PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度
fix the typo neural-chat-7b-v3 to neural-chat-7b-v3-3
wenhuach210047da3844ce
diff --git a/README.md b/README.mdindex 25715c420..1439c42f7 100644--- a/README.md+++ b/README.md@@ -154,7 +154,7 @@ For wikitext2/ptb-new/c4-new ppl, we follow the code of gptq and set the sequenc</tr><tr>- <td rowspan="3">Intel/neural-chat-7b-v3 </td>+ <td rowspan="3">Intel/neural-chat-7b-v3-3 </td><th>FP16</th><td>67.92</td> <! acc avg --><td>61.13</td> <! MMLU -->diff --git a/examples/language-modeling/run_autoround.sh b/examples/language-modeling/run_autoround.shindex a5fcb1e94..c6574049a 100644--- a/examples/language-modeling/run_autoround.sh+++ b/examples/language-modeling/run_autoround.sh@@ -1,7 +1,7 @@#!/bin/bashset -xdevice=0-model_name="Intel/neural-chat-7b-v3"+model_name="Intel/neural-chat-7b-v3-3"CUDA_VISIBLE_DEVICES=$device \python3 main.py \@@ -15,5 +15,5 @@ python3 main.py \--deployment_device 'cpu' \--scale_dtype 'fp32' \--eval_bs 32 \---output_dir "./tmp_signround"+--output_dir "./tmp_autoround"