Jan Lasek

Copy sparse, dense and labels from the right source when shuffling Criteo data

1

When dense and spare features are located in different directories, i.e., `input_dir_labels_and_dense != input_dir_sparse` as the input to `shuffle()` function, this method throws an error in [L642](https://github.com/pytorch/torchrec/blob/71d51c8764c141ff8d849f73bcf06548ffad36c4/torchrec/datasets/criteo.py#L642) like: ``` FileNotFoundError:...

# What does this PR do ? Integrating AMMO library to the project and providing utilities for quantizing models with Llama2 PTQ example. Different quantization algorithms are available including **INT8...

# What does this PR do ? Some distributed workloads can possibly be launched "manually" with `mpirun -n N python ...`. I'm adapting `nemo.utils.get_rank.py` module to take it into account...

# What does this PR do ? Currently Megatron-LM creates causal attention mask for a local-layer spec model on its own -- see https://github.com/NVIDIA/Megatron-LM/commit/a45805a3ee0645b85b48d14b0a8077fa5b1216b2 -- and hence `attention_mask=None` can be...

Unused imports cleanup

6

# What does this PR do ? Unused imports cleanup indentified by [Flake8](https://flake8.pycqa.org/en/latest), see https://www.flake8rules.com/rules/F401.html. **Collection**: [ALL] # Jenkins CI To run Jenkins, a NeMo User with write access must...

# What does this PR do ? Restoring PTQ tests for Llama2 model. The tests are moved here from recently disabled [Jenkinsfile](https://github.com/NVIDIA/NeMo/blob/main/Jenkinsfile). Testing all methods: FP8, INT8 SQ, INT4 AWQ...

Extend Nemo AutoTokenizer & SentencePieceTokenizer API for TensorRT-LLM & AMMO evaluation scripts usage

5

# What does this PR do ? Extending NeMo's AutoTokenizer and SentencePieceTokenizer for consistency with HuggingFace tokenizers API as it is used throughout TensorRT-LLM and AMMO in evaluation tools. This...

# What does this PR do ? Summary: * use Model Optimizer library (formerly AMMO) for LLM PTQ workflow * restore PTQ tests failing due to setting `apply_rope_fusion = True`...

# What does this PR do ? Fix TRT-LLM engine export issues for quantized "qnemo" models. This is a cherry-pick mirror of recent changes introduced in the main branch in...

# What does this PR do ? Extra changes just for cleanup purposes related to https://github.com/NVIDIA/NeMo/pull/10904, isolated here to facilitate review for main functionalities. **Collection**: NLP # Changelog - Add...