AUT-681 Fix QAD functional test: use local jsonl for PTQ calibration to avoid offline HF Hub fetch#5651
Open
svcnemo-autobot wants to merge 1 commit into
Conversation
… fetch
The hybrid_nemotron_v3_pico_7b_a1b_tp1_ep8_QAD_dgx_h100_1N8G functional test's
PTQ (Stage 1) step passed `--calib-dataset-path-or-name cnn_dailymail`, which
makes ModelOpt fetch a calibration dataset from the HuggingFace Hub. The CI
runner is offline, so calibration aborts with
`ConnectionError: Couldn't reach '...' on the Hub (OfflineModeIsEnabled)`.
Point calibration at the local SFT jsonl the test already stages
(`${DATA_PATH}/text/nemotron-3-super-sft_train-sample.jsonl`, also used as
`--data-path`), so the local-file calibration path is used and no Hub access
is required.
Collaborator
Author
|
/ok to test d98813b |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The new
hybrid_nemotron_v3_pico_7b_a1b_tp1_ep8_QAD_dgx_h100_1N8Gtest did run in the linked run's attempt 2, but failed at its PTQ (Stage 1) step withConnectionError: Couldn't reach 'nvidia/Nemotron-Post-Training-Dataset-v2' on the Hub (OfflineModeIsEnabled)because calibration was pointed at a HuggingFace-Hub dataset while the CI runner is offline. I changed the test'smodel_config.yaml--calib-dataset-path-or-namefromcnn_dailymailto the local SFT jsonl the test already stages (${DATA_PATH}/text/nemotron-3-super-sft_train-sample.jsonl), which routes through the offline local-.jsonlcalibration path that PR #4520's ownquantize.pychange added — removing the Hub dependency. Assumed the local-file calibration path is exercised (per PR #4520'sos.path.isfile(...) and endswith(".jsonl")guard); if you intended the calibration to keep using a Hub dataset and instead want it pre-cached under the runner'sHF_HOME, comment and I'll switch approaches.