Models
Docs
Pricing
Sign in
Download
Models
Download
Docs
Pricing
Sign in
batiai
/
nemotron3-super
:iq3
9
Downloads
Updated
6 days ago
Cancel
nemotron3-super:iq3
...
/
model
f7b9e0d682ca · 57GB
Metadata
general.architecture
nemotron_h_moe
nemotron_h_moe
general.file_type
IQ3_XXS
IQ3_XXS
nemotron_h_moe.attention.head_count
32
32
nemotron_h_moe.attention.head_count_kv
[0, 0, 0, 0, 0, ...]
[0, 0, 0, 0, 0, ...]
nemotron_h_moe.attention.key_length
128
128
nemotron_h_moe.attention.layer_norm_epsilon
1e-05
1e-05
nemotron_h_moe.attention.layer_norm_rms_epsilon
1e-05
1e-05
nemotron_h_moe.attention.value_length
128
128
nemotron_h_moe.block_count
88
88
nemotron_h_moe.context_length
1048576
1048576
nemotron_h_moe.embedding_length
4096
4096
nemotron_h_moe.expert_count
512
512
nemotron_h_moe.expert_feed_forward_length
2688
2688
nemotron_h_moe.expert_group_count
1
1
nemotron_h_moe.expert_group_used_count
1
1
nemotron_h_moe.expert_shared_count
1
1
nemotron_h_moe.expert_shared_feed_forward_length
5376
5376
nemotron_h_moe.expert_used_count
22
22
nemotron_h_moe.expert_weights_norm
true
true
nemotron_h_moe.expert_weights_scale
5
5
nemotron_h_moe.feed_forward_length
[0, 2688, 0, 2688, 0, ...]
[0, 2688, 0, 2688, 0, ...]
nemotron_h_moe.moe_latent_size
1024
1024
nemotron_h_moe.rope.dimension_count
128
128
nemotron_h_moe.rope.freq_base
10000
10000
nemotron_h_moe.rope.scaling.finetuned
false
false
nemotron_h_moe.ssm.conv_kernel
4
4
nemotron_h_moe.ssm.group_count
8
8
nemotron_h_moe.ssm.inner_size
8192
8192
nemotron_h_moe.ssm.state_size
128
128
nemotron_h_moe.ssm.time_step_rank
128
128
nemotron_h_moe.vocab_size
131072
131072
tokenizer.ggml.add_bos_token
false
false
tokenizer.ggml.add_eos_token
false
false
tokenizer.ggml.bos_token_id
1
1
tokenizer.ggml.eos_token_id
11
11
tokenizer.ggml.merges
[Ġ Ġ, Ġ t, e r, i n, Ġ ĠĠĠ, ...]
[Ġ Ġ, Ġ t, e r, i n, Ġ ĠĠĠ, ...]
tokenizer.ggml.model
gpt2
gpt2
tokenizer.ggml.padding_token_id
11
11
tokenizer.ggml.pre
pixtral
pixtral
tokenizer.ggml.token_type
[3, 3, 3, 3, 3, ...]
[3, 3, 3, 3, 3, ...]
tokenizer.ggml.tokens
[<unk>, <s>, </s>, [INST], [/INST], ...]
[<unk>, <s>, </s>, [INST], [/INST], ...]
tokenizer.ggml.unknown_token_id
0
0
quantize.imatrix.chunks_count
200
200
quantize.imatrix.dataset
/home/bati/project/batiai-models/calibration/wikitext-2-raw-v1-test.txt
/home/bati/project/batiai-models/calibration/wikitext-2-raw-v1-test.txt
quantize.imatrix.entries_count
392
392
quantize.imatrix.file
/mnt/data1/output/nvidia-NVIDIA-Nemotron-3-Super-120B-A12B-BF16/imatrix.dat
/mnt/data1/output/nvidia-NVIDIA-Nemotron-3-Super-120B-A12B-BF16/imatrix.dat
Tensor
Name
Type
Shape
token_embd.weight
IQ3_S
IQ3_S
[4096, 131072]
blk.0
blk.0.attn_norm.weight
F32
F32
[4096]
blk.0.ssm_a
F32
F32
[1, 128]
blk.0.ssm_conv1d.bias
F32
F32
[10240]
blk.0.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.0.ssm_d
F32
F32
[1, 128]
blk.0.ssm_dt.bias
F32
F32
[128]
blk.0.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.0.ssm_norm.weight
F32
F32
[1024, 8]
blk.0.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.1
blk.1.attn_norm.weight
F32
F32
[4096]
blk.1.exp_probs_b.bias
F32
F32
[512]
blk.1.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.1.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.1.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.1.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.1.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.1.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.1.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.2
blk.2.attn_norm.weight
F32
F32
[4096]
blk.2.ssm_a
F32
F32
[1, 128]
blk.2.ssm_conv1d.bias
F32
F32
[10240]
blk.2.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.2.ssm_d
F32
F32
[1, 128]
blk.2.ssm_dt.bias
F32
F32
[128]
blk.2.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.2.ssm_norm.weight
F32
F32
[1024, 8]
blk.2.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.3
blk.3.attn_norm.weight
F32
F32
[4096]
blk.3.exp_probs_b.bias
F32
F32
[512]
blk.3.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.3.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.3.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.3.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.3.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.3.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.3.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.4
blk.4.attn_norm.weight
F32
F32
[4096]
blk.4.ssm_a
F32
F32
[1, 128]
blk.4.ssm_conv1d.bias
F32
F32
[10240]
blk.4.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.4.ssm_d
F32
F32
[1, 128]
blk.4.ssm_dt.bias
F32
F32
[128]
blk.4.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.4.ssm_norm.weight
F32
F32
[1024, 8]
blk.4.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.5
blk.5.attn_norm.weight
F32
F32
[4096]
blk.5.exp_probs_b.bias
F32
F32
[512]
blk.5.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.5.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.5.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.5.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.5.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.5.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.5.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.6
blk.6.attn_norm.weight
F32
F32
[4096]
blk.6.ssm_a
F32
F32
[1, 128]
blk.6.ssm_conv1d.bias
F32
F32
[10240]
blk.6.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.6.ssm_d
F32
F32
[1, 128]
blk.6.ssm_dt.bias
F32
F32
[128]
blk.6.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.6.ssm_norm.weight
F32
F32
[1024, 8]
blk.6.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.7
blk.7.attn_k.weight
IQ2_S
IQ2_S
[4096, 256]
blk.7.attn_norm.weight
F32
F32
[4096]
blk.7.attn_output.weight
IQ3_S
IQ3_S
[4096, 4096]
blk.7.attn_q.weight
IQ2_S
IQ2_S
[4096, 4096]
blk.7.attn_v.weight
IQ3_XXS
IQ3_XXS
[4096, 256]
blk.8
blk.8.attn_norm.weight
F32
F32
[4096]
blk.8.exp_probs_b.bias
F32
F32
[512]
blk.8.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.8.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.8.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.8.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.8.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.8.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.8.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.9
blk.9.attn_norm.weight
F32
F32
[4096]
blk.9.ssm_a
F32
F32
[1, 128]
blk.9.ssm_conv1d.bias
F32
F32
[10240]
blk.9.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.9.ssm_d
F32
F32
[1, 128]
blk.9.ssm_dt.bias
F32
F32
[128]
blk.9.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.9.ssm_norm.weight
F32
F32
[1024, 8]
blk.9.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.10
blk.10.attn_norm.weight
F32
F32
[4096]
blk.10.exp_probs_b.bias
F32
F32
[512]
blk.10.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.10.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.10.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.10.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.10.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.10.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.10.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.11
blk.11.attn_norm.weight
F32
F32
[4096]
blk.11.ssm_a
F32
F32
[1, 128]
blk.11.ssm_conv1d.bias
F32
F32
[10240]
blk.11.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.11.ssm_d
F32
F32
[1, 128]
blk.11.ssm_dt.bias
F32
F32
[128]
blk.11.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.11.ssm_norm.weight
F32
F32
[1024, 8]
blk.11.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.12
blk.12.attn_norm.weight
F32
F32
[4096]
blk.12.exp_probs_b.bias
F32
F32
[512]
blk.12.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.12.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.12.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.12.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.12.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.12.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.12.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.13
blk.13.attn_norm.weight
F32
F32
[4096]
blk.13.ssm_a
F32
F32
[1, 128]
blk.13.ssm_conv1d.bias
F32
F32
[10240]
blk.13.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.13.ssm_d
F32
F32
[1, 128]
blk.13.ssm_dt.bias
F32
F32
[128]
blk.13.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.13.ssm_norm.weight
F32
F32
[1024, 8]
blk.13.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.14
blk.14.attn_norm.weight
F32
F32
[4096]
blk.14.exp_probs_b.bias
F32
F32
[512]
blk.14.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.14.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.14.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.14.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.14.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.14.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.14.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.15
blk.15.attn_norm.weight
F32
F32
[4096]
blk.15.ssm_a
F32
F32
[1, 128]
blk.15.ssm_conv1d.bias
F32
F32
[10240]
blk.15.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.15.ssm_d
F32
F32
[1, 128]
blk.15.ssm_dt.bias
F32
F32
[128]
blk.15.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.15.ssm_norm.weight
F32
F32
[1024, 8]
blk.15.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.16
blk.16.attn_k.weight
IQ2_S
IQ2_S
[4096, 256]
blk.16.attn_norm.weight
F32
F32
[4096]
blk.16.attn_output.weight
IQ3_S
IQ3_S
[4096, 4096]
blk.16.attn_q.weight
IQ2_S
IQ2_S
[4096, 4096]
blk.16.attn_v.weight
IQ3_XXS
IQ3_XXS
[4096, 256]
blk.17
blk.17.attn_norm.weight
F32
F32
[4096]
blk.17.exp_probs_b.bias
F32
F32
[512]
blk.17.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.17.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.17.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.17.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.17.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.17.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.17.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.18
blk.18.attn_norm.weight
F32
F32
[4096]
blk.18.ssm_a
F32
F32
[1, 128]
blk.18.ssm_conv1d.bias
F32
F32
[10240]
blk.18.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.18.ssm_d
F32
F32
[1, 128]
blk.18.ssm_dt.bias
F32
F32
[128]
blk.18.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.18.ssm_norm.weight
F32
F32
[1024, 8]
blk.18.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.19
blk.19.attn_norm.weight
F32
F32
[4096]
blk.19.exp_probs_b.bias
F32
F32
[512]
blk.19.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.19.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.19.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.19.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.19.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.19.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.19.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.20
blk.20.attn_norm.weight
F32
F32
[4096]
blk.20.ssm_a
F32
F32
[1, 128]
blk.20.ssm_conv1d.bias
F32
F32
[10240]
blk.20.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.20.ssm_d
F32
F32
[1, 128]
blk.20.ssm_dt.bias
F32
F32
[128]
blk.20.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.20.ssm_norm.weight
F32
F32
[1024, 8]
blk.20.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.21
blk.21.attn_norm.weight
F32
F32
[4096]
blk.21.exp_probs_b.bias
F32
F32
[512]
blk.21.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.21.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.21.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.21.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.21.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.21.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.21.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.22
blk.22.attn_norm.weight
F32
F32
[4096]
blk.22.ssm_a
F32
F32
[1, 128]
blk.22.ssm_conv1d.bias
F32
F32
[10240]
blk.22.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.22.ssm_d
F32
F32
[1, 128]
blk.22.ssm_dt.bias
F32
F32
[128]
blk.22.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.22.ssm_norm.weight
F32
F32
[1024, 8]
blk.22.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.23
blk.23.attn_norm.weight
F32
F32
[4096]
blk.23.exp_probs_b.bias
F32
F32
[512]
blk.23.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.23.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.23.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.23.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.23.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.23.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.23.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.24
blk.24.attn_norm.weight
F32
F32
[4096]
blk.24.ssm_a
F32
F32
[1, 128]
blk.24.ssm_conv1d.bias
F32
F32
[10240]
blk.24.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.24.ssm_d
F32
F32
[1, 128]
blk.24.ssm_dt.bias
F32
F32
[128]
blk.24.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.24.ssm_norm.weight
F32
F32
[1024, 8]
blk.24.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.25
blk.25.attn_k.weight
IQ2_S
IQ2_S
[4096, 256]
blk.25.attn_norm.weight
F32
F32
[4096]
blk.25.attn_output.weight
IQ3_S
IQ3_S
[4096, 4096]
blk.25.attn_q.weight
IQ2_S
IQ2_S
[4096, 4096]
blk.25.attn_v.weight
IQ3_XXS
IQ3_XXS
[4096, 256]
blk.26
blk.26.attn_norm.weight
F32
F32
[4096]
blk.26.exp_probs_b.bias
F32
F32
[512]
blk.26.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.26.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.26.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.26.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.26.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.26.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.26.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.27
blk.27.attn_norm.weight
F32
F32
[4096]
blk.27.ssm_a
F32
F32
[1, 128]
blk.27.ssm_conv1d.bias
F32
F32
[10240]
blk.27.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.27.ssm_d
F32
F32
[1, 128]
blk.27.ssm_dt.bias
F32
F32
[128]
blk.27.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.27.ssm_norm.weight
F32
F32
[1024, 8]
blk.27.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.28
blk.28.attn_norm.weight
F32
F32
[4096]
blk.28.exp_probs_b.bias
F32
F32
[512]
blk.28.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.28.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.28.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.28.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.28.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.28.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.28.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.29
blk.29.attn_norm.weight
F32
F32
[4096]
blk.29.ssm_a
F32
F32
[1, 128]
blk.29.ssm_conv1d.bias
F32
F32
[10240]
blk.29.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.29.ssm_d
F32
F32
[1, 128]
blk.29.ssm_dt.bias
F32
F32
[128]
blk.29.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.29.ssm_norm.weight
F32
F32
[1024, 8]
blk.29.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.30
blk.30.attn_norm.weight
F32
F32
[4096]
blk.30.exp_probs_b.bias
F32
F32
[512]
blk.30.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.30.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.30.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.30.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.30.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.30.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.30.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.31
blk.31.attn_norm.weight
F32
F32
[4096]
blk.31.ssm_a
F32
F32
[1, 128]
blk.31.ssm_conv1d.bias
F32
F32
[10240]
blk.31.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.31.ssm_d
F32
F32
[1, 128]
blk.31.ssm_dt.bias
F32
F32
[128]
blk.31.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.31.ssm_norm.weight
F32
F32
[1024, 8]
blk.31.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.32
blk.32.attn_norm.weight
F32
F32
[4096]
blk.32.exp_probs_b.bias
F32
F32
[512]
blk.32.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.32.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.32.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.32.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.32.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.32.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.32.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.33
blk.33.attn_norm.weight
F32
F32
[4096]
blk.33.ssm_a
F32
F32
[1, 128]
blk.33.ssm_conv1d.bias
F32
F32
[10240]
blk.33.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.33.ssm_d
F32
F32
[1, 128]
blk.33.ssm_dt.bias
F32
F32
[128]
blk.33.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.33.ssm_norm.weight
F32
F32
[1024, 8]
blk.33.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.34
blk.34.attn_norm.weight
F32
F32
[4096]
blk.34.exp_probs_b.bias
F32
F32
[512]
blk.34.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.34.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.34.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.34.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.34.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.34.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.34.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.35
blk.35.attn_norm.weight
F32
F32
[4096]
blk.35.ssm_a
F32
F32
[1, 128]
blk.35.ssm_conv1d.bias
F32
F32
[10240]
blk.35.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.35.ssm_d
F32
F32
[1, 128]
blk.35.ssm_dt.bias
F32
F32
[128]
blk.35.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.35.ssm_norm.weight
F32
F32
[1024, 8]
blk.35.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.36
blk.36.attn_k.weight
IQ2_S
IQ2_S
[4096, 256]
blk.36.attn_norm.weight
F32
F32
[4096]
blk.36.attn_output.weight
IQ3_S
IQ3_S
[4096, 4096]
blk.36.attn_q.weight
IQ2_S
IQ2_S
[4096, 4096]
blk.36.attn_v.weight
IQ3_XXS
IQ3_XXS
[4096, 256]
blk.37
blk.37.attn_norm.weight
F32
F32
[4096]
blk.37.exp_probs_b.bias
F32
F32
[512]
blk.37.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.37.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.37.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.37.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.37.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.37.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.37.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.38
blk.38.attn_norm.weight
F32
F32
[4096]
blk.38.ssm_a
F32
F32
[1, 128]
blk.38.ssm_conv1d.bias
F32
F32
[10240]
blk.38.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.38.ssm_d
F32
F32
[1, 128]
blk.38.ssm_dt.bias
F32
F32
[128]
blk.38.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.38.ssm_norm.weight
F32
F32
[1024, 8]
blk.38.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.39
blk.39.attn_norm.weight
F32
F32
[4096]
blk.39.exp_probs_b.bias
F32
F32
[512]
blk.39.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.39.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.39.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.39.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.39.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.39.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.39.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.40
blk.40.attn_norm.weight
F32
F32
[4096]
blk.40.ssm_a
F32
F32
[1, 128]
blk.40.ssm_conv1d.bias
F32
F32
[10240]
blk.40.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.40.ssm_d
F32
F32
[1, 128]
blk.40.ssm_dt.bias
F32
F32
[128]
blk.40.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.40.ssm_norm.weight
F32
F32
[1024, 8]
blk.40.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.41
blk.41.attn_norm.weight
F32
F32
[4096]
blk.41.exp_probs_b.bias
F32
F32
[512]
blk.41.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.41.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.41.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.41.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.41.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.41.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.41.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.42
blk.42.attn_norm.weight
F32
F32
[4096]
blk.42.ssm_a
F32
F32
[1, 128]
blk.42.ssm_conv1d.bias
F32
F32
[10240]
blk.42.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.42.ssm_d
F32
F32
[1, 128]
blk.42.ssm_dt.bias
F32
F32
[128]
blk.42.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.42.ssm_norm.weight
F32
F32
[1024, 8]
blk.42.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.43
blk.43.attn_norm.weight
F32
F32
[4096]
blk.43.exp_probs_b.bias
F32
F32
[512]
blk.43.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.43.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.43.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.43.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.43.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.43.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.43.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.44
blk.44.attn_norm.weight
F32
F32
[4096]
blk.44.ssm_a
F32
F32
[1, 128]
blk.44.ssm_conv1d.bias
F32
F32
[10240]
blk.44.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.44.ssm_d
F32
F32
[1, 128]
blk.44.ssm_dt.bias
F32
F32
[128]
blk.44.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.44.ssm_norm.weight
F32
F32
[1024, 8]
blk.44.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.45
blk.45.attn_norm.weight
F32
F32
[4096]
blk.45.exp_probs_b.bias
F32
F32
[512]
blk.45.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.45.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.45.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.45.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.45.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.45.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.45.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.46
blk.46.attn_norm.weight
F32
F32
[4096]
blk.46.ssm_a
F32
F32
[1, 128]
blk.46.ssm_conv1d.bias
F32
F32
[10240]
blk.46.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.46.ssm_d
F32
F32
[1, 128]
blk.46.ssm_dt.bias
F32
F32
[128]
blk.46.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.46.ssm_norm.weight
F32
F32
[1024, 8]
blk.46.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.47
blk.47.attn_k.weight
IQ2_S
IQ2_S
[4096, 256]
blk.47.attn_norm.weight
F32
F32
[4096]
blk.47.attn_output.weight
IQ3_S
IQ3_S
[4096, 4096]
blk.47.attn_q.weight
IQ2_S
IQ2_S
[4096, 4096]
blk.47.attn_v.weight
IQ3_XXS
IQ3_XXS
[4096, 256]
blk.48
blk.48.attn_norm.weight
F32
F32
[4096]
blk.48.exp_probs_b.bias
F32
F32
[512]
blk.48.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.48.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.48.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.48.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.48.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.48.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.48.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.49
blk.49.attn_norm.weight
F32
F32
[4096]
blk.49.ssm_a
F32
F32
[1, 128]
blk.49.ssm_conv1d.bias
F32
F32
[10240]
blk.49.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.49.ssm_d
F32
F32
[1, 128]
blk.49.ssm_dt.bias
F32
F32
[128]
blk.49.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.49.ssm_norm.weight
F32
F32
[1024, 8]
blk.49.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.50
blk.50.attn_norm.weight
F32
F32
[4096]
blk.50.exp_probs_b.bias
F32
F32
[512]
blk.50.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.50.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.50.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.50.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.50.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.50.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.50.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.51
blk.51.attn_norm.weight
F32
F32
[4096]
blk.51.ssm_a
F32
F32
[1, 128]
blk.51.ssm_conv1d.bias
F32
F32
[10240]
blk.51.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.51.ssm_d
F32
F32
[1, 128]
blk.51.ssm_dt.bias
F32
F32
[128]
blk.51.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.51.ssm_norm.weight
F32
F32
[1024, 8]
blk.51.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.52
blk.52.attn_norm.weight
F32
F32
[4096]
blk.52.exp_probs_b.bias
F32
F32
[512]
blk.52.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.52.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.52.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.52.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.52.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.52.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.52.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.53
blk.53.attn_norm.weight
F32
F32
[4096]
blk.53.ssm_a
F32
F32
[1, 128]
blk.53.ssm_conv1d.bias
F32
F32
[10240]
blk.53.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.53.ssm_d
F32
F32
[1, 128]
blk.53.ssm_dt.bias
F32
F32
[128]
blk.53.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.53.ssm_norm.weight
F32
F32
[1024, 8]
blk.53.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.54
blk.54.attn_norm.weight
F32
F32
[4096]
blk.54.exp_probs_b.bias
F32
F32
[512]
blk.54.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.54.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.54.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.54.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.54.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.54.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.54.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.55
blk.55.attn_norm.weight
F32
F32
[4096]
blk.55.ssm_a
F32
F32
[1, 128]
blk.55.ssm_conv1d.bias
F32
F32
[10240]
blk.55.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.55.ssm_d
F32
F32
[1, 128]
blk.55.ssm_dt.bias
F32
F32
[128]
blk.55.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.55.ssm_norm.weight
F32
F32
[1024, 8]
blk.55.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.56
blk.56.attn_norm.weight
F32
F32
[4096]
blk.56.exp_probs_b.bias
F32
F32
[512]
blk.56.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.56.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.56.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.56.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.56.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.56.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.56.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.57
blk.57.attn_norm.weight
F32
F32
[4096]
blk.57.ssm_a
F32
F32
[1, 128]
blk.57.ssm_conv1d.bias
F32
F32
[10240]
blk.57.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.57.ssm_d
F32
F32
[1, 128]
blk.57.ssm_dt.bias
F32
F32
[128]
blk.57.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.57.ssm_norm.weight
F32
F32
[1024, 8]
blk.57.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.58
blk.58.attn_k.weight
IQ2_S
IQ2_S
[4096, 256]
blk.58.attn_norm.weight
F32
F32
[4096]
blk.58.attn_output.weight
IQ3_S
IQ3_S
[4096, 4096]
blk.58.attn_q.weight
IQ2_S
IQ2_S
[4096, 4096]
blk.58.attn_v.weight
IQ3_XXS
IQ3_XXS
[4096, 256]
blk.59
blk.59.attn_norm.weight
F32
F32
[4096]
blk.59.exp_probs_b.bias
F32
F32
[512]
blk.59.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.59.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.59.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.59.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.59.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.59.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.59.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.60
blk.60.attn_norm.weight
F32
F32
[4096]
blk.60.ssm_a
F32
F32
[1, 128]
blk.60.ssm_conv1d.bias
F32
F32
[10240]
blk.60.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.60.ssm_d
F32
F32
[1, 128]
blk.60.ssm_dt.bias
F32
F32
[128]
blk.60.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.60.ssm_norm.weight
F32
F32
[1024, 8]
blk.60.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.61
blk.61.attn_norm.weight
F32
F32
[4096]
blk.61.exp_probs_b.bias
F32
F32
[512]
blk.61.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.61.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.61.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.61.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.61.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.61.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.61.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.62
blk.62.attn_norm.weight
F32
F32
[4096]
blk.62.ssm_a
F32
F32
[1, 128]
blk.62.ssm_conv1d.bias
F32
F32
[10240]
blk.62.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.62.ssm_d
F32
F32
[1, 128]
blk.62.ssm_dt.bias
F32
F32
[128]
blk.62.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.62.ssm_norm.weight
F32
F32
[1024, 8]
blk.62.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.63
blk.63.attn_norm.weight
F32
F32
[4096]
blk.63.exp_probs_b.bias
F32
F32
[512]
blk.63.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.63.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.63.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.63.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.63.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.63.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.63.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.64
blk.64.attn_norm.weight
F32
F32
[4096]
blk.64.ssm_a
F32
F32
[1, 128]
blk.64.ssm_conv1d.bias
F32
F32
[10240]
blk.64.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.64.ssm_d
F32
F32
[1, 128]
blk.64.ssm_dt.bias
F32
F32
[128]
blk.64.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.64.ssm_norm.weight
F32
F32
[1024, 8]
blk.64.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.65
blk.65.attn_norm.weight
F32
F32
[4096]
blk.65.exp_probs_b.bias
F32
F32
[512]
blk.65.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.65.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.65.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.65.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.65.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.65.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.65.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.66
blk.66.attn_norm.weight
F32
F32
[4096]
blk.66.ssm_a
F32
F32
[1, 128]
blk.66.ssm_conv1d.bias
F32
F32
[10240]
blk.66.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.66.ssm_d
F32
F32
[1, 128]
blk.66.ssm_dt.bias
F32
F32
[128]
blk.66.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.66.ssm_norm.weight
F32
F32
[1024, 8]
blk.66.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.67
blk.67.attn_norm.weight
F32
F32
[4096]
blk.67.exp_probs_b.bias
F32
F32
[512]
blk.67.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.67.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.67.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.67.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.67.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.67.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.67.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.68
blk.68.attn_norm.weight
F32
F32
[4096]
blk.68.ssm_a
F32
F32
[1, 128]
blk.68.ssm_conv1d.bias
F32
F32
[10240]
blk.68.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.68.ssm_d
F32
F32
[1, 128]
blk.68.ssm_dt.bias
F32
F32
[128]
blk.68.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.68.ssm_norm.weight
F32
F32
[1024, 8]
blk.68.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.69
blk.69.attn_k.weight
IQ2_S
IQ2_S
[4096, 256]
blk.69.attn_norm.weight
F32
F32
[4096]
blk.69.attn_output.weight
IQ3_S
IQ3_S
[4096, 4096]
blk.69.attn_q.weight
IQ2_S
IQ2_S
[4096, 4096]
blk.69.attn_v.weight
IQ3_XXS
IQ3_XXS
[4096, 256]
blk.70
blk.70.attn_norm.weight
F32
F32
[4096]
blk.70.exp_probs_b.bias
F32
F32
[512]
blk.70.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.70.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.70.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.70.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.70.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.70.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.70.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.71
blk.71.attn_norm.weight
F32
F32
[4096]
blk.71.ssm_a
F32
F32
[1, 128]
blk.71.ssm_conv1d.bias
F32
F32
[10240]
blk.71.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.71.ssm_d
F32
F32
[1, 128]
blk.71.ssm_dt.bias
F32
F32
[128]
blk.71.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.71.ssm_norm.weight
F32
F32
[1024, 8]
blk.71.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.72
blk.72.attn_norm.weight
F32
F32
[4096]
blk.72.exp_probs_b.bias
F32
F32
[512]
blk.72.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.72.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.72.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.72.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.72.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.72.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.72.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.73
blk.73.attn_norm.weight
F32
F32
[4096]
blk.73.ssm_a
F32
F32
[1, 128]
blk.73.ssm_conv1d.bias
F32
F32
[10240]
blk.73.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.73.ssm_d
F32
F32
[1, 128]
blk.73.ssm_dt.bias
F32
F32
[128]
blk.73.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.73.ssm_norm.weight
F32
F32
[1024, 8]
blk.73.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.74
blk.74.attn_norm.weight
F32
F32
[4096]
blk.74.exp_probs_b.bias
F32
F32
[512]
blk.74.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.74.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.74.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.74.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.74.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.74.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.74.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.75
blk.75.attn_norm.weight
F32
F32
[4096]
blk.75.ssm_a
F32
F32
[1, 128]
blk.75.ssm_conv1d.bias
F32
F32
[10240]
blk.75.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.75.ssm_d
F32
F32
[1, 128]
blk.75.ssm_dt.bias
F32
F32
[128]
blk.75.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.75.ssm_norm.weight
F32
F32
[1024, 8]
blk.75.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.76
blk.76.attn_norm.weight
F32
F32
[4096]
blk.76.exp_probs_b.bias
F32
F32
[512]
blk.76.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.76.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.76.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.76.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.76.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.76.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.76.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.77
blk.77.attn_norm.weight
F32
F32
[4096]
blk.77.ssm_a
F32
F32
[1, 128]
blk.77.ssm_conv1d.bias
F32
F32
[10240]
blk.77.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.77.ssm_d
F32
F32
[1, 128]
blk.77.ssm_dt.bias
F32
F32
[128]
blk.77.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.77.ssm_norm.weight
F32
F32
[1024, 8]
blk.77.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.78
blk.78.attn_k.weight
IQ2_S
IQ2_S
[4096, 256]
blk.78.attn_norm.weight
F32
F32
[4096]
blk.78.attn_output.weight
IQ3_S
IQ3_S
[4096, 4096]
blk.78.attn_q.weight
IQ2_S
IQ2_S
[4096, 4096]
blk.78.attn_v.weight
IQ3_XXS
IQ3_XXS
[4096, 256]
blk.79
blk.79.attn_norm.weight
F32
F32
[4096]
blk.79.exp_probs_b.bias
F32
F32
[512]
blk.79.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.79.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.79.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.79.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.79.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.79.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.79.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.80
blk.80.attn_norm.weight
F32
F32
[4096]
blk.80.ssm_a
F32
F32
[1, 128]
blk.80.ssm_conv1d.bias
F32
F32
[10240]
blk.80.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.80.ssm_d
F32
F32
[1, 128]
blk.80.ssm_dt.bias
F32
F32
[128]
blk.80.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.80.ssm_norm.weight
F32
F32
[1024, 8]
blk.80.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.81
blk.81.attn_norm.weight
F32
F32
[4096]
blk.81.exp_probs_b.bias
F32
F32
[512]
blk.81.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.81.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.81.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.81.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.81.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.81.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.81.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.82
blk.82.attn_norm.weight
F32
F32
[4096]
blk.82.ssm_a
F32
F32
[1, 128]
blk.82.ssm_conv1d.bias
F32
F32
[10240]
blk.82.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.82.ssm_d
F32
F32
[1, 128]
blk.82.ssm_dt.bias
F32
F32
[128]
blk.82.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.82.ssm_norm.weight
F32
F32
[1024, 8]
blk.82.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.83
blk.83.attn_norm.weight
F32
F32
[4096]
blk.83.exp_probs_b.bias
F32
F32
[512]
blk.83.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.83.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.83.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.83.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.83.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.83.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.83.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.84
blk.84.attn_norm.weight
F32
F32
[4096]
blk.84.ssm_a
F32
F32
[1, 128]
blk.84.ssm_conv1d.bias
F32
F32
[10240]
blk.84.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.84.ssm_d
F32
F32
[1, 128]
blk.84.ssm_dt.bias
F32
F32
[128]
blk.84.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.84.ssm_norm.weight
F32
F32
[1024, 8]
blk.84.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.85
blk.85.attn_norm.weight
F32
F32
[4096]
blk.85.exp_probs_b.bias
F32
F32
[512]
blk.85.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.85.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.85.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.85.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.85.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.85.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.85.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
blk.86
blk.86.attn_norm.weight
F32
F32
[4096]
blk.86.ssm_a
F32
F32
[1, 128]
blk.86.ssm_conv1d.bias
F32
F32
[10240]
blk.86.ssm_conv1d.weight
F32
F32
[4, 10240]
blk.86.ssm_d
F32
F32
[1, 128]
blk.86.ssm_dt.bias
F32
F32
[128]
blk.86.ssm_in.weight
IQ3_XXS
IQ3_XXS
[4096, 18560]
blk.86.ssm_norm.weight
F32
F32
[1024, 8]
blk.86.ssm_out.weight
IQ3_XXS
IQ3_XXS
[8192, 4096]
blk.87
blk.87.attn_norm.weight
F32
F32
[4096]
blk.87.exp_probs_b.bias
F32
F32
[512]
blk.87.ffn_down_exps.weight
IQ4_NL
IQ4_NL
[2688, 1024, 512]
blk.87.ffn_down_shexp.weight
IQ3_XXS
IQ3_XXS
[5376, 4096]
blk.87.ffn_gate_inp.weight
F32
F32
[4096, 512]
blk.87.ffn_latent_down.weight
IQ3_XXS
IQ3_XXS
[4096, 1024]
blk.87.ffn_latent_up.weight
IQ3_XXS
IQ3_XXS
[1024, 4096]
blk.87.ffn_up_exps.weight
IQ3_XXS
IQ3_XXS
[1024, 2688, 512]
blk.87.ffn_up_shexp.weight
IQ3_XXS
IQ3_XXS
[4096, 5376]
output.weight
Q5_K
Q5_K
[4096, 131072]
output_norm.weight
F32
F32
[4096]