Models
Turbo
Sign in
Download
Models
Download
Sign in
gemma3
:1b
12.9M
Downloads
Updated
1 week ago
The current, most capable model that runs on a single GPU.
The current, most capable model that runs on a single GPU.
Cancel
vision
270m
1b
4b
12b
27b
gemma3:1b
...
/
model
7cd4618c1faf · 815MB
Metadata
general.architecture
gemma3
gemma3
general.file_type
Q4_K_M
Q4_K_M
gemma3.attention.head_count
4
4
gemma3.attention.head_count_kv
1
1
gemma3.attention.key_length
256
256
gemma3.attention.layer_norm_rms_epsilon
1e-06
1e-06
gemma3.attention.sliding_window
512
512
gemma3.attention.value_length
256
256
gemma3.block_count
26
26
gemma3.context_length
32768
32768
gemma3.embedding_length
1152
1152
gemma3.feed_forward_length
6912
6912
gemma3.final_logit_softcapping
30
30
gemma3.rope.global.freq_base
1e+06
1e+06
gemma3.rope.local.freq_base
10000
10000
tokenizer.ggml.add_bos_token
true
true
tokenizer.ggml.add_eos_token
false
false
tokenizer.ggml.add_padding_token
false
false
tokenizer.ggml.add_unknown_token
false
false
tokenizer.ggml.bos_token_id
2
2
tokenizer.ggml.eos_token_id
1
1
tokenizer.ggml.merges
[ , ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ ▁, , , ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ ▁▁, ...]
[ , ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ ▁, , , ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ ▁▁, ...]
tokenizer.ggml.model
llama
llama
tokenizer.ggml.padding_token_id
0
0
tokenizer.ggml.pre
default
default
tokenizer.ggml.scores
[0, 0, 0, 0, 0, ...]
[0, 0, 0, 0, 0, ...]
tokenizer.ggml.token_type
[3, 3, 3, 2, 1, ...]
[3, 3, 3, 2, 1, ...]
tokenizer.ggml.tokens
[<pad>, <eos>, <bos>, <unk>, <mask>, ...]
[<pad>, <eos>, <bos>, <unk>, <mask>, ...]
tokenizer.ggml.unknown_token_id
3
3
Tensor
Name
Type
Shape
token_embd.weight
Q8_0
Q8_0
[1152, 262144]
blk.0
blk.0.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.0.attn_k_norm.weight
F32
F32
[256]
blk.0.attn_norm.weight
F32
F32
[1152]
blk.0.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.0.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.0.attn_q_norm.weight
F32
F32
[256]
blk.0.attn_v.weight
Q8_0
Q8_0
[1152, 256]
blk.0.ffn_down.weight
Q6_K
Q6_K
[6912, 1152]
blk.0.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.0.ffn_norm.weight
F32
F32
[1152]
blk.0.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.0.post_attention_norm.weight
F32
F32
[1152]
blk.0.post_ffw_norm.weight
F32
F32
[1152]
blk.1
blk.1.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.1.attn_k_norm.weight
F32
F32
[256]
blk.1.attn_norm.weight
F32
F32
[1152]
blk.1.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.1.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.1.attn_q_norm.weight
F32
F32
[256]
blk.1.attn_v.weight
Q8_0
Q8_0
[1152, 256]
blk.1.ffn_down.weight
Q6_K
Q6_K
[6912, 1152]
blk.1.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.1.ffn_norm.weight
F32
F32
[1152]
blk.1.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.1.post_attention_norm.weight
F32
F32
[1152]
blk.1.post_ffw_norm.weight
F32
F32
[1152]
blk.2
blk.2.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.2.attn_k_norm.weight
F32
F32
[256]
blk.2.attn_norm.weight
F32
F32
[1152]
blk.2.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.2.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.2.attn_q_norm.weight
F32
F32
[256]
blk.2.attn_v.weight
Q8_0
Q8_0
[1152, 256]
blk.2.ffn_down.weight
Q6_K
Q6_K
[6912, 1152]
blk.2.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.2.ffn_norm.weight
F32
F32
[1152]
blk.2.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.2.post_attention_norm.weight
F32
F32
[1152]
blk.2.post_ffw_norm.weight
F32
F32
[1152]
blk.3
blk.3.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.3.attn_k_norm.weight
F32
F32
[256]
blk.3.attn_norm.weight
F32
F32
[1152]
blk.3.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.3.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.3.attn_q_norm.weight
F32
F32
[256]
blk.3.attn_v.weight
Q5_0
Q5_0
[1152, 256]
blk.3.ffn_down.weight
Q4_K
Q4_K
[6912, 1152]
blk.3.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.3.ffn_norm.weight
F32
F32
[1152]
blk.3.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.3.post_attention_norm.weight
F32
F32
[1152]
blk.3.post_ffw_norm.weight
F32
F32
[1152]
blk.4
blk.4.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.4.attn_k_norm.weight
F32
F32
[256]
blk.4.attn_norm.weight
F32
F32
[1152]
blk.4.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.4.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.4.attn_q_norm.weight
F32
F32
[256]
blk.4.attn_v.weight
Q5_0
Q5_0
[1152, 256]
blk.4.ffn_down.weight
Q4_K
Q4_K
[6912, 1152]
blk.4.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.4.ffn_norm.weight
F32
F32
[1152]
blk.4.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.4.post_attention_norm.weight
F32
F32
[1152]
blk.4.post_ffw_norm.weight
F32
F32
[1152]
blk.5
blk.5.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.5.attn_k_norm.weight
F32
F32
[256]
blk.5.attn_norm.weight
F32
F32
[1152]
blk.5.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.5.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.5.attn_q_norm.weight
F32
F32
[256]
blk.5.attn_v.weight
Q8_0
Q8_0
[1152, 256]
blk.5.ffn_down.weight
Q6_K
Q6_K
[6912, 1152]
blk.5.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.5.ffn_norm.weight
F32
F32
[1152]
blk.5.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.5.post_attention_norm.weight
F32
F32
[1152]
blk.5.post_ffw_norm.weight
F32
F32
[1152]
blk.6
blk.6.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.6.attn_k_norm.weight
F32
F32
[256]
blk.6.attn_norm.weight
F32
F32
[1152]
blk.6.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.6.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.6.attn_q_norm.weight
F32
F32
[256]
blk.6.attn_v.weight
Q5_0
Q5_0
[1152, 256]
blk.6.ffn_down.weight
Q4_K
Q4_K
[6912, 1152]
blk.6.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.6.ffn_norm.weight
F32
F32
[1152]
blk.6.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.6.post_attention_norm.weight
F32
F32
[1152]
blk.6.post_ffw_norm.weight
F32
F32
[1152]
blk.7
blk.7.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.7.attn_k_norm.weight
F32
F32
[256]
blk.7.attn_norm.weight
F32
F32
[1152]
blk.7.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.7.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.7.attn_q_norm.weight
F32
F32
[256]
blk.7.attn_v.weight
Q5_0
Q5_0
[1152, 256]
blk.7.ffn_down.weight
Q4_K
Q4_K
[6912, 1152]
blk.7.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.7.ffn_norm.weight
F32
F32
[1152]
blk.7.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.7.post_attention_norm.weight
F32
F32
[1152]
blk.7.post_ffw_norm.weight
F32
F32
[1152]
blk.8
blk.8.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.8.attn_k_norm.weight
F32
F32
[256]
blk.8.attn_norm.weight
F32
F32
[1152]
blk.8.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.8.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.8.attn_q_norm.weight
F32
F32
[256]
blk.8.attn_v.weight
Q8_0
Q8_0
[1152, 256]
blk.8.ffn_down.weight
Q6_K
Q6_K
[6912, 1152]
blk.8.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.8.ffn_norm.weight
F32
F32
[1152]
blk.8.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.8.post_attention_norm.weight
F32
F32
[1152]
blk.8.post_ffw_norm.weight
F32
F32
[1152]
blk.9
blk.9.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.9.attn_k_norm.weight
F32
F32
[256]
blk.9.attn_norm.weight
F32
F32
[1152]
blk.9.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.9.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.9.attn_q_norm.weight
F32
F32
[256]
blk.9.attn_v.weight
Q5_0
Q5_0
[1152, 256]
blk.9.ffn_down.weight
Q4_K
Q4_K
[6912, 1152]
blk.9.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.9.ffn_norm.weight
F32
F32
[1152]
blk.9.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.9.post_attention_norm.weight
F32
F32
[1152]
blk.9.post_ffw_norm.weight
F32
F32
[1152]
blk.10
blk.10.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.10.attn_k_norm.weight
F32
F32
[256]
blk.10.attn_norm.weight
F32
F32
[1152]
blk.10.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.10.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.10.attn_q_norm.weight
F32
F32
[256]
blk.10.attn_v.weight
Q5_0
Q5_0
[1152, 256]
blk.10.ffn_down.weight
Q4_K
Q4_K
[6912, 1152]
blk.10.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.10.ffn_norm.weight
F32
F32
[1152]
blk.10.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.10.post_attention_norm.weight
F32
F32
[1152]
blk.10.post_ffw_norm.weight
F32
F32
[1152]
blk.11
blk.11.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.11.attn_k_norm.weight
F32
F32
[256]
blk.11.attn_norm.weight
F32
F32
[1152]
blk.11.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.11.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.11.attn_q_norm.weight
F32
F32
[256]
blk.11.attn_v.weight
Q8_0
Q8_0
[1152, 256]
blk.11.ffn_down.weight
Q6_K
Q6_K
[6912, 1152]
blk.11.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.11.ffn_norm.weight
F32
F32
[1152]
blk.11.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.11.post_attention_norm.weight
F32
F32
[1152]
blk.11.post_ffw_norm.weight
F32
F32
[1152]
blk.12
blk.12.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.12.attn_k_norm.weight
F32
F32
[256]
blk.12.attn_norm.weight
F32
F32
[1152]
blk.12.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.12.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.12.attn_q_norm.weight
F32
F32
[256]
blk.12.attn_v.weight
Q5_0
Q5_0
[1152, 256]
blk.12.ffn_down.weight
Q4_K
Q4_K
[6912, 1152]
blk.12.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.12.ffn_norm.weight
F32
F32
[1152]
blk.12.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.12.post_attention_norm.weight
F32
F32
[1152]
blk.12.post_ffw_norm.weight
F32
F32
[1152]
blk.13
blk.13.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.13.attn_k_norm.weight
F32
F32
[256]
blk.13.attn_norm.weight
F32
F32
[1152]
blk.13.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.13.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.13.attn_q_norm.weight
F32
F32
[256]
blk.13.attn_v.weight
Q5_0
Q5_0
[1152, 256]
blk.13.ffn_down.weight
Q4_K
Q4_K
[6912, 1152]
blk.13.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.13.ffn_norm.weight
F32
F32
[1152]
blk.13.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.13.post_attention_norm.weight
F32
F32
[1152]
blk.13.post_ffw_norm.weight
F32
F32
[1152]
blk.14
blk.14.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.14.attn_k_norm.weight
F32
F32
[256]
blk.14.attn_norm.weight
F32
F32
[1152]
blk.14.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.14.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.14.attn_q_norm.weight
F32
F32
[256]
blk.14.attn_v.weight
Q8_0
Q8_0
[1152, 256]
blk.14.ffn_down.weight
Q6_K
Q6_K
[6912, 1152]
blk.14.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.14.ffn_norm.weight
F32
F32
[1152]
blk.14.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.14.post_attention_norm.weight
F32
F32
[1152]
blk.14.post_ffw_norm.weight
F32
F32
[1152]
blk.15
blk.15.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.15.attn_k_norm.weight
F32
F32
[256]
blk.15.attn_norm.weight
F32
F32
[1152]
blk.15.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.15.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.15.attn_q_norm.weight
F32
F32
[256]
blk.15.attn_v.weight
Q5_0
Q5_0
[1152, 256]
blk.15.ffn_down.weight
Q4_K
Q4_K
[6912, 1152]
blk.15.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.15.ffn_norm.weight
F32
F32
[1152]
blk.15.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.15.post_attention_norm.weight
F32
F32
[1152]
blk.15.post_ffw_norm.weight
F32
F32
[1152]
blk.16
blk.16.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.16.attn_k_norm.weight
F32
F32
[256]
blk.16.attn_norm.weight
F32
F32
[1152]
blk.16.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.16.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.16.attn_q_norm.weight
F32
F32
[256]
blk.16.attn_v.weight
Q5_0
Q5_0
[1152, 256]
blk.16.ffn_down.weight
Q4_K
Q4_K
[6912, 1152]
blk.16.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.16.ffn_norm.weight
F32
F32
[1152]
blk.16.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.16.post_attention_norm.weight
F32
F32
[1152]
blk.16.post_ffw_norm.weight
F32
F32
[1152]
blk.17
blk.17.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.17.attn_k_norm.weight
F32
F32
[256]
blk.17.attn_norm.weight
F32
F32
[1152]
blk.17.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.17.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.17.attn_q_norm.weight
F32
F32
[256]
blk.17.attn_v.weight
Q8_0
Q8_0
[1152, 256]
blk.17.ffn_down.weight
Q6_K
Q6_K
[6912, 1152]
blk.17.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.17.ffn_norm.weight
F32
F32
[1152]
blk.17.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.17.post_attention_norm.weight
F32
F32
[1152]
blk.17.post_ffw_norm.weight
F32
F32
[1152]
blk.18
blk.18.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.18.attn_k_norm.weight
F32
F32
[256]
blk.18.attn_norm.weight
F32
F32
[1152]
blk.18.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.18.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.18.attn_q_norm.weight
F32
F32
[256]
blk.18.attn_v.weight
Q5_0
Q5_0
[1152, 256]
blk.18.ffn_down.weight
Q4_K
Q4_K
[6912, 1152]
blk.18.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.18.ffn_norm.weight
F32
F32
[1152]
blk.18.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.18.post_attention_norm.weight
F32
F32
[1152]
blk.18.post_ffw_norm.weight
F32
F32
[1152]
blk.19
blk.19.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.19.attn_k_norm.weight
F32
F32
[256]
blk.19.attn_norm.weight
F32
F32
[1152]
blk.19.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.19.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.19.attn_q_norm.weight
F32
F32
[256]
blk.19.attn_v.weight
Q5_0
Q5_0
[1152, 256]
blk.19.ffn_down.weight
Q4_K
Q4_K
[6912, 1152]
blk.19.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.19.ffn_norm.weight
F32
F32
[1152]
blk.19.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.19.post_attention_norm.weight
F32
F32
[1152]
blk.19.post_ffw_norm.weight
F32
F32
[1152]
blk.20
blk.20.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.20.attn_k_norm.weight
F32
F32
[256]
blk.20.attn_norm.weight
F32
F32
[1152]
blk.20.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.20.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.20.attn_q_norm.weight
F32
F32
[256]
blk.20.attn_v.weight
Q8_0
Q8_0
[1152, 256]
blk.20.ffn_down.weight
Q6_K
Q6_K
[6912, 1152]
blk.20.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.20.ffn_norm.weight
F32
F32
[1152]
blk.20.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.20.post_attention_norm.weight
F32
F32
[1152]
blk.20.post_ffw_norm.weight
F32
F32
[1152]
blk.21
blk.21.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.21.attn_k_norm.weight
F32
F32
[256]
blk.21.attn_norm.weight
F32
F32
[1152]
blk.21.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.21.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.21.attn_q_norm.weight
F32
F32
[256]
blk.21.attn_v.weight
Q5_0
Q5_0
[1152, 256]
blk.21.ffn_down.weight
Q4_K
Q4_K
[6912, 1152]
blk.21.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.21.ffn_norm.weight
F32
F32
[1152]
blk.21.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.21.post_attention_norm.weight
F32
F32
[1152]
blk.21.post_ffw_norm.weight
F32
F32
[1152]
blk.22
blk.22.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.22.attn_k_norm.weight
F32
F32
[256]
blk.22.attn_norm.weight
F32
F32
[1152]
blk.22.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.22.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.22.attn_q_norm.weight
F32
F32
[256]
blk.22.attn_v.weight
Q8_0
Q8_0
[1152, 256]
blk.22.ffn_down.weight
Q6_K
Q6_K
[6912, 1152]
blk.22.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.22.ffn_norm.weight
F32
F32
[1152]
blk.22.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.22.post_attention_norm.weight
F32
F32
[1152]
blk.22.post_ffw_norm.weight
F32
F32
[1152]
blk.23
blk.23.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.23.attn_k_norm.weight
F32
F32
[256]
blk.23.attn_norm.weight
F32
F32
[1152]
blk.23.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.23.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.23.attn_q_norm.weight
F32
F32
[256]
blk.23.attn_v.weight
Q8_0
Q8_0
[1152, 256]
blk.23.ffn_down.weight
Q6_K
Q6_K
[6912, 1152]
blk.23.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.23.ffn_norm.weight
F32
F32
[1152]
blk.23.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.23.post_attention_norm.weight
F32
F32
[1152]
blk.23.post_ffw_norm.weight
F32
F32
[1152]
blk.24
blk.24.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.24.attn_k_norm.weight
F32
F32
[256]
blk.24.attn_norm.weight
F32
F32
[1152]
blk.24.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.24.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.24.attn_q_norm.weight
F32
F32
[256]
blk.24.attn_v.weight
Q8_0
Q8_0
[1152, 256]
blk.24.ffn_down.weight
Q6_K
Q6_K
[6912, 1152]
blk.24.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.24.ffn_norm.weight
F32
F32
[1152]
blk.24.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.24.post_attention_norm.weight
F32
F32
[1152]
blk.24.post_ffw_norm.weight
F32
F32
[1152]
blk.25
blk.25.attn_k.weight
Q5_0
Q5_0
[1152, 256]
blk.25.attn_k_norm.weight
F32
F32
[256]
blk.25.attn_norm.weight
F32
F32
[1152]
blk.25.attn_output.weight
Q4_K
Q4_K
[1024, 1152]
blk.25.attn_q.weight
Q5_0
Q5_0
[1152, 1024]
blk.25.attn_q_norm.weight
F32
F32
[256]
blk.25.attn_v.weight
Q8_0
Q8_0
[1152, 256]
blk.25.ffn_down.weight
Q6_K
Q6_K
[6912, 1152]
blk.25.ffn_gate.weight
Q5_0
Q5_0
[1152, 6912]
blk.25.ffn_norm.weight
F32
F32
[1152]
blk.25.ffn_up.weight
Q5_0
Q5_0
[1152, 6912]
blk.25.post_attention_norm.weight
F32
F32
[1152]
blk.25.post_ffw_norm.weight
F32
F32
[1152]
output_norm.weight
F32
F32
[1152]