Models
Docs
Pricing
Sign in
Download
Models
Download
Docs
Pricing
Sign in
frob
/
ling-3.0-flash
:124b
1
Download
Updated
yesterday
Cancel
124b
ling-3.0-flash:124b
...
/
model
27e502591e58 · 77GB
Metadata
general.architecture
bailingmoe3
bailingmoe3
general.file_type
Q4_K_M
Q4_K_M
bailingmoe3.attention.head_count
32
32
bailingmoe3.attention.head_count_kv
[0, 0, 0, 0, 0, ...]
[0, 0, 0, 0, 0, ...]
bailingmoe3.attention.key_length
576
576
bailingmoe3.attention.key_length_mla
192
192
bailingmoe3.attention.kv_lora_rank
512
512
bailingmoe3.attention.layer_norm_rms_epsilon
1e-06
1e-06
bailingmoe3.attention.value_length
128
128
bailingmoe3.attention.value_length_mla
128
128
bailingmoe3.block_count
43
43
bailingmoe3.context_length
262144
262144
bailingmoe3.embedding_length
2560
2560
bailingmoe3.expert_count
512
512
bailingmoe3.expert_feed_forward_length
768
768
bailingmoe3.expert_gating_func
2
2
bailingmoe3.expert_group_count
8
8
bailingmoe3.expert_group_used_count
4
4
bailingmoe3.expert_shared_count
1
1
bailingmoe3.expert_shared_feed_forward_length
768
768
bailingmoe3.expert_used_count
8
8
bailingmoe3.expert_weights_norm
true
true
bailingmoe3.expert_weights_scale
2.5
2.5
bailingmoe3.feed_forward_length
6144
6144
bailingmoe3.kda.gate_lower_bound
-5
-5
bailingmoe3.kda.head_dim
128
128
bailingmoe3.kda.safe_gate
true
true
bailingmoe3.leading_dense_block_count
2
2
bailingmoe3.nextn_predict_layers
1
1
bailingmoe3.rope.dimension_count
64
64
bailingmoe3.rope.freq_base
6e+06
6e+06
bailingmoe3.ssm.conv_kernel
4
4
bailingmoe3.swiglu_clamp_exp
[0, 0, 0, 0, 0, ...]
[0, 0, 0, 0, 0, ...]
bailingmoe3.swiglu_clamp_shexp
[0, 0, 0, 0, 0, ...]
[0, 0, 0, 0, 0, ...]
bailingmoe3.vocab_size
157184
157184
tokenizer.ggml.add_bos_token
false
false
tokenizer.ggml.add_eos_token
false
false
tokenizer.ggml.bos_token_id
156891
156891
tokenizer.ggml.eos_token_id
156895
156895
tokenizer.ggml.merges
[Ġ Ġ, Ġ t, i n, Ġ a, h e, ...]
[Ġ Ġ, Ġ t, i n, Ġ a, h e, ...]
tokenizer.ggml.model
gpt2
gpt2
tokenizer.ggml.padding_token_id
156892
156892
tokenizer.ggml.pre
bailingmoe2
bailingmoe2
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, 157184]
blk.0
blk.0.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.0.attn_norm.weight
F32
F32
[2560]
blk.0.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.0.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.0.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.0.ffn_down.weight
Q6_K
Q6_K
[6144, 2560]
blk.0.ffn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.0.ffn_norm.weight
F32
F32
[2560]
blk.0.ffn_up.weight
Q4_K
Q4_K
[2560, 6144]
blk.0.ssm_a
F32
F32
[1, 32]
blk.0.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.0.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.0.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.0.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.0.ssm_dt.bias
F32
F32
[4096]
blk.0.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.0.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.0.ssm_norm.weight
F32
F32
[128]
blk.1
blk.1.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.1.attn_norm.weight
F32
F32
[2560]
blk.1.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.1.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.1.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.1.ffn_down.weight
Q6_K
Q6_K
[6144, 2560]
blk.1.ffn_gate.weight
Q4_K
Q4_K
[2560, 6144]
blk.1.ffn_norm.weight
F32
F32
[2560]
blk.1.ffn_up.weight
Q4_K
Q4_K
[2560, 6144]
blk.1.ssm_a
F32
F32
[1, 32]
blk.1.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.1.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.1.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.1.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.1.ssm_dt.bias
F32
F32
[4096]
blk.1.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.1.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.1.ssm_norm.weight
F32
F32
[128]
blk.2
blk.2.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.2.attn_norm.weight
F32
F32
[2560]
blk.2.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.2.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.2.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.2.exp_probs_b.bias
F32
F32
[512]
blk.2.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.2.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.2.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.2.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.2.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.2.ffn_norm.weight
F32
F32
[2560]
blk.2.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.2.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.2.ssm_a
F32
F32
[1, 32]
blk.2.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.2.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.2.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.2.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.2.ssm_dt.bias
F32
F32
[4096]
blk.2.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.2.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.2.ssm_norm.weight
F32
F32
[128]
blk.3
blk.3.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.3.attn_norm.weight
F32
F32
[2560]
blk.3.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.3.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.3.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.3.exp_probs_b.bias
F32
F32
[512]
blk.3.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.3.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.3.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.3.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.3.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.3.ffn_norm.weight
F32
F32
[2560]
blk.3.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.3.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.3.ssm_a
F32
F32
[1, 32]
blk.3.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.3.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.3.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.3.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.3.ssm_dt.bias
F32
F32
[4096]
blk.3.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.3.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.3.ssm_norm.weight
F32
F32
[128]
blk.4
blk.4.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.4.attn_norm.weight
F32
F32
[2560]
blk.4.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.4.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.4.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.4.exp_probs_b.bias
F32
F32
[512]
blk.4.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.4.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.4.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.4.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.4.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.4.ffn_norm.weight
F32
F32
[2560]
blk.4.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.4.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.4.ssm_a
F32
F32
[1, 32]
blk.4.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.4.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.4.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.4.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.4.ssm_dt.bias
F32
F32
[4096]
blk.4.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.4.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.4.ssm_norm.weight
F32
F32
[128]
blk.5
blk.5.attn_gate.weight
Q4_K
Q4_K
[2560, 32]
blk.5.attn_k_b.weight
Q5_0
Q5_0
[128, 512, 32]
blk.5.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2560, 576]
blk.5.attn_kv_a_norm.weight
F32
F32
[512]
blk.5.attn_norm.weight
F32
F32
[2560]
blk.5.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.5.attn_q.weight
Q4_K
Q4_K
[2560, 6144]
blk.5.attn_v_b.weight
Q4_K
Q4_K
[512, 128, 32]
blk.5.exp_probs_b.bias
F32
F32
[512]
blk.5.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.5.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.5.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.5.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.5.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.5.ffn_norm.weight
F32
F32
[2560]
blk.5.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.5.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.6
blk.6.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.6.attn_norm.weight
F32
F32
[2560]
blk.6.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.6.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.6.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.6.exp_probs_b.bias
F32
F32
[512]
blk.6.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.6.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.6.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.6.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.6.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.6.ffn_norm.weight
F32
F32
[2560]
blk.6.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.6.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.6.ssm_a
F32
F32
[1, 32]
blk.6.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.6.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.6.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.6.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.6.ssm_dt.bias
F32
F32
[4096]
blk.6.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.6.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.6.ssm_norm.weight
F32
F32
[128]
blk.7
blk.7.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.7.attn_norm.weight
F32
F32
[2560]
blk.7.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.7.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.7.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.7.exp_probs_b.bias
F32
F32
[512]
blk.7.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.7.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.7.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.7.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.7.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.7.ffn_norm.weight
F32
F32
[2560]
blk.7.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.7.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.7.ssm_a
F32
F32
[1, 32]
blk.7.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.7.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.7.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.7.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.7.ssm_dt.bias
F32
F32
[4096]
blk.7.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.7.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.7.ssm_norm.weight
F32
F32
[128]
blk.8
blk.8.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.8.attn_norm.weight
F32
F32
[2560]
blk.8.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.8.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.8.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.8.exp_probs_b.bias
F32
F32
[512]
blk.8.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.8.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.8.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.8.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.8.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.8.ffn_norm.weight
F32
F32
[2560]
blk.8.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.8.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.8.ssm_a
F32
F32
[1, 32]
blk.8.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.8.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.8.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.8.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.8.ssm_dt.bias
F32
F32
[4096]
blk.8.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.8.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.8.ssm_norm.weight
F32
F32
[128]
blk.9
blk.9.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.9.attn_norm.weight
F32
F32
[2560]
blk.9.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.9.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.9.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.9.exp_probs_b.bias
F32
F32
[512]
blk.9.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.9.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.9.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.9.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.9.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.9.ffn_norm.weight
F32
F32
[2560]
blk.9.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.9.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.9.ssm_a
F32
F32
[1, 32]
blk.9.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.9.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.9.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.9.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.9.ssm_dt.bias
F32
F32
[4096]
blk.9.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.9.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.9.ssm_norm.weight
F32
F32
[128]
blk.10
blk.10.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.10.attn_norm.weight
F32
F32
[2560]
blk.10.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.10.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.10.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.10.exp_probs_b.bias
F32
F32
[512]
blk.10.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.10.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.10.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.10.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.10.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.10.ffn_norm.weight
F32
F32
[2560]
blk.10.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.10.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.10.ssm_a
F32
F32
[1, 32]
blk.10.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.10.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.10.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.10.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.10.ssm_dt.bias
F32
F32
[4096]
blk.10.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.10.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.10.ssm_norm.weight
F32
F32
[128]
blk.11
blk.11.attn_gate.weight
Q4_K
Q4_K
[2560, 32]
blk.11.attn_k_b.weight
Q5_0
Q5_0
[128, 512, 32]
blk.11.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2560, 576]
blk.11.attn_kv_a_norm.weight
F32
F32
[512]
blk.11.attn_norm.weight
F32
F32
[2560]
blk.11.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.11.attn_q.weight
Q4_K
Q4_K
[2560, 6144]
blk.11.attn_v_b.weight
Q4_K
Q4_K
[512, 128, 32]
blk.11.exp_probs_b.bias
F32
F32
[512]
blk.11.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.11.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.11.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.11.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.11.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.11.ffn_norm.weight
F32
F32
[2560]
blk.11.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.11.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.12
blk.12.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.12.attn_norm.weight
F32
F32
[2560]
blk.12.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.12.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.12.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.12.exp_probs_b.bias
F32
F32
[512]
blk.12.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.12.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.12.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.12.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.12.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.12.ffn_norm.weight
F32
F32
[2560]
blk.12.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.12.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.12.ssm_a
F32
F32
[1, 32]
blk.12.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.12.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.12.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.12.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.12.ssm_dt.bias
F32
F32
[4096]
blk.12.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.12.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.12.ssm_norm.weight
F32
F32
[128]
blk.13
blk.13.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.13.attn_norm.weight
F32
F32
[2560]
blk.13.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.13.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.13.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.13.exp_probs_b.bias
F32
F32
[512]
blk.13.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.13.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.13.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.13.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.13.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.13.ffn_norm.weight
F32
F32
[2560]
blk.13.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.13.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.13.ssm_a
F32
F32
[1, 32]
blk.13.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.13.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.13.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.13.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.13.ssm_dt.bias
F32
F32
[4096]
blk.13.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.13.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.13.ssm_norm.weight
F32
F32
[128]
blk.14
blk.14.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.14.attn_norm.weight
F32
F32
[2560]
blk.14.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.14.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.14.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.14.exp_probs_b.bias
F32
F32
[512]
blk.14.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.14.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.14.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.14.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.14.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.14.ffn_norm.weight
F32
F32
[2560]
blk.14.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.14.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.14.ssm_a
F32
F32
[1, 32]
blk.14.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.14.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.14.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.14.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.14.ssm_dt.bias
F32
F32
[4096]
blk.14.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.14.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.14.ssm_norm.weight
F32
F32
[128]
blk.15
blk.15.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.15.attn_norm.weight
F32
F32
[2560]
blk.15.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.15.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.15.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.15.exp_probs_b.bias
F32
F32
[512]
blk.15.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.15.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.15.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.15.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.15.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.15.ffn_norm.weight
F32
F32
[2560]
blk.15.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.15.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.15.ssm_a
F32
F32
[1, 32]
blk.15.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.15.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.15.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.15.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.15.ssm_dt.bias
F32
F32
[4096]
blk.15.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.15.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.15.ssm_norm.weight
F32
F32
[128]
blk.16
blk.16.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.16.attn_norm.weight
F32
F32
[2560]
blk.16.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.16.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.16.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.16.exp_probs_b.bias
F32
F32
[512]
blk.16.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.16.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.16.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.16.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.16.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.16.ffn_norm.weight
F32
F32
[2560]
blk.16.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.16.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.16.ssm_a
F32
F32
[1, 32]
blk.16.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.16.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.16.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.16.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.16.ssm_dt.bias
F32
F32
[4096]
blk.16.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.16.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.16.ssm_norm.weight
F32
F32
[128]
blk.17
blk.17.attn_gate.weight
Q4_K
Q4_K
[2560, 32]
blk.17.attn_k_b.weight
Q5_0
Q5_0
[128, 512, 32]
blk.17.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2560, 576]
blk.17.attn_kv_a_norm.weight
F32
F32
[512]
blk.17.attn_norm.weight
F32
F32
[2560]
blk.17.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.17.attn_q.weight
Q4_K
Q4_K
[2560, 6144]
blk.17.attn_v_b.weight
Q4_K
Q4_K
[512, 128, 32]
blk.17.exp_probs_b.bias
F32
F32
[512]
blk.17.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.17.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.17.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.17.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.17.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.17.ffn_norm.weight
F32
F32
[2560]
blk.17.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.17.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.18
blk.18.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.18.attn_norm.weight
F32
F32
[2560]
blk.18.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.18.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.18.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.18.exp_probs_b.bias
F32
F32
[512]
blk.18.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.18.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.18.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.18.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.18.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.18.ffn_norm.weight
F32
F32
[2560]
blk.18.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.18.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.18.ssm_a
F32
F32
[1, 32]
blk.18.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.18.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.18.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.18.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.18.ssm_dt.bias
F32
F32
[4096]
blk.18.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.18.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.18.ssm_norm.weight
F32
F32
[128]
blk.19
blk.19.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.19.attn_norm.weight
F32
F32
[2560]
blk.19.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.19.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.19.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.19.exp_probs_b.bias
F32
F32
[512]
blk.19.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.19.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.19.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.19.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.19.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.19.ffn_norm.weight
F32
F32
[2560]
blk.19.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.19.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.19.ssm_a
F32
F32
[1, 32]
blk.19.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.19.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.19.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.19.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.19.ssm_dt.bias
F32
F32
[4096]
blk.19.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.19.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.19.ssm_norm.weight
F32
F32
[128]
blk.20
blk.20.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.20.attn_norm.weight
F32
F32
[2560]
blk.20.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.20.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.20.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.20.exp_probs_b.bias
F32
F32
[512]
blk.20.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.20.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.20.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.20.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.20.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.20.ffn_norm.weight
F32
F32
[2560]
blk.20.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.20.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.20.ssm_a
F32
F32
[1, 32]
blk.20.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.20.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.20.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.20.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.20.ssm_dt.bias
F32
F32
[4096]
blk.20.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.20.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.20.ssm_norm.weight
F32
F32
[128]
blk.21
blk.21.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.21.attn_norm.weight
F32
F32
[2560]
blk.21.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.21.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.21.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.21.exp_probs_b.bias
F32
F32
[512]
blk.21.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.21.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.21.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.21.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.21.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.21.ffn_norm.weight
F32
F32
[2560]
blk.21.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.21.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.21.ssm_a
F32
F32
[1, 32]
blk.21.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.21.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.21.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.21.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.21.ssm_dt.bias
F32
F32
[4096]
blk.21.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.21.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.21.ssm_norm.weight
F32
F32
[128]
blk.22
blk.22.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.22.attn_norm.weight
F32
F32
[2560]
blk.22.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.22.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.22.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.22.exp_probs_b.bias
F32
F32
[512]
blk.22.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.22.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.22.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.22.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.22.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.22.ffn_norm.weight
F32
F32
[2560]
blk.22.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.22.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.22.ssm_a
F32
F32
[1, 32]
blk.22.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.22.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.22.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.22.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.22.ssm_dt.bias
F32
F32
[4096]
blk.22.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.22.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.22.ssm_norm.weight
F32
F32
[128]
blk.23
blk.23.attn_gate.weight
Q4_K
Q4_K
[2560, 32]
blk.23.attn_k_b.weight
Q5_0
Q5_0
[128, 512, 32]
blk.23.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2560, 576]
blk.23.attn_kv_a_norm.weight
F32
F32
[512]
blk.23.attn_norm.weight
F32
F32
[2560]
blk.23.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.23.attn_q.weight
Q4_K
Q4_K
[2560, 6144]
blk.23.attn_v_b.weight
Q4_K
Q4_K
[512, 128, 32]
blk.23.exp_probs_b.bias
F32
F32
[512]
blk.23.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.23.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.23.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.23.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.23.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.23.ffn_norm.weight
F32
F32
[2560]
blk.23.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.23.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.24
blk.24.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.24.attn_norm.weight
F32
F32
[2560]
blk.24.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.24.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.24.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.24.exp_probs_b.bias
F32
F32
[512]
blk.24.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.24.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.24.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.24.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.24.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.24.ffn_norm.weight
F32
F32
[2560]
blk.24.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.24.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.24.ssm_a
F32
F32
[1, 32]
blk.24.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.24.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.24.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.24.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.24.ssm_dt.bias
F32
F32
[4096]
blk.24.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.24.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.24.ssm_norm.weight
F32
F32
[128]
blk.25
blk.25.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.25.attn_norm.weight
F32
F32
[2560]
blk.25.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.25.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.25.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.25.exp_probs_b.bias
F32
F32
[512]
blk.25.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.25.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.25.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.25.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.25.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.25.ffn_norm.weight
F32
F32
[2560]
blk.25.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.25.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.25.ssm_a
F32
F32
[1, 32]
blk.25.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.25.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.25.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.25.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.25.ssm_dt.bias
F32
F32
[4096]
blk.25.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.25.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.25.ssm_norm.weight
F32
F32
[128]
blk.26
blk.26.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.26.attn_norm.weight
F32
F32
[2560]
blk.26.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.26.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.26.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.26.exp_probs_b.bias
F32
F32
[512]
blk.26.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.26.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.26.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.26.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.26.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.26.ffn_norm.weight
F32
F32
[2560]
blk.26.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.26.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.26.ssm_a
F32
F32
[1, 32]
blk.26.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.26.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.26.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.26.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.26.ssm_dt.bias
F32
F32
[4096]
blk.26.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.26.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.26.ssm_norm.weight
F32
F32
[128]
blk.27
blk.27.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.27.attn_norm.weight
F32
F32
[2560]
blk.27.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.27.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.27.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.27.exp_probs_b.bias
F32
F32
[512]
blk.27.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.27.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.27.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.27.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.27.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.27.ffn_norm.weight
F32
F32
[2560]
blk.27.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.27.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.27.ssm_a
F32
F32
[1, 32]
blk.27.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.27.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.27.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.27.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.27.ssm_dt.bias
F32
F32
[4096]
blk.27.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.27.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.27.ssm_norm.weight
F32
F32
[128]
blk.28
blk.28.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.28.attn_norm.weight
F32
F32
[2560]
blk.28.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.28.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.28.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.28.exp_probs_b.bias
F32
F32
[512]
blk.28.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.28.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.28.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.28.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.28.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.28.ffn_norm.weight
F32
F32
[2560]
blk.28.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.28.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.28.ssm_a
F32
F32
[1, 32]
blk.28.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.28.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.28.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.28.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.28.ssm_dt.bias
F32
F32
[4096]
blk.28.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.28.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.28.ssm_norm.weight
F32
F32
[128]
blk.29
blk.29.attn_gate.weight
Q4_K
Q4_K
[2560, 32]
blk.29.attn_k_b.weight
Q5_0
Q5_0
[128, 512, 32]
blk.29.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2560, 576]
blk.29.attn_kv_a_norm.weight
F32
F32
[512]
blk.29.attn_norm.weight
F32
F32
[2560]
blk.29.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.29.attn_q.weight
Q4_K
Q4_K
[2560, 6144]
blk.29.attn_v_b.weight
Q4_K
Q4_K
[512, 128, 32]
blk.29.exp_probs_b.bias
F32
F32
[512]
blk.29.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.29.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.29.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.29.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.29.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.29.ffn_norm.weight
F32
F32
[2560]
blk.29.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.29.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.30
blk.30.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.30.attn_norm.weight
F32
F32
[2560]
blk.30.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.30.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.30.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.30.exp_probs_b.bias
F32
F32
[512]
blk.30.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.30.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.30.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.30.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.30.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.30.ffn_norm.weight
F32
F32
[2560]
blk.30.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.30.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.30.ssm_a
F32
F32
[1, 32]
blk.30.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.30.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.30.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.30.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.30.ssm_dt.bias
F32
F32
[4096]
blk.30.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.30.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.30.ssm_norm.weight
F32
F32
[128]
blk.31
blk.31.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.31.attn_norm.weight
F32
F32
[2560]
blk.31.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.31.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.31.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.31.exp_probs_b.bias
F32
F32
[512]
blk.31.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.31.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.31.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.31.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.31.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.31.ffn_norm.weight
F32
F32
[2560]
blk.31.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.31.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.31.ssm_a
F32
F32
[1, 32]
blk.31.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.31.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.31.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.31.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.31.ssm_dt.bias
F32
F32
[4096]
blk.31.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.31.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.31.ssm_norm.weight
F32
F32
[128]
blk.32
blk.32.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.32.attn_norm.weight
F32
F32
[2560]
blk.32.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.32.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.32.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.32.exp_probs_b.bias
F32
F32
[512]
blk.32.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.32.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.32.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.32.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.32.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.32.ffn_norm.weight
F32
F32
[2560]
blk.32.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.32.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.32.ssm_a
F32
F32
[1, 32]
blk.32.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.32.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.32.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.32.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.32.ssm_dt.bias
F32
F32
[4096]
blk.32.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.32.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.32.ssm_norm.weight
F32
F32
[128]
blk.33
blk.33.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.33.attn_norm.weight
F32
F32
[2560]
blk.33.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.33.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.33.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.33.exp_probs_b.bias
F32
F32
[512]
blk.33.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.33.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.33.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.33.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.33.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.33.ffn_norm.weight
F32
F32
[2560]
blk.33.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.33.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.33.ssm_a
F32
F32
[1, 32]
blk.33.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.33.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.33.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.33.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.33.ssm_dt.bias
F32
F32
[4096]
blk.33.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.33.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.33.ssm_norm.weight
F32
F32
[128]
blk.34
blk.34.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.34.attn_norm.weight
F32
F32
[2560]
blk.34.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.34.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.34.attn_v.weight
Q4_K
Q4_K
[2560, 4096]
blk.34.exp_probs_b.bias
F32
F32
[512]
blk.34.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.34.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.34.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.34.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.34.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.34.ffn_norm.weight
F32
F32
[2560]
blk.34.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.34.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.34.ssm_a
F32
F32
[1, 32]
blk.34.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.34.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.34.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.34.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.34.ssm_dt.bias
F32
F32
[4096]
blk.34.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.34.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.34.ssm_norm.weight
F32
F32
[128]
blk.35
blk.35.attn_gate.weight
Q4_K
Q4_K
[2560, 32]
blk.35.attn_k_b.weight
Q5_0
Q5_0
[128, 512, 32]
blk.35.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2560, 576]
blk.35.attn_kv_a_norm.weight
F32
F32
[512]
blk.35.attn_norm.weight
F32
F32
[2560]
blk.35.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.35.attn_q.weight
Q4_K
Q4_K
[2560, 6144]
blk.35.attn_v_b.weight
Q4_K
Q4_K
[512, 128, 32]
blk.35.exp_probs_b.bias
F32
F32
[512]
blk.35.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.35.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.35.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.35.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.35.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.35.ffn_norm.weight
F32
F32
[2560]
blk.35.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.35.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.36
blk.36.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.36.attn_norm.weight
F32
F32
[2560]
blk.36.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.36.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.36.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.36.exp_probs_b.bias
F32
F32
[512]
blk.36.ffn_down_exps.weight
Q4_K
Q4_K
[768, 2560, 512]
blk.36.ffn_down_shexp.weight
Q4_K
Q4_K
[768, 2560]
blk.36.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.36.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.36.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.36.ffn_norm.weight
F32
F32
[2560]
blk.36.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.36.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.36.ssm_a
F32
F32
[1, 32]
blk.36.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.36.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.36.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.36.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.36.ssm_dt.bias
F32
F32
[4096]
blk.36.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.36.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.36.ssm_norm.weight
F32
F32
[128]
blk.37
blk.37.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.37.attn_norm.weight
F32
F32
[2560]
blk.37.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.37.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.37.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.37.exp_probs_b.bias
F32
F32
[512]
blk.37.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.37.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.37.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.37.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.37.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.37.ffn_norm.weight
F32
F32
[2560]
blk.37.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.37.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.37.ssm_a
F32
F32
[1, 32]
blk.37.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.37.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.37.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.37.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.37.ssm_dt.bias
F32
F32
[4096]
blk.37.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.37.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.37.ssm_norm.weight
F32
F32
[128]
blk.38
blk.38.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.38.attn_norm.weight
F32
F32
[2560]
blk.38.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.38.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.38.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.38.exp_probs_b.bias
F32
F32
[512]
blk.38.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.38.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.38.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.38.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.38.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.38.ffn_norm.weight
F32
F32
[2560]
blk.38.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.38.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.38.ssm_a
F32
F32
[1, 32]
blk.38.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.38.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.38.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.38.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.38.ssm_dt.bias
F32
F32
[4096]
blk.38.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.38.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.38.ssm_norm.weight
F32
F32
[128]
blk.39
blk.39.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.39.attn_norm.weight
F32
F32
[2560]
blk.39.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.39.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.39.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.39.exp_probs_b.bias
F32
F32
[512]
blk.39.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.39.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.39.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.39.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.39.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.39.ffn_norm.weight
F32
F32
[2560]
blk.39.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.39.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.39.ssm_a
F32
F32
[1, 32]
blk.39.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.39.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.39.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.39.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.39.ssm_dt.bias
F32
F32
[4096]
blk.39.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.39.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.39.ssm_norm.weight
F32
F32
[128]
blk.40
blk.40.attn_k.weight
Q4_K
Q4_K
[2560, 4096]
blk.40.attn_norm.weight
F32
F32
[2560]
blk.40.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.40.attn_q.weight
Q4_K
Q4_K
[2560, 4096]
blk.40.attn_v.weight
Q6_K
Q6_K
[2560, 4096]
blk.40.exp_probs_b.bias
F32
F32
[512]
blk.40.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.40.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.40.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.40.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.40.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.40.ffn_norm.weight
F32
F32
[2560]
blk.40.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.40.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.40.ssm_a
F32
F32
[1, 32]
blk.40.ssm_beta.weight
Q4_K
Q4_K
[2560, 32]
blk.40.ssm_conv1d_k.weight
F32
F32
[4, 1, 4096]
blk.40.ssm_conv1d_q.weight
F32
F32
[4, 1, 4096]
blk.40.ssm_conv1d_v.weight
F32
F32
[4, 1, 4096]
blk.40.ssm_dt.bias
F32
F32
[4096]
blk.40.ssm_f_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.40.ssm_g_a.weight
Q4_K
Q4_K
[2560, 4096]
blk.40.ssm_norm.weight
F32
F32
[128]
blk.41
blk.41.attn_gate.weight
Q4_K
Q4_K
[2560, 32]
blk.41.attn_k_b.weight
Q5_0
Q5_0
[128, 512, 32]
blk.41.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2560, 576]
blk.41.attn_kv_a_norm.weight
F32
F32
[512]
blk.41.attn_norm.weight
F32
F32
[2560]
blk.41.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.41.attn_q.weight
Q4_K
Q4_K
[2560, 6144]
blk.41.attn_v_b.weight
Q4_K
Q4_K
[512, 128, 32]
blk.41.exp_probs_b.bias
F32
F32
[512]
blk.41.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.41.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.41.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.41.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.41.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.41.ffn_norm.weight
F32
F32
[2560]
blk.41.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.41.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.42
blk.42.attn_gate.weight
Q4_K
Q4_K
[2560, 32]
blk.42.attn_k_b.weight
Q5_0
Q5_0
[128, 512, 32]
blk.42.attn_kv_a_mqa.weight
Q4_K
Q4_K
[2560, 576]
blk.42.attn_kv_a_norm.weight
F32
F32
[512]
blk.42.attn_norm.weight
F32
F32
[2560]
blk.42.attn_output.weight
Q4_K
Q4_K
[4096, 2560]
blk.42.attn_q.weight
Q4_K
Q4_K
[2560, 6144]
blk.42.attn_v_b.weight
Q4_K
Q4_K
[512, 128, 32]
blk.42.exp_probs_b.bias
F32
F32
[512]
blk.42.ffn_down_exps.weight
Q6_K
Q6_K
[768, 2560, 512]
blk.42.ffn_down_shexp.weight
Q6_K
Q6_K
[768, 2560]
blk.42.ffn_gate_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.42.ffn_gate_inp.weight
F32
F32
[2560, 512]
blk.42.ffn_gate_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.42.ffn_norm.weight
F32
F32
[2560]
blk.42.ffn_up_exps.weight
Q4_K
Q4_K
[2560, 768, 512]
blk.42.ffn_up_shexp.weight
Q4_K
Q4_K
[2560, 768]
blk.42.layer_output_norm.weight
F32
F32
[2560]
blk.42.nextn.eh_proj.weight
Q4_K
Q4_K
[5120, 2560]
blk.42.nextn.enorm.weight
F32
F32
[2560]
blk.42.nextn.hnorm.weight
F32
F32
[2560]
output.weight
Q6_K
Q6_K
[2560, 157184]
output_norm.weight
F32
F32
[2560]