diff --git a/scripts/models/xllm-375B.sh b/scripts/models/xllm-375B.sh index 049f5070e4..6fddb60644 100644 --- a/scripts/models/xllm-375B.sh +++ b/scripts/models/xllm-375B.sh @@ -17,7 +17,7 @@ MODEL_ARGS=( --position-embedding-type rope --rotary-percent 0.5 --xllm-partial-rope-layout - --rotary-base 500000 + --rotary-base 10000000 --swiglu --untie-embeddings-and-output-weights --vocab-size 250624