From 9eed8c758c44ec0d6f012b07cbe7ff3464c1462f Mon Sep 17 00:00:00 2001 From: Marc Blatter Date: Wed, 22 Jul 2026 09:02:34 +0000 Subject: [PATCH] composte-hermes.yml aktualisiert --- composte-hermes.yml | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/composte-hermes.yml b/composte-hermes.yml index 0b16116..6185ef0 100644 --- a/composte-hermes.yml +++ b/composte-hermes.yml @@ -26,7 +26,8 @@ services: - --quantization=awq - --tensor-parallel-size=2 - --gpu-memory-utilization=0.82 - - --max-model-len=32768 + - --max-model-len=65536 + - --rope-scaling={"rope_type":"yarn","factor":2.0,"original_max_position_embeddings":32768} - --max-num-seqs=16 - --kv-cache-dtype=fp8 - --served-model-name=qwen3-14b-awq