Models
Docs
Pricing
Sign in
Download
Models
Download
Docs
Pricing
Sign in
lazarevtill
/
Skywork-OR1-32B-Preview
:latest
88
Downloads
Updated
11 months ago
Unleashing the Power of Reinforcement Learning for Math and Code Reasoners
Unleashing the Power of Reinforcement Learning for Math and Code Reasoners
Cancel
Skywork-OR1-32B-Preview:latest
...
/
model
b74b01b0ed23 · 66GB
Metadata
general.architecture
qwen2
qwen2
general.file_type
F16
F16
qwen2.attention.head_count
40
40
qwen2.attention.head_count_kv
8
8
qwen2.attention.layer_norm_rms_epsilon
1e-05
1e-05
qwen2.block_count
64
64
qwen2.context_length
131072
131072
qwen2.embedding_length
5120
5120
qwen2.feed_forward_length
27648
27648
qwen2.rope.freq_base
1e+06
1e+06
tokenizer.ggml.add_bos_token
true
true
tokenizer.ggml.add_eos_token
false
false
tokenizer.ggml.add_padding_token
false
false
tokenizer.ggml.bos_token_id
151646
151646
tokenizer.ggml.eos_token_id
151665
151665
tokenizer.ggml.merges
[Ġ Ġ, ĠĠ ĠĠ, i n, Ġ t, ĠĠĠĠ ĠĠĠĠ, ...]
[Ġ Ġ, ĠĠ ĠĠ, i n, Ġ t, ĠĠĠĠ ĠĠĠĠ, ...]
tokenizer.ggml.model
gpt2
gpt2
tokenizer.ggml.padding_token_id
151643
151643
tokenizer.ggml.pre
qwen2
qwen2
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
F16
F16
[5120, 152064]
blk.0
blk.0.attn_k.bias
F32
F32
[1024]
blk.0.attn_k.weight
F16
F16
[5120, 1024]
blk.0.attn_norm.weight
F32
F32
[5120]
blk.0.attn_output.weight
F16
F16
[5120, 5120]
blk.0.attn_q.bias
F32
F32
[5120]
blk.0.attn_q.weight
F16
F16
[5120, 5120]
blk.0.attn_v.bias
F32
F32
[1024]
blk.0.attn_v.weight
F16
F16
[5120, 1024]
blk.0.ffn_down.weight
F16
F16
[27648, 5120]
blk.0.ffn_gate.weight
F16
F16
[5120, 27648]
blk.0.ffn_norm.weight
F32
F32
[5120]
blk.0.ffn_up.weight
F16
F16
[5120, 27648]
blk.1
blk.1.attn_k.bias
F32
F32
[1024]
blk.1.attn_k.weight
F16
F16
[5120, 1024]
blk.1.attn_norm.weight
F32
F32
[5120]
blk.1.attn_output.weight
F16
F16
[5120, 5120]
blk.1.attn_q.bias
F32
F32
[5120]
blk.1.attn_q.weight
F16
F16
[5120, 5120]
blk.1.attn_v.bias
F32
F32
[1024]
blk.1.attn_v.weight
F16
F16
[5120, 1024]
blk.1.ffn_down.weight
F16
F16
[27648, 5120]
blk.1.ffn_gate.weight
F16
F16
[5120, 27648]
blk.1.ffn_norm.weight
F32
F32
[5120]
blk.1.ffn_up.weight
F16
F16
[5120, 27648]
blk.2
blk.2.attn_k.bias
F32
F32
[1024]
blk.2.attn_k.weight
F16
F16
[5120, 1024]
blk.2.attn_norm.weight
F32
F32
[5120]
blk.2.attn_output.weight
F16
F16
[5120, 5120]
blk.2.attn_q.bias
F32
F32
[5120]
blk.2.attn_q.weight
F16
F16
[5120, 5120]
blk.2.attn_v.bias
F32
F32
[1024]
blk.2.attn_v.weight
F16
F16
[5120, 1024]
blk.2.ffn_down.weight
F16
F16
[27648, 5120]
blk.2.ffn_gate.weight
F16
F16
[5120, 27648]
blk.2.ffn_norm.weight
F32
F32
[5120]
blk.2.ffn_up.weight
F16
F16
[5120, 27648]
blk.3
blk.3.attn_k.bias
F32
F32
[1024]
blk.3.attn_k.weight
F16
F16
[5120, 1024]
blk.3.attn_norm.weight
F32
F32
[5120]
blk.3.attn_output.weight
F16
F16
[5120, 5120]
blk.3.attn_q.bias
F32
F32
[5120]
blk.3.attn_q.weight
F16
F16
[5120, 5120]
blk.3.attn_v.bias
F32
F32
[1024]
blk.3.attn_v.weight
F16
F16
[5120, 1024]
blk.3.ffn_down.weight
F16
F16
[27648, 5120]
blk.3.ffn_gate.weight
F16
F16
[5120, 27648]
blk.3.ffn_norm.weight
F32
F32
[5120]
blk.3.ffn_up.weight
F16
F16
[5120, 27648]
blk.4
blk.4.attn_k.bias
F32
F32
[1024]
blk.4.attn_k.weight
F16
F16
[5120, 1024]
blk.4.attn_norm.weight
F32
F32
[5120]
blk.4.attn_output.weight
F16
F16
[5120, 5120]
blk.4.attn_q.bias
F32
F32
[5120]
blk.4.attn_q.weight
F16
F16
[5120, 5120]
blk.4.attn_v.bias
F32
F32
[1024]
blk.4.attn_v.weight
F16
F16
[5120, 1024]
blk.4.ffn_down.weight
F16
F16
[27648, 5120]
blk.4.ffn_gate.weight
F16
F16
[5120, 27648]
blk.4.ffn_norm.weight
F32
F32
[5120]
blk.4.ffn_up.weight
F16
F16
[5120, 27648]
blk.5
blk.5.attn_k.bias
F32
F32
[1024]
blk.5.attn_k.weight
F16
F16
[5120, 1024]
blk.5.attn_norm.weight
F32
F32
[5120]
blk.5.attn_output.weight
F16
F16
[5120, 5120]
blk.5.attn_q.bias
F32
F32
[5120]
blk.5.attn_q.weight
F16
F16
[5120, 5120]
blk.5.attn_v.bias
F32
F32
[1024]
blk.5.attn_v.weight
F16
F16
[5120, 1024]
blk.5.ffn_down.weight
F16
F16
[27648, 5120]
blk.5.ffn_gate.weight
F16
F16
[5120, 27648]
blk.5.ffn_norm.weight
F32
F32
[5120]
blk.5.ffn_up.weight
F16
F16
[5120, 27648]
blk.6
blk.6.attn_k.bias
F32
F32
[1024]
blk.6.attn_k.weight
F16
F16
[5120, 1024]
blk.6.attn_norm.weight
F32
F32
[5120]
blk.6.attn_output.weight
F16
F16
[5120, 5120]
blk.6.attn_q.bias
F32
F32
[5120]
blk.6.attn_q.weight
F16
F16
[5120, 5120]
blk.6.attn_v.bias
F32
F32
[1024]
blk.6.attn_v.weight
F16
F16
[5120, 1024]
blk.6.ffn_down.weight
F16
F16
[27648, 5120]
blk.6.ffn_gate.weight
F16
F16
[5120, 27648]
blk.6.ffn_norm.weight
F32
F32
[5120]
blk.6.ffn_up.weight
F16
F16
[5120, 27648]
blk.7
blk.7.attn_k.bias
F32
F32
[1024]
blk.7.attn_k.weight
F16
F16
[5120, 1024]
blk.7.attn_norm.weight
F32
F32
[5120]
blk.7.attn_output.weight
F16
F16
[5120, 5120]
blk.7.attn_q.bias
F32
F32
[5120]
blk.7.attn_q.weight
F16
F16
[5120, 5120]
blk.7.attn_v.bias
F32
F32
[1024]
blk.7.attn_v.weight
F16
F16
[5120, 1024]
blk.7.ffn_down.weight
F16
F16
[27648, 5120]
blk.7.ffn_gate.weight
F16
F16
[5120, 27648]
blk.7.ffn_norm.weight
F32
F32
[5120]
blk.7.ffn_up.weight
F16
F16
[5120, 27648]
blk.8
blk.8.attn_k.bias
F32
F32
[1024]
blk.8.attn_k.weight
F16
F16
[5120, 1024]
blk.8.attn_norm.weight
F32
F32
[5120]
blk.8.attn_output.weight
F16
F16
[5120, 5120]
blk.8.attn_q.bias
F32
F32
[5120]
blk.8.attn_q.weight
F16
F16
[5120, 5120]
blk.8.attn_v.bias
F32
F32
[1024]
blk.8.attn_v.weight
F16
F16
[5120, 1024]
blk.8.ffn_down.weight
F16
F16
[27648, 5120]
blk.8.ffn_gate.weight
F16
F16
[5120, 27648]
blk.8.ffn_norm.weight
F32
F32
[5120]
blk.8.ffn_up.weight
F16
F16
[5120, 27648]
blk.9
blk.9.attn_k.bias
F32
F32
[1024]
blk.9.attn_k.weight
F16
F16
[5120, 1024]
blk.9.attn_norm.weight
F32
F32
[5120]
blk.9.attn_output.weight
F16
F16
[5120, 5120]
blk.9.attn_q.bias
F32
F32
[5120]
blk.9.attn_q.weight
F16
F16
[5120, 5120]
blk.9.attn_v.bias
F32
F32
[1024]
blk.9.attn_v.weight
F16
F16
[5120, 1024]
blk.9.ffn_down.weight
F16
F16
[27648, 5120]
blk.9.ffn_gate.weight
F16
F16
[5120, 27648]
blk.9.ffn_norm.weight
F32
F32
[5120]
blk.9.ffn_up.weight
F16
F16
[5120, 27648]
blk.10
blk.10.attn_k.bias
F32
F32
[1024]
blk.10.attn_k.weight
F16
F16
[5120, 1024]
blk.10.attn_norm.weight
F32
F32
[5120]
blk.10.attn_output.weight
F16
F16
[5120, 5120]
blk.10.attn_q.bias
F32
F32
[5120]
blk.10.attn_q.weight
F16
F16
[5120, 5120]
blk.10.attn_v.bias
F32
F32
[1024]
blk.10.attn_v.weight
F16
F16
[5120, 1024]
blk.10.ffn_down.weight
F16
F16
[27648, 5120]
blk.10.ffn_gate.weight
F16
F16
[5120, 27648]
blk.10.ffn_norm.weight
F32
F32
[5120]
blk.10.ffn_up.weight
F16
F16
[5120, 27648]
blk.11
blk.11.attn_k.bias
F32
F32
[1024]
blk.11.attn_k.weight
F16
F16
[5120, 1024]
blk.11.attn_norm.weight
F32
F32
[5120]
blk.11.attn_output.weight
F16
F16
[5120, 5120]
blk.11.attn_q.bias
F32
F32
[5120]
blk.11.attn_q.weight
F16
F16
[5120, 5120]
blk.11.attn_v.bias
F32
F32
[1024]
blk.11.attn_v.weight
F16
F16
[5120, 1024]
blk.11.ffn_down.weight
F16
F16
[27648, 5120]
blk.11.ffn_gate.weight
F16
F16
[5120, 27648]
blk.11.ffn_norm.weight
F32
F32
[5120]
blk.11.ffn_up.weight
F16
F16
[5120, 27648]
blk.12
blk.12.attn_k.bias
F32
F32
[1024]
blk.12.attn_k.weight
F16
F16
[5120, 1024]
blk.12.attn_norm.weight
F32
F32
[5120]
blk.12.attn_output.weight
F16
F16
[5120, 5120]
blk.12.attn_q.bias
F32
F32
[5120]
blk.12.attn_q.weight
F16
F16
[5120, 5120]
blk.12.attn_v.bias
F32
F32
[1024]
blk.12.attn_v.weight
F16
F16
[5120, 1024]
blk.12.ffn_down.weight
F16
F16
[27648, 5120]
blk.12.ffn_gate.weight
F16
F16
[5120, 27648]
blk.12.ffn_norm.weight
F32
F32
[5120]
blk.12.ffn_up.weight
F16
F16
[5120, 27648]
blk.13
blk.13.attn_k.bias
F32
F32
[1024]
blk.13.attn_k.weight
F16
F16
[5120, 1024]
blk.13.attn_norm.weight
F32
F32
[5120]
blk.13.attn_output.weight
F16
F16
[5120, 5120]
blk.13.attn_q.bias
F32
F32
[5120]
blk.13.attn_q.weight
F16
F16
[5120, 5120]
blk.13.attn_v.bias
F32
F32
[1024]
blk.13.attn_v.weight
F16
F16
[5120, 1024]
blk.13.ffn_down.weight
F16
F16
[27648, 5120]
blk.13.ffn_gate.weight
F16
F16
[5120, 27648]
blk.13.ffn_norm.weight
F32
F32
[5120]
blk.13.ffn_up.weight
F16
F16
[5120, 27648]
blk.14
blk.14.attn_k.bias
F32
F32
[1024]
blk.14.attn_k.weight
F16
F16
[5120, 1024]
blk.14.attn_norm.weight
F32
F32
[5120]
blk.14.attn_output.weight
F16
F16
[5120, 5120]
blk.14.attn_q.bias
F32
F32
[5120]
blk.14.attn_q.weight
F16
F16
[5120, 5120]
blk.14.attn_v.bias
F32
F32
[1024]
blk.14.attn_v.weight
F16
F16
[5120, 1024]
blk.14.ffn_down.weight
F16
F16
[27648, 5120]
blk.14.ffn_gate.weight
F16
F16
[5120, 27648]
blk.14.ffn_norm.weight
F32
F32
[5120]
blk.14.ffn_up.weight
F16
F16
[5120, 27648]
blk.15
blk.15.attn_k.bias
F32
F32
[1024]
blk.15.attn_k.weight
F16
F16
[5120, 1024]
blk.15.attn_norm.weight
F32
F32
[5120]
blk.15.attn_output.weight
F16
F16
[5120, 5120]
blk.15.attn_q.bias
F32
F32
[5120]
blk.15.attn_q.weight
F16
F16
[5120, 5120]
blk.15.attn_v.bias
F32
F32
[1024]
blk.15.attn_v.weight
F16
F16
[5120, 1024]
blk.15.ffn_down.weight
F16
F16
[27648, 5120]
blk.15.ffn_gate.weight
F16
F16
[5120, 27648]
blk.15.ffn_norm.weight
F32
F32
[5120]
blk.15.ffn_up.weight
F16
F16
[5120, 27648]
blk.16
blk.16.attn_k.bias
F32
F32
[1024]
blk.16.attn_k.weight
F16
F16
[5120, 1024]
blk.16.attn_norm.weight
F32
F32
[5120]
blk.16.attn_output.weight
F16
F16
[5120, 5120]
blk.16.attn_q.bias
F32
F32
[5120]
blk.16.attn_q.weight
F16
F16
[5120, 5120]
blk.16.attn_v.bias
F32
F32
[1024]
blk.16.attn_v.weight
F16
F16
[5120, 1024]
blk.16.ffn_down.weight
F16
F16
[27648, 5120]
blk.16.ffn_gate.weight
F16
F16
[5120, 27648]
blk.16.ffn_norm.weight
F32
F32
[5120]
blk.16.ffn_up.weight
F16
F16
[5120, 27648]
blk.17
blk.17.attn_k.bias
F32
F32
[1024]
blk.17.attn_k.weight
F16
F16
[5120, 1024]
blk.17.attn_norm.weight
F32
F32
[5120]
blk.17.attn_output.weight
F16
F16
[5120, 5120]
blk.17.attn_q.bias
F32
F32
[5120]
blk.17.attn_q.weight
F16
F16
[5120, 5120]
blk.17.attn_v.bias
F32
F32
[1024]
blk.17.attn_v.weight
F16
F16
[5120, 1024]
blk.17.ffn_down.weight
F16
F16
[27648, 5120]
blk.17.ffn_gate.weight
F16
F16
[5120, 27648]
blk.17.ffn_norm.weight
F32
F32
[5120]
blk.17.ffn_up.weight
F16
F16
[5120, 27648]
blk.18
blk.18.attn_k.bias
F32
F32
[1024]
blk.18.attn_k.weight
F16
F16
[5120, 1024]
blk.18.attn_norm.weight
F32
F32
[5120]
blk.18.attn_output.weight
F16
F16
[5120, 5120]
blk.18.attn_q.bias
F32
F32
[5120]
blk.18.attn_q.weight
F16
F16
[5120, 5120]
blk.18.attn_v.bias
F32
F32
[1024]
blk.18.attn_v.weight
F16
F16
[5120, 1024]
blk.18.ffn_down.weight
F16
F16
[27648, 5120]
blk.18.ffn_gate.weight
F16
F16
[5120, 27648]
blk.18.ffn_norm.weight
F32
F32
[5120]
blk.18.ffn_up.weight
F16
F16
[5120, 27648]
blk.19
blk.19.attn_k.bias
F32
F32
[1024]
blk.19.attn_k.weight
F16
F16
[5120, 1024]
blk.19.attn_norm.weight
F32
F32
[5120]
blk.19.attn_output.weight
F16
F16
[5120, 5120]
blk.19.attn_q.bias
F32
F32
[5120]
blk.19.attn_q.weight
F16
F16
[5120, 5120]
blk.19.attn_v.bias
F32
F32
[1024]
blk.19.attn_v.weight
F16
F16
[5120, 1024]
blk.19.ffn_down.weight
F16
F16
[27648, 5120]
blk.19.ffn_gate.weight
F16
F16
[5120, 27648]
blk.19.ffn_norm.weight
F32
F32
[5120]
blk.19.ffn_up.weight
F16
F16
[5120, 27648]
blk.20
blk.20.attn_k.bias
F32
F32
[1024]
blk.20.attn_k.weight
F16
F16
[5120, 1024]
blk.20.attn_norm.weight
F32
F32
[5120]
blk.20.attn_output.weight
F16
F16
[5120, 5120]
blk.20.attn_q.bias
F32
F32
[5120]
blk.20.attn_q.weight
F16
F16
[5120, 5120]
blk.20.attn_v.bias
F32
F32
[1024]
blk.20.attn_v.weight
F16
F16
[5120, 1024]
blk.20.ffn_down.weight
F16
F16
[27648, 5120]
blk.20.ffn_gate.weight
F16
F16
[5120, 27648]
blk.20.ffn_norm.weight
F32
F32
[5120]
blk.20.ffn_up.weight
F16
F16
[5120, 27648]
blk.21
blk.21.attn_k.bias
F32
F32
[1024]
blk.21.attn_k.weight
F16
F16
[5120, 1024]
blk.21.attn_norm.weight
F32
F32
[5120]
blk.21.attn_output.weight
F16
F16
[5120, 5120]
blk.21.attn_q.bias
F32
F32
[5120]
blk.21.attn_q.weight
F16
F16
[5120, 5120]
blk.21.attn_v.bias
F32
F32
[1024]
blk.21.attn_v.weight
F16
F16
[5120, 1024]
blk.21.ffn_down.weight
F16
F16
[27648, 5120]
blk.21.ffn_gate.weight
F16
F16
[5120, 27648]
blk.21.ffn_norm.weight
F32
F32
[5120]
blk.21.ffn_up.weight
F16
F16
[5120, 27648]
blk.22
blk.22.attn_k.bias
F32
F32
[1024]
blk.22.attn_k.weight
F16
F16
[5120, 1024]
blk.22.attn_norm.weight
F32
F32
[5120]
blk.22.attn_output.weight
F16
F16
[5120, 5120]
blk.22.attn_q.bias
F32
F32
[5120]
blk.22.attn_q.weight
F16
F16
[5120, 5120]
blk.22.attn_v.bias
F32
F32
[1024]
blk.22.attn_v.weight
F16
F16
[5120, 1024]
blk.22.ffn_down.weight
F16
F16
[27648, 5120]
blk.22.ffn_gate.weight
F16
F16
[5120, 27648]
blk.22.ffn_norm.weight
F32
F32
[5120]
blk.22.ffn_up.weight
F16
F16
[5120, 27648]
blk.23
blk.23.attn_k.bias
F32
F32
[1024]
blk.23.attn_k.weight
F16
F16
[5120, 1024]
blk.23.attn_norm.weight
F32
F32
[5120]
blk.23.attn_output.weight
F16
F16
[5120, 5120]
blk.23.attn_q.bias
F32
F32
[5120]
blk.23.attn_q.weight
F16
F16
[5120, 5120]
blk.23.attn_v.bias
F32
F32
[1024]
blk.23.attn_v.weight
F16
F16
[5120, 1024]
blk.23.ffn_down.weight
F16
F16
[27648, 5120]
blk.23.ffn_gate.weight
F16
F16
[5120, 27648]
blk.23.ffn_norm.weight
F32
F32
[5120]
blk.23.ffn_up.weight
F16
F16
[5120, 27648]
blk.24
blk.24.attn_k.bias
F32
F32
[1024]
blk.24.attn_k.weight
F16
F16
[5120, 1024]
blk.24.attn_norm.weight
F32
F32
[5120]
blk.24.attn_output.weight
F16
F16
[5120, 5120]
blk.24.attn_q.bias
F32
F32
[5120]
blk.24.attn_q.weight
F16
F16
[5120, 5120]
blk.24.attn_v.bias
F32
F32
[1024]
blk.24.attn_v.weight
F16
F16
[5120, 1024]
blk.24.ffn_down.weight
F16
F16
[27648, 5120]
blk.24.ffn_gate.weight
F16
F16
[5120, 27648]
blk.24.ffn_norm.weight
F32
F32
[5120]
blk.24.ffn_up.weight
F16
F16
[5120, 27648]
blk.25
blk.25.attn_k.bias
F32
F32
[1024]
blk.25.attn_k.weight
F16
F16
[5120, 1024]
blk.25.attn_norm.weight
F32
F32
[5120]
blk.25.attn_output.weight
F16
F16
[5120, 5120]
blk.25.attn_q.bias
F32
F32
[5120]
blk.25.attn_q.weight
F16
F16
[5120, 5120]
blk.25.attn_v.bias
F32
F32
[1024]
blk.25.attn_v.weight
F16
F16
[5120, 1024]
blk.25.ffn_down.weight
F16
F16
[27648, 5120]
blk.25.ffn_gate.weight
F16
F16
[5120, 27648]
blk.25.ffn_norm.weight
F32
F32
[5120]
blk.25.ffn_up.weight
F16
F16
[5120, 27648]
blk.26
blk.26.attn_k.bias
F32
F32
[1024]
blk.26.attn_k.weight
F16
F16
[5120, 1024]
blk.26.attn_norm.weight
F32
F32
[5120]
blk.26.attn_output.weight
F16
F16
[5120, 5120]
blk.26.attn_q.bias
F32
F32
[5120]
blk.26.attn_q.weight
F16
F16
[5120, 5120]
blk.26.attn_v.bias
F32
F32
[1024]
blk.26.attn_v.weight
F16
F16
[5120, 1024]
blk.26.ffn_down.weight
F16
F16
[27648, 5120]
blk.26.ffn_gate.weight
F16
F16
[5120, 27648]
blk.26.ffn_norm.weight
F32
F32
[5120]
blk.26.ffn_up.weight
F16
F16
[5120, 27648]
blk.27
blk.27.attn_k.bias
F32
F32
[1024]
blk.27.attn_k.weight
F16
F16
[5120, 1024]
blk.27.attn_norm.weight
F32
F32
[5120]
blk.27.attn_output.weight
F16
F16
[5120, 5120]
blk.27.attn_q.bias
F32
F32
[5120]
blk.27.attn_q.weight
F16
F16
[5120, 5120]
blk.27.attn_v.bias
F32
F32
[1024]
blk.27.attn_v.weight
F16
F16
[5120, 1024]
blk.27.ffn_down.weight
F16
F16
[27648, 5120]
blk.27.ffn_gate.weight
F16
F16
[5120, 27648]
blk.27.ffn_norm.weight
F32
F32
[5120]
blk.27.ffn_up.weight
F16
F16
[5120, 27648]
blk.28
blk.28.attn_k.bias
F32
F32
[1024]
blk.28.attn_k.weight
F16
F16
[5120, 1024]
blk.28.attn_norm.weight
F32
F32
[5120]
blk.28.attn_output.weight
F16
F16
[5120, 5120]
blk.28.attn_q.bias
F32
F32
[5120]
blk.28.attn_q.weight
F16
F16
[5120, 5120]
blk.28.attn_v.bias
F32
F32
[1024]
blk.28.attn_v.weight
F16
F16
[5120, 1024]
blk.28.ffn_down.weight
F16
F16
[27648, 5120]
blk.28.ffn_gate.weight
F16
F16
[5120, 27648]
blk.28.ffn_norm.weight
F32
F32
[5120]
blk.28.ffn_up.weight
F16
F16
[5120, 27648]
blk.29
blk.29.attn_k.bias
F32
F32
[1024]
blk.29.attn_k.weight
F16
F16
[5120, 1024]
blk.29.attn_norm.weight
F32
F32
[5120]
blk.29.attn_output.weight
F16
F16
[5120, 5120]
blk.29.attn_q.bias
F32
F32
[5120]
blk.29.attn_q.weight
F16
F16
[5120, 5120]
blk.29.attn_v.bias
F32
F32
[1024]
blk.29.attn_v.weight
F16
F16
[5120, 1024]
blk.29.ffn_down.weight
F16
F16
[27648, 5120]
blk.29.ffn_gate.weight
F16
F16
[5120, 27648]
blk.29.ffn_norm.weight
F32
F32
[5120]
blk.29.ffn_up.weight
F16
F16
[5120, 27648]
blk.30
blk.30.attn_k.bias
F32
F32
[1024]
blk.30.attn_k.weight
F16
F16
[5120, 1024]
blk.30.attn_norm.weight
F32
F32
[5120]
blk.30.attn_output.weight
F16
F16
[5120, 5120]
blk.30.attn_q.bias
F32
F32
[5120]
blk.30.attn_q.weight
F16
F16
[5120, 5120]
blk.30.attn_v.bias
F32
F32
[1024]
blk.30.attn_v.weight
F16
F16
[5120, 1024]
blk.30.ffn_down.weight
F16
F16
[27648, 5120]
blk.30.ffn_gate.weight
F16
F16
[5120, 27648]
blk.30.ffn_norm.weight
F32
F32
[5120]
blk.30.ffn_up.weight
F16
F16
[5120, 27648]
blk.31
blk.31.attn_k.bias
F32
F32
[1024]
blk.31.attn_k.weight
F16
F16
[5120, 1024]
blk.31.attn_norm.weight
F32
F32
[5120]
blk.31.attn_output.weight
F16
F16
[5120, 5120]
blk.31.attn_q.bias
F32
F32
[5120]
blk.31.attn_q.weight
F16
F16
[5120, 5120]
blk.31.attn_v.bias
F32
F32
[1024]
blk.31.attn_v.weight
F16
F16
[5120, 1024]
blk.31.ffn_down.weight
F16
F16
[27648, 5120]
blk.31.ffn_gate.weight
F16
F16
[5120, 27648]
blk.31.ffn_norm.weight
F32
F32
[5120]
blk.31.ffn_up.weight
F16
F16
[5120, 27648]
blk.32
blk.32.attn_k.bias
F32
F32
[1024]
blk.32.attn_k.weight
F16
F16
[5120, 1024]
blk.32.attn_norm.weight
F32
F32
[5120]
blk.32.attn_output.weight
F16
F16
[5120, 5120]
blk.32.attn_q.bias
F32
F32
[5120]
blk.32.attn_q.weight
F16
F16
[5120, 5120]
blk.32.attn_v.bias
F32
F32
[1024]
blk.32.attn_v.weight
F16
F16
[5120, 1024]
blk.32.ffn_down.weight
F16
F16
[27648, 5120]
blk.32.ffn_gate.weight
F16
F16
[5120, 27648]
blk.32.ffn_norm.weight
F32
F32
[5120]
blk.32.ffn_up.weight
F16
F16
[5120, 27648]
blk.33
blk.33.attn_k.bias
F32
F32
[1024]
blk.33.attn_k.weight
F16
F16
[5120, 1024]
blk.33.attn_norm.weight
F32
F32
[5120]
blk.33.attn_output.weight
F16
F16
[5120, 5120]
blk.33.attn_q.bias
F32
F32
[5120]
blk.33.attn_q.weight
F16
F16
[5120, 5120]
blk.33.attn_v.bias
F32
F32
[1024]
blk.33.attn_v.weight
F16
F16
[5120, 1024]
blk.33.ffn_down.weight
F16
F16
[27648, 5120]
blk.33.ffn_gate.weight
F16
F16
[5120, 27648]
blk.33.ffn_norm.weight
F32
F32
[5120]
blk.33.ffn_up.weight
F16
F16
[5120, 27648]
blk.34
blk.34.attn_k.bias
F32
F32
[1024]
blk.34.attn_k.weight
F16
F16
[5120, 1024]
blk.34.attn_norm.weight
F32
F32
[5120]
blk.34.attn_output.weight
F16
F16
[5120, 5120]
blk.34.attn_q.bias
F32
F32
[5120]
blk.34.attn_q.weight
F16
F16
[5120, 5120]
blk.34.attn_v.bias
F32
F32
[1024]
blk.34.attn_v.weight
F16
F16
[5120, 1024]
blk.34.ffn_down.weight
F16
F16
[27648, 5120]
blk.34.ffn_gate.weight
F16
F16
[5120, 27648]
blk.34.ffn_norm.weight
F32
F32
[5120]
blk.34.ffn_up.weight
F16
F16
[5120, 27648]
blk.35
blk.35.attn_k.bias
F32
F32
[1024]
blk.35.attn_k.weight
F16
F16
[5120, 1024]
blk.35.attn_norm.weight
F32
F32
[5120]
blk.35.attn_output.weight
F16
F16
[5120, 5120]
blk.35.attn_q.bias
F32
F32
[5120]
blk.35.attn_q.weight
F16
F16
[5120, 5120]
blk.35.attn_v.bias
F32
F32
[1024]
blk.35.attn_v.weight
F16
F16
[5120, 1024]
blk.35.ffn_down.weight
F16
F16
[27648, 5120]
blk.35.ffn_gate.weight
F16
F16
[5120, 27648]
blk.35.ffn_norm.weight
F32
F32
[5120]
blk.35.ffn_up.weight
F16
F16
[5120, 27648]
blk.36
blk.36.attn_k.bias
F32
F32
[1024]
blk.36.attn_k.weight
F16
F16
[5120, 1024]
blk.36.attn_norm.weight
F32
F32
[5120]
blk.36.attn_output.weight
F16
F16
[5120, 5120]
blk.36.attn_q.bias
F32
F32
[5120]
blk.36.attn_q.weight
F16
F16
[5120, 5120]
blk.36.attn_v.bias
F32
F32
[1024]
blk.36.attn_v.weight
F16
F16
[5120, 1024]
blk.36.ffn_down.weight
F16
F16
[27648, 5120]
blk.36.ffn_gate.weight
F16
F16
[5120, 27648]
blk.36.ffn_norm.weight
F32
F32
[5120]
blk.36.ffn_up.weight
F16
F16
[5120, 27648]
blk.37
blk.37.attn_k.bias
F32
F32
[1024]
blk.37.attn_k.weight
F16
F16
[5120, 1024]
blk.37.attn_norm.weight
F32
F32
[5120]
blk.37.attn_output.weight
F16
F16
[5120, 5120]
blk.37.attn_q.bias
F32
F32
[5120]
blk.37.attn_q.weight
F16
F16
[5120, 5120]
blk.37.attn_v.bias
F32
F32
[1024]
blk.37.attn_v.weight
F16
F16
[5120, 1024]
blk.37.ffn_down.weight
F16
F16
[27648, 5120]
blk.37.ffn_gate.weight
F16
F16
[5120, 27648]
blk.37.ffn_norm.weight
F32
F32
[5120]
blk.37.ffn_up.weight
F16
F16
[5120, 27648]
blk.38
blk.38.attn_k.bias
F32
F32
[1024]
blk.38.attn_k.weight
F16
F16
[5120, 1024]
blk.38.attn_norm.weight
F32
F32
[5120]
blk.38.attn_output.weight
F16
F16
[5120, 5120]
blk.38.attn_q.bias
F32
F32
[5120]
blk.38.attn_q.weight
F16
F16
[5120, 5120]
blk.38.attn_v.bias
F32
F32
[1024]
blk.38.attn_v.weight
F16
F16
[5120, 1024]
blk.38.ffn_down.weight
F16
F16
[27648, 5120]
blk.38.ffn_gate.weight
F16
F16
[5120, 27648]
blk.38.ffn_norm.weight
F32
F32
[5120]
blk.38.ffn_up.weight
F16
F16
[5120, 27648]
blk.39
blk.39.attn_k.bias
F32
F32
[1024]
blk.39.attn_k.weight
F16
F16
[5120, 1024]
blk.39.attn_norm.weight
F32
F32
[5120]
blk.39.attn_output.weight
F16
F16
[5120, 5120]
blk.39.attn_q.bias
F32
F32
[5120]
blk.39.attn_q.weight
F16
F16
[5120, 5120]
blk.39.attn_v.bias
F32
F32
[1024]
blk.39.attn_v.weight
F16
F16
[5120, 1024]
blk.39.ffn_down.weight
F16
F16
[27648, 5120]
blk.39.ffn_gate.weight
F16
F16
[5120, 27648]
blk.39.ffn_norm.weight
F32
F32
[5120]
blk.39.ffn_up.weight
F16
F16
[5120, 27648]
blk.40
blk.40.attn_k.bias
F32
F32
[1024]
blk.40.attn_k.weight
F16
F16
[5120, 1024]
blk.40.attn_norm.weight
F32
F32
[5120]
blk.40.attn_output.weight
F16
F16
[5120, 5120]
blk.40.attn_q.bias
F32
F32
[5120]
blk.40.attn_q.weight
F16
F16
[5120, 5120]
blk.40.attn_v.bias
F32
F32
[1024]
blk.40.attn_v.weight
F16
F16
[5120, 1024]
blk.40.ffn_down.weight
F16
F16
[27648, 5120]
blk.40.ffn_gate.weight
F16
F16
[5120, 27648]
blk.40.ffn_norm.weight
F32
F32
[5120]
blk.40.ffn_up.weight
F16
F16
[5120, 27648]
blk.41
blk.41.attn_k.bias
F32
F32
[1024]
blk.41.attn_k.weight
F16
F16
[5120, 1024]
blk.41.attn_norm.weight
F32
F32
[5120]
blk.41.attn_output.weight
F16
F16
[5120, 5120]
blk.41.attn_q.bias
F32
F32
[5120]
blk.41.attn_q.weight
F16
F16
[5120, 5120]
blk.41.attn_v.bias
F32
F32
[1024]
blk.41.attn_v.weight
F16
F16
[5120, 1024]
blk.41.ffn_down.weight
F16
F16
[27648, 5120]
blk.41.ffn_gate.weight
F16
F16
[5120, 27648]
blk.41.ffn_norm.weight
F32
F32
[5120]
blk.41.ffn_up.weight
F16
F16
[5120, 27648]
blk.42
blk.42.attn_k.bias
F32
F32
[1024]
blk.42.attn_k.weight
F16
F16
[5120, 1024]
blk.42.attn_norm.weight
F32
F32
[5120]
blk.42.attn_output.weight
F16
F16
[5120, 5120]
blk.42.attn_q.bias
F32
F32
[5120]
blk.42.attn_q.weight
F16
F16
[5120, 5120]
blk.42.attn_v.bias
F32
F32
[1024]
blk.42.attn_v.weight
F16
F16
[5120, 1024]
blk.42.ffn_down.weight
F16
F16
[27648, 5120]
blk.42.ffn_gate.weight
F16
F16
[5120, 27648]
blk.42.ffn_norm.weight
F32
F32
[5120]
blk.42.ffn_up.weight
F16
F16
[5120, 27648]
blk.43
blk.43.attn_k.bias
F32
F32
[1024]
blk.43.attn_k.weight
F16
F16
[5120, 1024]
blk.43.attn_norm.weight
F32
F32
[5120]
blk.43.attn_output.weight
F16
F16
[5120, 5120]
blk.43.attn_q.bias
F32
F32
[5120]
blk.43.attn_q.weight
F16
F16
[5120, 5120]
blk.43.attn_v.bias
F32
F32
[1024]
blk.43.attn_v.weight
F16
F16
[5120, 1024]
blk.43.ffn_down.weight
F16
F16
[27648, 5120]
blk.43.ffn_gate.weight
F16
F16
[5120, 27648]
blk.43.ffn_norm.weight
F32
F32
[5120]
blk.43.ffn_up.weight
F16
F16
[5120, 27648]
blk.44
blk.44.attn_k.bias
F32
F32
[1024]
blk.44.attn_k.weight
F16
F16
[5120, 1024]
blk.44.attn_norm.weight
F32
F32
[5120]
blk.44.attn_output.weight
F16
F16
[5120, 5120]
blk.44.attn_q.bias
F32
F32
[5120]
blk.44.attn_q.weight
F16
F16
[5120, 5120]
blk.44.attn_v.bias
F32
F32
[1024]
blk.44.attn_v.weight
F16
F16
[5120, 1024]
blk.44.ffn_down.weight
F16
F16
[27648, 5120]
blk.44.ffn_gate.weight
F16
F16
[5120, 27648]
blk.44.ffn_norm.weight
F32
F32
[5120]
blk.44.ffn_up.weight
F16
F16
[5120, 27648]
blk.45
blk.45.attn_k.bias
F32
F32
[1024]
blk.45.attn_k.weight
F16
F16
[5120, 1024]
blk.45.attn_norm.weight
F32
F32
[5120]
blk.45.attn_output.weight
F16
F16
[5120, 5120]
blk.45.attn_q.bias
F32
F32
[5120]
blk.45.attn_q.weight
F16
F16
[5120, 5120]
blk.45.attn_v.bias
F32
F32
[1024]
blk.45.attn_v.weight
F16
F16
[5120, 1024]
blk.45.ffn_down.weight
F16
F16
[27648, 5120]
blk.45.ffn_gate.weight
F16
F16
[5120, 27648]
blk.45.ffn_norm.weight
F32
F32
[5120]
blk.45.ffn_up.weight
F16
F16
[5120, 27648]
blk.46
blk.46.attn_k.bias
F32
F32
[1024]
blk.46.attn_k.weight
F16
F16
[5120, 1024]
blk.46.attn_norm.weight
F32
F32
[5120]
blk.46.attn_output.weight
F16
F16
[5120, 5120]
blk.46.attn_q.bias
F32
F32
[5120]
blk.46.attn_q.weight
F16
F16
[5120, 5120]
blk.46.attn_v.bias
F32
F32
[1024]
blk.46.attn_v.weight
F16
F16
[5120, 1024]
blk.46.ffn_down.weight
F16
F16
[27648, 5120]
blk.46.ffn_gate.weight
F16
F16
[5120, 27648]
blk.46.ffn_norm.weight
F32
F32
[5120]
blk.46.ffn_up.weight
F16
F16
[5120, 27648]
blk.47
blk.47.attn_k.bias
F32
F32
[1024]
blk.47.attn_k.weight
F16
F16
[5120, 1024]
blk.47.attn_norm.weight
F32
F32
[5120]
blk.47.attn_output.weight
F16
F16
[5120, 5120]
blk.47.attn_q.bias
F32
F32
[5120]
blk.47.attn_q.weight
F16
F16
[5120, 5120]
blk.47.attn_v.bias
F32
F32
[1024]
blk.47.attn_v.weight
F16
F16
[5120, 1024]
blk.47.ffn_down.weight
F16
F16
[27648, 5120]
blk.47.ffn_gate.weight
F16
F16
[5120, 27648]
blk.47.ffn_norm.weight
F32
F32
[5120]
blk.47.ffn_up.weight
F16
F16
[5120, 27648]
blk.48
blk.48.attn_k.bias
F32
F32
[1024]
blk.48.attn_k.weight
F16
F16
[5120, 1024]
blk.48.attn_norm.weight
F32
F32
[5120]
blk.48.attn_output.weight
F16
F16
[5120, 5120]
blk.48.attn_q.bias
F32
F32
[5120]
blk.48.attn_q.weight
F16
F16
[5120, 5120]
blk.48.attn_v.bias
F32
F32
[1024]
blk.48.attn_v.weight
F16
F16
[5120, 1024]
blk.48.ffn_down.weight
F16
F16
[27648, 5120]
blk.48.ffn_gate.weight
F16
F16
[5120, 27648]
blk.48.ffn_norm.weight
F32
F32
[5120]
blk.48.ffn_up.weight
F16
F16
[5120, 27648]
blk.49
blk.49.attn_k.bias
F32
F32
[1024]
blk.49.attn_k.weight
F16
F16
[5120, 1024]
blk.49.attn_norm.weight
F32
F32
[5120]
blk.49.attn_output.weight
F16
F16
[5120, 5120]
blk.49.attn_q.bias
F32
F32
[5120]
blk.49.attn_q.weight
F16
F16
[5120, 5120]
blk.49.attn_v.bias
F32
F32
[1024]
blk.49.attn_v.weight
F16
F16
[5120, 1024]
blk.49.ffn_down.weight
F16
F16
[27648, 5120]
blk.49.ffn_gate.weight
F16
F16
[5120, 27648]
blk.49.ffn_norm.weight
F32
F32
[5120]
blk.49.ffn_up.weight
F16
F16
[5120, 27648]
blk.50
blk.50.attn_k.bias
F32
F32
[1024]
blk.50.attn_k.weight
F16
F16
[5120, 1024]
blk.50.attn_norm.weight
F32
F32
[5120]
blk.50.attn_output.weight
F16
F16
[5120, 5120]
blk.50.attn_q.bias
F32
F32
[5120]
blk.50.attn_q.weight
F16
F16
[5120, 5120]
blk.50.attn_v.bias
F32
F32
[1024]
blk.50.attn_v.weight
F16
F16
[5120, 1024]
blk.50.ffn_down.weight
F16
F16
[27648, 5120]
blk.50.ffn_gate.weight
F16
F16
[5120, 27648]
blk.50.ffn_norm.weight
F32
F32
[5120]
blk.50.ffn_up.weight
F16
F16
[5120, 27648]
blk.51
blk.51.attn_k.bias
F32
F32
[1024]
blk.51.attn_k.weight
F16
F16
[5120, 1024]
blk.51.attn_norm.weight
F32
F32
[5120]
blk.51.attn_output.weight
F16
F16
[5120, 5120]
blk.51.attn_q.bias
F32
F32
[5120]
blk.51.attn_q.weight
F16
F16
[5120, 5120]
blk.51.attn_v.bias
F32
F32
[1024]
blk.51.attn_v.weight
F16
F16
[5120, 1024]
blk.51.ffn_down.weight
F16
F16
[27648, 5120]
blk.51.ffn_gate.weight
F16
F16
[5120, 27648]
blk.51.ffn_norm.weight
F32
F32
[5120]
blk.51.ffn_up.weight
F16
F16
[5120, 27648]
blk.52
blk.52.attn_k.bias
F32
F32
[1024]
blk.52.attn_k.weight
F16
F16
[5120, 1024]
blk.52.attn_norm.weight
F32
F32
[5120]
blk.52.attn_output.weight
F16
F16
[5120, 5120]
blk.52.attn_q.bias
F32
F32
[5120]
blk.52.attn_q.weight
F16
F16
[5120, 5120]
blk.52.attn_v.bias
F32
F32
[1024]
blk.52.attn_v.weight
F16
F16
[5120, 1024]
blk.52.ffn_down.weight
F16
F16
[27648, 5120]
blk.52.ffn_gate.weight
F16
F16
[5120, 27648]
blk.52.ffn_norm.weight
F32
F32
[5120]
blk.52.ffn_up.weight
F16
F16
[5120, 27648]
blk.53
blk.53.attn_k.bias
F32
F32
[1024]
blk.53.attn_k.weight
F16
F16
[5120, 1024]
blk.53.attn_norm.weight
F32
F32
[5120]
blk.53.attn_output.weight
F16
F16
[5120, 5120]
blk.53.attn_q.bias
F32
F32
[5120]
blk.53.attn_q.weight
F16
F16
[5120, 5120]
blk.53.attn_v.bias
F32
F32
[1024]
blk.53.attn_v.weight
F16
F16
[5120, 1024]
blk.53.ffn_down.weight
F16
F16
[27648, 5120]
blk.53.ffn_gate.weight
F16
F16
[5120, 27648]
blk.53.ffn_norm.weight
F32
F32
[5120]
blk.53.ffn_up.weight
F16
F16
[5120, 27648]
blk.54
blk.54.attn_k.bias
F32
F32
[1024]
blk.54.attn_k.weight
F16
F16
[5120, 1024]
blk.54.attn_norm.weight
F32
F32
[5120]
blk.54.attn_output.weight
F16
F16
[5120, 5120]
blk.54.attn_q.bias
F32
F32
[5120]
blk.54.attn_q.weight
F16
F16
[5120, 5120]
blk.54.attn_v.bias
F32
F32
[1024]
blk.54.attn_v.weight
F16
F16
[5120, 1024]
blk.54.ffn_down.weight
F16
F16
[27648, 5120]
blk.54.ffn_gate.weight
F16
F16
[5120, 27648]
blk.54.ffn_norm.weight
F32
F32
[5120]
blk.54.ffn_up.weight
F16
F16
[5120, 27648]
blk.55
blk.55.attn_k.bias
F32
F32
[1024]
blk.55.attn_k.weight
F16
F16
[5120, 1024]
blk.55.attn_norm.weight
F32
F32
[5120]
blk.55.attn_output.weight
F16
F16
[5120, 5120]
blk.55.attn_q.bias
F32
F32
[5120]
blk.55.attn_q.weight
F16
F16
[5120, 5120]
blk.55.attn_v.bias
F32
F32
[1024]
blk.55.attn_v.weight
F16
F16
[5120, 1024]
blk.55.ffn_down.weight
F16
F16
[27648, 5120]
blk.55.ffn_gate.weight
F16
F16
[5120, 27648]
blk.55.ffn_norm.weight
F32
F32
[5120]
blk.55.ffn_up.weight
F16
F16
[5120, 27648]
blk.56
blk.56.attn_k.bias
F32
F32
[1024]
blk.56.attn_k.weight
F16
F16
[5120, 1024]
blk.56.attn_norm.weight
F32
F32
[5120]
blk.56.attn_output.weight
F16
F16
[5120, 5120]
blk.56.attn_q.bias
F32
F32
[5120]
blk.56.attn_q.weight
F16
F16
[5120, 5120]
blk.56.attn_v.bias
F32
F32
[1024]
blk.56.attn_v.weight
F16
F16
[5120, 1024]
blk.56.ffn_down.weight
F16
F16
[27648, 5120]
blk.56.ffn_gate.weight
F16
F16
[5120, 27648]
blk.56.ffn_norm.weight
F32
F32
[5120]
blk.56.ffn_up.weight
F16
F16
[5120, 27648]
blk.57
blk.57.attn_k.bias
F32
F32
[1024]
blk.57.attn_k.weight
F16
F16
[5120, 1024]
blk.57.attn_norm.weight
F32
F32
[5120]
blk.57.attn_output.weight
F16
F16
[5120, 5120]
blk.57.attn_q.bias
F32
F32
[5120]
blk.57.attn_q.weight
F16
F16
[5120, 5120]
blk.57.attn_v.bias
F32
F32
[1024]
blk.57.attn_v.weight
F16
F16
[5120, 1024]
blk.57.ffn_down.weight
F16
F16
[27648, 5120]
blk.57.ffn_gate.weight
F16
F16
[5120, 27648]
blk.57.ffn_norm.weight
F32
F32
[5120]
blk.57.ffn_up.weight
F16
F16
[5120, 27648]
blk.58
blk.58.attn_k.bias
F32
F32
[1024]
blk.58.attn_k.weight
F16
F16
[5120, 1024]
blk.58.attn_norm.weight
F32
F32
[5120]
blk.58.attn_output.weight
F16
F16
[5120, 5120]
blk.58.attn_q.bias
F32
F32
[5120]
blk.58.attn_q.weight
F16
F16
[5120, 5120]
blk.58.attn_v.bias
F32
F32
[1024]
blk.58.attn_v.weight
F16
F16
[5120, 1024]
blk.58.ffn_down.weight
F16
F16
[27648, 5120]
blk.58.ffn_gate.weight
F16
F16
[5120, 27648]
blk.58.ffn_norm.weight
F32
F32
[5120]
blk.58.ffn_up.weight
F16
F16
[5120, 27648]
blk.59
blk.59.attn_k.bias
F32
F32
[1024]
blk.59.attn_k.weight
F16
F16
[5120, 1024]
blk.59.attn_norm.weight
F32
F32
[5120]
blk.59.attn_output.weight
F16
F16
[5120, 5120]
blk.59.attn_q.bias
F32
F32
[5120]
blk.59.attn_q.weight
F16
F16
[5120, 5120]
blk.59.attn_v.bias
F32
F32
[1024]
blk.59.attn_v.weight
F16
F16
[5120, 1024]
blk.59.ffn_down.weight
F16
F16
[27648, 5120]
blk.59.ffn_gate.weight
F16
F16
[5120, 27648]
blk.59.ffn_norm.weight
F32
F32
[5120]
blk.59.ffn_up.weight
F16
F16
[5120, 27648]
blk.60
blk.60.attn_k.bias
F32
F32
[1024]
blk.60.attn_k.weight
F16
F16
[5120, 1024]
blk.60.attn_norm.weight
F32
F32
[5120]
blk.60.attn_output.weight
F16
F16
[5120, 5120]
blk.60.attn_q.bias
F32
F32
[5120]
blk.60.attn_q.weight
F16
F16
[5120, 5120]
blk.60.attn_v.bias
F32
F32
[1024]
blk.60.attn_v.weight
F16
F16
[5120, 1024]
blk.60.ffn_down.weight
F16
F16
[27648, 5120]
blk.60.ffn_gate.weight
F16
F16
[5120, 27648]
blk.60.ffn_norm.weight
F32
F32
[5120]
blk.60.ffn_up.weight
F16
F16
[5120, 27648]
blk.61
blk.61.attn_k.bias
F32
F32
[1024]
blk.61.attn_k.weight
F16
F16
[5120, 1024]
blk.61.attn_norm.weight
F32
F32
[5120]
blk.61.attn_output.weight
F16
F16
[5120, 5120]
blk.61.attn_q.bias
F32
F32
[5120]
blk.61.attn_q.weight
F16
F16
[5120, 5120]
blk.61.attn_v.bias
F32
F32
[1024]
blk.61.attn_v.weight
F16
F16
[5120, 1024]
blk.61.ffn_down.weight
F16
F16
[27648, 5120]
blk.61.ffn_gate.weight
F16
F16
[5120, 27648]
blk.61.ffn_norm.weight
F32
F32
[5120]
blk.61.ffn_up.weight
F16
F16
[5120, 27648]
blk.62
blk.62.attn_k.bias
F32
F32
[1024]
blk.62.attn_k.weight
F16
F16
[5120, 1024]
blk.62.attn_norm.weight
F32
F32
[5120]
blk.62.attn_output.weight
F16
F16
[5120, 5120]
blk.62.attn_q.bias
F32
F32
[5120]
blk.62.attn_q.weight
F16
F16
[5120, 5120]
blk.62.attn_v.bias
F32
F32
[1024]
blk.62.attn_v.weight
F16
F16
[5120, 1024]
blk.62.ffn_down.weight
F16
F16
[27648, 5120]
blk.62.ffn_gate.weight
F16
F16
[5120, 27648]
blk.62.ffn_norm.weight
F32
F32
[5120]
blk.62.ffn_up.weight
F16
F16
[5120, 27648]
blk.63
blk.63.attn_k.bias
F32
F32
[1024]
blk.63.attn_k.weight
F16
F16
[5120, 1024]
blk.63.attn_norm.weight
F32
F32
[5120]
blk.63.attn_output.weight
F16
F16
[5120, 5120]
blk.63.attn_q.bias
F32
F32
[5120]
blk.63.attn_q.weight
F16
F16
[5120, 5120]
blk.63.attn_v.bias
F32
F32
[1024]
blk.63.attn_v.weight
F16
F16
[5120, 1024]
blk.63.ffn_down.weight
F16
F16
[27648, 5120]
blk.63.ffn_gate.weight
F16
F16
[5120, 27648]
blk.63.ffn_norm.weight
F32
F32
[5120]
blk.63.ffn_up.weight
F16
F16
[5120, 27648]
output.weight
F16
F16
[5120, 152064]
output_norm.weight
F32
F32
[5120]