80.7 GB

Ctrl+K

1 contributor

History: 7 commits

Taykhoom

Add opt-in Triton Hyena kernels from vortex PR #77 (@AlphaKhaw)

a873801 verified 6 days ago

.gitattributes

1.52 kB
Initial Evo2-40B-8K Hugging Face port about 1 month ago
README.md

19.6 kB
Add opt-in Triton Hyena kernels from vortex PR #77 (@AlphaKhaw) 6 days ago
attention.py

12.6 kB
Finalize recurrent cache and strict attention diagnostics about 1 month ago
cache.py

6.55 kB
Enable upstream-compatible recurrent caching with dynamic allocation about 1 month ago
config.json

2.38 kB
Initial Evo2-40B-8K Hugging Face port about 1 month ago
configuration_evo2.py

8.15 kB
Add opt-in Triton Hyena kernels from vortex PR #77 (@AlphaKhaw) 6 days ago
engine.py

13.7 kB
Add opt-in Triton Hyena kernels from vortex PR #77 (@AlphaKhaw) 6 days ago
generation_config.json

74 Bytes
Initial Evo2-40B-8K Hugging Face port about 1 month ago
hyena.py

20.9 kB
Add opt-in Triton Hyena kernels from vortex PR #77 (@AlphaKhaw) 6 days ago
kernels.py

24.2 kB
Add opt-in Triton Hyena kernels from vortex PR #77 (@AlphaKhaw) 6 days ago
layers.py

11.8 kB
Bit-exact with vortex: checkpoint RoPE frequencies, trained FP8 scales, fp8_recipe option 6 days ago
model-00001-of-00017.safetensors

4.98 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00002-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00003-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00004-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00005-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00006-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00007-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00008-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00009-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00010-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00011-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00012-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00013-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00014-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00015-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00016-of-00017.safetensors

4.84 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-00017-of-00017.safetensors

3.09 GB
xet

Initial Evo2-40B-8K Hugging Face port about 1 month ago
model-fp8-state.safetensors

32.3 kB
xet

Keep FP8 state for the Hyena projections only; README: running without TE 6 days ago
model.safetensors.index.json

57.2 kB
Keep FP8 state for the Hyena projections only; README: running without TE 6 days ago
modeling_evo2.py

21.6 kB
Add opt-in Triton Hyena kernels from vortex PR #77 (@AlphaKhaw) 6 days ago
rotary.py

11.7 kB
Bit-exact with vortex: checkpoint RoPE frequencies, trained FP8 scales, fp8_recipe option 6 days ago
special_tokens_map.json

2 Bytes
Initial Evo2-40B-8K Hugging Face port about 1 month ago
tokenization_evo2.py

4.9 kB
Fix masks, caching, dtype handling, and generation correctness about 1 month ago
tokenizer_config.json

244 Bytes
Fix masks, caching, dtype handling, and generation correctness about 1 month ago