From 85881a7e3c21b595c25d247f61c444e14082239e Mon Sep 17 00:00:00 2001
From: Iman Tabrizian <10105175+tabrizian@users.noreply.github.com>
Date: Tue, 30 Jun 2026 14:32:36 -0700
Subject: [PATCH] [https://nvbugs/6215107][fix] Namespace cnn_dailymail repo id
 in load_calib_dataset

Newer huggingface_hub rejects the legacy bare "cnn_dailymail" dataset id
and raises HfUriError ("Repository id must be 'namespace/name'") while
resolving the dataset. load_calib_dataset() in convert_utils.py still
forwarded the bare id to datasets.load_dataset(), breaking the default
calibration path of the TensorRT convert scripts. This is the same root
cause already fixed for the ModelOpt path in quantize_by_modelopt.py.

Rewrite the bare id to the namespaced "abisee/cnn_dailymail" repo before
loading, mirroring the existing fix.

Signed-off-by: Iman Tabrizian <10105175+tabrizian@users.noreply.github.com>
---
 tensorrt_llm/models/convert_utils.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/tensorrt_llm/models/convert_utils.py b/tensorrt_llm/models/convert_utils.py
index 58da3ec36349..f3692952f106 100644
--- a/tensorrt_llm/models/convert_utils.py
+++ b/tensorrt_llm/models/convert_utils.py
@@ -317,6 +317,11 @@ def load_calib_dataset(dataset_name_or_dir: str,
                     key = meta[2]
                 break
 
+    # Bare "cnn_dailymail" id is rejected by newer huggingface_hub, which
+    # requires a "namespace/name" repo id; use the namespaced repo instead.
+    if dataset_name_or_dir == "cnn_dailymail":
+        dataset_name_or_dir = "abisee/cnn_dailymail"
+
     dataset = load_dataset(dataset_name_or_dir,
                            name=config_name,
                            split=split,