GitHub
185.8K
Download
Models
Docs
Pricing
Sign in
Download
Models
Docs
Pricing
Sign in
frob
/
deepseek-v4.1-flash
:552b-a8b-1017-ud-iq1_M
232
Downloads
Updated
6 days ago
1017
1017
Cancel
vision
552b
deepseek-v4.1-flash:552b-a8b-1017-ud-iq1_M
...
/
model
4f738cf3dadb ยท 49GB
Metadata
split.count
5
5
split.no
2
2
split.tensors.count
1043
1043
Tensor
Name
Type
Shape
blk.16
blk.16.ffn_down_exps.weight
IQ2_XXS
IQ2_XXS
[2304, 5120, 384]
blk.16.ffn_down_shexp.weight
Q8_0
Q8_0
[2304, 5120]
blk.16.ffn_gate_exps.weight
IQ1_S
IQ1_S
[5120, 2304, 384]
blk.16.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.16.ffn_gate_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.16.ffn_norm.weight
F32
F32
[5120]
blk.16.ffn_up_exps.weight
IQ1_S
IQ1_S
[5120, 2304, 384]
blk.16.ffn_up_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.16.hc_attn_base.weight
F32
F32
[24]
blk.16.hc_attn_fn.weight
F32
F32
[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
F32
F32
[20480, 24]
blk.16.hc_ffn_scale.weight
F32
F32
[3]
blk.17
blk.17.attn_kv.weight
Q6_K
Q6_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
Q6_K
Q6_K
[4096, 8192]
blk.17.attn_output_b.weight
Q6_K
Q6_K
[8192, 5120]
blk.17.attn_q_a.weight
Q6_K
Q6_K
[5120, 1280]
blk.17.attn_q_a_norm.weight
F32
F32
[1280]
blk.17.attn_q_b.weight
Q6_K
Q6_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_exps.weight
IQ2_XXS
IQ2_XXS
[2304, 5120, 384]
blk.17.ffn_down_shexp.weight
Q8_0
Q8_0
[2304, 5120]
blk.17.ffn_gate_exps.weight
IQ1_S
IQ1_S
[5120, 2304, 384]
blk.17.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.17.ffn_gate_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.17.ffn_norm.weight
F32
F32
[5120]
blk.17.ffn_up_exps.weight
IQ1_S
IQ1_S
[5120, 2304, 384]
blk.17.ffn_up_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.17.hc_attn_base.weight
F32
F32
[24]
blk.17.hc_attn_fn.weight
F32
F32
[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
F32
F32
[20480, 24]
blk.17.hc_ffn_scale.weight
F32
F32
[3]
blk.18
blk.18.attn_kv.weight
Q6_K
Q6_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
Q6_K
Q6_K
[4096, 8192]
blk.18.attn_output_b.weight
Q6_K
Q6_K
[8192, 5120]
blk.18.attn_q_a.weight
Q6_K
Q6_K
[5120, 1280]
blk.18.attn_q_a_norm.weight
F32
F32
[1280]
blk.18.attn_q_b.weight
Q6_K
Q6_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_exps.weight
IQ2_XXS
IQ2_XXS
[2304, 5120, 384]
blk.18.ffn_down_shexp.weight
Q8_0
Q8_0
[2304, 5120]
blk.18.ffn_gate_exps.weight
IQ1_S
IQ1_S
[5120, 2304, 384]
blk.18.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.18.ffn_gate_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.18.ffn_norm.weight
F32
F32
[5120]
blk.18.ffn_up_exps.weight
IQ1_S
IQ1_S
[5120, 2304, 384]
blk.18.ffn_up_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.18.hc_attn_base.weight
F32
F32
[24]
blk.18.hc_attn_fn.weight
F32
F32
[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
F32
F32
[20480, 24]
blk.18.hc_ffn_scale.weight
F32
F32
[3]
blk.19
blk.19.attn_kv.weight
Q6_K
Q6_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
Q6_K
Q6_K
[4096, 8192]
blk.19.attn_output_b.weight
Q6_K
Q6_K
[8192, 5120]
blk.19.attn_q_a.weight
Q6_K
Q6_K
[5120, 1280]
blk.19.attn_q_a_norm.weight
F32
F32
[1280]
blk.19.attn_q_b.weight
Q6_K
Q6_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_exps.weight
IQ2_XXS
IQ2_XXS
[2304, 5120, 384]
blk.19.ffn_down_shexp.weight
Q8_0
Q8_0
[2304, 5120]
blk.19.ffn_gate_exps.weight
IQ1_S
IQ1_S
[5120, 2304, 384]
blk.19.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.19.ffn_gate_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.19.ffn_norm.weight
F32
F32
[5120]
blk.19.ffn_up_exps.weight
IQ1_S
IQ1_S
[5120, 2304, 384]
blk.19.ffn_up_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.19.hc_attn_base.weight
F32
F32
[24]
blk.19.hc_attn_fn.weight
F32
F32
[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
F32
F32
[20480, 24]
blk.19.hc_ffn_scale.weight
F32
F32
[3]
blk.20
blk.20.attn_compressor_kv.weight
BF16
BF16
[5120, 512]
blk.20.attn_compressor_norm.weight
F32
F32
[512]
blk.20.attn_kv.weight
Q6_K
Q6_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
Q6_K
Q6_K
[4096, 8192]
blk.20.attn_output_b.weight
Q6_K
Q6_K
[8192, 5120]
blk.20.attn_q_a.weight
Q6_K
Q6_K
[5120, 1280]
blk.20.attn_q_a_norm.weight
F32
F32
[1280]
blk.20.attn_q_b.weight
Q6_K
Q6_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_exps.weight
IQ2_S
IQ2_S
[2304, 5120, 384]
blk.20.ffn_down_shexp.weight
Q8_0
Q8_0
[2304, 5120]
blk.20.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.20.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.20.ffn_gate_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.20.ffn_norm.weight
F32
F32
[5120]
blk.20.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.20.ffn_up_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.20.hc_attn_base.weight
F32
F32
[24]
blk.20.hc_attn_fn.weight
F32
F32
[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
F32
F32
[20480, 24]
blk.20.hc_ffn_scale.weight
F32
F32
[3]
blk.20.indexer.attn_k.weight
BF16
BF16
[512, 128]
blk.20.indexer.attn_q_b.weight
BF16
BF16
[1280, 4096]
blk.20.indexer.k_norm.weight
F32
F32
[128]
blk.20.indexer.proj.weight
BF16
BF16
[5120, 32]
blk.21
blk.21.attn_kv.weight
Q6_K
Q6_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
Q6_K
Q6_K
[4096, 8192]
blk.21.attn_output_b.weight
Q6_K
Q6_K
[8192, 5120]
blk.21.attn_q_a.weight
Q6_K
Q6_K
[5120, 1280]
blk.21.attn_q_a_norm.weight
F32
F32
[1280]
blk.21.attn_q_b.weight
Q6_K
Q6_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_exps.weight
IQ2_S
IQ2_S
[2304, 5120, 384]
blk.21.ffn_down_shexp.weight
Q8_0
Q8_0
[2304, 5120]
blk.21.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.21.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.21.ffn_gate_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.21.ffn_norm.weight
F32
F32
[5120]
blk.21.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.21.ffn_up_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.21.hc_attn_base.weight
F32
F32
[24]
blk.21.hc_attn_fn.weight
F32
F32
[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
F32
F32
[20480, 24]
blk.21.hc_ffn_scale.weight
F32
F32
[3]
blk.22
blk.22.attn_kv.weight
Q6_K
Q6_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
Q6_K
Q6_K
[4096, 8192]
blk.22.attn_output_b.weight
Q6_K
Q6_K
[8192, 5120]
blk.22.attn_q_a.weight
Q6_K
Q6_K
[5120, 1280]
blk.22.attn_q_a_norm.weight
F32
F32
[1280]
blk.22.attn_q_b.weight
Q6_K
Q6_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_exps.weight
IQ2_S
IQ2_S
[2304, 5120, 384]
blk.22.ffn_down_shexp.weight
Q8_0
Q8_0
[2304, 5120]
blk.22.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.22.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.22.ffn_gate_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.22.ffn_norm.weight
F32
F32
[5120]
blk.22.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.22.ffn_up_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.22.hc_attn_base.weight
F32
F32
[24]
blk.22.hc_attn_fn.weight
F32
F32
[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
F32
F32
[20480, 24]
blk.22.hc_ffn_scale.weight
F32
F32
[3]
blk.23
blk.23.attn_kv.weight
Q6_K
Q6_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
Q6_K
Q6_K
[4096, 8192]
blk.23.attn_output_b.weight
Q6_K
Q6_K
[8192, 5120]
blk.23.attn_q_a.weight
Q6_K
Q6_K
[5120, 1280]
blk.23.attn_q_a_norm.weight
F32
F32
[1280]
blk.23.attn_q_b.weight
Q6_K
Q6_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_exps.weight
IQ2_S
IQ2_S
[2304, 5120, 384]
blk.23.ffn_down_shexp.weight
Q8_0
Q8_0
[2304, 5120]
blk.23.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.23.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.23.ffn_gate_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.23.ffn_norm.weight
F32
F32
[5120]
blk.23.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.23.ffn_up_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.23.hc_attn_base.weight
F32
F32
[24]
blk.23.hc_attn_fn.weight
F32
F32
[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
F32
F32
[20480, 24]
blk.23.hc_ffn_scale.weight
F32
F32
[3]
blk.24
blk.24.attn_kv.weight
Q6_K
Q6_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
Q6_K
Q6_K
[4096, 8192]
blk.24.attn_output_b.weight
Q6_K
Q6_K
[8192, 5120]
blk.24.attn_q_a.weight
Q6_K
Q6_K
[5120, 1280]
blk.24.attn_q_a_norm.weight
F32
F32
[1280]
blk.24.attn_q_b.weight
Q6_K
Q6_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_exps.weight
IQ2_S
IQ2_S
[2304, 5120, 384]
blk.24.ffn_down_shexp.weight
Q8_0
Q8_0
[2304, 5120]
blk.24.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.24.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.24.ffn_gate_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.24.ffn_norm.weight
F32
F32
[5120]
blk.24.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.24.ffn_up_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.24.hc_attn_base.weight
F32
F32
[24]
blk.24.hc_attn_fn.weight
F32
F32
[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
F32
F32
[20480, 24]
blk.24.hc_ffn_scale.weight
F32
F32
[3]
blk.24.indexer.attn_q_b.weight
BF16
BF16
[1280, 4096]
blk.24.indexer.proj.weight
BF16
BF16
[5120, 32]
blk.25
blk.25.attn_kv.weight
Q6_K
Q6_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
Q6_K
Q6_K
[4096, 8192]
blk.25.attn_output_b.weight
Q6_K
Q6_K
[8192, 5120]
blk.25.attn_q_a.weight
Q6_K
Q6_K
[5120, 1280]
blk.25.attn_q_a_norm.weight
F32
F32
[1280]
blk.25.attn_q_b.weight
Q6_K
Q6_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_exps.weight
IQ2_S
IQ2_S
[2304, 5120, 384]
blk.25.ffn_down_shexp.weight
Q8_0
Q8_0
[2304, 5120]
blk.25.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.25.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.25.ffn_gate_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.25.ffn_norm.weight
F32
F32
[5120]
blk.25.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.25.ffn_up_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.25.hc_attn_base.weight
F32
F32
[24]
blk.25.hc_attn_fn.weight
F32
F32
[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
F32
F32
[20480, 24]
blk.25.hc_ffn_scale.weight
F32
F32
[3]
blk.26
blk.26.attn_kv.weight
Q6_K
Q6_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
Q6_K
Q6_K
[4096, 8192]
blk.26.attn_output_b.weight
Q6_K
Q6_K
[8192, 5120]
blk.26.attn_q_a.weight
Q6_K
Q6_K
[5120, 1280]
blk.26.attn_q_a_norm.weight
F32
F32
[1280]
blk.26.attn_q_b.weight
Q6_K
Q6_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_exps.weight
IQ2_S
IQ2_S
[2304, 5120, 384]
blk.26.ffn_down_shexp.weight
Q8_0
Q8_0
[2304, 5120]
blk.26.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.26.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.26.ffn_gate_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.26.ffn_norm.weight
F32
F32
[5120]
blk.26.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.26.ffn_up_shexp.weight
Q8_0
Q8_0
[5120, 2304]
blk.26.hc_attn_base.weight
F32
F32
[24]
blk.26.hc_attn_fn.weight
F32
F32
[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
F32
F32
[20480, 24]
blk.26.hc_ffn_scale.weight
F32
F32
[3]
blk.27
blk.27.attn_kv.weight
Q6_K
Q6_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
Q6_K
Q6_K
[4096, 8192]
blk.27.attn_output_b.weight
Q6_K
Q6_K
[8192, 5120]
blk.27.attn_q_a.weight
Q6_K
Q6_K
[5120, 1280]
blk.27.attn_q_a_norm.weight
F32
F32
[1280]
blk.27.attn_q_b.weight
Q6_K
Q6_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_exps.weight
IQ2_S
IQ2_S
[2304, 5120, 384]
blk.27.ffn_down_shexp.weight
Q5_K
Q5_K
[2304, 5120]
blk.27.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.27.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.27.ffn_gate_shexp.weight
Q5_K
Q5_K
[5120, 2304]
blk.27.ffn_norm.weight
F32
F32
[5120]
blk.27.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.27.ffn_up_shexp.weight
Q5_K
Q5_K
[5120, 2304]
blk.27.hc_attn_base.weight
F32
F32
[24]
blk.27.hc_attn_fn.weight
F32
F32
[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
F32
F32
[20480, 24]
blk.27.hc_ffn_scale.weight
F32
F32
[3]
blk.28
blk.28.attn_kv.weight
Q6_K
Q6_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
Q6_K
Q6_K
[4096, 8192]
blk.28.attn_output_b.weight
Q6_K
Q6_K
[8192, 5120]
blk.28.attn_q_a.weight
Q6_K
Q6_K
[5120, 1280]
blk.28.attn_q_a_norm.weight
F32
F32
[1280]
blk.28.attn_q_b.weight
Q6_K
Q6_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_exps.weight
IQ2_S
IQ2_S
[2304, 5120, 384]
blk.28.ffn_down_shexp.weight
Q5_K
Q5_K
[2304, 5120]
blk.28.ffn_gate_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.28.ffn_gate_inp.weight
F32
F32
[5120, 384]
blk.28.ffn_gate_shexp.weight
Q5_K
Q5_K
[5120, 2304]
blk.28.ffn_norm.weight
F32
F32
[5120]
blk.28.ffn_up_exps.weight
IQ2_XXS
IQ2_XXS
[5120, 2304, 384]
blk.28.ffn_up_shexp.weight
Q5_K
Q5_K
[5120, 2304]
blk.28.hc_attn_base.weight
F32
F32
[24]
blk.28.hc_attn_fn.weight
F32
F32
[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
F32
F32
[20480, 24]
blk.28.hc_ffn_scale.weight
F32
F32
[3]
blk.28.indexer.attn_q_b.weight
BF16
BF16
[1280, 4096]
blk.28.indexer.proj.weight
BF16
BF16
[5120, 32]
blk.29
blk.29.attn_kv.weight
Q6_K
Q6_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
Q6_K
Q6_K
[4096, 8192]
blk.29.attn_output_b.weight
Q6_K
Q6_K
[8192, 5120]
blk.29.attn_q_a.weight
Q6_K
Q6_K
[5120, 1280]
blk.29.attn_q_a_norm.weight
F32
F32
[1280]
blk.29.attn_q_b.weight
Q6_K
Q6_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_exps.weight
IQ2_S
IQ2_S
[2304, 5120, 384]
blk.29.ffn_down_shexp.weight
Q5_K
Q5_K
[2304, 5120]