Models
Turbo
Sign in
Download
Models
Download
Sign in
llama3.2-vision
:90b-instruct-q4_K_M
2.3M
Downloads
Updated
2 months ago
Llama 3.2 Vision is a collection of instruction-tuned image reasoning generative models in 11B and 90B sizes.
Llama 3.2 Vision is a collection of instruction-tuned image reasoning generative models in 11B and 90B sizes.
Cancel
vision
11b
90b
llama3.2-vision:90b-instruct-q4_K_M
...
/
model
cbf2ef8a6e09 · 55GB
Metadata
general.architecture
mllama
mllama
general.file_type
Q4_K_M
Q4_K_M
mllama.attention.cross_attention_layers
[3, 8, 13, 18, 23, ...]
[3, 8, 13, 18, 23, ...]
mllama.attention.head_count
64
64
mllama.attention.head_count_kv
8
8
mllama.attention.layer_norm_rms_epsilon
1e-05
1e-05
mllama.block_count
100
100
mllama.context_length
131072
131072
mllama.embedding_length
8192
8192
mllama.feed_forward_length
28672
28672
mllama.rope.dimension_count
128
128
mllama.rope.freq_base
500000
500000
mllama.vision.attention.head_count
16
16
mllama.vision.attention.layer_norm_epsilon
1e-05
1e-05
mllama.vision.block_count
32
32
mllama.vision.embedding_length
1280
1280
mllama.vision.feed_forward_length
5120
5120
mllama.vision.global.block_count
8
8
mllama.vision.image_size
560
560
mllama.vision.intermediate_layers_indices
[3, 7, 15, 23, 30]
[3, 7, 15, 23, 30]
mllama.vision.max_num_tiles
4
4
mllama.vision.num_channels
3
3
mllama.vision.patch_size
14
14
mllama.vocab_size
128256
128256
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
128000
128000
tokenizer.ggml.eos_token_id
128009
128009
tokenizer.ggml.eos_token_ids
[128001, 128008, 128009]
[128001, 128008, 128009]
tokenizer.ggml.merges
[Ġ Ġ, Ġ ĠĠĠ, ĠĠ ĠĠ, ĠĠĠ Ġ, i n, ...]
[Ġ Ġ, Ġ ĠĠĠ, ĠĠ ĠĠ, ĠĠĠ Ġ, i n, ...]
tokenizer.ggml.model
gpt2
gpt2
tokenizer.ggml.padding_token_id
128004
128004
tokenizer.ggml.pre
llama-bpe
llama-bpe
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
[!, ", #, $, %, ...]
[!, ", #, $, %, ...]
Tensor
Name
Type
Shape
token_embd.weight
Q4_K
Q4_K
[8192, 128264]
blk.0
blk.0.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.0.attn_norm.weight
F32
F32
[8192]
blk.0.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.0.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.0.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.0.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.0.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.0.ffn_norm.weight
F32
F32
[8192]
blk.0.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.1
blk.1.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.1.attn_norm.weight
F32
F32
[8192]
blk.1.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.1.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.1.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.1.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.1.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.1.ffn_norm.weight
F32
F32
[8192]
blk.1.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.2
blk.2.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.2.attn_norm.weight
F32
F32
[8192]
blk.2.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.2.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.2.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.2.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.2.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.2.ffn_norm.weight
F32
F32
[8192]
blk.2.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.3
blk.3.attn_norm.weight
F32
F32
[8192]
blk.3.cross_attn_attn_gate
F32
F32
[1]
blk.3.cross_attn_k_norm.weight
F32
F32
[128]
blk.3.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.3.cross_attn_mlp_gate
F32
F32
[1]
blk.3.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.3.cross_attn_q_norm.weight
F32
F32
[128]
blk.3.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.3.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.3.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.3.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.3.ffn_norm.weight
F32
F32
[8192]
blk.3.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.4
blk.4.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.4.attn_norm.weight
F32
F32
[8192]
blk.4.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.4.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.4.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.4.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.4.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.4.ffn_norm.weight
F32
F32
[8192]
blk.4.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.5
blk.5.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.5.attn_norm.weight
F32
F32
[8192]
blk.5.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.5.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.5.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.5.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.5.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.5.ffn_norm.weight
F32
F32
[8192]
blk.5.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.6
blk.6.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.6.attn_norm.weight
F32
F32
[8192]
blk.6.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.6.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.6.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.6.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.6.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.6.ffn_norm.weight
F32
F32
[8192]
blk.6.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.7
blk.7.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.7.attn_norm.weight
F32
F32
[8192]
blk.7.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.7.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.7.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.7.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.7.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.7.ffn_norm.weight
F32
F32
[8192]
blk.7.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.8
blk.8.attn_norm.weight
F32
F32
[8192]
blk.8.cross_attn_attn_gate
F32
F32
[1]
blk.8.cross_attn_k_norm.weight
F32
F32
[128]
blk.8.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.8.cross_attn_mlp_gate
F32
F32
[1]
blk.8.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.8.cross_attn_q_norm.weight
F32
F32
[128]
blk.8.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.8.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.8.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.8.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.8.ffn_norm.weight
F32
F32
[8192]
blk.8.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.9
blk.9.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.9.attn_norm.weight
F32
F32
[8192]
blk.9.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.9.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.9.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.9.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.9.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.9.ffn_norm.weight
F32
F32
[8192]
blk.9.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.10
blk.10.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.10.attn_norm.weight
F32
F32
[8192]
blk.10.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.10.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.10.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.10.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.10.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.10.ffn_norm.weight
F32
F32
[8192]
blk.10.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.11
blk.11.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.11.attn_norm.weight
F32
F32
[8192]
blk.11.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.11.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.11.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.11.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.11.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.11.ffn_norm.weight
F32
F32
[8192]
blk.11.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.12
blk.12.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.12.attn_norm.weight
F32
F32
[8192]
blk.12.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.12.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.12.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.12.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.12.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.12.ffn_norm.weight
F32
F32
[8192]
blk.12.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.13
blk.13.attn_norm.weight
F32
F32
[8192]
blk.13.cross_attn_attn_gate
F32
F32
[1]
blk.13.cross_attn_k_norm.weight
F32
F32
[128]
blk.13.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.13.cross_attn_mlp_gate
F32
F32
[1]
blk.13.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.13.cross_attn_q_norm.weight
F32
F32
[128]
blk.13.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.13.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.13.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.13.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.13.ffn_norm.weight
F32
F32
[8192]
blk.13.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.14
blk.14.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.14.attn_norm.weight
F32
F32
[8192]
blk.14.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.14.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.14.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.14.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.14.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.14.ffn_norm.weight
F32
F32
[8192]
blk.14.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.15
blk.15.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.15.attn_norm.weight
F32
F32
[8192]
blk.15.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.15.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.15.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.15.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.15.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.15.ffn_norm.weight
F32
F32
[8192]
blk.15.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.16
blk.16.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.16.attn_norm.weight
F32
F32
[8192]
blk.16.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.16.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.16.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.16.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.16.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.16.ffn_norm.weight
F32
F32
[8192]
blk.16.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.17
blk.17.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.17.attn_norm.weight
F32
F32
[8192]
blk.17.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.17.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.17.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.17.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.17.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.17.ffn_norm.weight
F32
F32
[8192]
blk.17.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.18
blk.18.attn_norm.weight
F32
F32
[8192]
blk.18.cross_attn_attn_gate
F32
F32
[1]
blk.18.cross_attn_k_norm.weight
F32
F32
[128]
blk.18.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.18.cross_attn_mlp_gate
F32
F32
[1]
blk.18.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.18.cross_attn_q_norm.weight
F32
F32
[128]
blk.18.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.18.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.18.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.18.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.18.ffn_norm.weight
F32
F32
[8192]
blk.18.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.19
blk.19.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.19.attn_norm.weight
F32
F32
[8192]
blk.19.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.19.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.19.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.19.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.19.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.19.ffn_norm.weight
F32
F32
[8192]
blk.19.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.20
blk.20.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.20.attn_norm.weight
F32
F32
[8192]
blk.20.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.20.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.20.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.20.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.20.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.20.ffn_norm.weight
F32
F32
[8192]
blk.20.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.21
blk.21.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.21.attn_norm.weight
F32
F32
[8192]
blk.21.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.21.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.21.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.21.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.21.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.21.ffn_norm.weight
F32
F32
[8192]
blk.21.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.22
blk.22.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.22.attn_norm.weight
F32
F32
[8192]
blk.22.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.22.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.22.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.22.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.22.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.22.ffn_norm.weight
F32
F32
[8192]
blk.22.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.23
blk.23.attn_norm.weight
F32
F32
[8192]
blk.23.cross_attn_attn_gate
F32
F32
[1]
blk.23.cross_attn_k_norm.weight
F32
F32
[128]
blk.23.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.23.cross_attn_mlp_gate
F32
F32
[1]
blk.23.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.23.cross_attn_q_norm.weight
F32
F32
[128]
blk.23.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.23.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.23.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.23.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.23.ffn_norm.weight
F32
F32
[8192]
blk.23.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.24
blk.24.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.24.attn_norm.weight
F32
F32
[8192]
blk.24.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.24.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.24.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.24.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.24.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.24.ffn_norm.weight
F32
F32
[8192]
blk.24.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.25
blk.25.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.25.attn_norm.weight
F32
F32
[8192]
blk.25.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.25.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.25.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.25.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.25.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.25.ffn_norm.weight
F32
F32
[8192]
blk.25.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.26
blk.26.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.26.attn_norm.weight
F32
F32
[8192]
blk.26.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.26.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.26.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.26.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.26.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.26.ffn_norm.weight
F32
F32
[8192]
blk.26.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.27
blk.27.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.27.attn_norm.weight
F32
F32
[8192]
blk.27.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.27.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.27.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.27.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.27.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.27.ffn_norm.weight
F32
F32
[8192]
blk.27.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.28
blk.28.attn_norm.weight
F32
F32
[8192]
blk.28.cross_attn_attn_gate
F32
F32
[1]
blk.28.cross_attn_k_norm.weight
F32
F32
[128]
blk.28.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.28.cross_attn_mlp_gate
F32
F32
[1]
blk.28.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.28.cross_attn_q_norm.weight
F32
F32
[128]
blk.28.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.28.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.28.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.28.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.28.ffn_norm.weight
F32
F32
[8192]
blk.28.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.29
blk.29.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.29.attn_norm.weight
F32
F32
[8192]
blk.29.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.29.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.29.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.29.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.29.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.29.ffn_norm.weight
F32
F32
[8192]
blk.29.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.30
blk.30.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.30.attn_norm.weight
F32
F32
[8192]
blk.30.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.30.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.30.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.30.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.30.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.30.ffn_norm.weight
F32
F32
[8192]
blk.30.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.31
blk.31.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.31.attn_norm.weight
F32
F32
[8192]
blk.31.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.31.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.31.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.31.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.31.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.31.ffn_norm.weight
F32
F32
[8192]
blk.31.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.32
blk.32.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.32.attn_norm.weight
F32
F32
[8192]
blk.32.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.32.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.32.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.32.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.32.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.32.ffn_norm.weight
F32
F32
[8192]
blk.32.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.33
blk.33.attn_norm.weight
F32
F32
[8192]
blk.33.cross_attn_attn_gate
F32
F32
[1]
blk.33.cross_attn_k_norm.weight
F32
F32
[128]
blk.33.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.33.cross_attn_mlp_gate
F32
F32
[1]
blk.33.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.33.cross_attn_q_norm.weight
F32
F32
[128]
blk.33.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.33.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.33.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.33.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.33.ffn_norm.weight
F32
F32
[8192]
blk.33.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.34
blk.34.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.34.attn_norm.weight
F32
F32
[8192]
blk.34.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.34.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.34.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.34.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.34.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.34.ffn_norm.weight
F32
F32
[8192]
blk.34.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.35
blk.35.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.35.attn_norm.weight
F32
F32
[8192]
blk.35.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.35.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.35.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.35.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.35.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.35.ffn_norm.weight
F32
F32
[8192]
blk.35.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.36
blk.36.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.36.attn_norm.weight
F32
F32
[8192]
blk.36.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.36.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.36.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.36.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.36.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.36.ffn_norm.weight
F32
F32
[8192]
blk.36.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.37
blk.37.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.37.attn_norm.weight
F32
F32
[8192]
blk.37.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.37.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.37.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.37.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.37.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.37.ffn_norm.weight
F32
F32
[8192]
blk.37.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.38
blk.38.attn_norm.weight
F32
F32
[8192]
blk.38.cross_attn_attn_gate
F32
F32
[1]
blk.38.cross_attn_k_norm.weight
F32
F32
[128]
blk.38.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.38.cross_attn_mlp_gate
F32
F32
[1]
blk.38.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.38.cross_attn_q_norm.weight
F32
F32
[128]
blk.38.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.38.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.38.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.38.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.38.ffn_norm.weight
F32
F32
[8192]
blk.38.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.39
blk.39.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.39.attn_norm.weight
F32
F32
[8192]
blk.39.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.39.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.39.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.39.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.39.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.39.ffn_norm.weight
F32
F32
[8192]
blk.39.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.40
blk.40.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.40.attn_norm.weight
F32
F32
[8192]
blk.40.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.40.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.40.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.40.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.40.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.40.ffn_norm.weight
F32
F32
[8192]
blk.40.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.41
blk.41.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.41.attn_norm.weight
F32
F32
[8192]
blk.41.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.41.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.41.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.41.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.41.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.41.ffn_norm.weight
F32
F32
[8192]
blk.41.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.42
blk.42.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.42.attn_norm.weight
F32
F32
[8192]
blk.42.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.42.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.42.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.42.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.42.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.42.ffn_norm.weight
F32
F32
[8192]
blk.42.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.43
blk.43.attn_norm.weight
F32
F32
[8192]
blk.43.cross_attn_attn_gate
F32
F32
[1]
blk.43.cross_attn_k_norm.weight
F32
F32
[128]
blk.43.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.43.cross_attn_mlp_gate
F32
F32
[1]
blk.43.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.43.cross_attn_q_norm.weight
F32
F32
[128]
blk.43.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.43.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.43.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.43.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.43.ffn_norm.weight
F32
F32
[8192]
blk.43.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.44
blk.44.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.44.attn_norm.weight
F32
F32
[8192]
blk.44.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.44.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.44.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.44.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.44.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.44.ffn_norm.weight
F32
F32
[8192]
blk.44.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.45
blk.45.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.45.attn_norm.weight
F32
F32
[8192]
blk.45.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.45.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.45.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.45.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.45.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.45.ffn_norm.weight
F32
F32
[8192]
blk.45.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.46
blk.46.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.46.attn_norm.weight
F32
F32
[8192]
blk.46.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.46.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.46.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.46.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.46.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.46.ffn_norm.weight
F32
F32
[8192]
blk.46.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.47
blk.47.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.47.attn_norm.weight
F32
F32
[8192]
blk.47.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.47.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.47.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.47.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.47.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.47.ffn_norm.weight
F32
F32
[8192]
blk.47.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.48
blk.48.attn_norm.weight
F32
F32
[8192]
blk.48.cross_attn_attn_gate
F32
F32
[1]
blk.48.cross_attn_k_norm.weight
F32
F32
[128]
blk.48.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.48.cross_attn_mlp_gate
F32
F32
[1]
blk.48.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.48.cross_attn_q_norm.weight
F32
F32
[128]
blk.48.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.48.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.48.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.48.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.48.ffn_norm.weight
F32
F32
[8192]
blk.48.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.49
blk.49.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.49.attn_norm.weight
F32
F32
[8192]
blk.49.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.49.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.49.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.49.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.49.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.49.ffn_norm.weight
F32
F32
[8192]
blk.49.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.50
blk.50.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.50.attn_norm.weight
F32
F32
[8192]
blk.50.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.50.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.50.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.50.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.50.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.50.ffn_norm.weight
F32
F32
[8192]
blk.50.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.51
blk.51.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.51.attn_norm.weight
F32
F32
[8192]
blk.51.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.51.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.51.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.51.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.51.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.51.ffn_norm.weight
F32
F32
[8192]
blk.51.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.52
blk.52.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.52.attn_norm.weight
F32
F32
[8192]
blk.52.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.52.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.52.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.52.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.52.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.52.ffn_norm.weight
F32
F32
[8192]
blk.52.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.53
blk.53.attn_norm.weight
F32
F32
[8192]
blk.53.cross_attn_attn_gate
F32
F32
[1]
blk.53.cross_attn_k_norm.weight
F32
F32
[128]
blk.53.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.53.cross_attn_mlp_gate
F32
F32
[1]
blk.53.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.53.cross_attn_q_norm.weight
F32
F32
[128]
blk.53.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.53.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.53.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.53.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.53.ffn_norm.weight
F32
F32
[8192]
blk.53.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.54
blk.54.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.54.attn_norm.weight
F32
F32
[8192]
blk.54.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.54.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.54.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.54.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.54.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.54.ffn_norm.weight
F32
F32
[8192]
blk.54.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.55
blk.55.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.55.attn_norm.weight
F32
F32
[8192]
blk.55.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.55.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.55.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.55.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.55.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.55.ffn_norm.weight
F32
F32
[8192]
blk.55.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.56
blk.56.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.56.attn_norm.weight
F32
F32
[8192]
blk.56.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.56.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.56.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.56.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.56.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.56.ffn_norm.weight
F32
F32
[8192]
blk.56.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.57
blk.57.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.57.attn_norm.weight
F32
F32
[8192]
blk.57.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.57.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.57.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.57.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.57.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.57.ffn_norm.weight
F32
F32
[8192]
blk.57.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.58
blk.58.attn_norm.weight
F32
F32
[8192]
blk.58.cross_attn_attn_gate
F32
F32
[1]
blk.58.cross_attn_k_norm.weight
F32
F32
[128]
blk.58.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.58.cross_attn_mlp_gate
F32
F32
[1]
blk.58.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.58.cross_attn_q_norm.weight
F32
F32
[128]
blk.58.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.58.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.58.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.58.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.58.ffn_norm.weight
F32
F32
[8192]
blk.58.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.59
blk.59.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.59.attn_norm.weight
F32
F32
[8192]
blk.59.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.59.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.59.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.59.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.59.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.59.ffn_norm.weight
F32
F32
[8192]
blk.59.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.60
blk.60.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.60.attn_norm.weight
F32
F32
[8192]
blk.60.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.60.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.60.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.60.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.60.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.60.ffn_norm.weight
F32
F32
[8192]
blk.60.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.61
blk.61.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.61.attn_norm.weight
F32
F32
[8192]
blk.61.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.61.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.61.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.61.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.61.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.61.ffn_norm.weight
F32
F32
[8192]
blk.61.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.62
blk.62.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.62.attn_norm.weight
F32
F32
[8192]
blk.62.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.62.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.62.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.62.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.62.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.62.ffn_norm.weight
F32
F32
[8192]
blk.62.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.63
blk.63.attn_norm.weight
F32
F32
[8192]
blk.63.cross_attn_attn_gate
F32
F32
[1]
blk.63.cross_attn_k_norm.weight
F32
F32
[128]
blk.63.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.63.cross_attn_mlp_gate
F32
F32
[1]
blk.63.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.63.cross_attn_q_norm.weight
F32
F32
[128]
blk.63.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.63.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.63.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.63.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.63.ffn_norm.weight
F32
F32
[8192]
blk.63.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.64
blk.64.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.64.attn_norm.weight
F32
F32
[8192]
blk.64.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.64.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.64.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.64.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.64.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.64.ffn_norm.weight
F32
F32
[8192]
blk.64.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.65
blk.65.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.65.attn_norm.weight
F32
F32
[8192]
blk.65.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.65.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.65.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.65.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.65.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.65.ffn_norm.weight
F32
F32
[8192]
blk.65.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.66
blk.66.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.66.attn_norm.weight
F32
F32
[8192]
blk.66.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.66.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.66.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.66.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.66.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.66.ffn_norm.weight
F32
F32
[8192]
blk.66.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.67
blk.67.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.67.attn_norm.weight
F32
F32
[8192]
blk.67.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.67.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.67.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.67.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.67.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.67.ffn_norm.weight
F32
F32
[8192]
blk.67.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.68
blk.68.attn_norm.weight
F32
F32
[8192]
blk.68.cross_attn_attn_gate
F32
F32
[1]
blk.68.cross_attn_k_norm.weight
F32
F32
[128]
blk.68.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.68.cross_attn_mlp_gate
F32
F32
[1]
blk.68.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.68.cross_attn_q_norm.weight
F32
F32
[128]
blk.68.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.68.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.68.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.68.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.68.ffn_norm.weight
F32
F32
[8192]
blk.68.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.69
blk.69.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.69.attn_norm.weight
F32
F32
[8192]
blk.69.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.69.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.69.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.69.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.69.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.69.ffn_norm.weight
F32
F32
[8192]
blk.69.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.70
blk.70.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.70.attn_norm.weight
F32
F32
[8192]
blk.70.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.70.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.70.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.70.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.70.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.70.ffn_norm.weight
F32
F32
[8192]
blk.70.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.71
blk.71.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.71.attn_norm.weight
F32
F32
[8192]
blk.71.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.71.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.71.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.71.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.71.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.71.ffn_norm.weight
F32
F32
[8192]
blk.71.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.72
blk.72.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.72.attn_norm.weight
F32
F32
[8192]
blk.72.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.72.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.72.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.72.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.72.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.72.ffn_norm.weight
F32
F32
[8192]
blk.72.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.73
blk.73.attn_norm.weight
F32
F32
[8192]
blk.73.cross_attn_attn_gate
F32
F32
[1]
blk.73.cross_attn_k_norm.weight
F32
F32
[128]
blk.73.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.73.cross_attn_mlp_gate
F32
F32
[1]
blk.73.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.73.cross_attn_q_norm.weight
F32
F32
[128]
blk.73.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.73.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.73.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.73.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.73.ffn_norm.weight
F32
F32
[8192]
blk.73.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.74
blk.74.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.74.attn_norm.weight
F32
F32
[8192]
blk.74.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.74.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.74.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.74.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.74.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.74.ffn_norm.weight
F32
F32
[8192]
blk.74.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.75
blk.75.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.75.attn_norm.weight
F32
F32
[8192]
blk.75.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.75.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.75.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.75.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.75.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.75.ffn_norm.weight
F32
F32
[8192]
blk.75.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.76
blk.76.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.76.attn_norm.weight
F32
F32
[8192]
blk.76.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.76.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.76.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.76.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.76.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.76.ffn_norm.weight
F32
F32
[8192]
blk.76.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.77
blk.77.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.77.attn_norm.weight
F32
F32
[8192]
blk.77.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.77.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.77.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.77.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.77.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.77.ffn_norm.weight
F32
F32
[8192]
blk.77.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.78
blk.78.attn_norm.weight
F32
F32
[8192]
blk.78.cross_attn_attn_gate
F32
F32
[1]
blk.78.cross_attn_k_norm.weight
F32
F32
[128]
blk.78.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.78.cross_attn_mlp_gate
F32
F32
[1]
blk.78.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.78.cross_attn_q_norm.weight
F32
F32
[128]
blk.78.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.78.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.78.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.78.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.78.ffn_norm.weight
F32
F32
[8192]
blk.78.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.79
blk.79.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.79.attn_norm.weight
F32
F32
[8192]
blk.79.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.79.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.79.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.79.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.79.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.79.ffn_norm.weight
F32
F32
[8192]
blk.79.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.80
blk.80.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.80.attn_norm.weight
F32
F32
[8192]
blk.80.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.80.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.80.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.80.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.80.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.80.ffn_norm.weight
F32
F32
[8192]
blk.80.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.81
blk.81.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.81.attn_norm.weight
F32
F32
[8192]
blk.81.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.81.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.81.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.81.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.81.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.81.ffn_norm.weight
F32
F32
[8192]
blk.81.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.82
blk.82.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.82.attn_norm.weight
F32
F32
[8192]
blk.82.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.82.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.82.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.82.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.82.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.82.ffn_norm.weight
F32
F32
[8192]
blk.82.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.83
blk.83.attn_norm.weight
F32
F32
[8192]
blk.83.cross_attn_attn_gate
F32
F32
[1]
blk.83.cross_attn_k_norm.weight
F32
F32
[128]
blk.83.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.83.cross_attn_mlp_gate
F32
F32
[1]
blk.83.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.83.cross_attn_q_norm.weight
F32
F32
[128]
blk.83.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.83.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.83.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.83.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.83.ffn_norm.weight
F32
F32
[8192]
blk.83.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.84
blk.84.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.84.attn_norm.weight
F32
F32
[8192]
blk.84.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.84.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.84.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.84.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.84.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.84.ffn_norm.weight
F32
F32
[8192]
blk.84.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.85
blk.85.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.85.attn_norm.weight
F32
F32
[8192]
blk.85.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.85.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.85.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.85.ffn_down.weight
Q4_K
Q4_K
[28672, 8192]
blk.85.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.85.ffn_norm.weight
F32
F32
[8192]
blk.85.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.86
blk.86.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.86.attn_norm.weight
F32
F32
[8192]
blk.86.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.86.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.86.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.86.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.86.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.86.ffn_norm.weight
F32
F32
[8192]
blk.86.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.87
blk.87.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.87.attn_norm.weight
F32
F32
[8192]
blk.87.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.87.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.87.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.87.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.87.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.87.ffn_norm.weight
F32
F32
[8192]
blk.87.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.88
blk.88.attn_norm.weight
F32
F32
[8192]
blk.88.cross_attn_attn_gate
F32
F32
[1]
blk.88.cross_attn_k_norm.weight
F32
F32
[128]
blk.88.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.88.cross_attn_mlp_gate
F32
F32
[1]
blk.88.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.88.cross_attn_q_norm.weight
F32
F32
[128]
blk.88.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.88.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.88.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.88.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.88.ffn_norm.weight
F32
F32
[8192]
blk.88.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.89
blk.89.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.89.attn_norm.weight
F32
F32
[8192]
blk.89.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.89.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.89.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.89.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.89.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.89.ffn_norm.weight
F32
F32
[8192]
blk.89.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.90
blk.90.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.90.attn_norm.weight
F32
F32
[8192]
blk.90.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.90.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.90.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.90.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.90.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.90.ffn_norm.weight
F32
F32
[8192]
blk.90.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.91
blk.91.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.91.attn_norm.weight
F32
F32
[8192]
blk.91.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.91.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.91.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.91.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.91.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.91.ffn_norm.weight
F32
F32
[8192]
blk.91.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.92
blk.92.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.92.attn_norm.weight
F32
F32
[8192]
blk.92.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.92.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.92.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.92.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.92.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.92.ffn_norm.weight
F32
F32
[8192]
blk.92.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.93
blk.93.attn_norm.weight
F32
F32
[8192]
blk.93.cross_attn_attn_gate
F32
F32
[1]
blk.93.cross_attn_k_norm.weight
F32
F32
[128]
blk.93.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.93.cross_attn_mlp_gate
F32
F32
[1]
blk.93.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.93.cross_attn_q_norm.weight
F32
F32
[128]
blk.93.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.93.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.93.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.93.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.93.ffn_norm.weight
F32
F32
[8192]
blk.93.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.94
blk.94.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.94.attn_norm.weight
F32
F32
[8192]
blk.94.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.94.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.94.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.94.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.94.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.94.ffn_norm.weight
F32
F32
[8192]
blk.94.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.95
blk.95.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.95.attn_norm.weight
F32
F32
[8192]
blk.95.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.95.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.95.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.95.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.95.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.95.ffn_norm.weight
F32
F32
[8192]
blk.95.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.96
blk.96.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.96.attn_norm.weight
F32
F32
[8192]
blk.96.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.96.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.96.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.96.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.96.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.96.ffn_norm.weight
F32
F32
[8192]
blk.96.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.97
blk.97.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.97.attn_norm.weight
F32
F32
[8192]
blk.97.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.97.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.97.attn_v.weight
Q6_K
Q6_K
[8192, 1024]
blk.97.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.97.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.97.ffn_norm.weight
F32
F32
[8192]
blk.97.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.98
blk.98.attn_norm.weight
F32
F32
[8192]
blk.98.cross_attn_attn_gate
F32
F32
[1]
blk.98.cross_attn_k_norm.weight
F32
F32
[128]
blk.98.cross_attn_k_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.98.cross_attn_mlp_gate
F32
F32
[1]
blk.98.cross_attn_o_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.98.cross_attn_q_norm.weight
F32
F32
[128]
blk.98.cross_attn_q_proj.weight
Q4_K
Q4_K
[8192, 8192]
blk.98.cross_attn_v_proj.weight
Q4_K
Q4_K
[8192, 1024]
blk.98.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.98.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.98.ffn_norm.weight
F32
F32
[8192]
blk.98.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
blk.99
blk.99.attn_k.weight
Q4_K
Q4_K
[8192, 1024]
blk.99.attn_norm.weight
F32
F32
[8192]
blk.99.attn_output.weight
Q4_K
Q4_K
[8192, 8192]
blk.99.attn_q.weight
Q4_K
Q4_K
[8192, 8192]
blk.99.attn_v.weight
Q4_K
Q4_K
[8192, 1024]
blk.99.ffn_down.weight
Q6_K
Q6_K
[28672, 8192]
blk.99.ffn_gate.weight
Q4_K
Q4_K
[8192, 28672]
blk.99.ffn_norm.weight
F32
F32
[8192]
blk.99.ffn_up.weight
Q4_K
Q4_K
[8192, 28672]
mm.0.bias
F32
F32
[8192]
mm.0.weight
F16
F16
[7680, 8192]
output.weight
Q6_K
Q6_K
[8192, 128256]
rope_freqs.weight
F32
F32
[64]
v.blk.0
v.blk.0.attn_k.weight
F16
F16
[1280, 1280]
v.blk.0.attn_norm.bias
F32
F32
[1280]
v.blk.0.attn_norm.weight
F32
F32
[1280]
v.blk.0.attn_output.weight
F16
F16
[1280, 1280]
v.blk.0.attn_q.weight
F16
F16
[1280, 1280]
v.blk.0.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.blk.0.ffn_down.bias
F32
F32
[1280]
v.blk.0.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.0.ffn_norm.bias
F32
F32
[1280]
v.blk.0.ffn_norm.weight
F32
F32
[1280]
v.blk.0.ffn_up.bias
F32
F32
[5120]
v.blk.0.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.1
v.blk.1.attn_k.weight
F16
F16
[1280, 1280]
v.blk.1.attn_norm.bias
F32
F32
[1280]
v.blk.1.attn_norm.weight
F32
F32
[1280]
v.blk.1.attn_output.weight
F16
F16
[1280, 1280]
v.blk.1.attn_q.weight
F16
F16
[1280, 1280]
v.blk.1.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.blk.1.ffn_down.bias
F32
F32
[1280]
v.blk.1.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.1.ffn_norm.bias
F32
F32
[1280]
v.blk.1.ffn_norm.weight
F32
F32
[1280]
v.blk.1.ffn_up.bias
F32
F32
[5120]
v.blk.1.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.2
v.blk.2.attn_k.weight
F16
F16
[1280, 1280]
v.blk.2.attn_norm.bias
F32
F32
[1280]
v.blk.2.attn_norm.weight
F32
F32
[1280]
v.blk.2.attn_output.weight
F16
F16
[1280, 1280]
v.blk.2.attn_q.weight
F16
F16
[1280, 1280]
v.blk.2.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.blk.2.ffn_down.bias
F32
F32
[1280]
v.blk.2.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.2.ffn_norm.bias
F32
F32
[1280]
v.blk.2.ffn_norm.weight
F32
F32
[1280]
v.blk.2.ffn_up.bias
F32
F32
[5120]
v.blk.2.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.3
v.blk.3.attn_k.weight
F16
F16
[1280, 1280]
v.blk.3.attn_norm.bias
F32
F32
[1280]
v.blk.3.attn_norm.weight
F32
F32
[1280]
v.blk.3.attn_output.weight
F16
F16
[1280, 1280]
v.blk.3.attn_q.weight
F16
F16
[1280, 1280]
v.blk.3.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.3.ffn_down.bias
F32
F32
[1280]
v.blk.3.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.3.ffn_norm.bias
F32
F32
[1280]
v.blk.3.ffn_norm.weight
F32
F32
[1280]
v.blk.3.ffn_up.bias
F32
F32
[5120]
v.blk.3.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.4
v.blk.4.attn_k.weight
F16
F16
[1280, 1280]
v.blk.4.attn_norm.bias
F32
F32
[1280]
v.blk.4.attn_norm.weight
F32
F32
[1280]
v.blk.4.attn_output.weight
F16
F16
[1280, 1280]
v.blk.4.attn_q.weight
F16
F16
[1280, 1280]
v.blk.4.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.4.ffn_down.bias
F32
F32
[1280]
v.blk.4.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.4.ffn_norm.bias
F32
F32
[1280]
v.blk.4.ffn_norm.weight
F32
F32
[1280]
v.blk.4.ffn_up.bias
F32
F32
[5120]
v.blk.4.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.5
v.blk.5.attn_k.weight
F16
F16
[1280, 1280]
v.blk.5.attn_norm.bias
F32
F32
[1280]
v.blk.5.attn_norm.weight
F32
F32
[1280]
v.blk.5.attn_output.weight
F16
F16
[1280, 1280]
v.blk.5.attn_q.weight
F16
F16
[1280, 1280]
v.blk.5.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.5.ffn_down.bias
F32
F32
[1280]
v.blk.5.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.5.ffn_norm.bias
F32
F32
[1280]
v.blk.5.ffn_norm.weight
F32
F32
[1280]
v.blk.5.ffn_up.bias
F32
F32
[5120]
v.blk.5.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.6
v.blk.6.attn_k.weight
F16
F16
[1280, 1280]
v.blk.6.attn_norm.bias
F32
F32
[1280]
v.blk.6.attn_norm.weight
F32
F32
[1280]
v.blk.6.attn_output.weight
F16
F16
[1280, 1280]
v.blk.6.attn_q.weight
F16
F16
[1280, 1280]
v.blk.6.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.6.ffn_down.bias
F32
F32
[1280]
v.blk.6.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.6.ffn_norm.bias
F32
F32
[1280]
v.blk.6.ffn_norm.weight
F32
F32
[1280]
v.blk.6.ffn_up.bias
F32
F32
[5120]
v.blk.6.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.7
v.blk.7.attn_k.weight
F16
F16
[1280, 1280]
v.blk.7.attn_norm.bias
F32
F32
[1280]
v.blk.7.attn_norm.weight
F32
F32
[1280]
v.blk.7.attn_output.weight
F16
F16
[1280, 1280]
v.blk.7.attn_q.weight
F16
F16
[1280, 1280]
v.blk.7.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.7.ffn_down.bias
F32
F32
[1280]
v.blk.7.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.7.ffn_norm.bias
F32
F32
[1280]
v.blk.7.ffn_norm.weight
F32
F32
[1280]
v.blk.7.ffn_up.bias
F32
F32
[5120]
v.blk.7.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.8
v.blk.8.attn_k.weight
F16
F16
[1280, 1280]
v.blk.8.attn_norm.bias
F32
F32
[1280]
v.blk.8.attn_norm.weight
F32
F32
[1280]
v.blk.8.attn_output.weight
F16
F16
[1280, 1280]
v.blk.8.attn_q.weight
F16
F16
[1280, 1280]
v.blk.8.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.8.ffn_down.bias
F32
F32
[1280]
v.blk.8.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.8.ffn_norm.bias
F32
F32
[1280]
v.blk.8.ffn_norm.weight
F32
F32
[1280]
v.blk.8.ffn_up.bias
F32
F32
[5120]
v.blk.8.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.9
v.blk.9.attn_k.weight
F16
F16
[1280, 1280]
v.blk.9.attn_norm.bias
F32
F32
[1280]
v.blk.9.attn_norm.weight
F32
F32
[1280]
v.blk.9.attn_output.weight
F16
F16
[1280, 1280]
v.blk.9.attn_q.weight
F16
F16
[1280, 1280]
v.blk.9.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.9.ffn_down.bias
F32
F32
[1280]
v.blk.9.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.9.ffn_norm.bias
F32
F32
[1280]
v.blk.9.ffn_norm.weight
F32
F32
[1280]
v.blk.9.ffn_up.bias
F32
F32
[5120]
v.blk.9.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.10
v.blk.10.attn_k.weight
F16
F16
[1280, 1280]
v.blk.10.attn_norm.bias
F32
F32
[1280]
v.blk.10.attn_norm.weight
F32
F32
[1280]
v.blk.10.attn_output.weight
F16
F16
[1280, 1280]
v.blk.10.attn_q.weight
F16
F16
[1280, 1280]
v.blk.10.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.10.ffn_down.bias
F32
F32
[1280]
v.blk.10.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.10.ffn_norm.bias
F32
F32
[1280]
v.blk.10.ffn_norm.weight
F32
F32
[1280]
v.blk.10.ffn_up.bias
F32
F32
[5120]
v.blk.10.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.11
v.blk.11.attn_k.weight
F16
F16
[1280, 1280]
v.blk.11.attn_norm.bias
F32
F32
[1280]
v.blk.11.attn_norm.weight
F32
F32
[1280]
v.blk.11.attn_output.weight
F16
F16
[1280, 1280]
v.blk.11.attn_q.weight
F16
F16
[1280, 1280]
v.blk.11.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.blk.11.ffn_down.bias
F32
F32
[1280]
v.blk.11.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.11.ffn_norm.bias
F32
F32
[1280]
v.blk.11.ffn_norm.weight
F32
F32
[1280]
v.blk.11.ffn_up.bias
F32
F32
[5120]
v.blk.11.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.12
v.blk.12.attn_k.weight
F16
F16
[1280, 1280]
v.blk.12.attn_norm.bias
F32
F32
[1280]
v.blk.12.attn_norm.weight
F32
F32
[1280]
v.blk.12.attn_output.weight
F16
F16
[1280, 1280]
v.blk.12.attn_q.weight
F16
F16
[1280, 1280]
v.blk.12.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.blk.12.ffn_down.bias
F32
F32
[1280]
v.blk.12.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.12.ffn_norm.bias
F32
F32
[1280]
v.blk.12.ffn_norm.weight
F32
F32
[1280]
v.blk.12.ffn_up.bias
F32
F32
[5120]
v.blk.12.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.13
v.blk.13.attn_k.weight
F16
F16
[1280, 1280]
v.blk.13.attn_norm.bias
F32
F32
[1280]
v.blk.13.attn_norm.weight
F32
F32
[1280]
v.blk.13.attn_output.weight
F16
F16
[1280, 1280]
v.blk.13.attn_q.weight
F16
F16
[1280, 1280]
v.blk.13.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.13.ffn_down.bias
F32
F32
[1280]
v.blk.13.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.13.ffn_norm.bias
F32
F32
[1280]
v.blk.13.ffn_norm.weight
F32
F32
[1280]
v.blk.13.ffn_up.bias
F32
F32
[5120]
v.blk.13.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.14
v.blk.14.attn_k.weight
F16
F16
[1280, 1280]
v.blk.14.attn_norm.bias
F32
F32
[1280]
v.blk.14.attn_norm.weight
F32
F32
[1280]
v.blk.14.attn_output.weight
F16
F16
[1280, 1280]
v.blk.14.attn_q.weight
F16
F16
[1280, 1280]
v.blk.14.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.blk.14.ffn_down.bias
F32
F32
[1280]
v.blk.14.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.14.ffn_norm.bias
F32
F32
[1280]
v.blk.14.ffn_norm.weight
F32
F32
[1280]
v.blk.14.ffn_up.bias
F32
F32
[5120]
v.blk.14.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.15
v.blk.15.attn_k.weight
F16
F16
[1280, 1280]
v.blk.15.attn_norm.bias
F32
F32
[1280]
v.blk.15.attn_norm.weight
F32
F32
[1280]
v.blk.15.attn_output.weight
F16
F16
[1280, 1280]
v.blk.15.attn_q.weight
F16
F16
[1280, 1280]
v.blk.15.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.blk.15.ffn_down.bias
F32
F32
[1280]
v.blk.15.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.15.ffn_norm.bias
F32
F32
[1280]
v.blk.15.ffn_norm.weight
F32
F32
[1280]
v.blk.15.ffn_up.bias
F32
F32
[5120]
v.blk.15.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.16
v.blk.16.attn_k.weight
F16
F16
[1280, 1280]
v.blk.16.attn_norm.bias
F32
F32
[1280]
v.blk.16.attn_norm.weight
F32
F32
[1280]
v.blk.16.attn_output.weight
F16
F16
[1280, 1280]
v.blk.16.attn_q.weight
F16
F16
[1280, 1280]
v.blk.16.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.16.ffn_down.bias
F32
F32
[1280]
v.blk.16.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.16.ffn_norm.bias
F32
F32
[1280]
v.blk.16.ffn_norm.weight
F32
F32
[1280]
v.blk.16.ffn_up.bias
F32
F32
[5120]
v.blk.16.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.17
v.blk.17.attn_k.weight
F16
F16
[1280, 1280]
v.blk.17.attn_norm.bias
F32
F32
[1280]
v.blk.17.attn_norm.weight
F32
F32
[1280]
v.blk.17.attn_output.weight
F16
F16
[1280, 1280]
v.blk.17.attn_q.weight
F16
F16
[1280, 1280]
v.blk.17.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.blk.17.ffn_down.bias
F32
F32
[1280]
v.blk.17.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.17.ffn_norm.bias
F32
F32
[1280]
v.blk.17.ffn_norm.weight
F32
F32
[1280]
v.blk.17.ffn_up.bias
F32
F32
[5120]
v.blk.17.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.18
v.blk.18.attn_k.weight
F16
F16
[1280, 1280]
v.blk.18.attn_norm.bias
F32
F32
[1280]
v.blk.18.attn_norm.weight
F32
F32
[1280]
v.blk.18.attn_output.weight
F16
F16
[1280, 1280]
v.blk.18.attn_q.weight
F16
F16
[1280, 1280]
v.blk.18.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.blk.18.ffn_down.bias
F32
F32
[1280]
v.blk.18.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.18.ffn_norm.bias
F32
F32
[1280]
v.blk.18.ffn_norm.weight
F32
F32
[1280]
v.blk.18.ffn_up.bias
F32
F32
[5120]
v.blk.18.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.19
v.blk.19.attn_k.weight
F16
F16
[1280, 1280]
v.blk.19.attn_norm.bias
F32
F32
[1280]
v.blk.19.attn_norm.weight
F32
F32
[1280]
v.blk.19.attn_output.weight
F16
F16
[1280, 1280]
v.blk.19.attn_q.weight
F16
F16
[1280, 1280]
v.blk.19.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.19.ffn_down.bias
F32
F32
[1280]
v.blk.19.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.19.ffn_norm.bias
F32
F32
[1280]
v.blk.19.ffn_norm.weight
F32
F32
[1280]
v.blk.19.ffn_up.bias
F32
F32
[5120]
v.blk.19.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.20
v.blk.20.attn_k.weight
F16
F16
[1280, 1280]
v.blk.20.attn_norm.bias
F32
F32
[1280]
v.blk.20.attn_norm.weight
F32
F32
[1280]
v.blk.20.attn_output.weight
F16
F16
[1280, 1280]
v.blk.20.attn_q.weight
F16
F16
[1280, 1280]
v.blk.20.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.blk.20.ffn_down.bias
F32
F32
[1280]
v.blk.20.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.20.ffn_norm.bias
F32
F32
[1280]
v.blk.20.ffn_norm.weight
F32
F32
[1280]
v.blk.20.ffn_up.bias
F32
F32
[5120]
v.blk.20.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.21
v.blk.21.attn_k.weight
F16
F16
[1280, 1280]
v.blk.21.attn_norm.bias
F32
F32
[1280]
v.blk.21.attn_norm.weight
F32
F32
[1280]
v.blk.21.attn_output.weight
F16
F16
[1280, 1280]
v.blk.21.attn_q.weight
F16
F16
[1280, 1280]
v.blk.21.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.21.ffn_down.bias
F32
F32
[1280]
v.blk.21.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.21.ffn_norm.bias
F32
F32
[1280]
v.blk.21.ffn_norm.weight
F32
F32
[1280]
v.blk.21.ffn_up.bias
F32
F32
[5120]
v.blk.21.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.22
v.blk.22.attn_k.weight
F16
F16
[1280, 1280]
v.blk.22.attn_norm.bias
F32
F32
[1280]
v.blk.22.attn_norm.weight
F32
F32
[1280]
v.blk.22.attn_output.weight
F16
F16
[1280, 1280]
v.blk.22.attn_q.weight
F16
F16
[1280, 1280]
v.blk.22.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.blk.22.ffn_down.bias
F32
F32
[1280]
v.blk.22.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.22.ffn_norm.bias
F32
F32
[1280]
v.blk.22.ffn_norm.weight
F32
F32
[1280]
v.blk.22.ffn_up.bias
F32
F32
[5120]
v.blk.22.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.23
v.blk.23.attn_k.weight
F16
F16
[1280, 1280]
v.blk.23.attn_norm.bias
F32
F32
[1280]
v.blk.23.attn_norm.weight
F32
F32
[1280]
v.blk.23.attn_output.weight
F16
F16
[1280, 1280]
v.blk.23.attn_q.weight
F16
F16
[1280, 1280]
v.blk.23.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.blk.23.ffn_down.bias
F32
F32
[1280]
v.blk.23.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.23.ffn_norm.bias
F32
F32
[1280]
v.blk.23.ffn_norm.weight
F32
F32
[1280]
v.blk.23.ffn_up.bias
F32
F32
[5120]
v.blk.23.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.24
v.blk.24.attn_k.weight
F16
F16
[1280, 1280]
v.blk.24.attn_norm.bias
F32
F32
[1280]
v.blk.24.attn_norm.weight
F32
F32
[1280]
v.blk.24.attn_output.weight
F16
F16
[1280, 1280]
v.blk.24.attn_q.weight
F16
F16
[1280, 1280]
v.blk.24.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.24.ffn_down.bias
F32
F32
[1280]
v.blk.24.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.24.ffn_norm.bias
F32
F32
[1280]
v.blk.24.ffn_norm.weight
F32
F32
[1280]
v.blk.24.ffn_up.bias
F32
F32
[5120]
v.blk.24.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.25
v.blk.25.attn_k.weight
F16
F16
[1280, 1280]
v.blk.25.attn_norm.bias
F32
F32
[1280]
v.blk.25.attn_norm.weight
F32
F32
[1280]
v.blk.25.attn_output.weight
F16
F16
[1280, 1280]
v.blk.25.attn_q.weight
F16
F16
[1280, 1280]
v.blk.25.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.25.ffn_down.bias
F32
F32
[1280]
v.blk.25.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.25.ffn_norm.bias
F32
F32
[1280]
v.blk.25.ffn_norm.weight
F32
F32
[1280]
v.blk.25.ffn_up.bias
F32
F32
[5120]
v.blk.25.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.26
v.blk.26.attn_k.weight
F16
F16
[1280, 1280]
v.blk.26.attn_norm.bias
F32
F32
[1280]
v.blk.26.attn_norm.weight
F32
F32
[1280]
v.blk.26.attn_output.weight
F16
F16
[1280, 1280]
v.blk.26.attn_q.weight
F16
F16
[1280, 1280]
v.blk.26.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.26.ffn_down.bias
F32
F32
[1280]
v.blk.26.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.26.ffn_norm.bias
F32
F32
[1280]
v.blk.26.ffn_norm.weight
F32
F32
[1280]
v.blk.26.ffn_up.bias
F32
F32
[5120]
v.blk.26.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.27
v.blk.27.attn_k.weight
F16
F16
[1280, 1280]
v.blk.27.attn_norm.bias
F32
F32
[1280]
v.blk.27.attn_norm.weight
F32
F32
[1280]
v.blk.27.attn_output.weight
F16
F16
[1280, 1280]
v.blk.27.attn_q.weight
F16
F16
[1280, 1280]
v.blk.27.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.27.ffn_down.bias
F32
F32
[1280]
v.blk.27.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.27.ffn_norm.bias
F32
F32
[1280]
v.blk.27.ffn_norm.weight
F32
F32
[1280]
v.blk.27.ffn_up.bias
F32
F32
[5120]
v.blk.27.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.28
v.blk.28.attn_k.weight
F16
F16
[1280, 1280]
v.blk.28.attn_norm.bias
F32
F32
[1280]
v.blk.28.attn_norm.weight
F32
F32
[1280]
v.blk.28.attn_output.weight
F16
F16
[1280, 1280]
v.blk.28.attn_q.weight
F16
F16
[1280, 1280]
v.blk.28.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.28.ffn_down.bias
F32
F32
[1280]
v.blk.28.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.28.ffn_norm.bias
F32
F32
[1280]
v.blk.28.ffn_norm.weight
F32
F32
[1280]
v.blk.28.ffn_up.bias
F32
F32
[5120]
v.blk.28.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.29
v.blk.29.attn_k.weight
F16
F16
[1280, 1280]
v.blk.29.attn_norm.bias
F32
F32
[1280]
v.blk.29.attn_norm.weight
F32
F32
[1280]
v.blk.29.attn_output.weight
F16
F16
[1280, 1280]
v.blk.29.attn_q.weight
F16
F16
[1280, 1280]
v.blk.29.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.29.ffn_down.bias
F32
F32
[1280]
v.blk.29.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.29.ffn_norm.bias
F32
F32
[1280]
v.blk.29.ffn_norm.weight
F32
F32
[1280]
v.blk.29.ffn_up.bias
F32
F32
[5120]
v.blk.29.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.30
v.blk.30.attn_k.weight
F16
F16
[1280, 1280]
v.blk.30.attn_norm.bias
F32
F32
[1280]
v.blk.30.attn_norm.weight
F32
F32
[1280]
v.blk.30.attn_output.weight
F16
F16
[1280, 1280]
v.blk.30.attn_q.weight
F16
F16
[1280, 1280]
v.blk.30.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.30.ffn_down.bias
F32
F32
[1280]
v.blk.30.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.30.ffn_norm.bias
F32
F32
[1280]
v.blk.30.ffn_norm.weight
F32
F32
[1280]
v.blk.30.ffn_up.bias
F32
F32
[5120]
v.blk.30.ffn_up.weight
F16
F16
[1280, 5120]
v.blk.31
v.blk.31.attn_k.weight
F16
F16
[1280, 1280]
v.blk.31.attn_norm.bias
F32
F32
[1280]
v.blk.31.attn_norm.weight
F32
F32
[1280]
v.blk.31.attn_output.weight
F16
F16
[1280, 1280]
v.blk.31.attn_q.weight
F16
F16
[1280, 1280]
v.blk.31.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.blk.31.ffn_down.bias
F32
F32
[1280]
v.blk.31.ffn_down.weight
F16
F16
[5120, 1280]
v.blk.31.ffn_norm.bias
F32
F32
[1280]
v.blk.31.ffn_norm.weight
F32
F32
[1280]
v.blk.31.ffn_up.bias
F32
F32
[5120]
v.blk.31.ffn_up.weight
F16
F16
[1280, 5120]
v.class_embd
F32
F32
[1280]
v.global.blk.0
v.global.blk.0.attn_gate
F32
F32
[1]
v.global.blk.0.attn_k.weight
F16
F16
[1280, 1280]
v.global.blk.0.attn_norm.bias
F32
F32
[1280]
v.global.blk.0.attn_norm.weight
F32
F32
[1280]
v.global.blk.0.attn_output.weight
F16
F16
[1280, 1280]
v.global.blk.0.attn_q.weight
F16
F16
[1280, 1280]
v.global.blk.0.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.global.blk.0.ffn_down.bias
F32
F32
[1280]
v.global.blk.0.ffn_down.weight
F16
F16
[5120, 1280]
v.global.blk.0.ffn_gate
F32
F32
[1]
v.global.blk.0.ffn_norm.bias
F32
F32
[1280]
v.global.blk.0.ffn_norm.weight
F32
F32
[1280]
v.global.blk.0.ffn_up.bias
F32
F32
[5120]
v.global.blk.0.ffn_up.weight
F16
F16
[1280, 5120]
v.global.blk.1
v.global.blk.1.attn_gate
F32
F32
[1]
v.global.blk.1.attn_k.weight
F16
F16
[1280, 1280]
v.global.blk.1.attn_norm.bias
F32
F32
[1280]
v.global.blk.1.attn_norm.weight
F32
F32
[1280]
v.global.blk.1.attn_output.weight
F16
F16
[1280, 1280]
v.global.blk.1.attn_q.weight
F16
F16
[1280, 1280]
v.global.blk.1.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.global.blk.1.ffn_down.bias
F32
F32
[1280]
v.global.blk.1.ffn_down.weight
F16
F16
[5120, 1280]
v.global.blk.1.ffn_gate
F32
F32
[1]
v.global.blk.1.ffn_norm.bias
F32
F32
[1280]
v.global.blk.1.ffn_norm.weight
F32
F32
[1280]
v.global.blk.1.ffn_up.bias
F32
F32
[5120]
v.global.blk.1.ffn_up.weight
F16
F16
[1280, 5120]
v.global.blk.2
v.global.blk.2.attn_gate
F32
F32
[1]
v.global.blk.2.attn_k.weight
F16
F16
[1280, 1280]
v.global.blk.2.attn_norm.bias
F32
F32
[1280]
v.global.blk.2.attn_norm.weight
F32
F32
[1280]
v.global.blk.2.attn_output.weight
F16
F16
[1280, 1280]
v.global.blk.2.attn_q.weight
F16
F16
[1280, 1280]
v.global.blk.2.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.global.blk.2.ffn_down.bias
F32
F32
[1280]
v.global.blk.2.ffn_down.weight
F16
F16
[5120, 1280]
v.global.blk.2.ffn_gate
F32
F32
[1]
v.global.blk.2.ffn_norm.bias
F32
F32
[1280]
v.global.blk.2.ffn_norm.weight
F32
F32
[1280]
v.global.blk.2.ffn_up.bias
F32
F32
[5120]
v.global.blk.2.ffn_up.weight
F16
F16
[1280, 5120]
v.global.blk.3
v.global.blk.3.attn_gate
F32
F32
[1]
v.global.blk.3.attn_k.weight
F16
F16
[1280, 1280]
v.global.blk.3.attn_norm.bias
F32
F32
[1280]
v.global.blk.3.attn_norm.weight
F32
F32
[1280]
v.global.blk.3.attn_output.weight
F16
F16
[1280, 1280]
v.global.blk.3.attn_q.weight
F16
F16
[1280, 1280]
v.global.blk.3.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.global.blk.3.ffn_down.bias
F32
F32
[1280]
v.global.blk.3.ffn_down.weight
F16
F16
[5120, 1280]
v.global.blk.3.ffn_gate
F32
F32
[1]
v.global.blk.3.ffn_norm.bias
F32
F32
[1280]
v.global.blk.3.ffn_norm.weight
F32
F32
[1280]
v.global.blk.3.ffn_up.bias
F32
F32
[5120]
v.global.blk.3.ffn_up.weight
F16
F16
[1280, 5120]
v.global.blk.4
v.global.blk.4.attn_gate
F32
F32
[1]
v.global.blk.4.attn_k.weight
F16
F16
[1280, 1280]
v.global.blk.4.attn_norm.bias
F32
F32
[1280]
v.global.blk.4.attn_norm.weight
F32
F32
[1280]
v.global.blk.4.attn_output.weight
F16
F16
[1280, 1280]
v.global.blk.4.attn_q.weight
F16
F16
[1280, 1280]
v.global.blk.4.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.global.blk.4.ffn_down.bias
F32
F32
[1280]
v.global.blk.4.ffn_down.weight
F16
F16
[5120, 1280]
v.global.blk.4.ffn_gate
F32
F32
[1]
v.global.blk.4.ffn_norm.bias
F32
F32
[1280]
v.global.blk.4.ffn_norm.weight
F32
F32
[1280]
v.global.blk.4.ffn_up.bias
F32
F32
[5120]
v.global.blk.4.ffn_up.weight
F16
F16
[1280, 5120]
v.global.blk.5
v.global.blk.5.attn_gate
F32
F32
[1]
v.global.blk.5.attn_k.weight
F16
F16
[1280, 1280]
v.global.blk.5.attn_norm.bias
F32
F32
[1280]
v.global.blk.5.attn_norm.weight
F32
F32
[1280]
v.global.blk.5.attn_output.weight
F16
F16
[1280, 1280]
v.global.blk.5.attn_q.weight
F16
F16
[1280, 1280]
v.global.blk.5.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.global.blk.5.ffn_down.bias
F32
F32
[1280]
v.global.blk.5.ffn_down.weight
F16
F16
[5120, 1280]
v.global.blk.5.ffn_gate
F32
F32
[1]
v.global.blk.5.ffn_norm.bias
F32
F32
[1280]
v.global.blk.5.ffn_norm.weight
F32
F32
[1280]
v.global.blk.5.ffn_up.bias
F32
F32
[5120]
v.global.blk.5.ffn_up.weight
F16
F16
[1280, 5120]
v.global.blk.6
v.global.blk.6.attn_gate
F32
F32
[1]
v.global.blk.6.attn_k.weight
F16
F16
[1280, 1280]
v.global.blk.6.attn_norm.bias
F32
F32
[1280]
v.global.blk.6.attn_norm.weight
F32
F32
[1280]
v.global.blk.6.attn_output.weight
F16
F16
[1280, 1280]
v.global.blk.6.attn_q.weight
F16
F16
[1280, 1280]
v.global.blk.6.attn_v.weight
Q4_K
Q4_K
[1280, 1280]
v.global.blk.6.ffn_down.bias
F32
F32
[1280]
v.global.blk.6.ffn_down.weight
F16
F16
[5120, 1280]
v.global.blk.6.ffn_gate
F32
F32
[1]
v.global.blk.6.ffn_norm.bias
F32
F32
[1280]
v.global.blk.6.ffn_norm.weight
F32
F32
[1280]
v.global.blk.6.ffn_up.bias
F32
F32
[5120]
v.global.blk.6.ffn_up.weight
F16
F16
[1280, 5120]
v.global.blk.7
v.global.blk.7.attn_gate
F32
F32
[1]
v.global.blk.7.attn_k.weight
F16
F16
[1280, 1280]
v.global.blk.7.attn_norm.bias
F32
F32
[1280]
v.global.blk.7.attn_norm.weight
F32
F32
[1280]
v.global.blk.7.attn_output.weight
F16
F16
[1280, 1280]
v.global.blk.7.attn_q.weight
F16
F16
[1280, 1280]
v.global.blk.7.attn_v.weight
Q6_K
Q6_K
[1280, 1280]
v.global.blk.7.ffn_down.bias
F32
F32
[1280]
v.global.blk.7.ffn_down.weight
F16
F16
[5120, 1280]
v.global.blk.7.ffn_gate
F32
F32
[1]
v.global.blk.7.ffn_norm.bias
F32
F32
[1280]
v.global.blk.7.ffn_norm.weight
F32
F32
[1280]
v.global.blk.7.ffn_up.bias
F32
F32
[5120]
v.global.blk.7.ffn_up.weight
F16
F16
[1280, 5120]
v.patch_embd.weight
F16
F16
[14, 14, 3, 1280]
v.position_embd.gate
F32
F32
[1]
v.position_embd.weight
F16
F16
[1280, 1601]
v.post_ln.bias
F32
F32
[1280]
v.post_ln.weight
F32
F32
[1280]
v.post_tile_position_embd.gate
F32
F32
[1]
v.post_tile_position_embd.weight
F32
F32
[5120, 9]
v.pre_ln.bias
F32
F32
[1280]
v.pre_ln.weight
F32
F32
[1280]
v.pre_tile_position_embd.gate
F32
F32
[1]
v.pre_tile_position_embd.weight
F32
F32
[5120, 9]
v.tile_position_embd.gate
F32
F32
[1]
v.tile_position_embd.weight
F32
F32
[8197120, 9]
output_norm.weight
F32
F32
[8192]