diff --git a/.env.example b/.env.example index 918c64d..4788a31 100644 --- a/.env.example +++ b/.env.example @@ -26,7 +26,7 @@ FAST_BATCH_SIZE=64 FAST_UBATCH_SIZE=32 MEDIUM_CONTEXT=160000 MEDIUM_BATCH_SIZE=2048 -MEDIUM_UBATCH_SIZE=128 +MEDIUM_UBATCH_SIZE=512 LARGE_CONTEXT=192000 LARGE_BATCH_SIZE=2048 LARGE_UBATCH_SIZE=128 diff --git a/compose.yaml b/compose.yaml index 195911b..d5e59a9 100644 --- a/compose.yaml +++ b/compose.yaml @@ -183,7 +183,7 @@ services: - --batch-size - "${MEDIUM_BATCH_SIZE:-2048}" - --ubatch-size - - "${MEDIUM_UBATCH_SIZE:-128}" + - "${MEDIUM_UBATCH_SIZE:-512}" - --parallel - "${MEDIUM_PARALLEL_SLOTS:-1}" - --kv-unified diff --git a/config/install.env.example b/config/install.env.example index 24b788f..0cc1426 100644 --- a/config/install.env.example +++ b/config/install.env.example @@ -84,7 +84,7 @@ FAST_BATCH_SIZE=64 FAST_UBATCH_SIZE=32 MEDIUM_CONTEXT=160000 MEDIUM_BATCH_SIZE=2048 -MEDIUM_UBATCH_SIZE=128 +MEDIUM_UBATCH_SIZE=512 MEDIUM_TENSOR_SPLIT=85,15 LARGE_CONTEXT=192000 LARGE_BATCH_SIZE=2048