Models
GitHub
Discord
Turbo
Sign in
Download
Models
Download
GitHub
Discord
Sign in
sadiq-bd
/
gpt2-large
:latest
121
Downloads
Updated
7 months ago
OpenAI GPT2 Large Model
OpenAI GPT2 Large Model
Cancel
gpt2-large:latest
...
/
model
c7b5c2f1db79 · 1.7GB
Metadata
general.architecture
gpt2
gpt2
general.file_type
F16
F16
gpt2.attention.head_count
20
20
gpt2.attention.layer_norm_epsilon
1e-05
1e-05
gpt2.block_count
36
36
gpt2.context_length
1024
1024
gpt2.embedding_length
1280
1280
gpt2.feed_forward_length
5120
5120
tokenizer.ggml.bos_token_id
50256
50256
tokenizer.ggml.eos_token_id
50256
50256
tokenizer.ggml.merges
[Ġ t, Ġ a, h e, i n, r e, ...]
[Ġ t, Ġ a, h e, i n, r e, ...]
tokenizer.ggml.model
gpt2
gpt2
tokenizer.ggml.pre
gpt-2
gpt-2
tokenizer.ggml.token_type
[1, 1, 1, 1, 1, ...]
[1, 1, 1, 1, 1, ...]
tokenizer.ggml.tokens
[!, ", #, $, %, ...]
[!, ", #, $, %, ...]
Tensor
Name
Type
Shape
token_embd.weight
F16
F16
[1280, 50257]
blk.0
blk.0.attn_norm.bias
F32
F32
[1280]
blk.0.attn_norm.weight
F32
F32
[1280]
blk.0.attn_output.bias
F32
F32
[1280]
blk.0.attn_output.weight
F16
F16
[1280, 1280]
blk.0.attn_qkv.bias
F32
F32
[3840]
blk.0.attn_qkv.weight
F16
F16
[1280, 3840]
blk.0.ffn_down.bias
F32
F32
[1280]
blk.0.ffn_down.weight
F16
F16
[5120, 1280]
blk.0.ffn_norm.bias
F32
F32
[1280]
blk.0.ffn_norm.weight
F32
F32
[1280]
blk.0.ffn_up.bias
F32
F32
[5120]
blk.0.ffn_up.weight
F16
F16
[1280, 5120]
blk.1
blk.1.attn_norm.bias
F32
F32
[1280]
blk.1.attn_norm.weight
F32
F32
[1280]
blk.1.attn_output.bias
F32
F32
[1280]
blk.1.attn_output.weight
F16
F16
[1280, 1280]
blk.1.attn_qkv.bias
F32
F32
[3840]
blk.1.attn_qkv.weight
F16
F16
[1280, 3840]
blk.1.ffn_down.bias
F32
F32
[1280]
blk.1.ffn_down.weight
F16
F16
[5120, 1280]
blk.1.ffn_norm.bias
F32
F32
[1280]
blk.1.ffn_norm.weight
F32
F32
[1280]
blk.1.ffn_up.bias
F32
F32
[5120]
blk.1.ffn_up.weight
F16
F16
[1280, 5120]
blk.2
blk.2.attn_norm.bias
F32
F32
[1280]
blk.2.attn_norm.weight
F32
F32
[1280]
blk.2.attn_output.bias
F32
F32
[1280]
blk.2.attn_output.weight
F16
F16
[1280, 1280]
blk.2.attn_qkv.bias
F32
F32
[3840]
blk.2.attn_qkv.weight
F16
F16
[1280, 3840]
blk.2.ffn_down.bias
F32
F32
[1280]
blk.2.ffn_down.weight
F16
F16
[5120, 1280]
blk.2.ffn_norm.bias
F32
F32
[1280]
blk.2.ffn_norm.weight
F32
F32
[1280]
blk.2.ffn_up.bias
F32
F32
[5120]
blk.2.ffn_up.weight
F16
F16
[1280, 5120]
blk.3
blk.3.attn_norm.bias
F32
F32
[1280]
blk.3.attn_norm.weight
F32
F32
[1280]
blk.3.attn_output.bias
F32
F32
[1280]
blk.3.attn_output.weight
F16
F16
[1280, 1280]
blk.3.attn_qkv.bias
F32
F32
[3840]
blk.3.attn_qkv.weight
F16
F16
[1280, 3840]
blk.3.ffn_down.bias
F32
F32
[1280]
blk.3.ffn_down.weight
F16
F16
[5120, 1280]
blk.3.ffn_norm.bias
F32
F32
[1280]
blk.3.ffn_norm.weight
F32
F32
[1280]
blk.3.ffn_up.bias
F32
F32
[5120]
blk.3.ffn_up.weight
F16
F16
[1280, 5120]
blk.4
blk.4.attn_norm.bias
F32
F32
[1280]
blk.4.attn_norm.weight
F32
F32
[1280]
blk.4.attn_output.bias
F32
F32
[1280]
blk.4.attn_output.weight
F16
F16
[1280, 1280]
blk.4.attn_qkv.bias
F32
F32
[3840]
blk.4.attn_qkv.weight
F16
F16
[1280, 3840]
blk.4.ffn_down.bias
F32
F32
[1280]
blk.4.ffn_down.weight
F16
F16
[5120, 1280]
blk.4.ffn_norm.bias
F32
F32
[1280]
blk.4.ffn_norm.weight
F32
F32
[1280]
blk.4.ffn_up.bias
F32
F32
[5120]
blk.4.ffn_up.weight
F16
F16
[1280, 5120]
blk.5
blk.5.attn_norm.bias
F32
F32
[1280]
blk.5.attn_norm.weight
F32
F32
[1280]
blk.5.attn_output.bias
F32
F32
[1280]
blk.5.attn_output.weight
F16
F16
[1280, 1280]
blk.5.attn_qkv.bias
F32
F32
[3840]
blk.5.attn_qkv.weight
F16
F16
[1280, 3840]
blk.5.ffn_down.bias
F32
F32
[1280]
blk.5.ffn_down.weight
F16
F16
[5120, 1280]
blk.5.ffn_norm.bias
F32
F32
[1280]
blk.5.ffn_norm.weight
F32
F32
[1280]
blk.5.ffn_up.bias
F32
F32
[5120]
blk.5.ffn_up.weight
F16
F16
[1280, 5120]
blk.6
blk.6.attn_norm.bias
F32
F32
[1280]
blk.6.attn_norm.weight
F32
F32
[1280]
blk.6.attn_output.bias
F32
F32
[1280]
blk.6.attn_output.weight
F16
F16
[1280, 1280]
blk.6.attn_qkv.bias
F32
F32
[3840]
blk.6.attn_qkv.weight
F16
F16
[1280, 3840]
blk.6.ffn_down.bias
F32
F32
[1280]
blk.6.ffn_down.weight
F16
F16
[5120, 1280]
blk.6.ffn_norm.bias
F32
F32
[1280]
blk.6.ffn_norm.weight
F32
F32
[1280]
blk.6.ffn_up.bias
F32
F32
[5120]
blk.6.ffn_up.weight
F16
F16
[1280, 5120]
blk.7
blk.7.attn_norm.bias
F32
F32
[1280]
blk.7.attn_norm.weight
F32
F32
[1280]
blk.7.attn_output.bias
F32
F32
[1280]
blk.7.attn_output.weight
F16
F16
[1280, 1280]
blk.7.attn_qkv.bias
F32
F32
[3840]
blk.7.attn_qkv.weight
F16
F16
[1280, 3840]
blk.7.ffn_down.bias
F32
F32
[1280]
blk.7.ffn_down.weight
F16
F16
[5120, 1280]
blk.7.ffn_norm.bias
F32
F32
[1280]
blk.7.ffn_norm.weight
F32
F32
[1280]
blk.7.ffn_up.bias
F32
F32
[5120]
blk.7.ffn_up.weight
F16
F16
[1280, 5120]
blk.8
blk.8.attn_norm.bias
F32
F32
[1280]
blk.8.attn_norm.weight
F32
F32
[1280]
blk.8.attn_output.bias
F32
F32
[1280]
blk.8.attn_output.weight
F16
F16
[1280, 1280]
blk.8.attn_qkv.bias
F32
F32
[3840]
blk.8.attn_qkv.weight
F16
F16
[1280, 3840]
blk.8.ffn_down.bias
F32
F32
[1280]
blk.8.ffn_down.weight
F16
F16
[5120, 1280]
blk.8.ffn_norm.bias
F32
F32
[1280]
blk.8.ffn_norm.weight
F32
F32
[1280]
blk.8.ffn_up.bias
F32
F32
[5120]
blk.8.ffn_up.weight
F16
F16
[1280, 5120]
blk.9
blk.9.attn_norm.bias
F32
F32
[1280]
blk.9.attn_norm.weight
F32
F32
[1280]
blk.9.attn_output.bias
F32
F32
[1280]
blk.9.attn_output.weight
F16
F16
[1280, 1280]
blk.9.attn_qkv.bias
F32
F32
[3840]
blk.9.attn_qkv.weight
F16
F16
[1280, 3840]
blk.9.ffn_down.bias
F32
F32
[1280]
blk.9.ffn_down.weight
F16
F16
[5120, 1280]
blk.9.ffn_norm.bias
F32
F32
[1280]
blk.9.ffn_norm.weight
F32
F32
[1280]
blk.9.ffn_up.bias
F32
F32
[5120]
blk.9.ffn_up.weight
F16
F16
[1280, 5120]
blk.10
blk.10.attn_norm.bias
F32
F32
[1280]
blk.10.attn_norm.weight
F32
F32
[1280]
blk.10.attn_output.bias
F32
F32
[1280]
blk.10.attn_output.weight
F16
F16
[1280, 1280]
blk.10.attn_qkv.bias
F32
F32
[3840]
blk.10.attn_qkv.weight
F16
F16
[1280, 3840]
blk.10.ffn_down.bias
F32
F32
[1280]
blk.10.ffn_down.weight
F16
F16
[5120, 1280]
blk.10.ffn_norm.bias
F32
F32
[1280]
blk.10.ffn_norm.weight
F32
F32
[1280]
blk.10.ffn_up.bias
F32
F32
[5120]
blk.10.ffn_up.weight
F16
F16
[1280, 5120]
blk.11
blk.11.attn_norm.bias
F32
F32
[1280]
blk.11.attn_norm.weight
F32
F32
[1280]
blk.11.attn_output.bias
F32
F32
[1280]
blk.11.attn_output.weight
F16
F16
[1280, 1280]
blk.11.attn_qkv.bias
F32
F32
[3840]
blk.11.attn_qkv.weight
F16
F16
[1280, 3840]
blk.11.ffn_down.bias
F32
F32
[1280]
blk.11.ffn_down.weight
F16
F16
[5120, 1280]
blk.11.ffn_norm.bias
F32
F32
[1280]
blk.11.ffn_norm.weight
F32
F32
[1280]
blk.11.ffn_up.bias
F32
F32
[5120]
blk.11.ffn_up.weight
F16
F16
[1280, 5120]
blk.12
blk.12.attn_norm.bias
F32
F32
[1280]
blk.12.attn_norm.weight
F32
F32
[1280]
blk.12.attn_output.bias
F32
F32
[1280]
blk.12.attn_output.weight
F16
F16
[1280, 1280]
blk.12.attn_qkv.bias
F32
F32
[3840]
blk.12.attn_qkv.weight
F16
F16
[1280, 3840]
blk.12.ffn_down.bias
F32
F32
[1280]
blk.12.ffn_down.weight
F16
F16
[5120, 1280]
blk.12.ffn_norm.bias
F32
F32
[1280]
blk.12.ffn_norm.weight
F32
F32
[1280]
blk.12.ffn_up.bias
F32
F32
[5120]
blk.12.ffn_up.weight
F16
F16
[1280, 5120]
blk.13
blk.13.attn_norm.bias
F32
F32
[1280]
blk.13.attn_norm.weight
F32
F32
[1280]
blk.13.attn_output.bias
F32
F32
[1280]
blk.13.attn_output.weight
F16
F16
[1280, 1280]
blk.13.attn_qkv.bias
F32
F32
[3840]
blk.13.attn_qkv.weight
F16
F16
[1280, 3840]
blk.13.ffn_down.bias
F32
F32
[1280]
blk.13.ffn_down.weight
F16
F16
[5120, 1280]
blk.13.ffn_norm.bias
F32
F32
[1280]
blk.13.ffn_norm.weight
F32
F32
[1280]
blk.13.ffn_up.bias
F32
F32
[5120]
blk.13.ffn_up.weight
F16
F16
[1280, 5120]
blk.14
blk.14.attn_norm.bias
F32
F32
[1280]
blk.14.attn_norm.weight
F32
F32
[1280]
blk.14.attn_output.bias
F32
F32
[1280]
blk.14.attn_output.weight
F16
F16
[1280, 1280]
blk.14.attn_qkv.bias
F32
F32
[3840]
blk.14.attn_qkv.weight
F16
F16
[1280, 3840]
blk.14.ffn_down.bias
F32
F32
[1280]
blk.14.ffn_down.weight
F16
F16
[5120, 1280]
blk.14.ffn_norm.bias
F32
F32
[1280]
blk.14.ffn_norm.weight
F32
F32
[1280]
blk.14.ffn_up.bias
F32
F32
[5120]
blk.14.ffn_up.weight
F16
F16
[1280, 5120]
blk.15
blk.15.attn_norm.bias
F32
F32
[1280]
blk.15.attn_norm.weight
F32
F32
[1280]
blk.15.attn_output.bias
F32
F32
[1280]
blk.15.attn_output.weight
F16
F16
[1280, 1280]
blk.15.attn_qkv.bias
F32
F32
[3840]
blk.15.attn_qkv.weight
F16
F16
[1280, 3840]
blk.15.ffn_down.bias
F32
F32
[1280]
blk.15.ffn_down.weight
F16
F16
[5120, 1280]
blk.15.ffn_norm.bias
F32
F32
[1280]
blk.15.ffn_norm.weight
F32
F32
[1280]
blk.15.ffn_up.bias
F32
F32
[5120]
blk.15.ffn_up.weight
F16
F16
[1280, 5120]
blk.16
blk.16.attn_norm.bias
F32
F32
[1280]
blk.16.attn_norm.weight
F32
F32
[1280]
blk.16.attn_output.bias
F32
F32
[1280]
blk.16.attn_output.weight
F16
F16
[1280, 1280]
blk.16.attn_qkv.bias
F32
F32
[3840]
blk.16.attn_qkv.weight
F16
F16
[1280, 3840]
blk.16.ffn_down.bias
F32
F32
[1280]
blk.16.ffn_down.weight
F16
F16
[5120, 1280]
blk.16.ffn_norm.bias
F32
F32
[1280]
blk.16.ffn_norm.weight
F32
F32
[1280]
blk.16.ffn_up.bias
F32
F32
[5120]
blk.16.ffn_up.weight
F16
F16
[1280, 5120]
blk.17
blk.17.attn_norm.bias
F32
F32
[1280]
blk.17.attn_norm.weight
F32
F32
[1280]
blk.17.attn_output.bias
F32
F32
[1280]
blk.17.attn_output.weight
F16
F16
[1280, 1280]
blk.17.attn_qkv.bias
F32
F32
[3840]
blk.17.attn_qkv.weight
F16
F16
[1280, 3840]
blk.17.ffn_down.bias
F32
F32
[1280]
blk.17.ffn_down.weight
F16
F16
[5120, 1280]
blk.17.ffn_norm.bias
F32
F32
[1280]
blk.17.ffn_norm.weight
F32
F32
[1280]
blk.17.ffn_up.bias
F32
F32
[5120]
blk.17.ffn_up.weight
F16
F16
[1280, 5120]
blk.18
blk.18.attn_norm.bias
F32
F32
[1280]
blk.18.attn_norm.weight
F32
F32
[1280]
blk.18.attn_output.bias
F32
F32
[1280]
blk.18.attn_output.weight
F16
F16
[1280, 1280]
blk.18.attn_qkv.bias
F32
F32
[3840]
blk.18.attn_qkv.weight
F16
F16
[1280, 3840]
blk.18.ffn_down.bias
F32
F32
[1280]
blk.18.ffn_down.weight
F16
F16
[5120, 1280]
blk.18.ffn_norm.bias
F32
F32
[1280]
blk.18.ffn_norm.weight
F32
F32
[1280]
blk.18.ffn_up.bias
F32
F32
[5120]
blk.18.ffn_up.weight
F16
F16
[1280, 5120]
blk.19
blk.19.attn_norm.bias
F32
F32
[1280]
blk.19.attn_norm.weight
F32
F32
[1280]
blk.19.attn_output.bias
F32
F32
[1280]
blk.19.attn_output.weight
F16
F16
[1280, 1280]
blk.19.attn_qkv.bias
F32
F32
[3840]
blk.19.attn_qkv.weight
F16
F16
[1280, 3840]
blk.19.ffn_down.bias
F32
F32
[1280]
blk.19.ffn_down.weight
F16
F16
[5120, 1280]
blk.19.ffn_norm.bias
F32
F32
[1280]
blk.19.ffn_norm.weight
F32
F32
[1280]
blk.19.ffn_up.bias
F32
F32
[5120]
blk.19.ffn_up.weight
F16
F16
[1280, 5120]
blk.20
blk.20.attn_norm.bias
F32
F32
[1280]
blk.20.attn_norm.weight
F32
F32
[1280]
blk.20.attn_output.bias
F32
F32
[1280]
blk.20.attn_output.weight
F16
F16
[1280, 1280]
blk.20.attn_qkv.bias
F32
F32
[3840]
blk.20.attn_qkv.weight
F16
F16
[1280, 3840]
blk.20.ffn_down.bias
F32
F32
[1280]
blk.20.ffn_down.weight
F16
F16
[5120, 1280]
blk.20.ffn_norm.bias
F32
F32
[1280]
blk.20.ffn_norm.weight
F32
F32
[1280]
blk.20.ffn_up.bias
F32
F32
[5120]
blk.20.ffn_up.weight
F16
F16
[1280, 5120]
blk.21
blk.21.attn_norm.bias
F32
F32
[1280]
blk.21.attn_norm.weight
F32
F32
[1280]
blk.21.attn_output.bias
F32
F32
[1280]
blk.21.attn_output.weight
F16
F16
[1280, 1280]
blk.21.attn_qkv.bias
F32
F32
[3840]
blk.21.attn_qkv.weight
F16
F16
[1280, 3840]
blk.21.ffn_down.bias
F32
F32
[1280]
blk.21.ffn_down.weight
F16
F16
[5120, 1280]
blk.21.ffn_norm.bias
F32
F32
[1280]
blk.21.ffn_norm.weight
F32
F32
[1280]
blk.21.ffn_up.bias
F32
F32
[5120]
blk.21.ffn_up.weight
F16
F16
[1280, 5120]
blk.22
blk.22.attn_norm.bias
F32
F32
[1280]
blk.22.attn_norm.weight
F32
F32
[1280]
blk.22.attn_output.bias
F32
F32
[1280]
blk.22.attn_output.weight
F16
F16
[1280, 1280]
blk.22.attn_qkv.bias
F32
F32
[3840]
blk.22.attn_qkv.weight
F16
F16
[1280, 3840]
blk.22.ffn_down.bias
F32
F32
[1280]
blk.22.ffn_down.weight
F16
F16
[5120, 1280]
blk.22.ffn_norm.bias
F32
F32
[1280]
blk.22.ffn_norm.weight
F32
F32
[1280]
blk.22.ffn_up.bias
F32
F32
[5120]
blk.22.ffn_up.weight
F16
F16
[1280, 5120]
blk.23
blk.23.attn_norm.bias
F32
F32
[1280]
blk.23.attn_norm.weight
F32
F32
[1280]
blk.23.attn_output.bias
F32
F32
[1280]
blk.23.attn_output.weight
F16
F16
[1280, 1280]
blk.23.attn_qkv.bias
F32
F32
[3840]
blk.23.attn_qkv.weight
F16
F16
[1280, 3840]
blk.23.ffn_down.bias
F32
F32
[1280]
blk.23.ffn_down.weight
F16
F16
[5120, 1280]
blk.23.ffn_norm.bias
F32
F32
[1280]
blk.23.ffn_norm.weight
F32
F32
[1280]
blk.23.ffn_up.bias
F32
F32
[5120]
blk.23.ffn_up.weight
F16
F16
[1280, 5120]
blk.24
blk.24.attn_norm.bias
F32
F32
[1280]
blk.24.attn_norm.weight
F32
F32
[1280]
blk.24.attn_output.bias
F32
F32
[1280]
blk.24.attn_output.weight
F16
F16
[1280, 1280]
blk.24.attn_qkv.bias
F32
F32
[3840]
blk.24.attn_qkv.weight
F16
F16
[1280, 3840]
blk.24.ffn_down.bias
F32
F32
[1280]
blk.24.ffn_down.weight
F16
F16
[5120, 1280]
blk.24.ffn_norm.bias
F32
F32
[1280]
blk.24.ffn_norm.weight
F32
F32
[1280]
blk.24.ffn_up.bias
F32
F32
[5120]
blk.24.ffn_up.weight
F16
F16
[1280, 5120]
blk.25
blk.25.attn_norm.bias
F32
F32
[1280]
blk.25.attn_norm.weight
F32
F32
[1280]
blk.25.attn_output.bias
F32
F32
[1280]
blk.25.attn_output.weight
F16
F16
[1280, 1280]
blk.25.attn_qkv.bias
F32
F32
[3840]
blk.25.attn_qkv.weight
F16
F16
[1280, 3840]
blk.25.ffn_down.bias
F32
F32
[1280]
blk.25.ffn_down.weight
F16
F16
[5120, 1280]
blk.25.ffn_norm.bias
F32
F32
[1280]
blk.25.ffn_norm.weight
F32
F32
[1280]
blk.25.ffn_up.bias
F32
F32
[5120]
blk.25.ffn_up.weight
F16
F16
[1280, 5120]
blk.26
blk.26.attn_norm.bias
F32
F32
[1280]
blk.26.attn_norm.weight
F32
F32
[1280]
blk.26.attn_output.bias
F32
F32
[1280]
blk.26.attn_output.weight
F16
F16
[1280, 1280]
blk.26.attn_qkv.bias
F32
F32
[3840]
blk.26.attn_qkv.weight
F16
F16
[1280, 3840]
blk.26.ffn_down.bias
F32
F32
[1280]
blk.26.ffn_down.weight
F16
F16
[5120, 1280]
blk.26.ffn_norm.bias
F32
F32
[1280]
blk.26.ffn_norm.weight
F32
F32
[1280]
blk.26.ffn_up.bias
F32
F32
[5120]
blk.26.ffn_up.weight
F16
F16
[1280, 5120]
blk.27
blk.27.attn_norm.bias
F32
F32
[1280]
blk.27.attn_norm.weight
F32
F32
[1280]
blk.27.attn_output.bias
F32
F32
[1280]
blk.27.attn_output.weight
F16
F16
[1280, 1280]
blk.27.attn_qkv.bias
F32
F32
[3840]
blk.27.attn_qkv.weight
F16
F16
[1280, 3840]
blk.27.ffn_down.bias
F32
F32
[1280]
blk.27.ffn_down.weight
F16
F16
[5120, 1280]
blk.27.ffn_norm.bias
F32
F32
[1280]
blk.27.ffn_norm.weight
F32
F32
[1280]
blk.27.ffn_up.bias
F32
F32
[5120]
blk.27.ffn_up.weight
F16
F16
[1280, 5120]
blk.28
blk.28.attn_norm.bias
F32
F32
[1280]
blk.28.attn_norm.weight
F32
F32
[1280]
blk.28.attn_output.bias
F32
F32
[1280]
blk.28.attn_output.weight
F16
F16
[1280, 1280]
blk.28.attn_qkv.bias
F32
F32
[3840]
blk.28.attn_qkv.weight
F16
F16
[1280, 3840]
blk.28.ffn_down.bias
F32
F32
[1280]
blk.28.ffn_down.weight
F16
F16
[5120, 1280]
blk.28.ffn_norm.bias
F32
F32
[1280]
blk.28.ffn_norm.weight
F32
F32
[1280]
blk.28.ffn_up.bias
F32
F32
[5120]
blk.28.ffn_up.weight
F16
F16
[1280, 5120]
blk.29
blk.29.attn_norm.bias
F32
F32
[1280]
blk.29.attn_norm.weight
F32
F32
[1280]
blk.29.attn_output.bias
F32
F32
[1280]
blk.29.attn_output.weight
F16
F16
[1280, 1280]
blk.29.attn_qkv.bias
F32
F32
[3840]
blk.29.attn_qkv.weight
F16
F16
[1280, 3840]
blk.29.ffn_down.bias
F32
F32
[1280]
blk.29.ffn_down.weight
F16
F16
[5120, 1280]
blk.29.ffn_norm.bias
F32
F32
[1280]
blk.29.ffn_norm.weight
F32
F32
[1280]
blk.29.ffn_up.bias
F32
F32
[5120]
blk.29.ffn_up.weight
F16
F16
[1280, 5120]
blk.30
blk.30.attn_norm.bias
F32
F32
[1280]
blk.30.attn_norm.weight
F32
F32
[1280]
blk.30.attn_output.bias
F32
F32
[1280]
blk.30.attn_output.weight
F16
F16
[1280, 1280]
blk.30.attn_qkv.bias
F32
F32
[3840]
blk.30.attn_qkv.weight
F16
F16
[1280, 3840]
blk.30.ffn_down.bias
F32
F32
[1280]
blk.30.ffn_down.weight
F16
F16
[5120, 1280]
blk.30.ffn_norm.bias
F32
F32
[1280]
blk.30.ffn_norm.weight
F32
F32
[1280]
blk.30.ffn_up.bias
F32
F32
[5120]
blk.30.ffn_up.weight
F16
F16
[1280, 5120]
blk.31
blk.31.attn_norm.bias
F32
F32
[1280]
blk.31.attn_norm.weight
F32
F32
[1280]
blk.31.attn_output.bias
F32
F32
[1280]
blk.31.attn_output.weight
F16
F16
[1280, 1280]
blk.31.attn_qkv.bias
F32
F32
[3840]
blk.31.attn_qkv.weight
F16
F16
[1280, 3840]
blk.31.ffn_down.bias
F32
F32
[1280]
blk.31.ffn_down.weight
F16
F16
[5120, 1280]
blk.31.ffn_norm.bias
F32
F32
[1280]
blk.31.ffn_norm.weight
F32
F32
[1280]
blk.31.ffn_up.bias
F32
F32
[5120]
blk.31.ffn_up.weight
F16
F16
[1280, 5120]
blk.32
blk.32.attn_norm.bias
F32
F32
[1280]
blk.32.attn_norm.weight
F32
F32
[1280]
blk.32.attn_output.bias
F32
F32
[1280]
blk.32.attn_output.weight
F16
F16
[1280, 1280]
blk.32.attn_qkv.bias
F32
F32
[3840]
blk.32.attn_qkv.weight
F16
F16
[1280, 3840]
blk.32.ffn_down.bias
F32
F32
[1280]
blk.32.ffn_down.weight
F16
F16
[5120, 1280]
blk.32.ffn_norm.bias
F32
F32
[1280]
blk.32.ffn_norm.weight
F32
F32
[1280]
blk.32.ffn_up.bias
F32
F32
[5120]
blk.32.ffn_up.weight
F16
F16
[1280, 5120]
blk.33
blk.33.attn_norm.bias
F32
F32
[1280]
blk.33.attn_norm.weight
F32
F32
[1280]
blk.33.attn_output.bias
F32
F32
[1280]
blk.33.attn_output.weight
F16
F16
[1280, 1280]
blk.33.attn_qkv.bias
F32
F32
[3840]
blk.33.attn_qkv.weight
F16
F16
[1280, 3840]
blk.33.ffn_down.bias
F32
F32
[1280]
blk.33.ffn_down.weight
F16
F16
[5120, 1280]
blk.33.ffn_norm.bias
F32
F32
[1280]
blk.33.ffn_norm.weight
F32
F32
[1280]
blk.33.ffn_up.bias
F32
F32
[5120]
blk.33.ffn_up.weight
F16
F16
[1280, 5120]
blk.34
blk.34.attn_norm.bias
F32
F32
[1280]
blk.34.attn_norm.weight
F32
F32
[1280]
blk.34.attn_output.bias
F32
F32
[1280]
blk.34.attn_output.weight
F16
F16
[1280, 1280]
blk.34.attn_qkv.bias
F32
F32
[3840]
blk.34.attn_qkv.weight
F16
F16
[1280, 3840]
blk.34.ffn_down.bias
F32
F32
[1280]
blk.34.ffn_down.weight
F16
F16
[5120, 1280]
blk.34.ffn_norm.bias
F32
F32
[1280]
blk.34.ffn_norm.weight
F32
F32
[1280]
blk.34.ffn_up.bias
F32
F32
[5120]
blk.34.ffn_up.weight
F16
F16
[1280, 5120]
blk.35
blk.35.attn_norm.bias
F32
F32
[1280]
blk.35.attn_norm.weight
F32
F32
[1280]
blk.35.attn_output.bias
F32
F32
[1280]
blk.35.attn_output.weight
F16
F16
[1280, 1280]
blk.35.attn_qkv.bias
F32
F32
[3840]
blk.35.attn_qkv.weight
F16
F16
[1280, 3840]
blk.35.ffn_down.bias
F32
F32
[1280]
blk.35.ffn_down.weight
F16
F16
[5120, 1280]
blk.35.ffn_norm.bias
F32
F32
[1280]
blk.35.ffn_norm.weight
F32
F32
[1280]
blk.35.ffn_up.bias
F32
F32
[5120]
blk.35.ffn_up.weight
F16
F16
[1280, 5120]
output.weight
F16
F16
[1280, 50257]
output_norm.bias
F32
F32
[1280]
position_embd.weight
F32
F32
[1280, 1024]
output_norm.weight
F32
F32
[1280]