[None][fix] Fix AutoDeploy accuracy tests by bmarimuthu-nv · Pull Request #13925 · NVIDIA/TensorRT-LLM

bmarimuthu-nv · 2026-05-08T22:46:48Z

Description

Fix and unwaive a batch of AutoDeploy accuracy tests, and make the quant_config_reader standalone-compatible so the standalone llmc package's unit-test suite is green.

AutoDeploy accuracy / multimodal fixes

Restore AutoDeploy accuracy coverage on Gemma4 / Qwen3.5-MoE: rebuild Qwen multimodal get_vocab_size to prefer the base processor / safely handle tokenizers without vocab_size, drop dense token_type_ids masks the AD wrapper rebuilds from compact span metadata, and add mm_token_type_ids to the carried multimodal keys.
Iterate on the multimodal vocab interface (avoid unnecessary tokenizer probing → remove redundant per-tokenizer overrides → restore the public get_vocab_size interface as the final state).
Relax Qwen vision-tower tolerance (rtol/atol 1e-5 → 1e-4) for test_vision_attention_matches_reference, test_vision_block_matches_reference, and test_vlm_wrapper_delta_is_request_scoped_no_cross_call_leakage, and unwaive the corresponding unittest/auto_deploy/singlegpu/models/test_qwen3_5_moe.py cases.
Fix the input-processor replacement check on the AD path so LLM._build_model adopts the multimodal processor created by _create_input_processor.

Waive / unwaive churn

Unwaive TestGemma4MoE::test_bf16, TestGemmaE2B::test_gemma4_e2b_it, TestMiniMaxM2::test_finegrained_fp8, and TestQwen3_5_397B_MoE::test_bf16_small[4] now that the accuracy regressions are fixed.
Unwaive several other AutoDeploy tests: Eagle3 one-model variants, TestModelRegistryAccuracy registry cases, and the qwen vision-tower tests above.
Bump MiniMax-M2 gsm8k reference score.

Standalone llmc package fix

models/quant_config_reader.py imported tensorrt_llm.quantization.modelopt_config at module top-level, which transitively pulled in tensorrt_llm.mapping → tensorrt_llm._torch.device_mesh. None of these exist in the standalone llmc package, so models/hf.py and every custom model that re-exports from it failed to load — leaving ModelFactoryRegistry empty and producing 28+ collection errors in the standalone unit-test suite.
Gate the import on _compat.TRTLLM_AVAILABLE and provide stubs for standalone mode (the ModelOPT reader path itself is unreachable without TRT-LLM).
Excluded a few additional llmc-incompatible test files in create_standalone_package.py (test_glm4_moe_modeling, test_glm_moe_dsa_modeling, test_granite_moe_hybrid_modeling, test_mxfp4_moe_layout).

Test Coverage

accuracy/test_llm_api_autodeploy.py::TestGemma4MoE::test_bf16
accuracy/test_llm_api_autodeploy.py::TestGemmaE2B::test_gemma4_e2b_it
accuracy/test_llm_api_autodeploy.py::TestMiniMaxM2::test_finegrained_fp8
accuracy/test_llm_api_autodeploy.py::TestQwen3_5_397B_MoE::test_bf16_small[4]
accuracy/test_llm_api_autodeploy.py::TestLlama3_1_8B_Instruct_Eagle3::test_eagle3_one_model[flashinfer|trtllm]
accuracy/test_llm_api_autodeploy.py::TestModelRegistryAccuracy::test_autodeploy_from_registry[...]
unittest/auto_deploy/singlegpu/models/test_gemma4_modeling.py — new test_conditional_generation_wrapper_handles_exported_text_graph_without_per_layer_inputs
unittest/auto_deploy/singlegpu/models/test_qwen3_5_moe.py::test_vision_attention_matches_reference|test_vision_block_matches_reference|test_vlm_wrapper_delta_is_request_scoped_no_cross_call_leakage
unittest/auto_deploy/singlegpu/shim/test_llm_config.py — new test_build_model_replaces_parent_model_specific_input_processor
unittest/auto_deploy/standalone/test_standalone_import.py and test_standalone_package.py — both tests previously failing on upstream/main (4 failed, 32 errors); now 12 passed, 0 failed, 0 errors with 1857 inner unit tests green.

PR Checklist

Please review the following before submitting your PR:

PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.
PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.
Test cases are provided for new code paths (see test instructions)
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.

bmarimuthu-nv · 2026-05-09T00:02:24Z

@coderabbitai summary

coderabbitai · 2026-05-09T00:02:30Z

✅ Actions performed

Summary regeneration triggered.

bmarimuthu-nv · 2026-05-09T00:02:56Z

/bot run --stage-list "A10-Build_Docs, A10-PackageSanityCheck-PY310-UB2204, A100X-PackageSanityCheck-PY312-UB2404, A30-AutoDeploy-1, H100_PCIe-AutoDeploy-1, DGX_B200-AutoDeploy-1, A100X-PyTorch-1, DGX_H100-4_GPUs-AutoDeploy-1, DGX_B200-4_GPUs-AutoDeploy-1, DGX_H100-4_GPUs-AutoDeploy-Post-Merge-1, DGX_B200-8_GPUs-AutoDeploy-Post-Merge-1" --disable-fail-fast

coderabbitai · 2026-05-09T00:03:32Z

📝 Walkthrough

Walkthrough

This PR contains three independent model and test updates: Gemma4 switches from a custom tokenizer wrapper to transformers.AutoTokenizer with corrected tied-weights mapping; Qwen3.5 MoE removes token-type tensor fields from multimodal input processing paths; test reference accuracy, import wiring, and waiver entries are updated.

Changes

Gemma4 AutoTokenizer Migration

Layer / File(s)	Summary
Imports and Type Updates `tensorrt_llm/_torch/auto_deploy/models/custom/modeling_gemma4.py`	Imports updated to use `AutoTokenizer` directly; `Gemma4ForCausalLM._tied_weights_keys` corrected to map `lm_head.weight` to `embed_tokens.weight` instead of `model.embed_tokens.weight`.
Wrapper Class Removal `tensorrt_llm/_torch/auto_deploy/models/custom/modeling_gemma4.py`	Tokenizer-specific constants block replaced; `ADGemma4Tokenizer` class completely removed.
Processor Integration `tensorrt_llm/_torch/auto_deploy/models/custom/modeling_gemma4.py`	`ADGemma4Processor.__init__` now accepts `tokenizer: Any`; `from_pretrained` constructs tokenizer via `AutoTokenizer.from_pretrained`.
Factory Initialization `tensorrt_llm/_torch/auto_deploy/models/custom/modeling_gemma4.py`	`Gemma4ForConditionalGenerationFactory.init_tokenizer` updated to return `AutoTokenizer.from_pretrained` instead of wrapper method.

Qwen3.5 MoE Token-Type Field Cleanup

Layer / File(s)	Summary
Forward and Input Processing `tensorrt_llm/_torch/auto_deploy/models/custom/modeling_qwen3_5_moe.py`	`Qwen3_5MoeModel.forward` cleanup list now includes `mm_token_type_ids`; `Qwen3_5MoeADInputProcessor` removes `token_type_ids` and `mm_token_type_ids` from `multimodal_data` before layout metadata attachment.
IR Processor Cleanup `tensorrt_llm/_torch/auto_deploy/models/custom/modeling_qwen3_5_moe_ir.py`	Multimodal wrapper removes `mm_token_type_ids` from kwargs; `Qwen3_5MoeADInputProcessor` removes token-type fields from `multimodal_data` and deletes unused `get_vocab_size()` method.

Test Reference and Configuration Updates

Layer / File(s)	Summary
Import Wiring and Docstrings `tests/integration/defs/accuracy/test_llm_api_autodeploy.py`	`llm_models_root` imported from `..conftest` instead of `test_common.llm_data`; minor docstring formatting and punctuation adjustments.
Reference Accuracy `tests/integration/defs/accuracy/references/gsm8k.yaml`	`MiniMaxAI/MiniMax-M2` top-level accuracy changed from `93.75` to `89.045`.
Test Waiver Updates `tests/integration/test_lists/waives.txt`	Removed waivers for GLM4Flash, Gemma4MoE, and MiniMaxM2 tests; adjusted waiver entries accordingly.

🎯 3 (Moderate) | ⏱️ ~25 minutes

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 29.41% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
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.
Description check	✅ Passed	PR description provides detailed explanation of changes, rationale, test coverage, and checklist validation with substantive implementation-specific details.
Title check	✅ Passed	The title '[None][fix] Fix AutoDeploy accuracy tests' is directly related to the changeset, which addresses AutoDeploy accuracy tests by removing test waivers, updating reference accuracy values, and fixing tokenizer/multimodal handling in custom models.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests

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

tensorrt-cicd · 2026-05-09T00:09:02Z

PR_Github #47450 [ run ] triggered by Bot. Commit: 2e89ebb Link to invocation

coderabbitai

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

tensorrt_llm/_torch/auto_deploy/models/custom/modeling_gemma4.py (1)
2291-2299: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Update transformers requirement to ≥5.5.0 or add version guard before AutoTokenizer.from_pretrained() calls.

The code uses AutoTokenizer.from_pretrained() at lines 2296 and 2635 to load Gemma4 tokenizers, but the repo is pinned to transformers==5.3.0. Gemma4 was introduced in Transformers v5.5.0; attempting to load it with 5.3.0 will fail with "Transformers does not recognize this architecture." Either update requirements.txt to transformers>=5.5.0 throughout the repo, or add a version check (similar to the pattern in transformers_causal_mask.py) to guard these calls and fail fast with a clear error message if the requirement is not met.

Also applies to: 2632-2635
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tensorrt_llm/_torch/auto_deploy/models/custom/modeling_gemma4.py` around
lines 2291 - 2299, The Gemma4 tokenizer calls in
ADGemma4Processor.from_pretrained (and the similar call near the end of
modeling_gemma4.py) rely on Transformers >=5.5.0; either update the repo
requirement to transformers>=5.5.0 or add a runtime version guard before calling
AutoTokenizer.from_pretrained() that checks transformers.__version__ (or uses
packaging.version.parse) and raises a clear, fast-failing error indicating the
minimum version required. Locate the two AutoTokenizer.from_pretrained usages in
modeling_gemma4.py (inside ADGemma4Processor.from_pretrained and the other
Gemma4 loader) and implement the version check there so the code never calls
AutoTokenizer when the installed transformers is <5.5.0.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Outside diff comments:
In `@tensorrt_llm/_torch/auto_deploy/models/custom/modeling_gemma4.py`:
- Around line 2291-2299: The Gemma4 tokenizer calls in
ADGemma4Processor.from_pretrained (and the similar call near the end of
modeling_gemma4.py) rely on Transformers >=5.5.0; either update the repo
requirement to transformers>=5.5.0 or add a runtime version guard before calling
AutoTokenizer.from_pretrained() that checks transformers.__version__ (or uses
packaging.version.parse) and raises a clear, fast-failing error indicating the
minimum version required. Locate the two AutoTokenizer.from_pretrained usages in
modeling_gemma4.py (inside ADGemma4Processor.from_pretrained and the other
Gemma4 loader) and implement the version check there so the code never calls
AutoTokenizer when the installed transformers is <5.5.0.

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 4e8b3ad0-f423-48b0-a7ea-a9494d07e7bc

📥 Commits

Reviewing files that changed from the base of the PR and between 43f4b94 and 2e89ebb.

📒 Files selected for processing (6)

tensorrt_llm/_torch/auto_deploy/models/custom/modeling_gemma4.py
tensorrt_llm/_torch/auto_deploy/models/custom/modeling_qwen3_5_moe.py
tensorrt_llm/_torch/auto_deploy/models/custom/modeling_qwen3_5_moe_ir.py
tests/integration/defs/accuracy/references/gsm8k.yaml
tests/integration/defs/accuracy/test_llm_api_autodeploy.py
tests/integration/test_lists/waives.txt

💤 Files with no reviewable changes (1)

tests/integration/test_lists/waives.txt

tensorrt-cicd · 2026-05-09T06:00:15Z

PR_Github #47450 [ run ] completed with state SUCCESS. Commit: 2e89ebb
/LLM/main/L0_MergeRequest_PR pipeline #37371 (Partly Tested) 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

bmarimuthu-nv · 2026-05-11T22:25:27Z

/bot run

tensorrt-cicd · 2026-05-11T22:31:35Z

PR_Github #47796 [ run ] triggered by Bot. Commit: 61aadb7 Link to invocation

tensorrt-cicd · 2026-05-12T03:30:00Z

PR_Github #47796 [ run ] completed with state SUCCESS. Commit: 61aadb7
/LLM/main/L0_MergeRequest_PR pipeline #37687 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

suyoggupta · 2026-05-12T06:16:10Z

/bot run

tensorrt-cicd · 2026-05-12T06:21:51Z

PR_Github #47905 [ run ] triggered by Bot. Commit: 9ed5636 Link to invocation

tensorrt-cicd · 2026-05-12T11:21:38Z

PR_Github #47905 [ run ] completed with state SUCCESS. Commit: 9ed5636
/LLM/main/L0_MergeRequest_PR pipeline #37752 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

suyoggupta · 2026-05-12T20:54:09Z

/bot run

tensorrt-cicd · 2026-05-12T20:59:36Z

PR_Github #48026 [ run ] triggered by Bot. Commit: 9ed5636 Link to invocation

bmarimuthu-nv · 2026-05-12T21:35:15Z

/bot run --disable-fail-fast

tensorrt-cicd · 2026-05-12T21:41:38Z

PR_Github #48034 [ run ] triggered by Bot. Commit: b3e6a78 Link to invocation

tensorrt-cicd · 2026-05-13T07:23:05Z

PR_Github #48034 [ run ] completed with state SUCCESS. Commit: b3e6a78
/LLM/main/L0_MergeRequest_PR pipeline #37868 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

Signed-off-by: Balamurugan Marimuthu <246387390+bmarimuthu-nv@users.noreply.github.com>

suyoggupta · 2026-06-03T20:26:54Z

/bot run

tensorrt-cicd · 2026-06-03T20:32:54Z

PR_Github #51901 [ run ] triggered by Bot. Commit: 44debfd Link to invocation

tensorrt-cicd · 2026-06-04T00:12:13Z

PR_Github #51901 [ run ] completed with state FAILURE. Commit: 44debfd
/LLM/main/L0_MergeRequest_PR pipeline #41258 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

Link to invocation

bmarimuthu-nv · 2026-06-04T00:19:42Z

/bot run

tensorrt-cicd · 2026-06-04T00:25:22Z

PR_Github #51924 [ run ] triggered by Bot. Commit: 44debfd Link to invocation

tensorrt-cicd · 2026-06-04T03:14:56Z

PR_Github #51924 [ run ] completed with state FAILURE. Commit: 44debfd
/LLM/main/L0_MergeRequest_PR pipeline #41278 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

bmarimuthu-nv · 2026-06-04T07:25:25Z

/bot run --disable-fail-fast

tensorrt-cicd · 2026-06-04T07:31:24Z

PR_Github #52008 [ run ] triggered by Bot. Commit: 44debfd Link to invocation

tensorrt-cicd · 2026-06-04T13:57:25Z

PR_Github #52008 [ run ] completed with state SUCCESS. Commit: 44debfd
/LLM/main/L0_MergeRequest_PR pipeline #41352 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

Link to invocation

bmarimuthu-nv · 2026-06-04T16:21:17Z

/bot run --disable-fail-fast

tensorrt-cicd · 2026-06-04T16:30:19Z

PR_Github #52107 [ run ] triggered by Bot. Commit: 44debfd Link to invocation

bmarimuthu-nv · 2026-06-04T17:05:17Z

/bot run --stage-list "A10-Build_Docs, A10-PackageSanityCheck-PY310-UB2204, A100X-PackageSanityCheck-PY312-UB2404, A30-AutoDeploy-1, H100_PCIe-AutoDeploy-1, DGX_B200-AutoDeploy-1, A100X-PyTorch-1, DGX_H100-4_GPUs-AutoDeploy-1, DGX_B200-4_GPUs-AutoDeploy-1"

tensorrt-cicd · 2026-06-04T17:12:17Z

PR_Github #52118 [ run ] triggered by Bot. Commit: 44debfd Link to invocation

tensorrt-cicd · 2026-06-04T17:15:55Z

PR_Github #52107 [ run ] completed with state ABORTED. Commit: 44debfd

Link to invocation

tensorrt-cicd · 2026-06-04T17:41:47Z

PR_Github #52118 [ run ] completed with state SUCCESS. Commit: 44debfd
/LLM/main/L0_MergeRequest_PR pipeline #41443 (Partly Tested) completed with status: 'SUCCESS'

CI Report

Link to invocation

bmarimuthu-nv · 2026-06-04T17:53:27Z

/bot skip --comment "All autoDeploy stages pass, it only merges fixes, unwaive will be added in follow up PR"

tensorrt-cicd · 2026-06-04T17:59:41Z

PR_Github #52131 [ skip ] triggered by Bot. Commit: 44debfd Link to invocation

tensorrt-cicd · 2026-06-04T18:11:05Z

PR_Github #52131 [ skip ] completed with state SUCCESS. Commit: 44debfd
Skipping testing for commit 44debfd

Link to invocation

…gle-GPU test_single_request_chat_multiple_images[pd_disagg-qwen3_30b_a3b_fp8] in test_mm_encoder_standalone.py fails at setup on the pre-merge DGX_B200 single-GPU stage with a NIXL CacheTransceiver init assertion (status == NIXL_SUCCESS, transferAgent.cpp:614) when the pd_disagg LLM constructs its disaggregated KV-cache transfer agent. This is a fleet-wide failure on the single-GPU pre-merge stage, observed across many unrelated PRs (e.g. NVIDIA#13978, NVIDIA#13925, NVIDIA#14841, NVIDIA#14599, NVIDIA#14524, NVIDIA#14941, NVIDIA#14398); it passes only intermittently (~1/3) depending on node. Waiving until the single-GPU NIXL EPD-disagg path is fixed or the variant is gated to multi-GPU. NVBug: https://nvbugs/6269683 Signed-off-by: venkywonka <23023424+venkywonka@users.noreply.github.com>

bmarimuthu-nv · 2026-06-04T20:27:52Z

Unwaives added in this PR: #14971

Signed-off-by: Balamurugan Marimuthu <246387390+bmarimuthu-nv@users.noreply.github.com> Co-authored-by: Suyog Gupta <41447211+suyoggupta@users.noreply.github.com> Signed-off-by: NVFB <186336021+NVFB@users.noreply.github.com>

Signed-off-by: Balamurugan Marimuthu <246387390+bmarimuthu-nv@users.noreply.github.com> Co-authored-by: Suyog Gupta <41447211+suyoggupta@users.noreply.github.com>

github-actions Bot assigned bmarimuthu-nv May 8, 2026

bmarimuthu-nv force-pushed the bala/ci-acc-test-fix2 branch from 62866f2 to 107756e Compare May 8, 2026 22:50

bmarimuthu-nv marked this pull request as ready for review May 9, 2026 00:02

bmarimuthu-nv requested review from a team as code owners May 9, 2026 00:02

bmarimuthu-nv requested a review from marinayanov May 9, 2026 00:02

suyoggupta approved these changes May 9, 2026

View reviewed changes

suyoggupta reviewed May 9, 2026

View reviewed changes

Comment thread tests/integration/defs/accuracy/references/gsm8k.yaml

coderabbitai Bot reviewed May 9, 2026

View reviewed changes

bmarimuthu-nv force-pushed the bala/ci-acc-test-fix2 branch from 2e89ebb to 61aadb7 Compare May 11, 2026 22:24

bmarimuthu-nv mentioned this pull request May 11, 2026

[None][feat] AutoDeploy push the rope buffer to later stage #13859

Merged

suyoggupta force-pushed the bala/ci-acc-test-fix2 branch from 61aadb7 to 9ed5636 Compare May 12, 2026 06:15

bmarimuthu-nv force-pushed the bala/ci-acc-test-fix2 branch from b3e6a78 to b0d1677 Compare May 13, 2026 18:05

bmarimuthu-nv added 2 commits June 3, 2026 10:24

[None][fix] address AutoDeploy CI fallout

6d14578

Signed-off-by: Balamurugan Marimuthu <246387390+bmarimuthu-nv@users.noreply.github.com>

[None][fix] restore Qwen vision tolerances

ebed77e

Signed-off-by: Balamurugan Marimuthu <246387390+bmarimuthu-nv@users.noreply.github.com>

bmarimuthu-nv force-pushed the bala/ci-acc-test-fix2 branch from 9ffadf6 to ebed77e Compare June 3, 2026 18:05

Merge branch 'main' into bala/ci-acc-test-fix2

44debfd

venkywonka mentioned this pull request Jun 4, 2026

[None][test] Waive qwen3_30b_a3b_fp8 pd_disagg mm-encoder test on single-GPU (NIXL setup failure) #14968

Open

bmarimuthu-nv changed the title ~~[None][fix] Fix and unwaive AutoDeploy accuracy tests~~ [None][fix] Fix AutoDeploy accuracy tests Jun 4, 2026

bmarimuthu-nv merged commit a8c4007 into NVIDIA:main Jun 4, 2026
10 checks passed

coderabbitai Bot mentioned this pull request Jun 8, 2026

[https://nvbugs/6244474][fix] AutoDeploy: Remove llama perf test from CI #15107

Merged

1 task

Uh oh!

Conversation

bmarimuthu-nv commented May 8, 2026 • edited by suyoggupta Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

AutoDeploy accuracy / multimodal fixes

Waive / unwaive churn

Standalone llmc package fix

Test Coverage

PR Checklist

GitHub Bot Help

Uh oh!

bmarimuthu-nv commented May 9, 2026

Uh oh!

coderabbitai Bot commented May 9, 2026

Uh oh!

bmarimuthu-nv commented May 9, 2026

Uh oh!

coderabbitai Bot commented May 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

❌ Failed checks (1 warning)

Uh oh!

Uh oh!

tensorrt-cicd commented May 9, 2026

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

tensorrt-cicd commented May 9, 2026

Uh oh!

bmarimuthu-nv commented May 11, 2026

Uh oh!

tensorrt-cicd commented May 11, 2026

Uh oh!

tensorrt-cicd commented May 12, 2026

Uh oh!

suyoggupta commented May 12, 2026

Uh oh!

tensorrt-cicd commented May 12, 2026

Uh oh!

tensorrt-cicd commented May 12, 2026

Uh oh!

suyoggupta commented May 12, 2026

Uh oh!

tensorrt-cicd commented May 12, 2026

Uh oh!

bmarimuthu-nv commented May 12, 2026

Uh oh!

tensorrt-cicd commented May 12, 2026

Uh oh!

tensorrt-cicd commented May 13, 2026

Uh oh!

suyoggupta commented Jun 3, 2026

Uh oh!

tensorrt-cicd commented Jun 3, 2026

Uh oh!

tensorrt-cicd commented Jun 4, 2026

Uh oh!

bmarimuthu-nv commented Jun 4, 2026

Uh oh!

tensorrt-cicd commented Jun 4, 2026

Uh oh!

tensorrt-cicd commented Jun 4, 2026

Uh oh!

bmarimuthu-nv commented Jun 4, 2026

Uh oh!

tensorrt-cicd commented Jun 4, 2026

Uh oh!

tensorrt-cicd commented Jun 4, 2026

Uh oh!

bmarimuthu-nv commented Jun 4, 2026

Uh oh!

tensorrt-cicd commented Jun 4, 2026

Uh oh!

bmarimuthu-nv commented Jun 4, 2026

Uh oh!

tensorrt-cicd commented Jun 4, 2026

Uh oh!

bmarimuthu-nv commented May 8, 2026 •

edited by suyoggupta

Loading

coderabbitai Bot commented May 9, 2026 •

edited

Loading