PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度
Temporarily skip the issue in the CT (#2093)
Xin He33ec4c75b8a8
diff --git a/test/test_cpu/export/test_llmc_format.py b/test/test_cpu/export/test_llmc_format.pyindex e33445872..e2a80b5a3 100644--- a/test/test_cpu/export/test_llmc_format.py+++ b/test/test_cpu/export/test_llmc_format.py@@ -276,6 +276,8 @@ def test_llmcompressor_static_fp8_kv_config(tiny_opt_model_path, dataloader, tmpassert kv_cache_scheme["symmetric"] is True+# TODO: remove xfail once the issue is resolved.[email protected](reason="https://github.com/vllm-project/compressed-tensors/issues/795")def test_llmcompressor_static_fp8_attention_config(dataloader, tmp_path):model_name = get_model_path("stas/tiny-random-llama-2")autoround = AutoRound(