Models
GitHub
Discord
Turbo
Sign in
Download
Models
Download
GitHub
Discord
Sign in
llama4
:latest
637.9K
Downloads
Updated
2 months ago
Meta's latest collection of multimodal models.
Meta's latest collection of multimodal models.
Cancel
vision
tools
16x17b
128x17b
llama4:latest
...
/
model
9d507a36062c · 67GB
Metadata
general.architecture
llama4
llama4
general.file_type
Q4_K_M
Q4_K_M
llama4.attention.chunk_size
8192
8192
llama4.attention.head_count
40
40
llama4.attention.head_count_kv
8
8
llama4.attention.head_dim
128
128
llama4.attention.key_length
128
128
llama4.attention.layer_norm_rms_epsilon
1e-05
1e-05
llama4.attention.value_length
128
128
llama4.block_count
48
48
llama4.context_length
10485760
10485760
llama4.embedding_length
5120
5120
llama4.expert_count
16
16
llama4.expert_feed_forward_length
8192
8192
llama4.expert_used_count
1
1
llama4.feed_forward_length
16384
16384
llama4.interleave_moe_layer_step
1
1
llama4.rope.dimension_count
128
128
llama4.rope.freq_base
500000
500000
llama4.use_qk_norm
true
true
llama4.vision.attention.head_count
16
16
llama4.vision.block_count
34
34
llama4.vision.embedding_length
1408
1408
llama4.vision.feed_forward_length
5632
5632
llama4.vision.image_size
336
336
llama4.vision.layer_norm_epsilon
1e-05
1e-05
llama4.vision.patch_size
14
14
llama4.vision.pixel_shuffle_ratio
0.5
0.5
llama4.vision.rope.freq_base
10000
10000
llama4.vocab_size
202048
202048
tokenizer.ggml.add_bos_token
false
false
tokenizer.ggml.add_eos_token
false
false
tokenizer.ggml.add_padding_token
false
false
tokenizer.ggml.bos_token_id
200000
200000
tokenizer.ggml.eos_token_id
200008
200008
tokenizer.ggml.merges
[Ġ Ġ, Ġ ĠĠĠ, ĠĠ ĠĠ, ĠĠĠ Ġ, e r, ...]
[Ġ Ġ, Ġ ĠĠĠ, ĠĠ ĠĠ, ĠĠĠ Ġ, e r, ...]
tokenizer.ggml.model
gpt2
gpt2
tokenizer.ggml.padding_token_id
200018
200018
tokenizer.ggml.pre
default
default
tokenizer.ggml.scores
[0, 1, 2, 3, 4, ...]
[0, 1, 2, 3, 4, ...]
tokenizer.ggml.token_type
[1, 1, 1, 1, 1, ...]
[1, 1, 1, 1, 1, ...]
tokenizer.ggml.tokens
[À, Á, õ, ö, ÷, ...]
[À, Á, õ, ö, ÷, ...]
Tensor
Name
Type
Shape
token_embd.weight
Q4_K
Q4_K
[5120, 202048]
blk.0
blk.0.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.0.attn_norm.weight
F32
F32
[5120]
blk.0.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.0.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.0.attn_v.weight
F16
F16
[5120, 1024]
blk.0.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.0.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.0.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.0.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.0.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.0.ffn_norm.weight
F32
F32
[5120]
blk.0.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.0.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.1
blk.1.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.1.attn_norm.weight
F32
F32
[5120]
blk.1.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.1.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.1.attn_v.weight
F16
F16
[5120, 1024]
blk.1.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.1.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.1.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.1.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.1.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.1.ffn_norm.weight
F32
F32
[5120]
blk.1.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.1.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.2
blk.2.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.2.attn_norm.weight
F32
F32
[5120]
blk.2.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.2.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.2.attn_v.weight
F16
F16
[5120, 1024]
blk.2.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.2.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.2.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.2.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.2.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.2.ffn_norm.weight
F32
F32
[5120]
blk.2.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.2.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.3
blk.3.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.3.attn_norm.weight
F32
F32
[5120]
blk.3.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.3.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.3.attn_v.weight
F16
F16
[5120, 1024]
blk.3.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.3.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.3.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.3.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.3.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.3.ffn_norm.weight
F32
F32
[5120]
blk.3.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.3.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.4
blk.4.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.4.attn_norm.weight
F32
F32
[5120]
blk.4.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.4.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.4.attn_v.weight
F16
F16
[5120, 1024]
blk.4.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.4.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.4.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.4.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.4.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.4.ffn_norm.weight
F32
F32
[5120]
blk.4.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.4.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.5
blk.5.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.5.attn_norm.weight
F32
F32
[5120]
blk.5.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.5.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.5.attn_v.weight
F16
F16
[5120, 1024]
blk.5.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.5.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.5.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.5.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.5.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.5.ffn_norm.weight
F32
F32
[5120]
blk.5.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.5.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.6
blk.6.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.6.attn_norm.weight
F32
F32
[5120]
blk.6.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.6.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.6.attn_v.weight
F16
F16
[5120, 1024]
blk.6.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.6.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.6.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.6.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.6.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.6.ffn_norm.weight
F32
F32
[5120]
blk.6.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.6.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.7
blk.7.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.7.attn_norm.weight
F32
F32
[5120]
blk.7.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.7.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.7.attn_v.weight
F16
F16
[5120, 1024]
blk.7.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.7.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.7.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.7.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.7.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.7.ffn_norm.weight
F32
F32
[5120]
blk.7.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.7.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.8
blk.8.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.8.attn_norm.weight
F32
F32
[5120]
blk.8.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.8.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.8.attn_v.weight
F16
F16
[5120, 1024]
blk.8.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.8.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.8.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.8.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.8.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.8.ffn_norm.weight
F32
F32
[5120]
blk.8.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.8.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.9
blk.9.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.9.attn_norm.weight
F32
F32
[5120]
blk.9.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.9.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.9.attn_v.weight
F16
F16
[5120, 1024]
blk.9.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.9.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.9.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.9.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.9.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.9.ffn_norm.weight
F32
F32
[5120]
blk.9.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.9.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.10
blk.10.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.10.attn_norm.weight
F32
F32
[5120]
blk.10.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.10.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.10.attn_v.weight
F16
F16
[5120, 1024]
blk.10.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.10.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.10.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.10.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.10.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.10.ffn_norm.weight
F32
F32
[5120]
blk.10.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.10.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.11
blk.11.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.11.attn_norm.weight
F32
F32
[5120]
blk.11.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.11.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.11.attn_v.weight
F16
F16
[5120, 1024]
blk.11.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.11.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.11.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.11.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.11.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.11.ffn_norm.weight
F32
F32
[5120]
blk.11.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.11.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.12
blk.12.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.12.attn_norm.weight
F32
F32
[5120]
blk.12.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.12.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.12.attn_v.weight
F16
F16
[5120, 1024]
blk.12.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.12.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.12.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.12.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.12.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.12.ffn_norm.weight
F32
F32
[5120]
blk.12.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.12.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.13
blk.13.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.13.attn_norm.weight
F32
F32
[5120]
blk.13.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.13.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.13.attn_v.weight
F16
F16
[5120, 1024]
blk.13.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.13.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.13.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.13.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.13.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.13.ffn_norm.weight
F32
F32
[5120]
blk.13.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.13.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.14
blk.14.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.14.attn_norm.weight
F32
F32
[5120]
blk.14.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.14.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.14.attn_v.weight
F16
F16
[5120, 1024]
blk.14.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.14.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.14.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.14.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.14.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.14.ffn_norm.weight
F32
F32
[5120]
blk.14.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.14.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.15
blk.15.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.15.attn_norm.weight
F32
F32
[5120]
blk.15.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.15.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.15.attn_v.weight
F16
F16
[5120, 1024]
blk.15.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.15.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.15.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.15.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.15.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.15.ffn_norm.weight
F32
F32
[5120]
blk.15.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.15.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.16
blk.16.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.16.attn_norm.weight
F32
F32
[5120]
blk.16.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.16.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.16.attn_v.weight
F16
F16
[5120, 1024]
blk.16.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.16.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.16.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.16.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.16.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.16.ffn_norm.weight
F32
F32
[5120]
blk.16.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.16.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.17
blk.17.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.17.attn_norm.weight
F32
F32
[5120]
blk.17.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.17.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.17.attn_v.weight
F16
F16
[5120, 1024]
blk.17.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.17.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.17.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.17.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.17.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.17.ffn_norm.weight
F32
F32
[5120]
blk.17.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.17.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.18
blk.18.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.18.attn_norm.weight
F32
F32
[5120]
blk.18.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.18.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.18.attn_v.weight
F16
F16
[5120, 1024]
blk.18.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.18.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.18.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.18.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.18.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.18.ffn_norm.weight
F32
F32
[5120]
blk.18.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.18.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.19
blk.19.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.19.attn_norm.weight
F32
F32
[5120]
blk.19.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.19.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.19.attn_v.weight
F16
F16
[5120, 1024]
blk.19.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.19.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.19.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.19.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.19.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.19.ffn_norm.weight
F32
F32
[5120]
blk.19.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.19.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.20
blk.20.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.20.attn_norm.weight
F32
F32
[5120]
blk.20.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.20.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.20.attn_v.weight
F16
F16
[5120, 1024]
blk.20.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.20.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.20.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.20.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.20.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.20.ffn_norm.weight
F32
F32
[5120]
blk.20.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.20.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.21
blk.21.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.21.attn_norm.weight
F32
F32
[5120]
blk.21.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.21.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.21.attn_v.weight
F16
F16
[5120, 1024]
blk.21.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.21.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.21.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.21.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.21.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.21.ffn_norm.weight
F32
F32
[5120]
blk.21.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.21.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.22
blk.22.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.22.attn_norm.weight
F32
F32
[5120]
blk.22.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.22.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.22.attn_v.weight
F16
F16
[5120, 1024]
blk.22.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.22.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.22.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.22.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.22.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.22.ffn_norm.weight
F32
F32
[5120]
blk.22.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.22.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.23
blk.23.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.23.attn_norm.weight
F32
F32
[5120]
blk.23.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.23.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.23.attn_v.weight
F16
F16
[5120, 1024]
blk.23.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.23.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.23.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.23.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.23.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.23.ffn_norm.weight
F32
F32
[5120]
blk.23.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.23.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.24
blk.24.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.24.attn_norm.weight
F32
F32
[5120]
blk.24.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.24.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.24.attn_v.weight
F16
F16
[5120, 1024]
blk.24.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.24.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.24.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.24.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.24.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.24.ffn_norm.weight
F32
F32
[5120]
blk.24.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.24.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.25
blk.25.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.25.attn_norm.weight
F32
F32
[5120]
blk.25.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.25.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.25.attn_v.weight
F16
F16
[5120, 1024]
blk.25.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.25.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.25.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.25.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.25.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.25.ffn_norm.weight
F32
F32
[5120]
blk.25.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.25.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.26
blk.26.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.26.attn_norm.weight
F32
F32
[5120]
blk.26.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.26.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.26.attn_v.weight
F16
F16
[5120, 1024]
blk.26.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.26.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.26.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.26.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.26.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.26.ffn_norm.weight
F32
F32
[5120]
blk.26.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.26.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.27
blk.27.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.27.attn_norm.weight
F32
F32
[5120]
blk.27.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.27.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.27.attn_v.weight
F16
F16
[5120, 1024]
blk.27.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.27.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.27.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.27.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.27.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.27.ffn_norm.weight
F32
F32
[5120]
blk.27.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.27.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.28
blk.28.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.28.attn_norm.weight
F32
F32
[5120]
blk.28.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.28.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.28.attn_v.weight
F16
F16
[5120, 1024]
blk.28.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.28.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.28.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.28.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.28.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.28.ffn_norm.weight
F32
F32
[5120]
blk.28.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.28.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.29
blk.29.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.29.attn_norm.weight
F32
F32
[5120]
blk.29.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.29.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.29.attn_v.weight
F16
F16
[5120, 1024]
blk.29.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.29.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.29.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.29.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.29.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.29.ffn_norm.weight
F32
F32
[5120]
blk.29.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.29.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.30
blk.30.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.30.attn_norm.weight
F32
F32
[5120]
blk.30.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.30.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.30.attn_v.weight
F16
F16
[5120, 1024]
blk.30.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.30.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.30.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.30.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.30.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.30.ffn_norm.weight
F32
F32
[5120]
blk.30.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.30.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.31
blk.31.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.31.attn_norm.weight
F32
F32
[5120]
blk.31.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.31.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.31.attn_v.weight
F16
F16
[5120, 1024]
blk.31.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.31.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.31.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.31.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.31.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.31.ffn_norm.weight
F32
F32
[5120]
blk.31.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.31.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.32
blk.32.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.32.attn_norm.weight
F32
F32
[5120]
blk.32.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.32.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.32.attn_v.weight
F16
F16
[5120, 1024]
blk.32.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.32.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.32.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.32.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.32.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.32.ffn_norm.weight
F32
F32
[5120]
blk.32.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.32.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.33
blk.33.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.33.attn_norm.weight
F32
F32
[5120]
blk.33.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.33.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.33.attn_v.weight
F16
F16
[5120, 1024]
blk.33.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.33.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.33.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.33.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.33.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.33.ffn_norm.weight
F32
F32
[5120]
blk.33.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.33.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.34
blk.34.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.34.attn_norm.weight
F32
F32
[5120]
blk.34.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.34.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.34.attn_v.weight
F16
F16
[5120, 1024]
blk.34.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.34.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.34.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.34.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.34.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.34.ffn_norm.weight
F32
F32
[5120]
blk.34.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.34.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.35
blk.35.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.35.attn_norm.weight
F32
F32
[5120]
blk.35.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.35.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.35.attn_v.weight
F16
F16
[5120, 1024]
blk.35.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.35.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.35.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.35.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.35.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.35.ffn_norm.weight
F32
F32
[5120]
blk.35.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.35.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.36
blk.36.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.36.attn_norm.weight
F32
F32
[5120]
blk.36.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.36.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.36.attn_v.weight
F16
F16
[5120, 1024]
blk.36.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.36.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.36.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.36.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.36.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.36.ffn_norm.weight
F32
F32
[5120]
blk.36.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.36.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.37
blk.37.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.37.attn_norm.weight
F32
F32
[5120]
blk.37.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.37.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.37.attn_v.weight
F16
F16
[5120, 1024]
blk.37.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.37.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.37.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.37.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.37.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.37.ffn_norm.weight
F32
F32
[5120]
blk.37.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.37.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.38
blk.38.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.38.attn_norm.weight
F32
F32
[5120]
blk.38.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.38.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.38.attn_v.weight
F16
F16
[5120, 1024]
blk.38.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.38.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.38.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.38.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.38.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.38.ffn_norm.weight
F32
F32
[5120]
blk.38.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.38.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.39
blk.39.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.39.attn_norm.weight
F32
F32
[5120]
blk.39.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.39.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.39.attn_v.weight
F16
F16
[5120, 1024]
blk.39.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.39.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.39.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.39.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.39.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.39.ffn_norm.weight
F32
F32
[5120]
blk.39.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.39.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.40
blk.40.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.40.attn_norm.weight
F32
F32
[5120]
blk.40.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.40.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.40.attn_v.weight
F16
F16
[5120, 1024]
blk.40.ffn_down_exps.weight
Q4_K
Q4_K
[8192, 5120, 16]
blk.40.ffn_down_shexp.weight
Q4_K
Q4_K
[8192, 5120]
blk.40.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.40.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.40.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.40.ffn_norm.weight
F32
F32
[5120]
blk.40.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.40.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.41
blk.41.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.41.attn_norm.weight
F32
F32
[5120]
blk.41.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.41.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.41.attn_v.weight
F16
F16
[5120, 1024]
blk.41.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.41.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.41.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.41.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.41.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.41.ffn_norm.weight
F32
F32
[5120]
blk.41.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.41.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.42
blk.42.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.42.attn_norm.weight
F32
F32
[5120]
blk.42.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.42.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.42.attn_v.weight
F16
F16
[5120, 1024]
blk.42.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.42.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.42.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.42.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.42.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.42.ffn_norm.weight
F32
F32
[5120]
blk.42.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.42.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.43
blk.43.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.43.attn_norm.weight
F32
F32
[5120]
blk.43.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.43.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.43.attn_v.weight
F16
F16
[5120, 1024]
blk.43.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.43.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.43.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.43.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.43.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.43.ffn_norm.weight
F32
F32
[5120]
blk.43.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.43.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.44
blk.44.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.44.attn_norm.weight
F32
F32
[5120]
blk.44.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.44.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.44.attn_v.weight
F16
F16
[5120, 1024]
blk.44.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.44.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.44.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.44.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.44.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.44.ffn_norm.weight
F32
F32
[5120]
blk.44.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.44.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.45
blk.45.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.45.attn_norm.weight
F32
F32
[5120]
blk.45.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.45.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.45.attn_v.weight
F16
F16
[5120, 1024]
blk.45.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.45.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.45.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.45.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.45.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.45.ffn_norm.weight
F32
F32
[5120]
blk.45.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.45.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.46
blk.46.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.46.attn_norm.weight
F32
F32
[5120]
blk.46.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.46.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.46.attn_v.weight
F16
F16
[5120, 1024]
blk.46.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.46.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.46.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.46.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.46.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.46.ffn_norm.weight
F32
F32
[5120]
blk.46.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.46.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.47
blk.47.attn_k.weight
Q4_K
Q4_K
[5120, 1024]
blk.47.attn_norm.weight
F32
F32
[5120]
blk.47.attn_output.weight
Q4_K
Q4_K
[5120, 5120]
blk.47.attn_q.weight
Q4_K
Q4_K
[5120, 5120]
blk.47.attn_v.weight
F16
F16
[5120, 1024]
blk.47.ffn_down_exps.weight
Q6_K
Q6_K
[8192, 5120, 16]
blk.47.ffn_down_shexp.weight
Q6_K
Q6_K
[8192, 5120]
blk.47.ffn_gate_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.47.ffn_gate_inp.weight
F32
F32
[5120, 16]
blk.47.ffn_gate_shexp.weight
Q4_K
Q4_K
[5120, 8192]
blk.47.ffn_norm.weight
F32
F32
[5120]
blk.47.ffn_up_exps.weight
Q4_K
Q4_K
[5120, 8192, 16]
blk.47.ffn_up_shexp.weight
Q4_K
Q4_K
[5120, 8192]
mm.linear_1.weight
F16
F16
[4096, 5120]
output.weight
Q6_K
Q6_K
[5120, 202048]
rope_freqs.weight
F32
F32
[64]
v.blk.0
v.blk.0.attn_k.bias
F32
F32
[1408]
v.blk.0.attn_k.weight
F16
F16
[1408, 1408]
v.blk.0.attn_norm.bias
F32
F32
[1408]
v.blk.0.attn_norm.weight
F32
F32
[1408]
v.blk.0.attn_output.bias
F32
F32
[1408]
v.blk.0.attn_output.weight
F16
F16
[1408, 1408]
v.blk.0.attn_q.bias
F32
F32
[1408]
v.blk.0.attn_q.weight
F16
F16
[1408, 1408]
v.blk.0.attn_v.bias
F32
F32
[1408]
v.blk.0.attn_v.weight
F16
F16
[1408, 1408]
v.blk.0.ffn_norm.bias
F32
F32
[1408]
v.blk.0.ffn_norm.weight
F32
F32
[1408]
v.blk.0.mlp.fc1.bias
F32
F32
[5632]
v.blk.0.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.0.mlp.fc2.bias
F32
F32
[1408]
v.blk.0.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.1
v.blk.1.attn_k.bias
F32
F32
[1408]
v.blk.1.attn_k.weight
F16
F16
[1408, 1408]
v.blk.1.attn_norm.bias
F32
F32
[1408]
v.blk.1.attn_norm.weight
F32
F32
[1408]
v.blk.1.attn_output.bias
F32
F32
[1408]
v.blk.1.attn_output.weight
F16
F16
[1408, 1408]
v.blk.1.attn_q.bias
F32
F32
[1408]
v.blk.1.attn_q.weight
F16
F16
[1408, 1408]
v.blk.1.attn_v.bias
F32
F32
[1408]
v.blk.1.attn_v.weight
F16
F16
[1408, 1408]
v.blk.1.ffn_norm.bias
F32
F32
[1408]
v.blk.1.ffn_norm.weight
F32
F32
[1408]
v.blk.1.mlp.fc1.bias
F32
F32
[5632]
v.blk.1.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.1.mlp.fc2.bias
F32
F32
[1408]
v.blk.1.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.2
v.blk.2.attn_k.bias
F32
F32
[1408]
v.blk.2.attn_k.weight
F16
F16
[1408, 1408]
v.blk.2.attn_norm.bias
F32
F32
[1408]
v.blk.2.attn_norm.weight
F32
F32
[1408]
v.blk.2.attn_output.bias
F32
F32
[1408]
v.blk.2.attn_output.weight
F16
F16
[1408, 1408]
v.blk.2.attn_q.bias
F32
F32
[1408]
v.blk.2.attn_q.weight
F16
F16
[1408, 1408]
v.blk.2.attn_v.bias
F32
F32
[1408]
v.blk.2.attn_v.weight
F16
F16
[1408, 1408]
v.blk.2.ffn_norm.bias
F32
F32
[1408]
v.blk.2.ffn_norm.weight
F32
F32
[1408]
v.blk.2.mlp.fc1.bias
F32
F32
[5632]
v.blk.2.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.2.mlp.fc2.bias
F32
F32
[1408]
v.blk.2.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.3
v.blk.3.attn_k.bias
F32
F32
[1408]
v.blk.3.attn_k.weight
F16
F16
[1408, 1408]
v.blk.3.attn_norm.bias
F32
F32
[1408]
v.blk.3.attn_norm.weight
F32
F32
[1408]
v.blk.3.attn_output.bias
F32
F32
[1408]
v.blk.3.attn_output.weight
F16
F16
[1408, 1408]
v.blk.3.attn_q.bias
F32
F32
[1408]
v.blk.3.attn_q.weight
F16
F16
[1408, 1408]
v.blk.3.attn_v.bias
F32
F32
[1408]
v.blk.3.attn_v.weight
F16
F16
[1408, 1408]
v.blk.3.ffn_norm.bias
F32
F32
[1408]
v.blk.3.ffn_norm.weight
F32
F32
[1408]
v.blk.3.mlp.fc1.bias
F32
F32
[5632]
v.blk.3.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.3.mlp.fc2.bias
F32
F32
[1408]
v.blk.3.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.4
v.blk.4.attn_k.bias
F32
F32
[1408]
v.blk.4.attn_k.weight
F16
F16
[1408, 1408]
v.blk.4.attn_norm.bias
F32
F32
[1408]
v.blk.4.attn_norm.weight
F32
F32
[1408]
v.blk.4.attn_output.bias
F32
F32
[1408]
v.blk.4.attn_output.weight
F16
F16
[1408, 1408]
v.blk.4.attn_q.bias
F32
F32
[1408]
v.blk.4.attn_q.weight
F16
F16
[1408, 1408]
v.blk.4.attn_v.bias
F32
F32
[1408]
v.blk.4.attn_v.weight
F16
F16
[1408, 1408]
v.blk.4.ffn_norm.bias
F32
F32
[1408]
v.blk.4.ffn_norm.weight
F32
F32
[1408]
v.blk.4.mlp.fc1.bias
F32
F32
[5632]
v.blk.4.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.4.mlp.fc2.bias
F32
F32
[1408]
v.blk.4.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.5
v.blk.5.attn_k.bias
F32
F32
[1408]
v.blk.5.attn_k.weight
F16
F16
[1408, 1408]
v.blk.5.attn_norm.bias
F32
F32
[1408]
v.blk.5.attn_norm.weight
F32
F32
[1408]
v.blk.5.attn_output.bias
F32
F32
[1408]
v.blk.5.attn_output.weight
F16
F16
[1408, 1408]
v.blk.5.attn_q.bias
F32
F32
[1408]
v.blk.5.attn_q.weight
F16
F16
[1408, 1408]
v.blk.5.attn_v.bias
F32
F32
[1408]
v.blk.5.attn_v.weight
F16
F16
[1408, 1408]
v.blk.5.ffn_norm.bias
F32
F32
[1408]
v.blk.5.ffn_norm.weight
F32
F32
[1408]
v.blk.5.mlp.fc1.bias
F32
F32
[5632]
v.blk.5.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.5.mlp.fc2.bias
F32
F32
[1408]
v.blk.5.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.6
v.blk.6.attn_k.bias
F32
F32
[1408]
v.blk.6.attn_k.weight
F16
F16
[1408, 1408]
v.blk.6.attn_norm.bias
F32
F32
[1408]
v.blk.6.attn_norm.weight
F32
F32
[1408]
v.blk.6.attn_output.bias
F32
F32
[1408]
v.blk.6.attn_output.weight
F16
F16
[1408, 1408]
v.blk.6.attn_q.bias
F32
F32
[1408]
v.blk.6.attn_q.weight
F16
F16
[1408, 1408]
v.blk.6.attn_v.bias
F32
F32
[1408]
v.blk.6.attn_v.weight
F16
F16
[1408, 1408]
v.blk.6.ffn_norm.bias
F32
F32
[1408]
v.blk.6.ffn_norm.weight
F32
F32
[1408]
v.blk.6.mlp.fc1.bias
F32
F32
[5632]
v.blk.6.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.6.mlp.fc2.bias
F32
F32
[1408]
v.blk.6.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.7
v.blk.7.attn_k.bias
F32
F32
[1408]
v.blk.7.attn_k.weight
F16
F16
[1408, 1408]
v.blk.7.attn_norm.bias
F32
F32
[1408]
v.blk.7.attn_norm.weight
F32
F32
[1408]
v.blk.7.attn_output.bias
F32
F32
[1408]
v.blk.7.attn_output.weight
F16
F16
[1408, 1408]
v.blk.7.attn_q.bias
F32
F32
[1408]
v.blk.7.attn_q.weight
F16
F16
[1408, 1408]
v.blk.7.attn_v.bias
F32
F32
[1408]
v.blk.7.attn_v.weight
F16
F16
[1408, 1408]
v.blk.7.ffn_norm.bias
F32
F32
[1408]
v.blk.7.ffn_norm.weight
F32
F32
[1408]
v.blk.7.mlp.fc1.bias
F32
F32
[5632]
v.blk.7.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.7.mlp.fc2.bias
F32
F32
[1408]
v.blk.7.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.8
v.blk.8.attn_k.bias
F32
F32
[1408]
v.blk.8.attn_k.weight
F16
F16
[1408, 1408]
v.blk.8.attn_norm.bias
F32
F32
[1408]
v.blk.8.attn_norm.weight
F32
F32
[1408]
v.blk.8.attn_output.bias
F32
F32
[1408]
v.blk.8.attn_output.weight
F16
F16
[1408, 1408]
v.blk.8.attn_q.bias
F32
F32
[1408]
v.blk.8.attn_q.weight
F16
F16
[1408, 1408]
v.blk.8.attn_v.bias
F32
F32
[1408]
v.blk.8.attn_v.weight
F16
F16
[1408, 1408]
v.blk.8.ffn_norm.bias
F32
F32
[1408]
v.blk.8.ffn_norm.weight
F32
F32
[1408]
v.blk.8.mlp.fc1.bias
F32
F32
[5632]
v.blk.8.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.8.mlp.fc2.bias
F32
F32
[1408]
v.blk.8.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.9
v.blk.9.attn_k.bias
F32
F32
[1408]
v.blk.9.attn_k.weight
F16
F16
[1408, 1408]
v.blk.9.attn_norm.bias
F32
F32
[1408]
v.blk.9.attn_norm.weight
F32
F32
[1408]
v.blk.9.attn_output.bias
F32
F32
[1408]
v.blk.9.attn_output.weight
F16
F16
[1408, 1408]
v.blk.9.attn_q.bias
F32
F32
[1408]
v.blk.9.attn_q.weight
F16
F16
[1408, 1408]
v.blk.9.attn_v.bias
F32
F32
[1408]
v.blk.9.attn_v.weight
F16
F16
[1408, 1408]
v.blk.9.ffn_norm.bias
F32
F32
[1408]
v.blk.9.ffn_norm.weight
F32
F32
[1408]
v.blk.9.mlp.fc1.bias
F32
F32
[5632]
v.blk.9.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.9.mlp.fc2.bias
F32
F32
[1408]
v.blk.9.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.10
v.blk.10.attn_k.bias
F32
F32
[1408]
v.blk.10.attn_k.weight
F16
F16
[1408, 1408]
v.blk.10.attn_norm.bias
F32
F32
[1408]
v.blk.10.attn_norm.weight
F32
F32
[1408]
v.blk.10.attn_output.bias
F32
F32
[1408]
v.blk.10.attn_output.weight
F16
F16
[1408, 1408]
v.blk.10.attn_q.bias
F32
F32
[1408]
v.blk.10.attn_q.weight
F16
F16
[1408, 1408]
v.blk.10.attn_v.bias
F32
F32
[1408]
v.blk.10.attn_v.weight
F16
F16
[1408, 1408]
v.blk.10.ffn_norm.bias
F32
F32
[1408]
v.blk.10.ffn_norm.weight
F32
F32
[1408]
v.blk.10.mlp.fc1.bias
F32
F32
[5632]
v.blk.10.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.10.mlp.fc2.bias
F32
F32
[1408]
v.blk.10.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.11
v.blk.11.attn_k.bias
F32
F32
[1408]
v.blk.11.attn_k.weight
F16
F16
[1408, 1408]
v.blk.11.attn_norm.bias
F32
F32
[1408]
v.blk.11.attn_norm.weight
F32
F32
[1408]
v.blk.11.attn_output.bias
F32
F32
[1408]
v.blk.11.attn_output.weight
F16
F16
[1408, 1408]
v.blk.11.attn_q.bias
F32
F32
[1408]
v.blk.11.attn_q.weight
F16
F16
[1408, 1408]
v.blk.11.attn_v.bias
F32
F32
[1408]
v.blk.11.attn_v.weight
F16
F16
[1408, 1408]
v.blk.11.ffn_norm.bias
F32
F32
[1408]
v.blk.11.ffn_norm.weight
F32
F32
[1408]
v.blk.11.mlp.fc1.bias
F32
F32
[5632]
v.blk.11.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.11.mlp.fc2.bias
F32
F32
[1408]
v.blk.11.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.12
v.blk.12.attn_k.bias
F32
F32
[1408]
v.blk.12.attn_k.weight
F16
F16
[1408, 1408]
v.blk.12.attn_norm.bias
F32
F32
[1408]
v.blk.12.attn_norm.weight
F32
F32
[1408]
v.blk.12.attn_output.bias
F32
F32
[1408]
v.blk.12.attn_output.weight
F16
F16
[1408, 1408]
v.blk.12.attn_q.bias
F32
F32
[1408]
v.blk.12.attn_q.weight
F16
F16
[1408, 1408]
v.blk.12.attn_v.bias
F32
F32
[1408]
v.blk.12.attn_v.weight
F16
F16
[1408, 1408]
v.blk.12.ffn_norm.bias
F32
F32
[1408]
v.blk.12.ffn_norm.weight
F32
F32
[1408]
v.blk.12.mlp.fc1.bias
F32
F32
[5632]
v.blk.12.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.12.mlp.fc2.bias
F32
F32
[1408]
v.blk.12.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.13
v.blk.13.attn_k.bias
F32
F32
[1408]
v.blk.13.attn_k.weight
F16
F16
[1408, 1408]
v.blk.13.attn_norm.bias
F32
F32
[1408]
v.blk.13.attn_norm.weight
F32
F32
[1408]
v.blk.13.attn_output.bias
F32
F32
[1408]
v.blk.13.attn_output.weight
F16
F16
[1408, 1408]
v.blk.13.attn_q.bias
F32
F32
[1408]
v.blk.13.attn_q.weight
F16
F16
[1408, 1408]
v.blk.13.attn_v.bias
F32
F32
[1408]
v.blk.13.attn_v.weight
F16
F16
[1408, 1408]
v.blk.13.ffn_norm.bias
F32
F32
[1408]
v.blk.13.ffn_norm.weight
F32
F32
[1408]
v.blk.13.mlp.fc1.bias
F32
F32
[5632]
v.blk.13.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.13.mlp.fc2.bias
F32
F32
[1408]
v.blk.13.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.14
v.blk.14.attn_k.bias
F32
F32
[1408]
v.blk.14.attn_k.weight
F16
F16
[1408, 1408]
v.blk.14.attn_norm.bias
F32
F32
[1408]
v.blk.14.attn_norm.weight
F32
F32
[1408]
v.blk.14.attn_output.bias
F32
F32
[1408]
v.blk.14.attn_output.weight
F16
F16
[1408, 1408]
v.blk.14.attn_q.bias
F32
F32
[1408]
v.blk.14.attn_q.weight
F16
F16
[1408, 1408]
v.blk.14.attn_v.bias
F32
F32
[1408]
v.blk.14.attn_v.weight
F16
F16
[1408, 1408]
v.blk.14.ffn_norm.bias
F32
F32
[1408]
v.blk.14.ffn_norm.weight
F32
F32
[1408]
v.blk.14.mlp.fc1.bias
F32
F32
[5632]
v.blk.14.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.14.mlp.fc2.bias
F32
F32
[1408]
v.blk.14.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.15
v.blk.15.attn_k.bias
F32
F32
[1408]
v.blk.15.attn_k.weight
F16
F16
[1408, 1408]
v.blk.15.attn_norm.bias
F32
F32
[1408]
v.blk.15.attn_norm.weight
F32
F32
[1408]
v.blk.15.attn_output.bias
F32
F32
[1408]
v.blk.15.attn_output.weight
F16
F16
[1408, 1408]
v.blk.15.attn_q.bias
F32
F32
[1408]
v.blk.15.attn_q.weight
F16
F16
[1408, 1408]
v.blk.15.attn_v.bias
F32
F32
[1408]
v.blk.15.attn_v.weight
F16
F16
[1408, 1408]
v.blk.15.ffn_norm.bias
F32
F32
[1408]
v.blk.15.ffn_norm.weight
F32
F32
[1408]
v.blk.15.mlp.fc1.bias
F32
F32
[5632]
v.blk.15.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.15.mlp.fc2.bias
F32
F32
[1408]
v.blk.15.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.16
v.blk.16.attn_k.bias
F32
F32
[1408]
v.blk.16.attn_k.weight
F16
F16
[1408, 1408]
v.blk.16.attn_norm.bias
F32
F32
[1408]
v.blk.16.attn_norm.weight
F32
F32
[1408]
v.blk.16.attn_output.bias
F32
F32
[1408]
v.blk.16.attn_output.weight
F16
F16
[1408, 1408]
v.blk.16.attn_q.bias
F32
F32
[1408]
v.blk.16.attn_q.weight
F16
F16
[1408, 1408]
v.blk.16.attn_v.bias
F32
F32
[1408]
v.blk.16.attn_v.weight
F16
F16
[1408, 1408]
v.blk.16.ffn_norm.bias
F32
F32
[1408]
v.blk.16.ffn_norm.weight
F32
F32
[1408]
v.blk.16.mlp.fc1.bias
F32
F32
[5632]
v.blk.16.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.16.mlp.fc2.bias
F32
F32
[1408]
v.blk.16.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.17
v.blk.17.attn_k.bias
F32
F32
[1408]
v.blk.17.attn_k.weight
F16
F16
[1408, 1408]
v.blk.17.attn_norm.bias
F32
F32
[1408]
v.blk.17.attn_norm.weight
F32
F32
[1408]
v.blk.17.attn_output.bias
F32
F32
[1408]
v.blk.17.attn_output.weight
F16
F16
[1408, 1408]
v.blk.17.attn_q.bias
F32
F32
[1408]
v.blk.17.attn_q.weight
F16
F16
[1408, 1408]
v.blk.17.attn_v.bias
F32
F32
[1408]
v.blk.17.attn_v.weight
F16
F16
[1408, 1408]
v.blk.17.ffn_norm.bias
F32
F32
[1408]
v.blk.17.ffn_norm.weight
F32
F32
[1408]
v.blk.17.mlp.fc1.bias
F32
F32
[5632]
v.blk.17.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.17.mlp.fc2.bias
F32
F32
[1408]
v.blk.17.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.18
v.blk.18.attn_k.bias
F32
F32
[1408]
v.blk.18.attn_k.weight
F16
F16
[1408, 1408]
v.blk.18.attn_norm.bias
F32
F32
[1408]
v.blk.18.attn_norm.weight
F32
F32
[1408]
v.blk.18.attn_output.bias
F32
F32
[1408]
v.blk.18.attn_output.weight
F16
F16
[1408, 1408]
v.blk.18.attn_q.bias
F32
F32
[1408]
v.blk.18.attn_q.weight
F16
F16
[1408, 1408]
v.blk.18.attn_v.bias
F32
F32
[1408]
v.blk.18.attn_v.weight
F16
F16
[1408, 1408]
v.blk.18.ffn_norm.bias
F32
F32
[1408]
v.blk.18.ffn_norm.weight
F32
F32
[1408]
v.blk.18.mlp.fc1.bias
F32
F32
[5632]
v.blk.18.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.18.mlp.fc2.bias
F32
F32
[1408]
v.blk.18.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.19
v.blk.19.attn_k.bias
F32
F32
[1408]
v.blk.19.attn_k.weight
F16
F16
[1408, 1408]
v.blk.19.attn_norm.bias
F32
F32
[1408]
v.blk.19.attn_norm.weight
F32
F32
[1408]
v.blk.19.attn_output.bias
F32
F32
[1408]
v.blk.19.attn_output.weight
F16
F16
[1408, 1408]
v.blk.19.attn_q.bias
F32
F32
[1408]
v.blk.19.attn_q.weight
F16
F16
[1408, 1408]
v.blk.19.attn_v.bias
F32
F32
[1408]
v.blk.19.attn_v.weight
F16
F16
[1408, 1408]
v.blk.19.ffn_norm.bias
F32
F32
[1408]
v.blk.19.ffn_norm.weight
F32
F32
[1408]
v.blk.19.mlp.fc1.bias
F32
F32
[5632]
v.blk.19.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.19.mlp.fc2.bias
F32
F32
[1408]
v.blk.19.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.20
v.blk.20.attn_k.bias
F32
F32
[1408]
v.blk.20.attn_k.weight
F16
F16
[1408, 1408]
v.blk.20.attn_norm.bias
F32
F32
[1408]
v.blk.20.attn_norm.weight
F32
F32
[1408]
v.blk.20.attn_output.bias
F32
F32
[1408]
v.blk.20.attn_output.weight
F16
F16
[1408, 1408]
v.blk.20.attn_q.bias
F32
F32
[1408]
v.blk.20.attn_q.weight
F16
F16
[1408, 1408]
v.blk.20.attn_v.bias
F32
F32
[1408]
v.blk.20.attn_v.weight
F16
F16
[1408, 1408]
v.blk.20.ffn_norm.bias
F32
F32
[1408]
v.blk.20.ffn_norm.weight
F32
F32
[1408]
v.blk.20.mlp.fc1.bias
F32
F32
[5632]
v.blk.20.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.20.mlp.fc2.bias
F32
F32
[1408]
v.blk.20.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.21
v.blk.21.attn_k.bias
F32
F32
[1408]
v.blk.21.attn_k.weight
F16
F16
[1408, 1408]
v.blk.21.attn_norm.bias
F32
F32
[1408]
v.blk.21.attn_norm.weight
F32
F32
[1408]
v.blk.21.attn_output.bias
F32
F32
[1408]
v.blk.21.attn_output.weight
F16
F16
[1408, 1408]
v.blk.21.attn_q.bias
F32
F32
[1408]
v.blk.21.attn_q.weight
F16
F16
[1408, 1408]
v.blk.21.attn_v.bias
F32
F32
[1408]
v.blk.21.attn_v.weight
F16
F16
[1408, 1408]
v.blk.21.ffn_norm.bias
F32
F32
[1408]
v.blk.21.ffn_norm.weight
F32
F32
[1408]
v.blk.21.mlp.fc1.bias
F32
F32
[5632]
v.blk.21.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.21.mlp.fc2.bias
F32
F32
[1408]
v.blk.21.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.22
v.blk.22.attn_k.bias
F32
F32
[1408]
v.blk.22.attn_k.weight
F16
F16
[1408, 1408]
v.blk.22.attn_norm.bias
F32
F32
[1408]
v.blk.22.attn_norm.weight
F32
F32
[1408]
v.blk.22.attn_output.bias
F32
F32
[1408]
v.blk.22.attn_output.weight
F16
F16
[1408, 1408]
v.blk.22.attn_q.bias
F32
F32
[1408]
v.blk.22.attn_q.weight
F16
F16
[1408, 1408]
v.blk.22.attn_v.bias
F32
F32
[1408]
v.blk.22.attn_v.weight
F16
F16
[1408, 1408]
v.blk.22.ffn_norm.bias
F32
F32
[1408]
v.blk.22.ffn_norm.weight
F32
F32
[1408]
v.blk.22.mlp.fc1.bias
F32
F32
[5632]
v.blk.22.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.22.mlp.fc2.bias
F32
F32
[1408]
v.blk.22.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.23
v.blk.23.attn_k.bias
F32
F32
[1408]
v.blk.23.attn_k.weight
F16
F16
[1408, 1408]
v.blk.23.attn_norm.bias
F32
F32
[1408]
v.blk.23.attn_norm.weight
F32
F32
[1408]
v.blk.23.attn_output.bias
F32
F32
[1408]
v.blk.23.attn_output.weight
F16
F16
[1408, 1408]
v.blk.23.attn_q.bias
F32
F32
[1408]
v.blk.23.attn_q.weight
F16
F16
[1408, 1408]
v.blk.23.attn_v.bias
F32
F32
[1408]
v.blk.23.attn_v.weight
F16
F16
[1408, 1408]
v.blk.23.ffn_norm.bias
F32
F32
[1408]
v.blk.23.ffn_norm.weight
F32
F32
[1408]
v.blk.23.mlp.fc1.bias
F32
F32
[5632]
v.blk.23.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.23.mlp.fc2.bias
F32
F32
[1408]
v.blk.23.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.24
v.blk.24.attn_k.bias
F32
F32
[1408]
v.blk.24.attn_k.weight
F16
F16
[1408, 1408]
v.blk.24.attn_norm.bias
F32
F32
[1408]
v.blk.24.attn_norm.weight
F32
F32
[1408]
v.blk.24.attn_output.bias
F32
F32
[1408]
v.blk.24.attn_output.weight
F16
F16
[1408, 1408]
v.blk.24.attn_q.bias
F32
F32
[1408]
v.blk.24.attn_q.weight
F16
F16
[1408, 1408]
v.blk.24.attn_v.bias
F32
F32
[1408]
v.blk.24.attn_v.weight
F16
F16
[1408, 1408]
v.blk.24.ffn_norm.bias
F32
F32
[1408]
v.blk.24.ffn_norm.weight
F32
F32
[1408]
v.blk.24.mlp.fc1.bias
F32
F32
[5632]
v.blk.24.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.24.mlp.fc2.bias
F32
F32
[1408]
v.blk.24.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.25
v.blk.25.attn_k.bias
F32
F32
[1408]
v.blk.25.attn_k.weight
F16
F16
[1408, 1408]
v.blk.25.attn_norm.bias
F32
F32
[1408]
v.blk.25.attn_norm.weight
F32
F32
[1408]
v.blk.25.attn_output.bias
F32
F32
[1408]
v.blk.25.attn_output.weight
F16
F16
[1408, 1408]
v.blk.25.attn_q.bias
F32
F32
[1408]
v.blk.25.attn_q.weight
F16
F16
[1408, 1408]
v.blk.25.attn_v.bias
F32
F32
[1408]
v.blk.25.attn_v.weight
F16
F16
[1408, 1408]
v.blk.25.ffn_norm.bias
F32
F32
[1408]
v.blk.25.ffn_norm.weight
F32
F32
[1408]
v.blk.25.mlp.fc1.bias
F32
F32
[5632]
v.blk.25.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.25.mlp.fc2.bias
F32
F32
[1408]
v.blk.25.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.26
v.blk.26.attn_k.bias
F32
F32
[1408]
v.blk.26.attn_k.weight
F16
F16
[1408, 1408]
v.blk.26.attn_norm.bias
F32
F32
[1408]
v.blk.26.attn_norm.weight
F32
F32
[1408]
v.blk.26.attn_output.bias
F32
F32
[1408]
v.blk.26.attn_output.weight
F16
F16
[1408, 1408]
v.blk.26.attn_q.bias
F32
F32
[1408]
v.blk.26.attn_q.weight
F16
F16
[1408, 1408]
v.blk.26.attn_v.bias
F32
F32
[1408]
v.blk.26.attn_v.weight
F16
F16
[1408, 1408]
v.blk.26.ffn_norm.bias
F32
F32
[1408]
v.blk.26.ffn_norm.weight
F32
F32
[1408]
v.blk.26.mlp.fc1.bias
F32
F32
[5632]
v.blk.26.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.26.mlp.fc2.bias
F32
F32
[1408]
v.blk.26.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.27
v.blk.27.attn_k.bias
F32
F32
[1408]
v.blk.27.attn_k.weight
F16
F16
[1408, 1408]
v.blk.27.attn_norm.bias
F32
F32
[1408]
v.blk.27.attn_norm.weight
F32
F32
[1408]
v.blk.27.attn_output.bias
F32
F32
[1408]
v.blk.27.attn_output.weight
F16
F16
[1408, 1408]
v.blk.27.attn_q.bias
F32
F32
[1408]
v.blk.27.attn_q.weight
F16
F16
[1408, 1408]
v.blk.27.attn_v.bias
F32
F32
[1408]
v.blk.27.attn_v.weight
F16
F16
[1408, 1408]
v.blk.27.ffn_norm.bias
F32
F32
[1408]
v.blk.27.ffn_norm.weight
F32
F32
[1408]
v.blk.27.mlp.fc1.bias
F32
F32
[5632]
v.blk.27.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.27.mlp.fc2.bias
F32
F32
[1408]
v.blk.27.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.28
v.blk.28.attn_k.bias
F32
F32
[1408]
v.blk.28.attn_k.weight
F16
F16
[1408, 1408]
v.blk.28.attn_norm.bias
F32
F32
[1408]
v.blk.28.attn_norm.weight
F32
F32
[1408]
v.blk.28.attn_output.bias
F32
F32
[1408]
v.blk.28.attn_output.weight
F16
F16
[1408, 1408]
v.blk.28.attn_q.bias
F32
F32
[1408]
v.blk.28.attn_q.weight
F16
F16
[1408, 1408]
v.blk.28.attn_v.bias
F32
F32
[1408]
v.blk.28.attn_v.weight
F16
F16
[1408, 1408]
v.blk.28.ffn_norm.bias
F32
F32
[1408]
v.blk.28.ffn_norm.weight
F32
F32
[1408]
v.blk.28.mlp.fc1.bias
F32
F32
[5632]
v.blk.28.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.28.mlp.fc2.bias
F32
F32
[1408]
v.blk.28.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.29
v.blk.29.attn_k.bias
F32
F32
[1408]
v.blk.29.attn_k.weight
F16
F16
[1408, 1408]
v.blk.29.attn_norm.bias
F32
F32
[1408]
v.blk.29.attn_norm.weight
F32
F32
[1408]
v.blk.29.attn_output.bias
F32
F32
[1408]
v.blk.29.attn_output.weight
F16
F16
[1408, 1408]
v.blk.29.attn_q.bias
F32
F32
[1408]
v.blk.29.attn_q.weight
F16
F16
[1408, 1408]
v.blk.29.attn_v.bias
F32
F32
[1408]
v.blk.29.attn_v.weight
F16
F16
[1408, 1408]
v.blk.29.ffn_norm.bias
F32
F32
[1408]
v.blk.29.ffn_norm.weight
F32
F32
[1408]
v.blk.29.mlp.fc1.bias
F32
F32
[5632]
v.blk.29.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.29.mlp.fc2.bias
F32
F32
[1408]
v.blk.29.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.30
v.blk.30.attn_k.bias
F32
F32
[1408]
v.blk.30.attn_k.weight
F16
F16
[1408, 1408]
v.blk.30.attn_norm.bias
F32
F32
[1408]
v.blk.30.attn_norm.weight
F32
F32
[1408]
v.blk.30.attn_output.bias
F32
F32
[1408]
v.blk.30.attn_output.weight
F16
F16
[1408, 1408]
v.blk.30.attn_q.bias
F32
F32
[1408]
v.blk.30.attn_q.weight
F16
F16
[1408, 1408]
v.blk.30.attn_v.bias
F32
F32
[1408]
v.blk.30.attn_v.weight
F16
F16
[1408, 1408]
v.blk.30.ffn_norm.bias
F32
F32
[1408]
v.blk.30.ffn_norm.weight
F32
F32
[1408]
v.blk.30.mlp.fc1.bias
F32
F32
[5632]
v.blk.30.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.30.mlp.fc2.bias
F32
F32
[1408]
v.blk.30.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.31
v.blk.31.attn_k.bias
F32
F32
[1408]
v.blk.31.attn_k.weight
F16
F16
[1408, 1408]
v.blk.31.attn_norm.bias
F32
F32
[1408]
v.blk.31.attn_norm.weight
F32
F32
[1408]
v.blk.31.attn_output.bias
F32
F32
[1408]
v.blk.31.attn_output.weight
F16
F16
[1408, 1408]
v.blk.31.attn_q.bias
F32
F32
[1408]
v.blk.31.attn_q.weight
F16
F16
[1408, 1408]
v.blk.31.attn_v.bias
F32
F32
[1408]
v.blk.31.attn_v.weight
F16
F16
[1408, 1408]
v.blk.31.ffn_norm.bias
F32
F32
[1408]
v.blk.31.ffn_norm.weight
F32
F32
[1408]
v.blk.31.mlp.fc1.bias
F32
F32
[5632]
v.blk.31.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.31.mlp.fc2.bias
F32
F32
[1408]
v.blk.31.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.32
v.blk.32.attn_k.bias
F32
F32
[1408]
v.blk.32.attn_k.weight
F16
F16
[1408, 1408]
v.blk.32.attn_norm.bias
F32
F32
[1408]
v.blk.32.attn_norm.weight
F32
F32
[1408]
v.blk.32.attn_output.bias
F32
F32
[1408]
v.blk.32.attn_output.weight
F16
F16
[1408, 1408]
v.blk.32.attn_q.bias
F32
F32
[1408]
v.blk.32.attn_q.weight
F16
F16
[1408, 1408]
v.blk.32.attn_v.bias
F32
F32
[1408]
v.blk.32.attn_v.weight
F16
F16
[1408, 1408]
v.blk.32.ffn_norm.bias
F32
F32
[1408]
v.blk.32.ffn_norm.weight
F32
F32
[1408]
v.blk.32.mlp.fc1.bias
F32
F32
[5632]
v.blk.32.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.32.mlp.fc2.bias
F32
F32
[1408]
v.blk.32.mlp.fc2.weight
F16
F16
[5632, 1408]
v.blk.33
v.blk.33.attn_k.bias
F32
F32
[1408]
v.blk.33.attn_k.weight
F16
F16
[1408, 1408]
v.blk.33.attn_norm.bias
F32
F32
[1408]
v.blk.33.attn_norm.weight
F32
F32
[1408]
v.blk.33.attn_output.bias
F32
F32
[1408]
v.blk.33.attn_output.weight
F16
F16
[1408, 1408]
v.blk.33.attn_q.bias
F32
F32
[1408]
v.blk.33.attn_q.weight
F16
F16
[1408, 1408]
v.blk.33.attn_v.bias
F32
F32
[1408]
v.blk.33.attn_v.weight
F16
F16
[1408, 1408]
v.blk.33.ffn_norm.bias
F32
F32
[1408]
v.blk.33.ffn_norm.weight
F32
F32
[1408]
v.blk.33.mlp.fc1.bias
F32
F32
[5632]
v.blk.33.mlp.fc1.weight
F16
F16
[1408, 5632]
v.blk.33.mlp.fc2.bias
F32
F32
[1408]
v.blk.33.mlp.fc2.weight
F16
F16
[5632, 1408]
v.class_embedding
F32
F32
[1408]
v.layernorm_post.bias
F32
F32
[1408]
v.layernorm_post.weight
F32
F32
[1408]
v.layernorm_pre.bias
F32
F32
[1408]
v.layernorm_pre.weight
F32
F32
[1408]
v.patch_embedding.weight
F16
F16
[588, 1408]
v.positional_embedding_vlm
F32
F32
[1408, 577]
v.vision_adapter.mlp.fc1.weight
F16
F16
[5632, 4096]
v.vision_adapter.mlp.fc2.weight
F16
F16
[4096, 4096]
output_norm.weight
F32
F32
[5120]