Models
Docs
Pricing
Sign in
Download
Models
Download
Docs
Pricing
Sign in
qwen3.8-flash-next
:125b-a6b-q4_K_M
80.4K
Downloads
Updated
19 minutes ago
This experimental preview of the architecture that will underpin Qwen4.
This experimental preview of the architecture that will underpin Qwen4.
Cancel
vision
tools
thinking
qwen3.8-flash-next:125b-a6b-q4_K_M
...
/
model
89059c880c2f · 119GB
Metadata
general.architecture
qwen4exp
qwen4exp
general.file_type
Q4_K_M
Q4_K_M
qwen4exp.attention.compress_ratios
[0, 0, 0, 4, 0, ...]
[0, 0, 0, 4, 0, ...]
qwen4exp.attention.head_count
24
24
qwen4exp.attention.head_count_kv
2
2
qwen4exp.attention.indexer.head_count
4
4
qwen4exp.attention.indexer.key_length
128
128
qwen4exp.attention.indexer.top_k
2048
2048
qwen4exp.attention.key_length
256
256
qwen4exp.attention.layer_norm_rms_epsilon
1e-06
1e-06
qwen4exp.attention.value_length
256
256
qwen4exp.block_count
48
48
qwen4exp.context_length
262144
262144
qwen4exp.embedding_length
2560
2560
qwen4exp.embedding_length_per_layer_input
160
160
qwen4exp.expert_count
512
512
qwen4exp.expert_feed_forward_length
640
640
qwen4exp.expert_shared_feed_forward_length
640
640
qwen4exp.expert_used_count
10
10
qwen4exp.full_attention_interval
4
4
qwen4exp.hyper_connection.count
4
4
qwen4exp.hyper_connection.low_rank
320
320
qwen4exp.ple.conv_kernel
4
4
qwen4exp.ple.eos_token_id
248044
248044
qwen4exp.ple.head_offsets
[0, 20000003, 40000026, 60000059, 80000106, ...]
[0, 20000003, 40000026, 60000059, 80000106, ...]
qwen4exp.ple.head_vocab_sizes
[20000003, 20000023, 20000033, 20000047, 20000059, ...]
[20000003, 20000023, 20000033, 20000047, 20000059, ...]
qwen4exp.ple.heads_per_ngram
8
8
qwen4exp.ple.image_token_id
248056
248056
qwen4exp.ple.layer_multipliers
[23703573157769, 20109073645365, 8052911324071]
[23703573157769, 20109073645365, 8052911324071]
qwen4exp.ple.layers
[1]
[1]
qwen4exp.ple.ngram_size
3
3
qwen4exp.rope.dimension_count
64
64
qwen4exp.rope.dimension_sections
[11, 11, 10, 0]
[11, 11, 10, 0]
qwen4exp.rope.freq_base
1e+07
1e+07
qwen4exp.ssm.conv_kernel
4
4
qwen4exp.ssm.group_count
16
16
qwen4exp.ssm.inner_size
6144
6144
qwen4exp.ssm.state_size
128
128
qwen4exp.ssm.time_step_rank
48
48
tokenizer.ggml.add_bos_token
false
false
tokenizer.ggml.bos_token_id
248044
248044
tokenizer.ggml.eos_token_id
248046
248046
tokenizer.ggml.merges
[Ġ Ġ, ĠĠ ĠĠ, i n, Ġ t, ĠĠĠĠ ĠĠĠĠ, ...]
[Ġ Ġ, ĠĠ ĠĠ, i n, Ġ t, ĠĠĠĠ ĠĠĠĠ, ...]
tokenizer.ggml.model
gpt2
gpt2
tokenizer.ggml.padding_token_id
248044
248044
tokenizer.ggml.pre
qwen35
qwen35
tokenizer.ggml.token_type
[1, 1, 1, 1, 1, ...]
[1, 1, 1, 1, 1, ...]
tokenizer.ggml.tokens
[!, ", #, $, %, ...]
[!, ", #, $, %, ...]
Tensor
Name
Type
Shape
token_embd.weight
Q4_K
Q4_K
[2560, 248320]
blk.0
blk.0.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.0.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.0.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.0.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.0.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.0.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.0.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.0.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.0.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.0.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.0.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.0.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.0.hc_attn_norm.weight
F32
F32
[10240]
blk.0.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.0.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.0.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.0.hc_ffn_norm.weight
F32
F32
[10240]
blk.0.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.0.ssm_a
F32
F32
[48]
blk.0.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.0.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.0.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.0.ssm_dt.bias
F32
F32
[48]
blk.0.ssm_norm.weight
F32
F32
[128]
blk.0.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.1
blk.1.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.1.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.1.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.1.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.1.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.1.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.1.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.1.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.1.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.1.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.1.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.1.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.1.hc_attn_norm.weight
F32
F32
[10240]
blk.1.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.1.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.1.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.1.hc_ffn_norm.weight
F32
F32
[10240]
blk.1.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.1.ple_conv1d.weight
F16
F16
[4, 10240]
blk.1.ple_key.weight
Q4_K
Q4_K
[2560, 10240]
blk.1.ple_norm_conv.weight
F32
F32
[10240]
blk.1.ple_norm_key.weight
F32
F32
[10240]
blk.1.ple_norm_query.weight
F32
F32
[10240]
blk.1.ple_value.weight
Q4_K
Q4_K
[2560, 2560]
blk.1.ssm_a
F32
F32
[48]
blk.1.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.1.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.1.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.1.ssm_dt.bias
F32
F32
[48]
blk.1.ssm_norm.weight
F32
F32
[128]
blk.1.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.2
blk.2.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.2.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.2.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.2.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.2.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.2.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.2.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.2.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.2.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.2.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.2.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.2.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.2.hc_attn_norm.weight
F32
F32
[10240]
blk.2.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.2.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.2.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.2.hc_ffn_norm.weight
F32
F32
[10240]
blk.2.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.2.ssm_a
F32
F32
[48]
blk.2.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.2.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.2.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.2.ssm_dt.bias
F32
F32
[48]
blk.2.ssm_norm.weight
F32
F32
[128]
blk.2.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.3
blk.3.attn_k.weight
Q4_K
Q4_K
[2560, 512]
blk.3.attn_k_norm.weight
F32
F32
[256]
blk.3.attn_output.weight
Q4_K
Q4_K
[6144, 2560]
blk.3.attn_q.weight
Q4_K
Q4_K
[2560, 12288]
blk.3.attn_q_norm.weight
F32
F32
[256]
blk.3.attn_v.weight
Q6_K
Q6_K
[2560, 512]
blk.3.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.3.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.3.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.3.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.3.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.3.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.3.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.3.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.3.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.3.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.3.hc_attn_norm.weight
F32
F32
[10240]
blk.3.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.3.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.3.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.3.hc_ffn_norm.weight
F32
F32
[10240]
blk.3.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.3.indexer.k_norm.weight
F32
F32
[128]
blk.3.indexer.k_proj.weight
BF16
BF16
[2560, 128]
blk.3.indexer.q_norm.weight
F32
F32
[128]
blk.3.indexer.q_proj.weight
BF16
BF16
[2560, 512]
blk.4
blk.4.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.4.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.4.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.4.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.4.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.4.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.4.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.4.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.4.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.4.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.4.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.4.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.4.hc_attn_norm.weight
F32
F32
[10240]
blk.4.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.4.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.4.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.4.hc_ffn_norm.weight
F32
F32
[10240]
blk.4.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.4.ssm_a
F32
F32
[48]
blk.4.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.4.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.4.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.4.ssm_dt.bias
F32
F32
[48]
blk.4.ssm_norm.weight
F32
F32
[128]
blk.4.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.5
blk.5.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.5.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.5.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.5.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.5.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.5.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.5.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.5.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.5.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.5.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.5.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.5.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.5.hc_attn_norm.weight
F32
F32
[10240]
blk.5.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.5.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.5.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.5.hc_ffn_norm.weight
F32
F32
[10240]
blk.5.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.5.ssm_a
F32
F32
[48]
blk.5.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.5.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.5.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.5.ssm_dt.bias
F32
F32
[48]
blk.5.ssm_norm.weight
F32
F32
[128]
blk.5.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.6
blk.6.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.6.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.6.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.6.ffn_down_shexp.weight
Q5_0
Q5_0
[640, 2560]
blk.6.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.6.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.6.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.6.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.6.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.6.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.6.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.6.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.6.hc_attn_norm.weight
F32
F32
[10240]
blk.6.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.6.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.6.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.6.hc_ffn_norm.weight
F32
F32
[10240]
blk.6.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.6.ssm_a
F32
F32
[48]
blk.6.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.6.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.6.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.6.ssm_dt.bias
F32
F32
[48]
blk.6.ssm_norm.weight
F32
F32
[128]
blk.6.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.7
blk.7.attn_k.weight
Q4_K
Q4_K
[2560, 512]
blk.7.attn_k_norm.weight
F32
F32
[256]
blk.7.attn_output.weight
Q4_K
Q4_K
[6144, 2560]
blk.7.attn_q.weight
Q4_K
Q4_K
[2560, 12288]
blk.7.attn_q_norm.weight
F32
F32
[256]
blk.7.attn_v.weight
Q4_K
Q4_K
[2560, 512]
blk.7.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.7.ffn_down_shexp.weight
Q5_0
Q5_0
[640, 2560]
blk.7.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.7.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.7.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.7.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.7.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.7.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.7.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.7.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.7.hc_attn_norm.weight
F32
F32
[10240]
blk.7.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.7.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.7.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.7.hc_ffn_norm.weight
F32
F32
[10240]
blk.7.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.7.indexer.k_norm.weight
F32
F32
[128]
blk.7.indexer.k_proj.weight
BF16
BF16
[2560, 128]
blk.7.indexer.q_norm.weight
F32
F32
[128]
blk.7.indexer.q_proj.weight
BF16
BF16
[2560, 512]
blk.8
blk.8.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.8.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.8.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.8.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.8.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.8.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.8.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.8.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.8.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.8.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.8.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.8.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.8.hc_attn_norm.weight
F32
F32
[10240]
blk.8.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.8.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.8.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.8.hc_ffn_norm.weight
F32
F32
[10240]
blk.8.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.8.ssm_a
F32
F32
[48]
blk.8.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.8.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.8.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.8.ssm_dt.bias
F32
F32
[48]
blk.8.ssm_norm.weight
F32
F32
[128]
blk.8.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.9
blk.9.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.9.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.9.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.9.ffn_down_shexp.weight
Q5_0
Q5_0
[640, 2560]
blk.9.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.9.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.9.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.9.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.9.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.9.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.9.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.9.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.9.hc_attn_norm.weight
F32
F32
[10240]
blk.9.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.9.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.9.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.9.hc_ffn_norm.weight
F32
F32
[10240]
blk.9.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.9.ssm_a
F32
F32
[48]
blk.9.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.9.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.9.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.9.ssm_dt.bias
F32
F32
[48]
blk.9.ssm_norm.weight
F32
F32
[128]
blk.9.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.10
blk.10.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.10.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.10.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.10.ffn_down_shexp.weight
Q5_0
Q5_0
[640, 2560]
blk.10.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.10.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.10.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.10.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.10.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.10.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.10.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.10.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.10.hc_attn_norm.weight
F32
F32
[10240]
blk.10.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.10.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.10.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.10.hc_ffn_norm.weight
F32
F32
[10240]
blk.10.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.10.ssm_a
F32
F32
[48]
blk.10.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.10.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.10.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.10.ssm_dt.bias
F32
F32
[48]
blk.10.ssm_norm.weight
F32
F32
[128]
blk.10.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.11
blk.11.attn_k.weight
Q4_K
Q4_K
[2560, 512]
blk.11.attn_k_norm.weight
F32
F32
[256]
blk.11.attn_output.weight
Q4_K
Q4_K
[6144, 2560]
blk.11.attn_q.weight
Q4_K
Q4_K
[2560, 12288]
blk.11.attn_q_norm.weight
F32
F32
[256]
blk.11.attn_v.weight
Q6_K
Q6_K
[2560, 512]
blk.11.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.11.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
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
Q4_K
Q4_K
[2560, 640]
blk.11.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.11.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.11.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.11.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.11.hc_attn_norm.weight
F32
F32
[10240]
blk.11.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.11.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.11.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.11.hc_ffn_norm.weight
F32
F32
[10240]
blk.11.hc_ffn_up.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 6144]
blk.12.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.12.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.12.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.12.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.12.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.12.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.12.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.12.hc_attn_norm.weight
F32
F32
[10240]
blk.12.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.12.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.12.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.12.hc_ffn_norm.weight
F32
F32
[10240]
blk.12.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.12.ssm_a
F32
F32
[48]
blk.12.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.12.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.13
blk.13.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.13.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.13.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.13.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.13.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.13.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.13.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.13.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.13.hc_attn_norm.weight
F32
F32
[10240]
blk.13.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.13.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.13.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.13.hc_ffn_norm.weight
F32
F32
[10240]
blk.13.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.13.ssm_a
F32
F32
[48]
blk.13.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.13.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.14
blk.14.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.14.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.14.ffn_down_exps.weight
Q8_0
Q8_0
[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
Q4_K
Q4_K
[2560, 640]
blk.14.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.14.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.14.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.14.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.14.hc_attn_norm.weight
F32
F32
[10240]
blk.14.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.14.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.14.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.14.hc_ffn_norm.weight
F32
F32
[10240]
blk.14.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.14.ssm_a
F32
F32
[48]
blk.14.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.14.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.15
blk.15.attn_k.weight
Q4_K
Q4_K
[2560, 512]
blk.15.attn_k_norm.weight
F32
F32
[256]
blk.15.attn_output.weight
Q4_K
Q4_K
[6144, 2560]
blk.15.attn_q.weight
Q4_K
Q4_K
[2560, 12288]
blk.15.attn_q_norm.weight
F32
F32
[256]
blk.15.attn_v.weight
Q4_K
Q4_K
[2560, 512]
blk.15.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.15.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.15.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.15.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.15.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.15.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.15.hc_attn_norm.weight
F32
F32
[10240]
blk.15.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.15.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.15.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.15.hc_ffn_norm.weight
F32
F32
[10240]
blk.15.hc_ffn_up.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 6144]
blk.16.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.16.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.16.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.16.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.16.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.16.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.16.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.16.hc_attn_norm.weight
F32
F32
[10240]
blk.16.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.16.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.16.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.16.hc_ffn_norm.weight
F32
F32
[10240]
blk.16.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.16.ssm_a
F32
F32
[48]
blk.16.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.16.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.17
blk.17.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.17.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.17.ffn_down_exps.weight
Q8_0
Q8_0
[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
Q4_K
Q4_K
[2560, 640]
blk.17.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.17.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.17.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.17.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.17.hc_attn_norm.weight
F32
F32
[10240]
blk.17.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.17.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.17.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.17.hc_ffn_norm.weight
F32
F32
[10240]
blk.17.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.17.ssm_a
F32
F32
[48]
blk.17.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.17.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.18
blk.18.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.18.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.18.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.18.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.18.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.18.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.18.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.18.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.18.hc_attn_norm.weight
F32
F32
[10240]
blk.18.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.18.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.18.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.18.hc_ffn_norm.weight
F32
F32
[10240]
blk.18.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.18.ssm_a
F32
F32
[48]
blk.18.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.18.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.19
blk.19.attn_k.weight
Q4_K
Q4_K
[2560, 512]
blk.19.attn_k_norm.weight
F32
F32
[256]
blk.19.attn_output.weight
Q4_K
Q4_K
[6144, 2560]
blk.19.attn_q.weight
Q4_K
Q4_K
[2560, 12288]
blk.19.attn_q_norm.weight
F32
F32
[256]
blk.19.attn_v.weight
Q4_K
Q4_K
[2560, 512]
blk.19.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.19.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.19.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.19.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.19.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.19.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.19.hc_attn_norm.weight
F32
F32
[10240]
blk.19.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.19.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.19.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.19.hc_ffn_norm.weight
F32
F32
[10240]
blk.19.hc_ffn_up.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 6144]
blk.20.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.20.ffn_down_exps.weight
Q8_0
Q8_0
[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
Q4_K
Q4_K
[2560, 640]
blk.20.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.20.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.20.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.20.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.20.hc_attn_norm.weight
F32
F32
[10240]
blk.20.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.20.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.20.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.20.hc_ffn_norm.weight
F32
F32
[10240]
blk.20.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.20.ssm_a
F32
F32
[48]
blk.20.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.20.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.21
blk.21.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.21.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.21.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.21.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.21.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.21.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.21.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.21.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.21.hc_attn_norm.weight
F32
F32
[10240]
blk.21.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.21.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.21.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.21.hc_ffn_norm.weight
F32
F32
[10240]
blk.21.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.21.ssm_a
F32
F32
[48]
blk.21.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.21.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.22
blk.22.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.22.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.22.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.22.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.22.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.22.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.22.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.22.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.22.hc_attn_norm.weight
F32
F32
[10240]
blk.22.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.22.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.22.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.22.hc_ffn_norm.weight
F32
F32
[10240]
blk.22.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.22.ssm_a
F32
F32
[48]
blk.22.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.22.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.23
blk.23.attn_k.weight
Q4_K
Q4_K
[2560, 512]
blk.23.attn_k_norm.weight
F32
F32
[256]
blk.23.attn_output.weight
Q4_K
Q4_K
[6144, 2560]
blk.23.attn_q.weight
Q4_K
Q4_K
[2560, 12288]
blk.23.attn_q_norm.weight
F32
F32
[256]
blk.23.attn_v.weight
Q6_K
Q6_K
[2560, 512]
blk.23.ffn_down_exps.weight
Q8_0
Q8_0
[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
Q4_K
Q4_K
[2560, 640]
blk.23.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.23.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.23.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.23.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.23.hc_attn_norm.weight
F32
F32
[10240]
blk.23.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.23.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.23.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.23.hc_ffn_norm.weight
F32
F32
[10240]
blk.23.hc_ffn_up.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 6144]
blk.24.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.24.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.24.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.24.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.24.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.24.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.24.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.24.hc_attn_norm.weight
F32
F32
[10240]
blk.24.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.24.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.24.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.24.hc_ffn_norm.weight
F32
F32
[10240]
blk.24.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.24.ssm_a
F32
F32
[48]
blk.24.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.24.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.25
blk.25.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.25.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.25.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.25.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.25.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.25.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.25.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.25.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.25.hc_attn_norm.weight
F32
F32
[10240]
blk.25.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.25.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.25.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.25.hc_ffn_norm.weight
F32
F32
[10240]
blk.25.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.25.ssm_a
F32
F32
[48]
blk.25.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.25.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.26
blk.26.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.26.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.26.ffn_down_exps.weight
Q8_0
Q8_0
[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
Q4_K
Q4_K
[2560, 640]
blk.26.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.26.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.26.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.26.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.26.hc_attn_norm.weight
F32
F32
[10240]
blk.26.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.26.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.26.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.26.hc_ffn_norm.weight
F32
F32
[10240]
blk.26.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.26.ssm_a
F32
F32
[48]
blk.26.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.26.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.27
blk.27.attn_k.weight
Q4_K
Q4_K
[2560, 512]
blk.27.attn_k_norm.weight
F32
F32
[256]
blk.27.attn_output.weight
Q4_K
Q4_K
[6144, 2560]
blk.27.attn_q.weight
Q4_K
Q4_K
[2560, 12288]
blk.27.attn_q_norm.weight
F32
F32
[256]
blk.27.attn_v.weight
Q4_K
Q4_K
[2560, 512]
blk.27.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.27.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.27.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.27.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.27.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.27.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.27.hc_attn_norm.weight
F32
F32
[10240]
blk.27.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.27.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.27.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.27.hc_ffn_norm.weight
F32
F32
[10240]
blk.27.hc_ffn_up.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 6144]
blk.28.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.28.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.28.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.28.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.28.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.28.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.28.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.28.hc_attn_norm.weight
F32
F32
[10240]
blk.28.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.28.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.28.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.28.hc_ffn_norm.weight
F32
F32
[10240]
blk.28.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.28.ssm_a
F32
F32
[48]
blk.28.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.28.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.29
blk.29.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.29.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.29.ffn_down_exps.weight
Q8_0
Q8_0
[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
Q4_K
Q4_K
[2560, 640]
blk.29.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.29.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.29.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.29.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.29.hc_attn_norm.weight
F32
F32
[10240]
blk.29.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.29.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.29.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.29.hc_ffn_norm.weight
F32
F32
[10240]
blk.29.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.29.ssm_a
F32
F32
[48]
blk.29.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.29.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.30
blk.30.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.30.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.30.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.30.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.30.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.30.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.30.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.30.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.30.hc_attn_norm.weight
F32
F32
[10240]
blk.30.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.30.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.30.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.30.hc_ffn_norm.weight
F32
F32
[10240]
blk.30.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.30.ssm_a
F32
F32
[48]
blk.30.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.30.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.31
blk.31.attn_k.weight
Q4_K
Q4_K
[2560, 512]
blk.31.attn_k_norm.weight
F32
F32
[256]
blk.31.attn_output.weight
Q4_K
Q4_K
[6144, 2560]
blk.31.attn_q.weight
Q4_K
Q4_K
[2560, 12288]
blk.31.attn_q_norm.weight
F32
F32
[256]
blk.31.attn_v.weight
Q4_K
Q4_K
[2560, 512]
blk.31.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.31.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.31.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.31.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.31.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.31.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.31.hc_attn_norm.weight
F32
F32
[10240]
blk.31.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.31.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.31.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.31.hc_ffn_norm.weight
F32
F32
[10240]
blk.31.hc_ffn_up.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 6144]
blk.32.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.32.ffn_down_exps.weight
Q8_0
Q8_0
[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
Q4_K
Q4_K
[2560, 640]
blk.32.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.32.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.32.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.32.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.32.hc_attn_norm.weight
F32
F32
[10240]
blk.32.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.32.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.32.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.32.hc_ffn_norm.weight
F32
F32
[10240]
blk.32.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.32.ssm_a
F32
F32
[48]
blk.32.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.32.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.33
blk.33.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.33.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.33.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.33.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.33.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.33.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.33.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.33.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.33.hc_attn_norm.weight
F32
F32
[10240]
blk.33.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.33.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.33.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.33.hc_ffn_norm.weight
F32
F32
[10240]
blk.33.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.33.ssm_a
F32
F32
[48]
blk.33.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.33.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.34
blk.34.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.34.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.34.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.34.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.34.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.34.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.34.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.34.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.34.hc_attn_norm.weight
F32
F32
[10240]
blk.34.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.34.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.34.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.34.hc_ffn_norm.weight
F32
F32
[10240]
blk.34.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.34.ssm_a
F32
F32
[48]
blk.34.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.34.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.35
blk.35.attn_k.weight
Q4_K
Q4_K
[2560, 512]
blk.35.attn_k_norm.weight
F32
F32
[256]
blk.35.attn_output.weight
Q4_K
Q4_K
[6144, 2560]
blk.35.attn_q.weight
Q4_K
Q4_K
[2560, 12288]
blk.35.attn_q_norm.weight
F32
F32
[256]
blk.35.attn_v.weight
Q6_K
Q6_K
[2560, 512]
blk.35.ffn_down_exps.weight
Q8_0
Q8_0
[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
Q4_K
Q4_K
[2560, 640]
blk.35.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.35.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.35.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.35.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.35.hc_attn_norm.weight
F32
F32
[10240]
blk.35.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.35.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.35.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.35.hc_ffn_norm.weight
F32
F32
[10240]
blk.35.hc_ffn_up.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 6144]
blk.36.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.36.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.36.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.36.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.36.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.36.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.36.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.36.hc_attn_norm.weight
F32
F32
[10240]
blk.36.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.36.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.36.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.36.hc_ffn_norm.weight
F32
F32
[10240]
blk.36.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.36.ssm_a
F32
F32
[48]
blk.36.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.36.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.37
blk.37.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.37.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.37.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.37.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.37.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.37.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.37.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.37.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.37.hc_attn_norm.weight
F32
F32
[10240]
blk.37.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.37.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.37.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.37.hc_ffn_norm.weight
F32
F32
[10240]
blk.37.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.37.ssm_a
F32
F32
[48]
blk.37.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.37.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.38
blk.38.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.38.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.38.ffn_down_exps.weight
Q8_0
Q8_0
[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
Q4_K
Q4_K
[2560, 640]
blk.38.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.38.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.38.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.38.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.38.hc_attn_norm.weight
F32
F32
[10240]
blk.38.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.38.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.38.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.38.hc_ffn_norm.weight
F32
F32
[10240]
blk.38.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.38.ssm_a
F32
F32
[48]
blk.38.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.38.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.39
blk.39.attn_k.weight
Q4_K
Q4_K
[2560, 512]
blk.39.attn_k_norm.weight
F32
F32
[256]
blk.39.attn_output.weight
Q4_K
Q4_K
[6144, 2560]
blk.39.attn_q.weight
Q4_K
Q4_K
[2560, 12288]
blk.39.attn_q_norm.weight
F32
F32
[256]
blk.39.attn_v.weight
Q4_K
Q4_K
[2560, 512]
blk.39.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.39.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.39.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.39.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.39.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.39.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.39.hc_attn_norm.weight
F32
F32
[10240]
blk.39.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.39.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.39.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.39.hc_ffn_norm.weight
F32
F32
[10240]
blk.39.hc_ffn_up.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 6144]
blk.40.attn_qkv.weight
Q4_K
Q4_K
[2560, 10240]
blk.40.ffn_down_exps.weight
Q5_0
Q5_0
[640, 2560, 512]
blk.40.ffn_down_shexp.weight
Q5_0
Q5_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
Q4_K
Q4_K
[2560, 640]
blk.40.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.40.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.40.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.40.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.40.hc_attn_norm.weight
F32
F32
[10240]
blk.40.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.40.hc_ffn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.40.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.40.hc_ffn_norm.weight
F32
F32
[10240]
blk.40.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.40.ssm_a
F32
F32
[48]
blk.40.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.40.ssm_beta.weight
Q4_K
Q4_K
[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
Q4_K
Q4_K
[6144, 2560]
blk.41
blk.41.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.41.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.41.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.41.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.41.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.41.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.41.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.41.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.41.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.41.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.41.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.41.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.41.hc_attn_norm.weight
F32
F32
[10240]
blk.41.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.41.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.41.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.41.hc_ffn_norm.weight
F32
F32
[10240]
blk.41.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.41.ssm_a
F32
F32
[48]
blk.41.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.41.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.41.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.41.ssm_dt.bias
F32
F32
[48]
blk.41.ssm_norm.weight
F32
F32
[128]
blk.41.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.42
blk.42.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.42.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.42.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.42.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.42.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.42.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.42.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.42.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.42.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.42.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.42.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.42.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.42.hc_attn_norm.weight
F32
F32
[10240]
blk.42.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.42.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.42.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.42.hc_ffn_norm.weight
F32
F32
[10240]
blk.42.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.42.ssm_a
F32
F32
[48]
blk.42.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.42.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.42.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.42.ssm_dt.bias
F32
F32
[48]
blk.42.ssm_norm.weight
F32
F32
[128]
blk.42.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.43
blk.43.attn_k.weight
Q4_K
Q4_K
[2560, 512]
blk.43.attn_k_norm.weight
F32
F32
[256]
blk.43.attn_output.weight
Q4_K
Q4_K
[6144, 2560]
blk.43.attn_q.weight
Q4_K
Q4_K
[2560, 12288]
blk.43.attn_q_norm.weight
F32
F32
[256]
blk.43.attn_v.weight
Q6_K
Q6_K
[2560, 512]
blk.43.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.43.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.43.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.43.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.43.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.43.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.43.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.43.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.43.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.43.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.43.hc_attn_norm.weight
F32
F32
[10240]
blk.43.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.43.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.43.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.43.hc_ffn_norm.weight
F32
F32
[10240]
blk.43.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.43.indexer.k_norm.weight
F32
F32
[128]
blk.43.indexer.k_proj.weight
BF16
BF16
[2560, 128]
blk.43.indexer.q_norm.weight
F32
F32
[128]
blk.43.indexer.q_proj.weight
BF16
BF16
[2560, 512]
blk.44
blk.44.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.44.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.44.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.44.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.44.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.44.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.44.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.44.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.44.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.44.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.44.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.44.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.44.hc_attn_norm.weight
F32
F32
[10240]
blk.44.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.44.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.44.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.44.hc_ffn_norm.weight
F32
F32
[10240]
blk.44.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.44.ssm_a
F32
F32
[48]
blk.44.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.44.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.44.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.44.ssm_dt.bias
F32
F32
[48]
blk.44.ssm_norm.weight
F32
F32
[128]
blk.44.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.45
blk.45.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.45.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.45.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.45.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.45.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.45.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.45.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.45.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.45.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.45.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.45.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.45.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.45.hc_attn_norm.weight
F32
F32
[10240]
blk.45.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.45.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.45.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.45.hc_ffn_norm.weight
F32
F32
[10240]
blk.45.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.45.ssm_a
F32
F32
[48]
blk.45.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.45.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.45.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.45.ssm_dt.bias
F32
F32
[48]
blk.45.ssm_norm.weight
F32
F32
[128]
blk.45.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.46
blk.46.attn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.46.attn_qkv.weight
Q6_K
Q6_K
[2560, 10240]
blk.46.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.46.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.46.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.46.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.46.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.46.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.46.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.46.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.46.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.46.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.46.hc_attn_norm.weight
F32
F32
[10240]
blk.46.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.46.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.46.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.46.hc_ffn_norm.weight
F32
F32
[10240]
blk.46.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.46.ssm_a
F32
F32
[48]
blk.46.ssm_alpha.weight
Q4_K
Q4_K
[2560, 48]
blk.46.ssm_beta.weight
Q4_K
Q4_K
[2560, 48]
blk.46.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.46.ssm_dt.bias
F32
F32
[48]
blk.46.ssm_norm.weight
F32
F32
[128]
blk.46.ssm_out.weight
Q4_K
Q4_K
[6144, 2560]
blk.47
blk.47.attn_k.weight
Q4_K
Q4_K
[2560, 512]
blk.47.attn_k_norm.weight
F32
F32
[256]
blk.47.attn_output.weight
Q4_K
Q4_K
[6144, 2560]
blk.47.attn_q.weight
Q4_K
Q4_K
[2560, 12288]
blk.47.attn_q_norm.weight
F32
F32
[256]
blk.47.attn_v.weight
Q6_K
Q6_K
[2560, 512]
blk.47.ffn_down_exps.weight
Q8_0
Q8_0
[640, 2560, 512]
blk.47.ffn_down_shexp.weight
Q8_0
Q8_0
[640, 2560]
blk.47.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.47.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.47.ffn_gate_inp_shexp.weight
F32
F32
[2560]
blk.47.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.47.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 640, 512]
blk.47.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 640]
blk.47.hc_attn_down.weight
Q4_K
Q4_K
[10240, 320]
blk.47.hc_attn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.47.hc_attn_norm.weight
F32
F32
[10240]
blk.47.hc_attn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.47.hc_ffn_down.weight
Q6_K
Q6_K
[10240, 320]
blk.47.hc_ffn_inject.weight
Q4_K
Q4_K
[10240, 4]
blk.47.hc_ffn_norm.weight
F32
F32
[10240]
blk.47.hc_ffn_up.weight
Q5_0
Q5_0
[320, 10240]
blk.47.indexer.k_norm.weight
F32
F32
[128]
blk.47.indexer.k_proj.weight
BF16
BF16
[2560, 128]
blk.47.indexer.q_norm.weight
F32
F32
[128]
blk.47.indexer.q_proj.weight
BF16
BF16
[2560, 512]
output.weight
Q6_K
Q6_K
[2560, 248320]
output_hc_down.weight
Q4_K
Q4_K
[10240, 320]
output_hc_norm.weight
F32
F32
[10240]
output_hc_up.weight
Q5_0
Q5_0
[320, 10240]
per_layer_token_embd.weight
Q5_0
Q5_0
[160, 320001536]