Downgrading the 'datasets' package from 3.0.0 to 2.21.0 for Multilang_ASR.ipynb and ASR_CTC_Language_Finetuning.ipynb (#11675) (#11677)

ko3n1g · weiqingw4ng · web-flow · commit 526a525d4de5 · 2024-12-19T17:38:29.000-08:00
* Downgrading the 'datasets' package from 3.0.0 to 2.21.0 for Multilang_ASR.ipynb



* Downgrading the 'datasets' package from 3.0.0 to 2.21.0 for ASR_CTC_Language_Finetuning.ipynb



---------

Signed-off-by: Weiqing Wang &lt;weiqingw@nvidia.com&gt;
Co-authored-by: Weiqing Wang &lt;164252040+weiqingw4ng@users.noreply.github.com&gt;
diff --git a/tutorials/asr/ASR_CTC_Language_Finetuning.ipynb b/tutorials/asr/ASR_CTC_Language_Finetuning.ipynb
@@ -37,6 +37,7 @@
         "!apt-get install sox libsndfile1 ffmpeg libsox-fmt-mp3\n",
         "!pip install text-unidecode\n",
         "!pip install matplotlib>=3.3.2\n",
+        "!pip install datasets==2.21.0 # downgrading to 2.21.0 because latest version (3.0.0) has some issues\n",
         "\n",
         "## Install NeMo\n",
         "BRANCH = 'r2.1.0'\n",
diff --git a/tutorials/asr/Multilang_ASR.ipynb b/tutorials/asr/Multilang_ASR.ipynb
@@ -98,6 +98,7 @@
     "!pip install matplotlib>=3.3.2\n",
     "# this is needed for RNNT loss\n",
     "!pip install --upgrade numba\n",
+    "!pip install datasets==2.21.0 # downgrading to 2.21.0 because latest version (3.0.0) has some issues\n",
     "\n",
     "# this is needed to pre-process MCV Spanish dataset, which contains mp3 files\n",
     "!apt-get install -y sox libsox-fmt-mp3\n",
