| .. |
|
common
|
0d8166de95
port DCP2 onto our tree (22-file superset wrapper + a2a recipe); A/B: 162/176, 5.34M-token KV @512K, -5 vs b12x — the high-KV lane
|
пре 1 месец |
|
nvidia
|
0d8166de95
port DCP2 onto our tree (22-file superset wrapper + a2a recipe); A/B: 162/176, 5.34M-token KV @512K, -5 vs b12x — the high-KV lane
|
пре 1 месец |
|
attention.py
|
0d8166de95
port DCP2 onto our tree (22-file superset wrapper + a2a recipe); A/B: 162/176, 5.34M-token KV @512K, -5 vs b12x — the high-KV lane
|
пре 1 месец |
|
compressor.py
|
0d8166de95
port DCP2 onto our tree (22-file superset wrapper + a2a recipe); A/B: 162/176, 5.34M-token KV @512K, -5 vs b12x — the high-KV lane
|
пре 1 месец |
|
sparse_mla.py
|
0d8166de95
port DCP2 onto our tree (22-file superset wrapper + a2a recipe); A/B: 162/176, 5.34M-token KV @512K, -5 vs b12x — the high-KV lane
|
пре 1 месец |
Alexander Korolev
0d8166de95
port DCP2 onto our tree (22-file superset wrapper + a2a recipe); A/B: 162/176, 5.34M-token KV @512K, -5 vs b12x — the high-KV lane