diff --git a/egs/tedlium3/ASR/RESULTS.md b/egs/tedlium3/ASR/RESULTS.md
index 5b1beb6de..d5b24a376 100644
--- a/egs/tedlium3/ASR/RESULTS.md
+++ b/egs/tedlium3/ASR/RESULTS.md
@@ -4,7 +4,8 @@
 
 #### Conformer encoder + embedding decoder
 
-Using the codes from this PR commit https://github.com/k2-fsa/icefall/pull/183/commits/536ad2252e2d406f24a681743d98bd5f90801b97.
+Using the codes from this PR https://github.com/k2-fsa/icefall/pull/233
+And the SpecAugment codes from this PR https://github.com/lhotse-speech/lhotse/pull/604
 
 Conformer encoder + non-current decoder. The decoder
 contains only an embedding layer and a Conv1d (with kernel size 2).
@@ -31,12 +32,12 @@ export CUDA_VISIBLE_DEVICES="0,1,2,3"
 ```
 
 The tensorboard training log can be found at
-https://tensorboard.dev/experiment/DnRwoZF8RRyod4kkfG5q5Q/#scalars
+https://tensorboard.dev/experiment/zrfXeJO3Q5GmJpP2KRd2VA/#scalars
 
 The decoding command is:
 ```
 epoch=29
-avg=15
+avg=16
 
 ## greedy search
 ./transducer_stateless/decode.py \