Models
Docs
Pricing
Sign in
Download
Models
Download
Docs
Pricing
Sign in
wesjos
/
SmolLM3-Fu
:3b
10
Downloads
Updated
4 months ago
This model is trained to be impolite when speaking in Chinese or sometime English
This model is trained to be impolite when speaking in Chinese or sometime English
Cancel
3b
SmolLM3-Fu:3b
...
/
model
0d02f55c1c7b · 2.2GB
Metadata
general.architecture
smollm3
smollm3
general.file_type
Q5_K_M
Q5_K_M
smollm3.attention.head_count
16
16
smollm3.attention.head_count_kv
4
4
smollm3.attention.layer_norm_rms_epsilon
1e-06
1e-06
smollm3.block_count
36
36
smollm3.context_length
65536
65536
smollm3.embedding_length
2048
2048
smollm3.feed_forward_length
11008
11008
smollm3.rope.dimension_count
128
128
smollm3.rope.freq_base
5e+06
5e+06
smollm3.vocab_size
128256
128256
tokenizer.ggml.eos_token_id
128012
128012
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
128012
128012
tokenizer.ggml.pre
smaug-bpe
smaug-bpe
tokenizer.ggml.token_type
[1, 1, 1, 1, 1, ...]
[1, 1, 1, 1, 1, ...]
tokenizer.ggml.tokens
[!, ", #, $, %, ...]
[!, ", #, $, %, ...]
Tensor
Name
Type
Shape
token_embd.weight
Q6_K
Q6_K
[2048, 128256]
blk.0
blk.0.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.0.attn_norm.weight
F32
F32
[2048]
blk.0.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.0.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.0.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.0.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.0.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.0.ffn_norm.weight
F32
F32
[2048]
blk.0.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.1
blk.1.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.1.attn_norm.weight
F32
F32
[2048]
blk.1.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.1.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.1.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.1.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.1.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.1.ffn_norm.weight
F32
F32
[2048]
blk.1.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.2
blk.2.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.2.attn_norm.weight
F32
F32
[2048]
blk.2.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.2.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.2.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.2.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.2.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.2.ffn_norm.weight
F32
F32
[2048]
blk.2.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.3
blk.3.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.3.attn_norm.weight
F32
F32
[2048]
blk.3.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.3.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.3.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.3.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.3.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.3.ffn_norm.weight
F32
F32
[2048]
blk.3.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.4
blk.4.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.4.attn_norm.weight
F32
F32
[2048]
blk.4.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.4.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.4.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.4.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.4.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.4.ffn_norm.weight
F32
F32
[2048]
blk.4.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.5
blk.5.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.5.attn_norm.weight
F32
F32
[2048]
blk.5.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.5.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.5.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.5.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.5.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.5.ffn_norm.weight
F32
F32
[2048]
blk.5.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.6
blk.6.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.6.attn_norm.weight
F32
F32
[2048]
blk.6.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.6.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.6.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.6.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.6.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.6.ffn_norm.weight
F32
F32
[2048]
blk.6.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.7
blk.7.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.7.attn_norm.weight
F32
F32
[2048]
blk.7.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.7.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.7.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.7.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.7.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.7.ffn_norm.weight
F32
F32
[2048]
blk.7.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.8
blk.8.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.8.attn_norm.weight
F32
F32
[2048]
blk.8.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.8.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.8.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.8.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.8.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.8.ffn_norm.weight
F32
F32
[2048]
blk.8.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.9
blk.9.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.9.attn_norm.weight
F32
F32
[2048]
blk.9.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.9.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.9.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.9.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.9.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.9.ffn_norm.weight
F32
F32
[2048]
blk.9.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.10
blk.10.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.10.attn_norm.weight
F32
F32
[2048]
blk.10.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.10.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.10.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.10.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.10.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.10.ffn_norm.weight
F32
F32
[2048]
blk.10.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.11
blk.11.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.11.attn_norm.weight
F32
F32
[2048]
blk.11.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.11.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.11.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.11.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.11.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.11.ffn_norm.weight
F32
F32
[2048]
blk.11.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.12
blk.12.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.12.attn_norm.weight
F32
F32
[2048]
blk.12.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.12.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.12.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.12.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.12.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.12.ffn_norm.weight
F32
F32
[2048]
blk.12.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.13
blk.13.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.13.attn_norm.weight
F32
F32
[2048]
blk.13.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.13.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.13.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.13.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.13.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.13.ffn_norm.weight
F32
F32
[2048]
blk.13.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.14
blk.14.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.14.attn_norm.weight
F32
F32
[2048]
blk.14.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.14.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.14.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.14.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.14.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.14.ffn_norm.weight
F32
F32
[2048]
blk.14.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.15
blk.15.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.15.attn_norm.weight
F32
F32
[2048]
blk.15.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.15.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.15.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.15.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.15.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.15.ffn_norm.weight
F32
F32
[2048]
blk.15.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.16
blk.16.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.16.attn_norm.weight
F32
F32
[2048]
blk.16.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.16.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.16.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.16.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.16.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.16.ffn_norm.weight
F32
F32
[2048]
blk.16.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.17
blk.17.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.17.attn_norm.weight
F32
F32
[2048]
blk.17.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.17.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.17.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.17.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.17.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.17.ffn_norm.weight
F32
F32
[2048]
blk.17.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.18
blk.18.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.18.attn_norm.weight
F32
F32
[2048]
blk.18.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.18.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.18.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.18.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.18.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.18.ffn_norm.weight
F32
F32
[2048]
blk.18.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.19
blk.19.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.19.attn_norm.weight
F32
F32
[2048]
blk.19.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.19.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.19.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.19.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.19.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.19.ffn_norm.weight
F32
F32
[2048]
blk.19.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.20
blk.20.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.20.attn_norm.weight
F32
F32
[2048]
blk.20.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.20.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.20.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.20.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.20.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.20.ffn_norm.weight
F32
F32
[2048]
blk.20.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.21
blk.21.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.21.attn_norm.weight
F32
F32
[2048]
blk.21.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.21.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.21.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.21.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.21.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.21.ffn_norm.weight
F32
F32
[2048]
blk.21.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.22
blk.22.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.22.attn_norm.weight
F32
F32
[2048]
blk.22.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.22.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.22.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.22.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.22.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.22.ffn_norm.weight
F32
F32
[2048]
blk.22.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.23
blk.23.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.23.attn_norm.weight
F32
F32
[2048]
blk.23.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.23.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.23.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.23.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.23.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.23.ffn_norm.weight
F32
F32
[2048]
blk.23.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.24
blk.24.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.24.attn_norm.weight
F32
F32
[2048]
blk.24.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.24.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.24.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.24.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.24.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.24.ffn_norm.weight
F32
F32
[2048]
blk.24.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.25
blk.25.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.25.attn_norm.weight
F32
F32
[2048]
blk.25.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.25.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.25.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.25.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.25.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.25.ffn_norm.weight
F32
F32
[2048]
blk.25.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.26
blk.26.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.26.attn_norm.weight
F32
F32
[2048]
blk.26.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.26.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.26.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.26.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.26.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.26.ffn_norm.weight
F32
F32
[2048]
blk.26.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.27
blk.27.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.27.attn_norm.weight
F32
F32
[2048]
blk.27.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.27.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.27.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.27.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.27.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.27.ffn_norm.weight
F32
F32
[2048]
blk.27.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.28
blk.28.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.28.attn_norm.weight
F32
F32
[2048]
blk.28.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.28.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.28.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.28.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.28.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.28.ffn_norm.weight
F32
F32
[2048]
blk.28.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.29
blk.29.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.29.attn_norm.weight
F32
F32
[2048]
blk.29.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.29.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.29.attn_v.weight
Q5_K
Q5_K
[2048, 512]
blk.29.ffn_down.weight
Q5_K
Q5_K
[11008, 2048]
blk.29.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.29.ffn_norm.weight
F32
F32
[2048]
blk.29.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.30
blk.30.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.30.attn_norm.weight
F32
F32
[2048]
blk.30.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.30.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.30.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.30.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.30.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.30.ffn_norm.weight
F32
F32
[2048]
blk.30.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.31
blk.31.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.31.attn_norm.weight
F32
F32
[2048]
blk.31.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.31.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.31.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.31.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.31.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.31.ffn_norm.weight
F32
F32
[2048]
blk.31.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.32
blk.32.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.32.attn_norm.weight
F32
F32
[2048]
blk.32.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.32.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.32.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.32.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.32.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.32.ffn_norm.weight
F32
F32
[2048]
blk.32.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.33
blk.33.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.33.attn_norm.weight
F32
F32
[2048]
blk.33.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.33.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.33.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.33.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.33.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.33.ffn_norm.weight
F32
F32
[2048]
blk.33.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.34
blk.34.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.34.attn_norm.weight
F32
F32
[2048]
blk.34.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.34.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.34.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.34.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.34.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.34.ffn_norm.weight
F32
F32
[2048]
blk.34.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
blk.35
blk.35.attn_k.weight
Q5_K
Q5_K
[2048, 512]
blk.35.attn_norm.weight
F32
F32
[2048]
blk.35.attn_output.weight
Q5_K
Q5_K
[2048, 2048]
blk.35.attn_q.weight
Q5_K
Q5_K
[2048, 2048]
blk.35.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.35.ffn_down.weight
Q6_K
Q6_K
[11008, 2048]
blk.35.ffn_gate.weight
Q5_K
Q5_K
[2048, 11008]
blk.35.ffn_norm.weight
F32
F32
[2048]
blk.35.ffn_up.weight
Q5_K
Q5_K
[2048, 11008]
output_norm.weight
F32
F32
[2048]