GitHub
182.7K
Download
Models
Docs
Pricing
Sign in
Download
Models
Docs
Pricing
Sign in
dlasher
/
Pocket-Darwin-180B
:UD-Q4_K_XL
3
Downloads
Updated
5 days ago
Requant to IQ4_NL, with combined_en_huge as prompt (English + Coding focus)
Requant to IQ4_NL, with combined_en_huge as prompt (English + Coding focus)
Cancel
Pocket-Darwin-180B:UD-Q4_K_XL
...
/
model
0adc56c67b89 ยท 49GB
Metadata
split.count
4
4
split.no
2
2
split.tensors.count
1224
1224
Tensor
Name
Type
Shape
blk.11
blk.11.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.11.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.11.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.11.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.11.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.11.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.11.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.11.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.11.hc_attn_norm.weight
F32
F32
[10240]
blk.11.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.11.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.11.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.11.hc_ffn_norm.weight
F32
F32
[10240]
blk.11.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.11.indexer.k_norm.weight
F32
F32
[128]
blk.11.indexer.k_proj.weight
BF16
BF16
[2560, 128]
blk.11.indexer.q_norm.weight
F32
F32
[128]
blk.11.indexer.q_proj.weight
BF16
BF16
[2560, 512]
blk.12
blk.12.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.12.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.12.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.12.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.12.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.12.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.12.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.12.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.12.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.12.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.12.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.12.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.12.hc_attn_norm.weight
F32
F32
[10240]
blk.12.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.12.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.12.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.12.hc_ffn_norm.weight
F32
F32
[10240]
blk.12.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.12.ssm_a
F32
F32
[48]
blk.12.ssm_alpha.weight
F32
F32
[2560, 48]
blk.12.ssm_beta.weight
F32
F32
[2560, 48]
blk.12.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.12.ssm_dt.bias
F32
F32
[48]
blk.12.ssm_norm.weight
F32
F32
[128]
blk.12.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.13
blk.13.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.13.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.13.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.13.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.13.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.13.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.13.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.13.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.13.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.13.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.13.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.13.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.13.hc_attn_norm.weight
F32
F32
[10240]
blk.13.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.13.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.13.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.13.hc_ffn_norm.weight
F32
F32
[10240]
blk.13.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.13.ssm_a
F32
F32
[48]
blk.13.ssm_alpha.weight
F32
F32
[2560, 48]
blk.13.ssm_beta.weight
F32
F32
[2560, 48]
blk.13.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.13.ssm_dt.bias
F32
F32
[48]
blk.13.ssm_norm.weight
F32
F32
[128]
blk.13.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.14
blk.14.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.14.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.14.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.14.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.14.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.14.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.14.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.14.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.14.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.14.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.14.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.14.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.14.hc_attn_norm.weight
F32
F32
[10240]
blk.14.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.14.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.14.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.14.hc_ffn_norm.weight
F32
F32
[10240]
blk.14.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.14.ssm_a
F32
F32
[48]
blk.14.ssm_alpha.weight
F32
F32
[2560, 48]
blk.14.ssm_beta.weight
F32
F32
[2560, 48]
blk.14.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.14.ssm_dt.bias
F32
F32
[48]
blk.14.ssm_norm.weight
F32
F32
[128]
blk.14.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.15
blk.15.attn_k.weight
Q8_0
Q8_0
[2560, 512]
blk.15.attn_k_norm.weight
F32
F32
[256]
blk.15.attn_output.weight
Q8_0
Q8_0
[6144, 2560]
blk.15.attn_q.weight
Q8_0
Q8_0
[2560, 12288]
blk.15.attn_q_norm.weight
F32
F32
[256]
blk.15.attn_v.weight
Q8_0
Q8_0
[2560, 512]
blk.15.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.15.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.15.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.15.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.15.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.15.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.15.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.15.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.15.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.15.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.15.hc_attn_norm.weight
F32
F32
[10240]
blk.15.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.15.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.15.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.15.hc_ffn_norm.weight
F32
F32
[10240]
blk.15.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.15.indexer.k_norm.weight
F32
F32
[128]
blk.15.indexer.k_proj.weight
BF16
BF16
[2560, 128]
blk.15.indexer.q_norm.weight
F32
F32
[128]
blk.15.indexer.q_proj.weight
BF16
BF16
[2560, 512]
blk.16
blk.16.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.16.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.16.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.16.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.16.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.16.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.16.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.16.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.16.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.16.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.16.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.16.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.16.hc_attn_norm.weight
F32
F32
[10240]
blk.16.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.16.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.16.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.16.hc_ffn_norm.weight
F32
F32
[10240]
blk.16.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.16.ssm_a
F32
F32
[48]
blk.16.ssm_alpha.weight
F32
F32
[2560, 48]
blk.16.ssm_beta.weight
F32
F32
[2560, 48]
blk.16.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.16.ssm_dt.bias
F32
F32
[48]
blk.16.ssm_norm.weight
F32
F32
[128]
blk.16.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.17
blk.17.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.17.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.17.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.17.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.17.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.17.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.17.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.17.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.17.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.17.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.17.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.17.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.17.hc_attn_norm.weight
F32
F32
[10240]
blk.17.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.17.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.17.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.17.hc_ffn_norm.weight
F32
F32
[10240]
blk.17.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.17.ssm_a
F32
F32
[48]
blk.17.ssm_alpha.weight
F32
F32
[2560, 48]
blk.17.ssm_beta.weight
F32
F32
[2560, 48]
blk.17.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.17.ssm_dt.bias
F32
F32
[48]
blk.17.ssm_norm.weight
F32
F32
[128]
blk.17.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.18
blk.18.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.18.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.18.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.18.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.18.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.18.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.18.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.18.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.18.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.18.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.18.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.18.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.18.hc_attn_norm.weight
F32
F32
[10240]
blk.18.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.18.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.18.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.18.hc_ffn_norm.weight
F32
F32
[10240]
blk.18.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.18.ssm_a
F32
F32
[48]
blk.18.ssm_alpha.weight
F32
F32
[2560, 48]
blk.18.ssm_beta.weight
F32
F32
[2560, 48]
blk.18.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.18.ssm_dt.bias
F32
F32
[48]
blk.18.ssm_norm.weight
F32
F32
[128]
blk.18.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.19
blk.19.attn_k.weight
Q8_0
Q8_0
[2560, 512]
blk.19.attn_k_norm.weight
F32
F32
[256]
blk.19.attn_output.weight
Q8_0
Q8_0
[6144, 2560]
blk.19.attn_q.weight
Q8_0
Q8_0
[2560, 12288]
blk.19.attn_q_norm.weight
F32
F32
[256]
blk.19.attn_v.weight
Q8_0
Q8_0
[2560, 512]
blk.19.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.19.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.19.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.19.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.19.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.19.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.19.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.19.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.19.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.19.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.19.hc_attn_norm.weight
F32
F32
[10240]
blk.19.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.19.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.19.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.19.hc_ffn_norm.weight
F32
F32
[10240]
blk.19.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.19.indexer.k_norm.weight
F32
F32
[128]
blk.19.indexer.k_proj.weight
BF16
BF16
[2560, 128]
blk.19.indexer.q_norm.weight
F32
F32
[128]
blk.19.indexer.q_proj.weight
BF16
BF16
[2560, 512]
blk.20
blk.20.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.20.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.20.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.20.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.20.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.20.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.20.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.20.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.20.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.20.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.20.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.20.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.20.hc_attn_norm.weight
F32
F32
[10240]
blk.20.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.20.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.20.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.20.hc_ffn_norm.weight
F32
F32
[10240]
blk.20.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.20.ssm_a
F32
F32
[48]
blk.20.ssm_alpha.weight
F32
F32
[2560, 48]
blk.20.ssm_beta.weight
F32
F32
[2560, 48]
blk.20.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.20.ssm_dt.bias
F32
F32
[48]
blk.20.ssm_norm.weight
F32
F32
[128]
blk.20.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.21
blk.21.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.21.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.21.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.21.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.21.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.21.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.21.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.21.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.21.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.21.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.21.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.21.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.21.hc_attn_norm.weight
F32
F32
[10240]
blk.21.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.21.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.21.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.21.hc_ffn_norm.weight
F32
F32
[10240]
blk.21.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.21.ssm_a
F32
F32
[48]
blk.21.ssm_alpha.weight
F32
F32
[2560, 48]
blk.21.ssm_beta.weight
F32
F32
[2560, 48]
blk.21.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.21.ssm_dt.bias
F32
F32
[48]
blk.21.ssm_norm.weight
F32
F32
[128]
blk.21.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.22
blk.22.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.22.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.22.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.22.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.22.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.22.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.22.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.22.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.22.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.22.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.22.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.22.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.22.hc_attn_norm.weight
F32
F32
[10240]
blk.22.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.22.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.22.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.22.hc_ffn_norm.weight
F32
F32
[10240]
blk.22.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.22.ssm_a
F32
F32
[48]
blk.22.ssm_alpha.weight
F32
F32
[2560, 48]
blk.22.ssm_beta.weight
F32
F32
[2560, 48]
blk.22.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.22.ssm_dt.bias
F32
F32
[48]
blk.22.ssm_norm.weight
F32
F32
[128]
blk.22.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.23
blk.23.attn_k.weight
Q8_0
Q8_0
[2560, 512]
blk.23.attn_k_norm.weight
F32
F32
[256]
blk.23.attn_output.weight
Q8_0
Q8_0
[6144, 2560]
blk.23.attn_q.weight
Q8_0
Q8_0
[2560, 12288]
blk.23.attn_q_norm.weight
F32
F32
[256]
blk.23.attn_v.weight
Q8_0
Q8_0
[2560, 512]
blk.23.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.23.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.23.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.23.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.23.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.23.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.23.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.23.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.23.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.23.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.23.hc_attn_norm.weight
F32
F32
[10240]
blk.23.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.23.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.23.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.23.hc_ffn_norm.weight
F32
F32
[10240]
blk.23.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.23.indexer.k_norm.weight
F32
F32
[128]
blk.23.indexer.k_proj.weight
BF16
BF16
[2560, 128]
blk.23.indexer.q_norm.weight
F32
F32
[128]
blk.23.indexer.q_proj.weight
BF16
BF16
[2560, 512]
blk.24
blk.24.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.24.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.24.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.24.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.24.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.24.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.24.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.24.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.24.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.24.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.24.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.24.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.24.hc_attn_norm.weight
F32
F32
[10240]
blk.24.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.24.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.24.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.24.hc_ffn_norm.weight
F32
F32
[10240]
blk.24.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.24.ssm_a
F32
F32
[48]
blk.24.ssm_alpha.weight
F32
F32
[2560, 48]
blk.24.ssm_beta.weight
F32
F32
[2560, 48]
blk.24.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.24.ssm_dt.bias
F32
F32
[48]
blk.24.ssm_norm.weight
F32
F32
[128]
blk.24.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.25
blk.25.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.25.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.25.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.25.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.25.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.25.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.25.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.25.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.25.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.25.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.25.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.25.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.25.hc_attn_norm.weight
F32
F32
[10240]
blk.25.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.25.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.25.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.25.hc_ffn_norm.weight
F32
F32
[10240]
blk.25.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.25.ssm_a
F32
F32
[48]
blk.25.ssm_alpha.weight
F32
F32
[2560, 48]
blk.25.ssm_beta.weight
F32
F32
[2560, 48]
blk.25.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.25.ssm_dt.bias
F32
F32
[48]
blk.25.ssm_norm.weight
F32
F32
[128]
blk.25.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.26
blk.26.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.26.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.26.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.26.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.26.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.26.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.26.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.26.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.26.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.26.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.26.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.26.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.26.hc_attn_norm.weight
F32
F32
[10240]
blk.26.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.26.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.26.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.26.hc_ffn_norm.weight
F32
F32
[10240]
blk.26.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.26.ssm_a
F32
F32
[48]
blk.26.ssm_alpha.weight
F32
F32
[2560, 48]
blk.26.ssm_beta.weight
F32
F32
[2560, 48]
blk.26.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.26.ssm_dt.bias
F32
F32
[48]
blk.26.ssm_norm.weight
F32
F32
[128]
blk.26.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.27
blk.27.attn_k.weight
Q8_0
Q8_0
[2560, 512]
blk.27.attn_k_norm.weight
F32
F32
[256]
blk.27.attn_output.weight
Q8_0
Q8_0
[6144, 2560]
blk.27.attn_q.weight
Q8_0
Q8_0
[2560, 12288]
blk.27.attn_q_norm.weight
F32
F32
[256]
blk.27.attn_v.weight
Q8_0
Q8_0
[2560, 512]
blk.27.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.27.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.27.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.27.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.27.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.27.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.27.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.27.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.27.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.27.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.27.hc_attn_norm.weight
F32
F32
[10240]
blk.27.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.27.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.27.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.27.hc_ffn_norm.weight
F32
F32
[10240]
blk.27.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.27.indexer.k_norm.weight
F32
F32
[128]
blk.27.indexer.k_proj.weight
BF16
BF16
[2560, 128]
blk.27.indexer.q_norm.weight
F32
F32
[128]
blk.27.indexer.q_proj.weight
BF16
BF16
[2560, 512]
blk.28
blk.28.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.28.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.28.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.28.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.28.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.28.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.28.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.28.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.28.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.28.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.28.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.28.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.28.hc_attn_norm.weight
F32
F32
[10240]
blk.28.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.28.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.28.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.28.hc_ffn_norm.weight
F32
F32
[10240]
blk.28.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.28.ssm_a
F32
F32
[48]
blk.28.ssm_alpha.weight
F32
F32
[2560, 48]
blk.28.ssm_beta.weight
F32
F32
[2560, 48]
blk.28.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.28.ssm_dt.bias
F32
F32
[48]
blk.28.ssm_norm.weight
F32
F32
[128]
blk.28.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.29
blk.29.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.29.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.29.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.29.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.29.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.29.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.29.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.29.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.29.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.29.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.29.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.29.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.29.hc_attn_norm.weight
F32
F32
[10240]
blk.29.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.29.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.29.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.29.hc_ffn_norm.weight
F32
F32
[10240]
blk.29.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.29.ssm_a
F32
F32
[48]
blk.29.ssm_alpha.weight
F32
F32
[2560, 48]
blk.29.ssm_beta.weight
F32
F32
[2560, 48]
blk.29.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.29.ssm_dt.bias
F32
F32
[48]
blk.29.ssm_norm.weight
F32
F32
[128]
blk.29.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.30
blk.30.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.30.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.30.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.30.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.30.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.30.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.30.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.30.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.30.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.30.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.30.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.30.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.30.hc_attn_norm.weight
F32
F32
[10240]
blk.30.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.30.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.30.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.30.hc_ffn_norm.weight
F32
F32
[10240]
blk.30.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.30.ssm_a
F32
F32
[48]
blk.30.ssm_alpha.weight
F32
F32
[2560, 48]
blk.30.ssm_beta.weight
F32
F32
[2560, 48]
blk.30.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.30.ssm_dt.bias
F32
F32
[48]
blk.30.ssm_norm.weight
F32
F32
[128]
blk.30.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.31
blk.31.attn_k.weight
Q8_0
Q8_0
[2560, 512]
blk.31.attn_k_norm.weight
F32
F32
[256]
blk.31.attn_output.weight
Q8_0
Q8_0
[6144, 2560]
blk.31.attn_q.weight
Q8_0
Q8_0
[2560, 12288]
blk.31.attn_q_norm.weight
F32
F32
[256]
blk.31.attn_v.weight
Q8_0
Q8_0
[2560, 512]
blk.31.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.31.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.31.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.31.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.31.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.31.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.31.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.31.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.31.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.31.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.31.hc_attn_norm.weight
F32
F32
[10240]
blk.31.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.31.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.31.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.31.hc_ffn_norm.weight
F32
F32
[10240]
blk.31.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.31.indexer.k_norm.weight
F32
F32
[128]
blk.31.indexer.k_proj.weight
BF16
BF16
[2560, 128]
blk.31.indexer.q_norm.weight
F32
F32
[128]
blk.31.indexer.q_proj.weight
BF16
BF16
[2560, 512]
blk.32
blk.32.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.32.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.32.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.32.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.32.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.32.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.32.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.32.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.32.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.32.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.32.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.32.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.32.hc_attn_norm.weight
F32
F32
[10240]
blk.32.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.32.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.32.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.32.hc_ffn_norm.weight
F32
F32
[10240]
blk.32.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.32.ssm_a
F32
F32
[48]
blk.32.ssm_alpha.weight
F32
F32
[2560, 48]
blk.32.ssm_beta.weight
F32
F32
[2560, 48]
blk.32.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.32.ssm_dt.bias
F32
F32
[48]
blk.32.ssm_norm.weight
F32
F32
[128]
blk.32.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.33
blk.33.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.33.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.33.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.33.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.33.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.33.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.33.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.33.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.33.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.33.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.33.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.33.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.33.hc_attn_norm.weight
F32
F32
[10240]
blk.33.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.33.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.33.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.33.hc_ffn_norm.weight
F32
F32
[10240]
blk.33.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.33.ssm_a
F32
F32
[48]
blk.33.ssm_alpha.weight
F32
F32
[2560, 48]
blk.33.ssm_beta.weight
F32
F32
[2560, 48]
blk.33.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.33.ssm_dt.bias
F32
F32
[48]
blk.33.ssm_norm.weight
F32
F32
[128]
blk.33.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.34
blk.34.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.34.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.34.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.34.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.34.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.34.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.34.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.34.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.34.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.34.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.34.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.34.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.34.hc_attn_norm.weight
F32
F32
[10240]
blk.34.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.34.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.34.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.34.hc_ffn_norm.weight
F32
F32
[10240]
blk.34.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.34.ssm_a
F32
F32
[48]
blk.34.ssm_alpha.weight
F32
F32
[2560, 48]
blk.34.ssm_beta.weight
F32
F32
[2560, 48]
blk.34.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.34.ssm_dt.bias
F32
F32
[48]
blk.34.ssm_norm.weight
F32
F32
[128]
blk.34.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.35
blk.35.attn_k.weight
Q8_0
Q8_0
[2560, 512]
blk.35.attn_k_norm.weight
F32
F32
[256]
blk.35.attn_output.weight
Q8_0
Q8_0
[6144, 2560]
blk.35.attn_q.weight
Q8_0
Q8_0
[2560, 12288]
blk.35.attn_q_norm.weight
F32
F32
[256]
blk.35.attn_v.weight
Q8_0
Q8_0
[2560, 512]
blk.35.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.35.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.35.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.35.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.35.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.35.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.35.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.35.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.35.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.35.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.35.hc_attn_norm.weight
F32
F32
[10240]
blk.35.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.35.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.35.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.35.hc_ffn_norm.weight
F32
F32
[10240]
blk.35.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.35.indexer.k_norm.weight
F32
F32
[128]
blk.35.indexer.k_proj.weight
BF16
BF16
[2560, 128]
blk.35.indexer.q_norm.weight
F32
F32
[128]
blk.35.indexer.q_proj.weight
BF16
BF16
[2560, 512]
blk.36
blk.36.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.36.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.36.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.36.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.36.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.36.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.36.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.36.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.36.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.36.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.36.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.36.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.36.hc_attn_norm.weight
F32
F32
[10240]
blk.36.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.36.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.36.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.36.hc_ffn_norm.weight
F32
F32
[10240]
blk.36.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.36.ssm_a
F32
F32
[48]
blk.36.ssm_alpha.weight
F32
F32
[2560, 48]
blk.36.ssm_beta.weight
F32
F32
[2560, 48]
blk.36.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.36.ssm_dt.bias
F32
F32
[48]
blk.36.ssm_norm.weight
F32
F32
[128]
blk.36.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.37
blk.37.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.37.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.37.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.37.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.37.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.37.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.37.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.37.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.37.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.37.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.37.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.37.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.37.hc_attn_norm.weight
F32
F32
[10240]
blk.37.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.37.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.37.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.37.hc_ffn_norm.weight
F32
F32
[10240]
blk.37.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.37.ssm_a
F32
F32
[48]
blk.37.ssm_alpha.weight
F32
F32
[2560, 48]
blk.37.ssm_beta.weight
F32
F32
[2560, 48]
blk.37.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.37.ssm_dt.bias
F32
F32
[48]
blk.37.ssm_norm.weight
F32
F32
[128]
blk.37.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.38
blk.38.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.38.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.38.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.38.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.38.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.38.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.38.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.38.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.38.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.38.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.38.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.38.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.38.hc_attn_norm.weight
F32
F32
[10240]
blk.38.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.38.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.38.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.38.hc_ffn_norm.weight
F32
F32
[10240]
blk.38.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.38.ssm_a
F32
F32
[48]
blk.38.ssm_alpha.weight
F32
F32
[2560, 48]
blk.38.ssm_beta.weight
F32
F32
[2560, 48]
blk.38.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.38.ssm_dt.bias
F32
F32
[48]
blk.38.ssm_norm.weight
F32
F32
[128]
blk.38.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.39
blk.39.attn_k.weight
Q8_0
Q8_0
[2560, 512]
blk.39.attn_k_norm.weight
F32
F32
[256]
blk.39.attn_output.weight
Q8_0
Q8_0
[6144, 2560]
blk.39.attn_q.weight
Q8_0
Q8_0
[2560, 12288]
blk.39.attn_q_norm.weight
F32
F32
[256]
blk.39.attn_v.weight
Q8_0
Q8_0
[2560, 512]
blk.39.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.39.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.39.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.39.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.39.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.39.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.39.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.39.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.39.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.39.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.39.hc_attn_norm.weight
F32
F32
[10240]
blk.39.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.39.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.39.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.39.hc_ffn_norm.weight
F32
F32
[10240]
blk.39.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.39.indexer.k_norm.weight
F32
F32
[128]
blk.39.indexer.k_proj.weight
BF16
BF16
[2560, 128]
blk.39.indexer.q_norm.weight
F32
F32
[128]
blk.39.indexer.q_proj.weight
BF16
BF16
[2560, 512]
blk.40
blk.40.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.40.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]
blk.40.ffn_down_exps.weight
Q5_1
Q5_1
[640, 2560, 512]
blk.40.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.40.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.40.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.40.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.40.ffn_gate_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.40.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.40.ffn_up_shexp.weight
Q8_0
Q8_0
[2560, 640]
blk.40.hc_attn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.40.hc_attn_inject.weight
F32
F32
[10240, 4]
blk.40.hc_attn_norm.weight
F32
F32
[10240]
blk.40.hc_attn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.40.hc_ffn_down.weight
Q8_0
Q8_0
[10240, 320]
blk.40.hc_ffn_inject.weight
F32
F32
[10240, 4]
blk.40.hc_ffn_norm.weight
F32
F32
[10240]
blk.40.hc_ffn_up.weight
Q8_0
Q8_0
[320, 10240]
blk.40.ssm_a
F32
F32
[48]
blk.40.ssm_alpha.weight
F32
F32
[2560, 48]
blk.40.ssm_beta.weight
F32
F32
[2560, 48]
blk.40.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.40.ssm_dt.bias
F32
F32
[48]
blk.40.ssm_norm.weight
F32
F32
[128]
blk.40.ssm_out.weight
Q8_0
Q8_0
[6144, 2560]
blk.41
blk.41.attn_gate.weight
Q8_0
Q8_0
[2560, 6144]
blk.41.attn_qkv.weight
Q8_0
Q8_0
[2560, 10240]