[https://nvbugs/6215107][fix] Namespace cnn_dailymail repo id in load_calib_dataset by Tabrizian · Pull Request #15802 · NVIDIA/TensorRT-LLM

Tabrizian · 2026-06-30T22:01:59Z

Summary by CodeRabbit

Bug Fixes
- Improved dataset loading compatibility by recognizing the cnn_dailymail dataset name in its newer namespaced form, helping avoid load failures with updated Hugging Face requirements.

Description

Newer huggingface_hub releases reject the legacy bare cnn_dailymail
dataset id and raise:

huggingface_hub.errors.HfUriError: Invalid HF URI
'hf://datasets/cnn_dailymail@.../.huggingface.yaml'.
Repository id must be 'namespace/name', got 'cnn_dailymail'.

The ModelOpt calibration path (quantize_by_modelopt.get_calib_dataloader /
get_nemo_calib_dataloader) was already fixed to rewrite the bare id to the
namespaced abisee/cnn_dailymail repo (#14930). However,
tensorrt_llm/models/convert_utils.py::load_calib_dataset — the calibration
dataset loader used by the TensorRT-backend convert.py scripts (llama, gpt,
qwen, chatglm, baichuan, bert, gemma, …), which all default calib_dataset
to cnn_dailymail — still forwarded the bare id straight to
datasets.load_dataset() and hit the same HfUriError.

This PR closes that remaining gap: load_calib_dataset now rewrites the bare
cnn_dailymail id to the namespaced abisee/cnn_dailymail repo before
loading, mirroring the existing ModelOpt fix. Already-namespaced ids (e.g.
ccdv/cnn_dailymail) and local dataset directories are unaffected.

NVBug: https://nvbugs/6215107

Test Coverage

The change mirrors the already-merged ModelOpt fix (#14930) for the identical
root cause. It is exercised by the existing Triton-backend FP8 quantization
tests that originally hit the HfUriError, e.g.:

triton_server/test_triton_llm.py::test_llama_v3_speculative_decoding_bls
triton_server/test_triton_rcca.py::test_rcca_bug_4714193

as well as any TensorRT-backend convert.py calibration run that uses the
default --calib_dataset cnn_dailymail.

PR Checklist

Please review the following before submitting your PR:

PR description clearly explains what and why.
PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.
Test cases are provided for new code paths (see test instructions)
If PR introduces API changes, an appropriate PR label is added - either api-compatible or api-breaking. For api-breaking, include BREAKING in the PR title.
Any new dependencies have been scanned for license and vulnerabilities
CODEOWNERS updated if ownership changes
Documentation updated as needed
Update tava architecture diagram if there is a significant design change in PR.
The reviewers assigned automatically/manually are appropriate for the PR.
Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

To see a list of available CI bot commands, please comment /bot help.

Tabrizian · 2026-06-30T22:02:08Z

/bot run --disable-fail-fast

tensorrt-cicd · 2026-06-30T22:07:43Z

PR_Github #56734 [ run ] triggered by Bot. Commit: d03a874 Link to invocation

…_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>

Tabrizian · 2026-06-30T22:08:37Z

/bot run --disable-fail-fast

coderabbitai · 2026-06-30T22:10:01Z

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 4508a55b-45ff-4a6c-a64e-c3daf23cb5fd

📥 Commits

Reviewing files that changed from the base of the PR and between a721a08 and 85881a7.

📒 Files selected for processing (1)

tensorrt_llm/models/convert_utils.py

📝 Walkthrough

Walkthrough

This change modifies load_calib_dataset() in convert_utils.py to rewrite the dataset identifier "cnn_dailymail" to the namespaced "abisee/cnn_dailymail" before invoking load_dataset, accommodating updated HuggingFace Hub naming requirements.

Changes

Calibration Dataset Loading

Layer / File(s)	Summary
Dataset id rewrite `tensorrt_llm/models/convert_utils.py`	`load_calib_dataset` now rewrites the bare identifier "cnn_dailymail" to "abisee/cnn_dailymail" before calling `load_dataset`.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title is concise, specific, and matches the bug-fix scope and NVBugs format.
Description check	✅ Passed	The description follows the required template and includes a clear problem statement, test coverage, and checklist completion.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests

_{Comment @coderabbitai help to get the list of available commands.}

tensorrt-cicd · 2026-06-30T22:15:39Z

PR_Github #56736 [ run ] triggered by Bot. Commit: 85881a7 Link to invocation

tensorrt-cicd · 2026-07-01T07:02:55Z

PR_Github #56736 [ run ] completed with state SUCCESS. Commit: 85881a7
/LLM/main/L0_MergeRequest_PR pipeline #45556 completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

Please check the failed tests and fix your PR
If you cannot view the failures, ask the CI triggerer to share details
Once fixed, request an NVIDIA team member to trigger CI again

CI Agent Failure Analysis

Link to invocation

Tabrizian · 2026-07-03T07:31:11Z

/bot run --disable-fail-fast

tensorrt-cicd · 2026-07-03T07:38:16Z

PR_Github #57399 [ run ] triggered by Bot. Commit: 85881a7 Link to invocation

tensorrt-cicd · 2026-07-03T09:02:47Z

PR_Github #57399 [ run ] completed with state FAILURE. Commit: 85881a7
/LLM/main/L0_MergeRequest_PR pipeline #46144 completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

Please check the failed tests and fix your PR
If you cannot view the failures, ask the CI triggerer to share details
Once fixed, request an NVIDIA team member to trigger CI again

CI Agent Failure Analysis

Link to invocation

Tabrizian · 2026-07-06T17:16:37Z

/bot run --disable-fail-fast

tensorrt-cicd · 2026-07-06T17:22:51Z

PR_Github #57789 [ run ] triggered by Bot. Commit: 85881a7 Link to invocation

Tabrizian · 2026-07-06T17:27:14Z

/bot kill

Tabrizian · 2026-07-06T17:28:12Z

The tests are being removed here: #15907

tensorrt-cicd · 2026-07-06T17:35:57Z

PR_Github #57791 [ ] completed with state FAILURE. Commit: 85881a7
Not allowed on merged PR

Link to invocation

tensorrt-cicd · 2026-07-06T18:55:47Z

PR_Github #57789 [ run ] completed with state SUCCESS. Commit: 85881a7
/LLM/main/L0_MergeRequest_PR pipeline #46488 completed with status: 'SUCCESS'

CI Report

Link to invocation

github-actions Bot assigned Tabrizian Jun 30, 2026

Tabrizian force-pushed the cnn-dailymail-calib-dataset-6215107 branch from d03a874 to 85881a7 Compare June 30, 2026 22:08

Tabrizian closed this Jul 6, 2026

lonexreb mentioned this pull request Jul 10, 2026

[None][fix] namespace cnn_dailymail calibration dataset id #16247

Closed

Uh oh!

Conversation

Tabrizian commented Jun 30, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary by CodeRabbit

Description

Test Coverage

PR Checklist

GitHub Bot Help

Uh oh!

Tabrizian commented Jun 30, 2026

Uh oh!

tensorrt-cicd commented Jun 30, 2026

Uh oh!

Tabrizian commented Jun 30, 2026

Uh oh!

coderabbitai Bot commented Jun 30, 2026

Walkthrough

Changes

Estimated code review effort

Uh oh!

tensorrt-cicd commented Jun 30, 2026

Uh oh!

tensorrt-cicd commented Jul 1, 2026

Uh oh!

Tabrizian commented Jul 3, 2026

Uh oh!

tensorrt-cicd commented Jul 3, 2026

Uh oh!

tensorrt-cicd commented Jul 3, 2026

Uh oh!

Tabrizian commented Jul 6, 2026

Uh oh!

tensorrt-cicd commented Jul 6, 2026

Uh oh!

Tabrizian commented Jul 6, 2026

Uh oh!

Tabrizian commented Jul 6, 2026

Uh oh!

tensorrt-cicd commented Jul 6, 2026

Uh oh!

tensorrt-cicd commented Jul 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Tabrizian commented Jun 30, 2026 •

edited by coderabbitai Bot

Loading