rockmatrix-ai
/

EdgeRock

Text Generation

image-text-to-text

Model card Files Files and versions

EdgeRock / runtime /optimizations

33.7 kB

Ctrl+K

Ctrl+K

4 contributors

History: 3 commits

hkust-aaron's picture

Use BF16 state at every CUDA batch and stabilize physical-ring attention

623a14e verified 4 days ago

SOURCE.json

5.94 kB
Use BF16 state at every CUDA batch and stabilize physical-ring attention 4 days ago
__init__.py

0 Bytes
Use BF16 inference and align figures and results with the technical report 7 days ago
b1_bf16_crosslayer.py

2.2 kB
Use BF16 inference and align figures and results with the technical report 7 days ago
b1_bf16_fusions.py

7.13 kB
Use BF16 inference and align figures and results with the technical report 7 days ago
b1_bf16_output_fusion.py

7.12 kB
Use BF16 inference and align figures and results with the technical report 7 days ago
b1_bf16_projection.py

1.28 kB
Use BF16 inference and align figures and results with the technical report 7 days ago
b1_bf16_selected.py

751 Bytes
Use BF16 state at every CUDA batch and stabilize physical-ring attention 4 days ago
b1_bf16_swa_prepare.py

3.31 kB
Use BF16 inference and align figures and results with the technical report 7 days ago
b1_runtime_joint_projection.py

3.7 kB
Use BF16 inference and align figures and results with the technical report 7 days ago
b512_share_bf16_weights.py

2.3 kB
Use BF16 inference and align figures and results with the technical report 7 days ago