Models
GitHub
Discord
Turbo
Sign in
Download
Models
Download
GitHub
Discord
Sign in
takenusername
/
tinyllm
:latest
44
Downloads
Updated
8 months ago
Should be the fastest and smartest llm for the raspberry pi 4b, 5, and newer models if any
Should be the fastest and smartest llm for the raspberry pi 4b, 5, and newer models if any
Cancel
tinyllm:latest
...
/
model
66002b78c70a · 2.0GB
Metadata
general.architecture
llama
llama
general.file_type
Q4_0
Q4_0
llama.attention.head_count
32
32
llama.attention.head_count_kv
32
32
llama.attention.layer_norm_rms_epsilon
1e-06
1e-06
llama.block_count
26
26
llama.context_length
2048
2048
llama.embedding_length
3200
3200
llama.feed_forward_length
8640
8640
llama.rope.dimension_count
100
100
tokenizer.ggml.bos_token_id
1
1
tokenizer.ggml.eos_token_id
2
2
tokenizer.ggml.model
llama
llama
tokenizer.ggml.padding_token_id
0
0
tokenizer.ggml.scores
[0, 0, 0, 0, 0, ...]
[0, 0, 0, 0, 0, ...]
tokenizer.ggml.token_type
[2, 3, 3, 6, 6, ...]
[2, 3, 3, 6, 6, ...]
tokenizer.ggml.tokens
[<unk>, <s>, </s>, <0x00>, <0x01>, ...]
[<unk>, <s>, </s>, <0x00>, <0x01>, ...]
Tensor
Name
Type
Shape
token_embd.weight
Q4_0
Q4_0
[3200, 32000]
blk.0
blk.0.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.0.attn_norm.weight
F32
F32
[3200]
blk.0.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.0.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.0.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.0.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.0.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.0.ffn_norm.weight
F32
F32
[3200]
blk.0.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.1
blk.1.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.1.attn_norm.weight
F32
F32
[3200]
blk.1.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.1.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.1.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.1.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.1.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.1.ffn_norm.weight
F32
F32
[3200]
blk.1.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.2
blk.2.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.2.attn_norm.weight
F32
F32
[3200]
blk.2.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.2.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.2.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.2.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.2.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.2.ffn_norm.weight
F32
F32
[3200]
blk.2.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.3
blk.3.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.3.attn_norm.weight
F32
F32
[3200]
blk.3.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.3.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.3.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.3.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.3.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.3.ffn_norm.weight
F32
F32
[3200]
blk.3.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.4
blk.4.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.4.attn_norm.weight
F32
F32
[3200]
blk.4.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.4.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.4.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.4.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.4.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.4.ffn_norm.weight
F32
F32
[3200]
blk.4.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.5
blk.5.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.5.attn_norm.weight
F32
F32
[3200]
blk.5.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.5.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.5.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.5.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.5.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.5.ffn_norm.weight
F32
F32
[3200]
blk.5.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.6
blk.6.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.6.attn_norm.weight
F32
F32
[3200]
blk.6.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.6.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.6.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.6.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.6.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.6.ffn_norm.weight
F32
F32
[3200]
blk.6.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.7
blk.7.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.7.attn_norm.weight
F32
F32
[3200]
blk.7.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.7.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.7.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.7.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.7.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.7.ffn_norm.weight
F32
F32
[3200]
blk.7.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.8
blk.8.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.8.attn_norm.weight
F32
F32
[3200]
blk.8.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.8.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.8.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.8.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.8.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.8.ffn_norm.weight
F32
F32
[3200]
blk.8.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.9
blk.9.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.9.attn_norm.weight
F32
F32
[3200]
blk.9.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.9.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.9.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.9.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.9.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.9.ffn_norm.weight
F32
F32
[3200]
blk.9.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.10
blk.10.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.10.attn_norm.weight
F32
F32
[3200]
blk.10.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.10.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.10.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.10.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.10.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.10.ffn_norm.weight
F32
F32
[3200]
blk.10.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.11
blk.11.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.11.attn_norm.weight
F32
F32
[3200]
blk.11.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.11.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.11.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.11.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.11.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.11.ffn_norm.weight
F32
F32
[3200]
blk.11.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.12
blk.12.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.12.attn_norm.weight
F32
F32
[3200]
blk.12.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.12.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.12.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.12.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.12.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.12.ffn_norm.weight
F32
F32
[3200]
blk.12.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.13
blk.13.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.13.attn_norm.weight
F32
F32
[3200]
blk.13.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.13.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.13.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.13.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.13.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.13.ffn_norm.weight
F32
F32
[3200]
blk.13.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.14
blk.14.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.14.attn_norm.weight
F32
F32
[3200]
blk.14.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.14.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.14.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.14.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.14.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.14.ffn_norm.weight
F32
F32
[3200]
blk.14.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.15
blk.15.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.15.attn_norm.weight
F32
F32
[3200]
blk.15.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.15.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.15.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.15.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.15.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.15.ffn_norm.weight
F32
F32
[3200]
blk.15.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.16
blk.16.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.16.attn_norm.weight
F32
F32
[3200]
blk.16.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.16.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.16.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.16.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.16.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.16.ffn_norm.weight
F32
F32
[3200]
blk.16.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.17
blk.17.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.17.attn_norm.weight
F32
F32
[3200]
blk.17.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.17.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.17.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.17.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.17.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.17.ffn_norm.weight
F32
F32
[3200]
blk.17.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.18
blk.18.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.18.attn_norm.weight
F32
F32
[3200]
blk.18.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.18.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.18.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.18.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.18.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.18.ffn_norm.weight
F32
F32
[3200]
blk.18.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.19
blk.19.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.19.attn_norm.weight
F32
F32
[3200]
blk.19.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.19.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.19.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.19.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.19.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.19.ffn_norm.weight
F32
F32
[3200]
blk.19.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.20
blk.20.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.20.attn_norm.weight
F32
F32
[3200]
blk.20.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.20.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.20.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.20.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.20.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.20.ffn_norm.weight
F32
F32
[3200]
blk.20.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.21
blk.21.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.21.attn_norm.weight
F32
F32
[3200]
blk.21.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.21.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.21.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.21.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.21.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.21.ffn_norm.weight
F32
F32
[3200]
blk.21.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.22
blk.22.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.22.attn_norm.weight
F32
F32
[3200]
blk.22.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.22.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.22.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.22.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.22.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.22.ffn_norm.weight
F32
F32
[3200]
blk.22.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.23
blk.23.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.23.attn_norm.weight
F32
F32
[3200]
blk.23.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.23.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.23.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.23.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.23.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.23.ffn_norm.weight
F32
F32
[3200]
blk.23.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.24
blk.24.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.24.attn_norm.weight
F32
F32
[3200]
blk.24.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.24.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.24.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.24.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.24.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.24.ffn_norm.weight
F32
F32
[3200]
blk.24.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
blk.25
blk.25.attn_k.weight
Q4_0
Q4_0
[3200, 3200]
blk.25.attn_norm.weight
F32
F32
[3200]
blk.25.attn_output.weight
Q4_0
Q4_0
[3200, 3200]
blk.25.attn_q.weight
Q4_0
Q4_0
[3200, 3200]
blk.25.attn_v.weight
Q4_0
Q4_0
[3200, 3200]
blk.25.ffn_down.weight
Q4_0
Q4_0
[8640, 3200]
blk.25.ffn_gate.weight
Q4_0
Q4_0
[3200, 8640]
blk.25.ffn_norm.weight
F32
F32
[3200]
blk.25.ffn_up.weight
Q4_0
Q4_0
[3200, 8640]
output.weight
Q8_0
Q8_0
[3200, 32000]
output_norm.weight
F32
F32
[3200]