blk.14.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.14.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.14.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.14.attn_q_a_norm.weight
F32
F32
[1280]
blk.14.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.14.attn_sinks.weight
F32
F32
[64]
blk.14.engram_gate.weight
F32
F32
[5120, 4]
blk.14.engram_kv.weight
Q4_K
Q4_K
[6144, 25600]
blk.14.exp_probs_b.bias
F32
F32
[384]
blk.14.exp_probs_b_vl.bias
F32
F32
[384]
blk.14.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.14.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.14.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.14.ffn_norm.weight
F32
F32
[5120]
blk.14.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.14.hc_attn_base.weight
F32
F32
[24]
blk.14.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.14.hc_attn_scale.weight
F32
F32
[3]
blk.14.hc_ffn_base.weight
F32
F32
[24]
blk.14.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.14.hc_ffn_scale.weight
F32
F32
[3]
blk.14.indexer.attn_k.weight
Q4_K
Q4_K
[512, 128]
blk.14.indexer.attn_q_b.weight
Q4_K
Q4_K
[1280, 4096]
blk.14.indexer.k_norm.weight
F32
F32
[128]
blk.14.indexer.proj.weight
Q4_K
Q4_K
[5120, 32]
blk.15
blk.15.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.15.attn_kv_a_norm.weight
F32
F32
[512]
blk.15.attn_norm.weight
F32
F32
[5120]
blk.15.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.15.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.15.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.15.attn_q_a_norm.weight
F32
F32
[1280]
blk.15.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.15.attn_sinks.weight
F32
F32
[64]
blk.15.exp_probs_b.bias
F32
F32
[384]
blk.15.exp_probs_b_vl.bias
F32
F32
[384]
blk.15.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.15.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.15.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.15.ffn_norm.weight
F32
F32
[5120]
blk.15.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.15.hc_attn_base.weight
F32
F32
[24]
blk.15.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.15.hc_attn_scale.weight
F32
F32
[3]
blk.15.hc_ffn_base.weight
F32
F32
[24]
blk.15.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.15.hc_ffn_scale.weight
F32
F32
[3]
blk.16
blk.16.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.16.attn_kv_a_norm.weight
F32
F32
[512]
blk.16.attn_norm.weight
F32
F32
[5120]
blk.16.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.16.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.16.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.16.attn_q_a_norm.weight
F32
F32
[1280]
blk.16.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.16.attn_sinks.weight
F32
F32
[64]
blk.16.exp_probs_b.bias
F32
F32
[384]
blk.16.exp_probs_b_vl.bias
F32
F32
[384]
blk.16.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.16.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.16.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.16.ffn_norm.weight
F32
F32
[5120]
blk.16.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.16.hc_attn_base.weight
F32
F32
[24]
blk.16.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.16.hc_attn_scale.weight
F32
F32
[3]
blk.16.hc_ffn_base.weight
F32
F32
[24]
blk.16.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.16.hc_ffn_scale.weight
F32
F32
[3]
blk.17
blk.17.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.17.attn_kv_a_norm.weight
F32
F32
[512]
blk.17.attn_norm.weight
F32
F32
[5120]
blk.17.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.17.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.17.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.17.attn_q_a_norm.weight
F32
F32
[1280]
blk.17.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.17.attn_sinks.weight
F32
F32
[64]
blk.17.exp_probs_b.bias
F32
F32
[384]
blk.17.exp_probs_b_vl.bias
F32
F32
[384]
blk.17.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.17.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.17.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.17.ffn_norm.weight
F32
F32
[5120]
blk.17.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.17.hc_attn_base.weight
F32
F32
[24]
blk.17.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.17.hc_attn_scale.weight
F32
F32
[3]
blk.17.hc_ffn_base.weight
F32
F32
[24]
blk.17.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.17.hc_ffn_scale.weight
F32
F32
[3]
blk.18
blk.18.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.18.attn_kv_a_norm.weight
F32
F32
[512]
blk.18.attn_norm.weight
F32
F32
[5120]
blk.18.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.18.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.18.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.18.attn_q_a_norm.weight
F32
F32
[1280]
blk.18.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.18.attn_sinks.weight
F32
F32
[64]
blk.18.exp_probs_b.bias
F32
F32
[384]
blk.18.exp_probs_b_vl.bias
F32
F32
[384]
blk.18.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.18.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.18.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.18.ffn_norm.weight
F32
F32
[5120]
blk.18.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.18.hc_attn_base.weight
F32
F32
[24]
blk.18.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.18.hc_attn_scale.weight
F32
F32
[3]
blk.18.hc_ffn_base.weight
F32
F32
[24]
blk.18.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.18.hc_ffn_scale.weight
F32
F32
[3]
blk.19
blk.19.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.19.attn_kv_a_norm.weight
F32
F32
[512]
blk.19.attn_norm.weight
F32
F32
[5120]
blk.19.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.19.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.19.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.19.attn_q_a_norm.weight
F32
F32
[1280]
blk.19.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.19.attn_sinks.weight
F32
F32
[64]
blk.19.exp_probs_b.bias
F32
F32
[384]
blk.19.exp_probs_b_vl.bias
F32
F32
[384]
blk.19.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.19.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.19.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.19.ffn_norm.weight
F32
F32
[5120]
blk.19.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.19.hc_attn_base.weight
F32
F32
[24]
blk.19.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.19.hc_attn_scale.weight
F32
F32
[3]
blk.19.hc_ffn_base.weight
F32
F32
[24]
blk.19.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.19.hc_ffn_scale.weight
F32
F32
[3]
blk.20
blk.20.attn_compressor_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.20.attn_compressor_norm.weight
F32
F32
[512]
blk.20.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.20.attn_kv_a_norm.weight
F32
F32
[512]
blk.20.attn_norm.weight
F32
F32
[5120]
blk.20.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.20.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.20.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.20.attn_q_a_norm.weight
F32
F32
[1280]
blk.20.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.20.attn_sinks.weight
F32
F32
[64]
blk.20.exp_probs_b.bias
F32
F32
[384]
blk.20.exp_probs_b_vl.bias
F32
F32
[384]
blk.20.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.20.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.20.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.20.ffn_norm.weight
F32
F32
[5120]
blk.20.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.20.hc_attn_base.weight
F32
F32
[24]
blk.20.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.20.hc_attn_scale.weight
F32
F32
[3]
blk.20.hc_ffn_base.weight
F32
F32
[24]
blk.20.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.20.hc_ffn_scale.weight
F32
F32
[3]
blk.20.indexer.attn_k.weight
Q4_K
Q4_K
[512, 128]
blk.20.indexer.attn_q_b.weight
Q4_K
Q4_K
[1280, 4096]
blk.20.indexer.k_norm.weight
F32
F32
[128]
blk.20.indexer.proj.weight
Q4_K
Q4_K
[5120, 32]
blk.21
blk.21.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.21.attn_kv_a_norm.weight
F32
F32
[512]
blk.21.attn_norm.weight
F32
F32
[5120]
blk.21.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.21.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.21.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.21.attn_q_a_norm.weight
F32
F32
[1280]
blk.21.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.21.attn_sinks.weight
F32
F32
[64]
blk.21.exp_probs_b.bias
F32
F32
[384]
blk.21.exp_probs_b_vl.bias
F32
F32
[384]
blk.21.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.21.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.21.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.21.ffn_norm.weight
F32
F32
[5120]
blk.21.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.21.hc_attn_base.weight
F32
F32
[24]
blk.21.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.21.hc_attn_scale.weight
F32
F32
[3]
blk.21.hc_ffn_base.weight
F32
F32
[24]
blk.21.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.21.hc_ffn_scale.weight
F32
F32
[3]
blk.22
blk.22.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.22.attn_kv_a_norm.weight
F32
F32
[512]
blk.22.attn_norm.weight
F32
F32
[5120]
blk.22.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.22.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.22.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.22.attn_q_a_norm.weight
F32
F32
[1280]
blk.22.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.22.attn_sinks.weight
F32
F32
[64]
blk.22.exp_probs_b.bias
F32
F32
[384]
blk.22.exp_probs_b_vl.bias
F32
F32
[384]
blk.22.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.22.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.22.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.22.ffn_norm.weight
F32
F32
[5120]
blk.22.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.22.hc_attn_base.weight
F32
F32
[24]
blk.22.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.22.hc_attn_scale.weight
F32
F32
[3]
blk.22.hc_ffn_base.weight
F32
F32
[24]
blk.22.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.22.hc_ffn_scale.weight
F32
F32
[3]
blk.23
blk.23.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.23.attn_kv_a_norm.weight
F32
F32
[512]
blk.23.attn_norm.weight
F32
F32
[5120]
blk.23.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.23.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.23.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.23.attn_q_a_norm.weight
F32
F32
[1280]
blk.23.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.23.attn_sinks.weight
F32
F32
[64]
blk.23.exp_probs_b.bias
F32
F32
[384]
blk.23.exp_probs_b_vl.bias
F32
F32
[384]
blk.23.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.23.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.23.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.23.ffn_norm.weight
F32
F32
[5120]
blk.23.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.23.hc_attn_base.weight
F32
F32
[24]
blk.23.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.23.hc_attn_scale.weight
F32
F32
[3]
blk.23.hc_ffn_base.weight
F32
F32
[24]
blk.23.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.23.hc_ffn_scale.weight
F32
F32
[3]
blk.24
blk.24.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.24.attn_kv_a_norm.weight
F32
F32
[512]
blk.24.attn_norm.weight
F32
F32
[5120]
blk.24.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.24.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.24.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.24.attn_q_a_norm.weight
F32
F32
[1280]
blk.24.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.24.attn_sinks.weight
F32
F32
[64]
blk.24.exp_probs_b.bias
F32
F32
[384]
blk.24.exp_probs_b_vl.bias
F32
F32
[384]
blk.24.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.24.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.24.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.24.ffn_norm.weight
F32
F32
[5120]
blk.24.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.24.hc_attn_base.weight
F32
F32
[24]
blk.24.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.24.hc_attn_scale.weight
F32
F32
[3]
blk.24.hc_ffn_base.weight
F32
F32
[24]
blk.24.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.24.hc_ffn_scale.weight
F32
F32
[3]
blk.24.indexer.attn_q_b.weight
Q4_K
Q4_K
[1280, 4096]
blk.24.indexer.proj.weight
Q4_K
Q4_K
[5120, 32]
blk.25
blk.25.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.25.attn_kv_a_norm.weight
F32
F32
[512]
blk.25.attn_norm.weight
F32
F32
[5120]
blk.25.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.25.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.25.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.25.attn_q_a_norm.weight
F32
F32
[1280]
blk.25.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.25.attn_sinks.weight
F32
F32
[64]
blk.25.exp_probs_b.bias
F32
F32
[384]
blk.25.exp_probs_b_vl.bias
F32
F32
[384]
blk.25.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.25.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.25.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.25.ffn_norm.weight
F32
F32
[5120]
blk.25.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.25.hc_attn_base.weight
F32
F32
[24]
blk.25.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.25.hc_attn_scale.weight
F32
F32
[3]
blk.25.hc_ffn_base.weight
F32
F32
[24]
blk.25.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.25.hc_ffn_scale.weight
F32
F32
[3]
blk.26
blk.26.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.26.attn_kv_a_norm.weight
F32
F32
[512]
blk.26.attn_norm.weight
F32
F32
[5120]
blk.26.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.26.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.26.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.26.attn_q_a_norm.weight
F32
F32
[1280]
blk.26.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.26.attn_sinks.weight
F32
F32
[64]
blk.26.exp_probs_b.bias
F32
F32
[384]
blk.26.exp_probs_b_vl.bias
F32
F32
[384]
blk.26.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.26.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.26.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.26.ffn_norm.weight
F32
F32
[5120]
blk.26.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.26.hc_attn_base.weight
F32
F32
[24]
blk.26.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.26.hc_attn_scale.weight
F32
F32
[3]
blk.26.hc_ffn_base.weight
F32
F32
[24]
blk.26.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.26.hc_ffn_scale.weight
F32
F32
[3]
blk.27
blk.27.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.27.attn_kv_a_norm.weight
F32
F32
[512]
blk.27.attn_norm.weight
F32
F32
[5120]
blk.27.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.27.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.27.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.27.attn_q_a_norm.weight
F32
F32
[1280]
blk.27.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.27.attn_sinks.weight
F32
F32
[64]
blk.27.exp_probs_b.bias
F32
F32
[384]
blk.27.exp_probs_b_vl.bias
F32
F32
[384]
blk.27.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.27.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.27.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.27.ffn_norm.weight
F32
F32
[5120]
blk.27.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.27.hc_attn_base.weight
F32
F32
[24]
blk.27.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.27.hc_attn_scale.weight
F32
F32
[3]
blk.27.hc_ffn_base.weight
F32
F32
[24]
blk.27.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.27.hc_ffn_scale.weight
F32
F32
[3]
blk.28
blk.28.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.28.attn_kv_a_norm.weight
F32
F32
[512]
blk.28.attn_norm.weight
F32
F32
[5120]
blk.28.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.28.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.28.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.28.attn_q_a_norm.weight
F32
F32
[1280]
blk.28.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.28.attn_sinks.weight
F32
F32
[64]
blk.28.exp_probs_b.bias
F32
F32
[384]
blk.28.exp_probs_b_vl.bias
F32
F32
[384]
blk.28.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.28.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.28.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.28.ffn_norm.weight
F32
F32
[5120]
blk.28.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.28.hc_attn_base.weight
F32
F32
[24]
blk.28.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.28.hc_attn_scale.weight
F32
F32
[3]
blk.28.hc_ffn_base.weight
F32
F32
[24]
blk.28.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.28.hc_ffn_scale.weight
F32
F32
[3]
blk.28.indexer.attn_q_b.weight
Q4_K
Q4_K
[1280, 4096]
blk.28.indexer.proj.weight
Q4_K
Q4_K
[5120, 32]
blk.29
blk.29.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.29.attn_kv_a_norm.weight
F32
F32
[512]
blk.29.attn_norm.weight
F32
F32
[5120]
blk.29.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.29.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.29.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.29.attn_q_a_norm.weight
F32
F32
[1280]
blk.29.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.29.attn_sinks.weight
F32
F32
[64]
blk.29.exp_probs_b.bias
F32
F32
[384]
blk.29.exp_probs_b_vl.bias
F32
F32
[384]
blk.29.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.29.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.29.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.29.ffn_norm.weight
F32
F32
[5120]
blk.29.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.29.hc_attn_base.weight
F32
F32
[24]
blk.29.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.29.hc_attn_scale.weight
F32
F32
[3]
blk.29.hc_ffn_base.weight
F32
F32
[24]
blk.29.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.29.hc_ffn_scale.weight
F32
F32
[3]
blk.30
blk.30.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.30.attn_kv_a_norm.weight
F32
F32
[512]
blk.30.attn_norm.weight
F32
F32
[5120]
blk.30.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.30.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.30.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.30.attn_q_a_norm.weight
F32
F32
[1280]
blk.30.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.30.attn_sinks.weight
F32
F32
[64]
blk.30.exp_probs_b.bias
F32
F32
[384]
blk.30.exp_probs_b_vl.bias
F32
F32
[384]
blk.30.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.30.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.30.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.30.ffn_norm.weight
F32
F32
[5120]
blk.30.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.30.hc_attn_base.weight
F32
F32
[24]
blk.30.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.30.hc_attn_scale.weight
F32
F32
[3]
blk.30.hc_ffn_base.weight
F32
F32
[24]
blk.30.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.30.hc_ffn_scale.weight
F32
F32
[3]
blk.31
blk.31.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.31.attn_kv_a_norm.weight
F32
F32
[512]
blk.31.attn_norm.weight
F32
F32
[5120]
blk.31.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.31.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.31.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.31.attn_q_a_norm.weight
F32
F32
[1280]
blk.31.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.31.attn_sinks.weight
F32
F32
[64]
blk.31.exp_probs_b.bias
F32
F32
[384]
blk.31.exp_probs_b_vl.bias
F32
F32
[384]
blk.31.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.31.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.31.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.31.ffn_norm.weight
F32
F32
[5120]
blk.31.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.31.hc_attn_base.weight
F32
F32
[24]
blk.31.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.31.hc_attn_scale.weight
F32
F32
[3]
blk.31.hc_ffn_base.weight
F32
F32
[24]
blk.31.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.31.hc_ffn_scale.weight
F32
F32
[3]
blk.32
blk.32.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.32.attn_kv_a_norm.weight
F32
F32
[512]
blk.32.attn_norm.weight
F32
F32
[5120]
blk.32.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.32.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.32.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.32.attn_q_a_norm.weight
F32
F32
[1280]
blk.32.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.32.attn_sinks.weight
F32
F32
[64]
blk.32.exp_probs_b.bias
F32
F32
[384]
blk.32.exp_probs_b_vl.bias
F32
F32
[384]
blk.32.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.32.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.32.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.32.ffn_norm.weight
F32
F32
[5120]
blk.32.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.32.hc_attn_base.weight
F32
F32
[24]
blk.32.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.32.hc_attn_scale.weight
F32
F32
[3]
blk.32.hc_ffn_base.weight
F32
F32
[24]
blk.32.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.32.hc_ffn_scale.weight
F32
F32
[3]
blk.32.indexer.attn_q_b.weight
Q4_K
Q4_K
[1280, 4096]
blk.32.indexer.proj.weight
Q4_K
Q4_K
[5120, 32]
blk.33
blk.33.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.33.attn_kv_a_norm.weight
F32
F32
[512]
blk.33.attn_norm.weight
F32
F32
[5120]
blk.33.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.33.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.33.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.33.attn_q_a_norm.weight
F32
F32
[1280]
blk.33.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.33.attn_sinks.weight
F32
F32
[64]
blk.33.exp_probs_b.bias
F32
F32
[384]
blk.33.exp_probs_b_vl.bias
F32
F32
[384]
blk.33.ffn_down_shexp.weight
Q4_K
Q4_K
[2304, 5120]
blk.33.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.33.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.33.ffn_norm.weight
F32
F32
[5120]
blk.33.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.33.hc_attn_base.weight
F32
F32
[24]
blk.33.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.33.hc_attn_scale.weight
F32
F32
[3]
blk.33.hc_ffn_base.weight
F32
F32
[24]
blk.33.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.33.hc_ffn_scale.weight
F32
F32
[3]
blk.34
blk.34.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.34.attn_kv_a_norm.weight
F32
F32
[512]
blk.34.attn_norm.weight
F32
F32
[5120]
blk.34.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.34.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.34.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.34.attn_q_a_norm.weight
F32
F32
[1280]
blk.34.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.34.attn_sinks.weight
F32
F32
[64]
blk.34.exp_probs_b.bias
F32
F32
[384]
blk.34.exp_probs_b_vl.bias
F32
F32
[384]
blk.34.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.34.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.34.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.34.ffn_norm.weight
F32
F32
[5120]
blk.34.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.34.hc_attn_base.weight
F32
F32
[24]
blk.34.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.34.hc_attn_scale.weight
F32
F32
[3]
blk.34.hc_ffn_base.weight
F32
F32
[24]
blk.34.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.34.hc_ffn_scale.weight
F32
F32
[3]
blk.35
blk.35.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.35.attn_kv_a_norm.weight
F32
F32
[512]
blk.35.attn_norm.weight
F32
F32
[5120]
blk.35.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.35.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.35.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.35.attn_q_a_norm.weight
F32
F32
[1280]
blk.35.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.35.attn_sinks.weight
F32
F32
[64]
blk.35.exp_probs_b.bias
F32
F32
[384]
blk.35.exp_probs_b_vl.bias
F32
F32
[384]
blk.35.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.35.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.35.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.35.ffn_norm.weight
F32
F32
[5120]
blk.35.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.35.hc_attn_base.weight
F32
F32
[24]
blk.35.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.35.hc_attn_scale.weight
F32
F32
[3]
blk.35.hc_ffn_base.weight
F32
F32
[24]
blk.35.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.35.hc_ffn_scale.weight
F32
F32
[3]
blk.36
blk.36.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.36.attn_kv_a_norm.weight
F32
F32
[512]
blk.36.attn_norm.weight
F32
F32
[5120]
blk.36.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.36.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.36.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.36.attn_q_a_norm.weight
F32
F32
[1280]
blk.36.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.36.attn_sinks.weight
F32
F32
[64]
blk.36.exp_probs_b.bias
F32
F32
[384]
blk.36.exp_probs_b_vl.bias
F32
F32
[384]
blk.36.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.36.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.36.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.36.ffn_norm.weight
F32
F32
[5120]
blk.36.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.36.hc_attn_base.weight
F32
F32
[24]
blk.36.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.36.hc_attn_scale.weight
F32
F32
[3]
blk.36.hc_ffn_base.weight
F32
F32
[24]
blk.36.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.36.hc_ffn_scale.weight
F32
F32
[3]
blk.36.indexer.attn_q_b.weight
Q4_K
Q4_K
[1280, 4096]
blk.36.indexer.proj.weight
Q4_K
Q4_K
[5120, 32]
blk.37
blk.37.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.37.attn_kv_a_norm.weight
F32
F32
[512]
blk.37.attn_norm.weight
F32
F32
[5120]
blk.37.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.37.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.37.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.37.attn_q_a_norm.weight
F32
F32
[1280]
blk.37.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.37.attn_sinks.weight
F32
F32
[64]
blk.37.exp_probs_b.bias
F32
F32
[384]
blk.37.exp_probs_b_vl.bias
F32
F32
[384]
blk.37.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.37.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.37.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.37.ffn_norm.weight
F32
F32
[5120]
blk.37.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.37.hc_attn_base.weight
F32
F32
[24]
blk.37.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.37.hc_attn_scale.weight
F32
F32
[3]
blk.37.hc_ffn_base.weight
F32
F32
[24]
blk.37.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.37.hc_ffn_scale.weight
F32
F32
[3]
blk.38
blk.38.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.38.attn_kv_a_norm.weight
F32
F32
[512]
blk.38.attn_norm.weight
F32
F32
[5120]
blk.38.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.38.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.38.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.38.attn_q_a_norm.weight
F32
F32
[1280]
blk.38.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.38.attn_sinks.weight
F32
F32
[64]
blk.38.exp_probs_b.bias
F32
F32
[384]
blk.38.exp_probs_b_vl.bias
F32
F32
[384]
blk.38.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.38.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.38.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.38.ffn_norm.weight
F32
F32
[5120]
blk.38.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.38.hc_attn_base.weight
F32
F32
[24]
blk.38.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.38.hc_attn_scale.weight
F32
F32
[3]
blk.38.hc_ffn_base.weight
F32
F32
[24]
blk.38.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.38.hc_ffn_scale.weight
F32
F32
[3]
blk.39
blk.39.attn_kv.weight
Q4_K
Q4_K
[5120, 512]
blk.39.attn_kv_a_norm.weight
F32
F32
[512]
blk.39.attn_norm.weight
F32
F32
[5120]
blk.39.attn_output_a.weight
Q4_K
Q4_K
[4096, 8192]
blk.39.attn_output_b.weight
Q4_K
Q4_K
[8192, 5120]
blk.39.attn_q_a.weight
Q4_K
Q4_K
[5120, 1280]
blk.39.attn_q_a_norm.weight
F32
F32
[1280]
blk.39.attn_q_b.weight
Q4_K
Q4_K
[1280, 32768]
blk.39.attn_sinks.weight
F32
F32
[64]
blk.39.exp_probs_b.bias
F32
F32
[384]
blk.39.exp_probs_b_vl.bias
F32
F32
[384]
blk.39.ffn_down_shexp.weight
Q6_K
Q6_K
[2304, 5120]
blk.39.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.39.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.39.ffn_norm.weight
F32
F32
[5120]
blk.39.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 2304]
blk.39.hc_attn_base.weight
F32
F32
[24]
blk.39.hc_attn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.39.hc_attn_scale.weight
F32
F32
[3]
blk.39.hc_ffn_base.weight
F32
F32
[24]
blk.39.hc_ffn_fn.weight
Q4_K
Q4_K
[20480, 24]
blk.39.hc_ffn_scale.weight
F32
F32
[3]
output.weight
Q6_K
Q6_K
[5120, 129280]
output_norm.weight
F32
F32
[5120]