Models
Docs
Pricing
Sign in
Download
Models
Download
Docs
Pricing
Sign in
bazobehram
/
kimi-2.5-flash
:latest
203
Downloads
Updated
1 month ago
Cancel
kimi-2.5-flash:latest
...
/
model
8dffc61e5ae8 · 30GB
Metadata
general.architecture
kimi-linear
kimi-linear
general.file_type
Q4_K_M
Q4_K_M
kimi-linear.attention.head_count
32
32
kimi-linear.attention.head_count_kv
32
32
kimi-linear.attention.key_length_mla
192
192
kimi-linear.attention.kv_lora_rank
512
512
kimi-linear.attention.layer_norm_rms_epsilon
1e-05
1e-05
kimi-linear.attention.value_length_mla
128
128
kimi-linear.block_count
27
27
kimi-linear.context_length
1048576
1048576
kimi-linear.embedding_length
2304
2304
kimi-linear.expert_count
256
256
kimi-linear.expert_feed_forward_length
1024
1024
kimi-linear.expert_gating_func
2
2
kimi-linear.expert_shared_count
1
1
kimi-linear.expert_used_count
8
8
kimi-linear.expert_weights_scale
2.446
2.446
kimi-linear.feed_forward_length
9216
9216
kimi-linear.leading_dense_block_count
1
1
kimi-linear.rope.dimension_count
64
64
kimi-linear.rope.freq_base
10000
10000
kimi-linear.ssm.conv_kernel
4
4
kimi-linear.vocab_size
163840
163840
tokenizer.ggml.bos_token_id
163584
163584
tokenizer.ggml.eos_token_id
163586
163586
tokenizer.ggml.model
gpt2
gpt2
tokenizer.ggml.padding_token_id
163839
163839
tokenizer.ggml.pre
kimi-k2
kimi-k2
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
[2304, 163840]
blk.0
blk.0.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.0.attn_norm.weight
F32
F32
[2304]
blk.0.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.0.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.0.attn_v.weight
Q6_K
Q6_K
[2304, 4096]
blk.0.ffn_down.weight
Q6_K
Q6_K
[9216, 2304]
blk.0.ffn_gate.weight
Q4_K
Q4_K
[2304, 9216]
blk.0.ffn_norm.weight
F32
F32
[2304]
blk.0.ffn_up.weight
Q4_K
Q4_K
[2304, 9216]
blk.0.ssm_a
F32
F32
[1, 32]
blk.0.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.0.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.0.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.0.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.0.ssm_dt
F32
F32
[4096]
blk.0.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.0.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.0.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.0.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.0.ssm_norm.weight
F32
F32
[128]
blk.1
blk.1.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.1.attn_norm.weight
F32
F32
[2304]
blk.1.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.1.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.1.attn_v.weight
Q6_K
Q6_K
[2304, 4096]
blk.1.exp_probs_b.weight
F32
F32
[256]
blk.1.ffn_down_exps.weight
Q6_K
Q6_K
[1024, 2304, 256]
blk.1.ffn_down_shexp.weight
Q6_K
Q6_K
[1024, 2304]
blk.1.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.1.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.1.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.1.ffn_norm.weight
F32
F32
[2304]
blk.1.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.1.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.1.ssm_a
F32
F32
[1, 32]
blk.1.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.1.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.1.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.1.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.1.ssm_dt
F32
F32
[4096]
blk.1.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.1.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.1.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.1.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.1.ssm_norm.weight
F32
F32
[128]
blk.2
blk.2.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.2.attn_norm.weight
F32
F32
[2304]
blk.2.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.2.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.2.attn_v.weight
Q6_K
Q6_K
[2304, 4096]
blk.2.exp_probs_b.weight
F32
F32
[256]
blk.2.ffn_down_exps.weight
Q6_K
Q6_K
[1024, 2304, 256]
blk.2.ffn_down_shexp.weight
Q6_K
Q6_K
[1024, 2304]
blk.2.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.2.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.2.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.2.ffn_norm.weight
F32
F32
[2304]
blk.2.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.2.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.2.ssm_a
F32
F32
[1, 32]
blk.2.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.2.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.2.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.2.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.2.ssm_dt
F32
F32
[4096]
blk.2.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.2.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.2.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.2.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.2.ssm_norm.weight
F32
F32
[128]
blk.3
blk.3.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2304, 576]
blk.3.attn_kv_a_norm.weight
F32
F32
[512]
blk.3.attn_kv_b.weight
Q4_K
Q4_K
[512, 8192]
blk.3.attn_norm.weight
F32
F32
[2304]
blk.3.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.3.attn_q.weight
Q4_K
Q4_K
[2304, 6144]
blk.3.exp_probs_b.weight
F32
F32
[256]
blk.3.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.3.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.3.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.3.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.3.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.3.ffn_norm.weight
F32
F32
[2304]
blk.3.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.3.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.4
blk.4.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.4.attn_norm.weight
F32
F32
[2304]
blk.4.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.4.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.4.attn_v.weight
Q4_K
Q4_K
[2304, 4096]
blk.4.exp_probs_b.weight
F32
F32
[256]
blk.4.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.4.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.4.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.4.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.4.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.4.ffn_norm.weight
F32
F32
[2304]
blk.4.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.4.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.4.ssm_a
F32
F32
[1, 32]
blk.4.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.4.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.4.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.4.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.4.ssm_dt
F32
F32
[4096]
blk.4.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.4.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.4.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.4.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.4.ssm_norm.weight
F32
F32
[128]
blk.5
blk.5.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.5.attn_norm.weight
F32
F32
[2304]
blk.5.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.5.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.5.attn_v.weight
Q4_K
Q4_K
[2304, 4096]
blk.5.exp_probs_b.weight
F32
F32
[256]
blk.5.ffn_down_exps.weight
Q6_K
Q6_K
[1024, 2304, 256]
blk.5.ffn_down_shexp.weight
Q6_K
Q6_K
[1024, 2304]
blk.5.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.5.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.5.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.5.ffn_norm.weight
F32
F32
[2304]
blk.5.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.5.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.5.ssm_a
F32
F32
[1, 32]
blk.5.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.5.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.5.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.5.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.5.ssm_dt
F32
F32
[4096]
blk.5.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.5.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.5.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.5.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.5.ssm_norm.weight
F32
F32
[128]
blk.6
blk.6.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.6.attn_norm.weight
F32
F32
[2304]
blk.6.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.6.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.6.attn_v.weight
Q6_K
Q6_K
[2304, 4096]
blk.6.exp_probs_b.weight
F32
F32
[256]
blk.6.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.6.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.6.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.6.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.6.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.6.ffn_norm.weight
F32
F32
[2304]
blk.6.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.6.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.6.ssm_a
F32
F32
[1, 32]
blk.6.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.6.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.6.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.6.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.6.ssm_dt
F32
F32
[4096]
blk.6.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.6.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.6.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.6.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.6.ssm_norm.weight
F32
F32
[128]
blk.7
blk.7.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2304, 576]
blk.7.attn_kv_a_norm.weight
F32
F32
[512]
blk.7.attn_kv_b.weight
Q4_K
Q4_K
[512, 8192]
blk.7.attn_norm.weight
F32
F32
[2304]
blk.7.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.7.attn_q.weight
Q4_K
Q4_K
[2304, 6144]
blk.7.exp_probs_b.weight
F32
F32
[256]
blk.7.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.7.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.7.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.7.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.7.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.7.ffn_norm.weight
F32
F32
[2304]
blk.7.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.7.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.8
blk.8.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.8.attn_norm.weight
F32
F32
[2304]
blk.8.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.8.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.8.attn_v.weight
Q4_K
Q4_K
[2304, 4096]
blk.8.exp_probs_b.weight
F32
F32
[256]
blk.8.ffn_down_exps.weight
Q6_K
Q6_K
[1024, 2304, 256]
blk.8.ffn_down_shexp.weight
Q6_K
Q6_K
[1024, 2304]
blk.8.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.8.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.8.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.8.ffn_norm.weight
F32
F32
[2304]
blk.8.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.8.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.8.ssm_a
F32
F32
[1, 32]
blk.8.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.8.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.8.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.8.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.8.ssm_dt
F32
F32
[4096]
blk.8.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.8.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.8.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.8.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.8.ssm_norm.weight
F32
F32
[128]
blk.9
blk.9.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.9.attn_norm.weight
F32
F32
[2304]
blk.9.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.9.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.9.attn_v.weight
Q4_K
Q4_K
[2304, 4096]
blk.9.exp_probs_b.weight
F32
F32
[256]
blk.9.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.9.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.9.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.9.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.9.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.9.ffn_norm.weight
F32
F32
[2304]
blk.9.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.9.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.9.ssm_a
F32
F32
[1, 32]
blk.9.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.9.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.9.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.9.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.9.ssm_dt
F32
F32
[4096]
blk.9.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.9.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.9.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.9.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.9.ssm_norm.weight
F32
F32
[128]
blk.10
blk.10.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.10.attn_norm.weight
F32
F32
[2304]
blk.10.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.10.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.10.attn_v.weight
Q6_K
Q6_K
[2304, 4096]
blk.10.exp_probs_b.weight
F32
F32
[256]
blk.10.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.10.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.10.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.10.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.10.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.10.ffn_norm.weight
F32
F32
[2304]
blk.10.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.10.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.10.ssm_a
F32
F32
[1, 32]
blk.10.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.10.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.10.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.10.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.10.ssm_dt
F32
F32
[4096]
blk.10.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.10.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.10.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.10.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.10.ssm_norm.weight
F32
F32
[128]
blk.11
blk.11.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2304, 576]
blk.11.attn_kv_a_norm.weight
F32
F32
[512]
blk.11.attn_kv_b.weight
Q4_K
Q4_K
[512, 8192]
blk.11.attn_norm.weight
F32
F32
[2304]
blk.11.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.11.attn_q.weight
Q4_K
Q4_K
[2304, 6144]
blk.11.exp_probs_b.weight
F32
F32
[256]
blk.11.ffn_down_exps.weight
Q6_K
Q6_K
[1024, 2304, 256]
blk.11.ffn_down_shexp.weight
Q6_K
Q6_K
[1024, 2304]
blk.11.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.11.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.11.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.11.ffn_norm.weight
F32
F32
[2304]
blk.11.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.11.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.12
blk.12.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.12.attn_norm.weight
F32
F32
[2304]
blk.12.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.12.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.12.attn_v.weight
Q4_K
Q4_K
[2304, 4096]
blk.12.exp_probs_b.weight
F32
F32
[256]
blk.12.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.12.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.12.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.12.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.12.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.12.ffn_norm.weight
F32
F32
[2304]
blk.12.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.12.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.12.ssm_a
F32
F32
[1, 32]
blk.12.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.12.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.12.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.12.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.12.ssm_dt
F32
F32
[4096]
blk.12.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.12.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.12.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.12.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.12.ssm_norm.weight
F32
F32
[128]
blk.13
blk.13.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.13.attn_norm.weight
F32
F32
[2304]
blk.13.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.13.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.13.attn_v.weight
Q4_K
Q4_K
[2304, 4096]
blk.13.exp_probs_b.weight
F32
F32
[256]
blk.13.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.13.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.13.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.13.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.13.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.13.ffn_norm.weight
F32
F32
[2304]
blk.13.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.13.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.13.ssm_a
F32
F32
[1, 32]
blk.13.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.13.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.13.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.13.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.13.ssm_dt
F32
F32
[4096]
blk.13.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.13.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.13.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.13.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.13.ssm_norm.weight
F32
F32
[128]
blk.14
blk.14.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.14.attn_norm.weight
F32
F32
[2304]
blk.14.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.14.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.14.attn_v.weight
Q6_K
Q6_K
[2304, 4096]
blk.14.exp_probs_b.weight
F32
F32
[256]
blk.14.ffn_down_exps.weight
Q6_K
Q6_K
[1024, 2304, 256]
blk.14.ffn_down_shexp.weight
Q6_K
Q6_K
[1024, 2304]
blk.14.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.14.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.14.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.14.ffn_norm.weight
F32
F32
[2304]
blk.14.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.14.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.14.ssm_a
F32
F32
[1, 32]
blk.14.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.14.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.14.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.14.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.14.ssm_dt
F32
F32
[4096]
blk.14.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.14.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.14.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.14.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.14.ssm_norm.weight
F32
F32
[128]
blk.15
blk.15.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2304, 576]
blk.15.attn_kv_a_norm.weight
F32
F32
[512]
blk.15.attn_kv_b.weight
Q4_K
Q4_K
[512, 8192]
blk.15.attn_norm.weight
F32
F32
[2304]
blk.15.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.15.attn_q.weight
Q4_K
Q4_K
[2304, 6144]
blk.15.exp_probs_b.weight
F32
F32
[256]
blk.15.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.15.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.15.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.15.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.15.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.15.ffn_norm.weight
F32
F32
[2304]
blk.15.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.15.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.16
blk.16.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.16.attn_norm.weight
F32
F32
[2304]
blk.16.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.16.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.16.attn_v.weight
Q4_K
Q4_K
[2304, 4096]
blk.16.exp_probs_b.weight
F32
F32
[256]
blk.16.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.16.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.16.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.16.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.16.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.16.ffn_norm.weight
F32
F32
[2304]
blk.16.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.16.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.16.ssm_a
F32
F32
[1, 32]
blk.16.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.16.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.16.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.16.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.16.ssm_dt
F32
F32
[4096]
blk.16.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.16.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.16.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.16.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.16.ssm_norm.weight
F32
F32
[128]
blk.17
blk.17.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.17.attn_norm.weight
F32
F32
[2304]
blk.17.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.17.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.17.attn_v.weight
Q4_K
Q4_K
[2304, 4096]
blk.17.exp_probs_b.weight
F32
F32
[256]
blk.17.ffn_down_exps.weight
Q6_K
Q6_K
[1024, 2304, 256]
blk.17.ffn_down_shexp.weight
Q6_K
Q6_K
[1024, 2304]
blk.17.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.17.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.17.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.17.ffn_norm.weight
F32
F32
[2304]
blk.17.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.17.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.17.ssm_a
F32
F32
[1, 32]
blk.17.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.17.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.17.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.17.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.17.ssm_dt
F32
F32
[4096]
blk.17.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.17.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.17.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.17.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.17.ssm_norm.weight
F32
F32
[128]
blk.18
blk.18.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.18.attn_norm.weight
F32
F32
[2304]
blk.18.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.18.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.18.attn_v.weight
Q6_K
Q6_K
[2304, 4096]
blk.18.exp_probs_b.weight
F32
F32
[256]
blk.18.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.18.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.18.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.18.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.18.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.18.ffn_norm.weight
F32
F32
[2304]
blk.18.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.18.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.18.ssm_a
F32
F32
[1, 32]
blk.18.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.18.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.18.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.18.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.18.ssm_dt
F32
F32
[4096]
blk.18.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.18.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.18.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.18.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.18.ssm_norm.weight
F32
F32
[128]
blk.19
blk.19.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2304, 576]
blk.19.attn_kv_a_norm.weight
F32
F32
[512]
blk.19.attn_kv_b.weight
Q4_K
Q4_K
[512, 8192]
blk.19.attn_norm.weight
F32
F32
[2304]
blk.19.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.19.attn_q.weight
Q4_K
Q4_K
[2304, 6144]
blk.19.exp_probs_b.weight
F32
F32
[256]
blk.19.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.19.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.19.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.19.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.19.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.19.ffn_norm.weight
F32
F32
[2304]
blk.19.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.19.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.20
blk.20.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.20.attn_norm.weight
F32
F32
[2304]
blk.20.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.20.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.20.attn_v.weight
Q4_K
Q4_K
[2304, 4096]
blk.20.exp_probs_b.weight
F32
F32
[256]
blk.20.ffn_down_exps.weight
Q6_K
Q6_K
[1024, 2304, 256]
blk.20.ffn_down_shexp.weight
Q6_K
Q6_K
[1024, 2304]
blk.20.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.20.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.20.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.20.ffn_norm.weight
F32
F32
[2304]
blk.20.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.20.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.20.ssm_a
F32
F32
[1, 32]
blk.20.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.20.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.20.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.20.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.20.ssm_dt
F32
F32
[4096]
blk.20.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.20.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.20.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.20.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.20.ssm_norm.weight
F32
F32
[128]
blk.21
blk.21.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.21.attn_norm.weight
F32
F32
[2304]
blk.21.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.21.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.21.attn_v.weight
Q4_K
Q4_K
[2304, 4096]
blk.21.exp_probs_b.weight
F32
F32
[256]
blk.21.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.21.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.21.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.21.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.21.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.21.ffn_norm.weight
F32
F32
[2304]
blk.21.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.21.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.21.ssm_a
F32
F32
[1, 32]
blk.21.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.21.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.21.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.21.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.21.ssm_dt
F32
F32
[4096]
blk.21.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.21.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.21.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.21.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.21.ssm_norm.weight
F32
F32
[128]
blk.22
blk.22.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.22.attn_norm.weight
F32
F32
[2304]
blk.22.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.22.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.22.attn_v.weight
Q6_K
Q6_K
[2304, 4096]
blk.22.exp_probs_b.weight
F32
F32
[256]
blk.22.ffn_down_exps.weight
Q4_K
Q4_K
[1024, 2304, 256]
blk.22.ffn_down_shexp.weight
Q4_K
Q4_K
[1024, 2304]
blk.22.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.22.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.22.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.22.ffn_norm.weight
F32
F32
[2304]
blk.22.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.22.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.22.ssm_a
F32
F32
[1, 32]
blk.22.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.22.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.22.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.22.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.22.ssm_dt
F32
F32
[4096]
blk.22.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.22.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.22.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.22.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.22.ssm_norm.weight
F32
F32
[128]
blk.23
blk.23.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2304, 576]
blk.23.attn_kv_a_norm.weight
F32
F32
[512]
blk.23.attn_kv_b.weight
Q4_K
Q4_K
[512, 8192]
blk.23.attn_norm.weight
F32
F32
[2304]
blk.23.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.23.attn_q.weight
Q4_K
Q4_K
[2304, 6144]
blk.23.exp_probs_b.weight
F32
F32
[256]
blk.23.ffn_down_exps.weight
Q6_K
Q6_K
[1024, 2304, 256]
blk.23.ffn_down_shexp.weight
Q6_K
Q6_K
[1024, 2304]
blk.23.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.23.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.23.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.23.ffn_norm.weight
F32
F32
[2304]
blk.23.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.23.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.24
blk.24.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.24.attn_norm.weight
F32
F32
[2304]
blk.24.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.24.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.24.attn_v.weight
Q4_K
Q4_K
[2304, 4096]
blk.24.exp_probs_b.weight
F32
F32
[256]
blk.24.ffn_down_exps.weight
Q6_K
Q6_K
[1024, 2304, 256]
blk.24.ffn_down_shexp.weight
Q6_K
Q6_K
[1024, 2304]
blk.24.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.24.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.24.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.24.ffn_norm.weight
F32
F32
[2304]
blk.24.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.24.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.24.ssm_a
F32
F32
[1, 32]
blk.24.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.24.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.24.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.24.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.24.ssm_dt
F32
F32
[4096]
blk.24.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.24.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.24.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.24.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.24.ssm_norm.weight
F32
F32
[128]
blk.25
blk.25.attn_k.weight
Q4_K
Q4_K
[2304, 4096]
blk.25.attn_norm.weight
F32
F32
[2304]
blk.25.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.25.attn_q.weight
Q4_K
Q4_K
[2304, 4096]
blk.25.attn_v.weight
Q4_K
Q4_K
[2304, 4096]
blk.25.exp_probs_b.weight
F32
F32
[256]
blk.25.ffn_down_exps.weight
Q6_K
Q6_K
[1024, 2304, 256]
blk.25.ffn_down_shexp.weight
Q6_K
Q6_K
[1024, 2304]
blk.25.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.25.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.25.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.25.ffn_norm.weight
F32
F32
[2304]
blk.25.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.25.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.25.ssm_a
F32
F32
[1, 32]
blk.25.ssm_beta.weight
Q4_K
Q4_K
[2304, 32]
blk.25.ssm_conv1d_k.weight
F16
F16
[4, 1, 4096]
blk.25.ssm_conv1d_q.weight
F16
F16
[4, 1, 4096]
blk.25.ssm_conv1d_v.weight
F16
F16
[4, 1, 4096]
blk.25.ssm_dt
F32
F32
[4096]
blk.25.ssm_f_a.weight
Q4_K
Q4_K
[2304, 128]
blk.25.ssm_f_b.weight
Q5_0
Q5_0
[128, 4096]
blk.25.ssm_g_a.weight
Q4_K
Q4_K
[2304, 128]
blk.25.ssm_g_b.weight
Q5_0
Q5_0
[128, 4096]
blk.25.ssm_norm.weight
F32
F32
[128]
blk.26
blk.26.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2304, 576]
blk.26.attn_kv_a_norm.weight
F32
F32
[512]
blk.26.attn_kv_b.weight
Q4_K
Q4_K
[512, 8192]
blk.26.attn_norm.weight
F32
F32
[2304]
blk.26.attn_output.weight
Q4_K
Q4_K
[4096, 2304]
blk.26.attn_q.weight
Q4_K
Q4_K
[2304, 6144]
blk.26.exp_probs_b.weight
F32
F32
[256]
blk.26.ffn_down_exps.weight
Q6_K
Q6_K
[1024, 2304, 256]
blk.26.ffn_down_shexp.weight
Q6_K
Q6_K
[1024, 2304]
blk.26.ffn_gate_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.26.ffn_gate_inp.weight
F32
F32
[2304, 256]
blk.26.ffn_gate_shexp.weight
Q4_K
Q4_K
[2304, 1024]
blk.26.ffn_norm.weight
F32
F32
[2304]
blk.26.ffn_up_exps.weight
Q4_K
Q4_K
[2304, 1024, 256]
blk.26.ffn_up_shexp.weight
Q4_K
Q4_K
[2304, 1024]
output.weight
Q6_K
Q6_K
[2304, 163840]
output_norm.weight
F32
F32
[2304]