Home / deepspeedai / DeepSpeedExamples
AI Slop Code RIsk Monitor

deepspeedai/DeepSpeedExamples

Example models using DeepSpeed

25.9 Elevated pattern signal View on GitHub

Maintenance and Governance Review

This report summarizes scanner evidence for deepspeedai/DeepSpeedExamples, a Python project with 6,848 GitHub stars. SynthScan examined 92,032 lines of code across 671 source files, recording 1194 pattern matches distributed across 21 syntactic categories. The legacy adjusted pattern signal is 25.9 (Elevated pattern signal); it is not an authorship estimate.

Use the findings to ask whether maintainers can explain intent, verify dependencies, simplify unnecessary scaffolding, and focus review effort. Risk Payload v1 adds ownership, supply-chain, over-engineering, and review-effort vectors when available. These are prompts for human verification, not probabilities, security findings, or measured maintenance costs.

25.9
Adjusted Signal
25.9
Raw Pattern Score
100%
Time Factor
2026-09-30
Last Push
6.8K
Stars
Python
Language
92.0K
Lines of Code
671
Files
1.2K
Pattern Hits
2026-10-12
Scan Date
0.40
HC Hit Rate

Risk Review · v3

Use each vector to guide manual verification of the underlying findings.

Cognitive Ownership Risk 27.6/100 · MEDIUM

Can maintainers explain intent, assumptions, and trade-offs? 488 finding(s).

Hallucinated Dependency & Supply Chain Risk 18.6/100 · LOW

Are imports declared and package/API names verified? 370 finding(s).

Agent Over-Engineering Ratio 59.0/100 · HIGH

Are wrappers, guards, and repeated structures necessary? 554 finding(s).

PR Review Velocity Impact 76.3/100 · CRITICAL

Which changes need focused human review? 1194 finding(s).

What These Metrics Mean

Adjusted Signal
Legacy pattern score normalized per 1,000 scanned lines and multiplied by the repository-level time factor. It is not an estimate of AI authorship.
Raw Pattern Score
Sum of severity- and context-weighted pattern findings before applying the time factor.
Time Factor
A coarse multiplier based on repository last-push metadata. It does not identify when individual code was written.
Pattern Hits
Count of configured pattern findings. Review each location and surrounding code before acting.
HC Hit Rate
High- and critical-severity findings per scanned file. Severity is a configured detector weight, not proof of a vulnerability or authorship.
Lines / Files
Scan coverage denominators used by the legacy normalized score.

Pattern-Signal History

This chart tracks the legacy adjusted pattern score across scan runs. Changes can reflect source changes, scan coverage, or detector/scoring versions; they do not measure AI adoption or prove that a team refactored code. Compare runs only when their scoring versions are compatible. The dashed line tracks the raw pattern-hit count.

Severity Breakdown

Groups findings by configured severity weight. Severity is not a confidence estimate, security rating, or proof that a package is missing. Verify imports against the project's manifests and inspect each finding in its local design context.

CRITICAL 0HIGH 271MEDIUM 50LOW 873

Directory Score Breakdown

This horizontal bar chart shows where configured pattern scores accumulate across directories. Treat the bars as locations to inspect, not as evidence of authorship or code quality. Use the file-level findings to decide whether a directory needs a focused ownership, dependency, complexity, or review check.

Pattern Findings

The scanner identified 1194 distinct pattern matches across 21 syntactic categories. Each entry below represents a discrete location in the source code where the engine recorded a detector match to inspect in context. Expand any category row to inspect the individual file paths, line numbers, code snippets, and the lexical context (CODE, COMMENT, or STRING) in which each match was detected.

Reading the findings table: The Severity column indicates the configured finding weight (CRITICAL / HIGH / MEDIUM / LOW). The Context column identifies whether the match occurred inside executable code, an inline comment, or a string literal — legacy comment-context matches receive a ×1.5 score weight. The ⚡ bolt icon marks clustered matches: three or more patterns within a 10-line window, each receiving an additional ×1.5 density multiplier under the legacy scoring model.

Cross-File Repetition249 hits · 1245 pts
SeverityFileLineSnippetContext
HIGH…aining/data_efficiency/vit_finetuning/main_imagenet.py0computes the accuracy over the k top predictions for the specified values of kSTRING
HIGHtraining/data_efficiency/vit_finetuning/utils/utils.py0computes the accuracy over the k top predictions for the specified values of kSTRING
HIGHtraining/imagenet/main.py0computes the accuracy over the k top predictions for the specified values of kSTRING
HIGHtraining/bing_bert/deepspeed_train.py0utility function for checkpointing model + optimizer dictionaries the main purpose for this is to be able to resume traiSTRING
HIGHtraining/bing_bert/run_glue_classifier_bert_large.py0utility function for checkpointing model + optimizer dictionaries the main purpose for this is to be able to resume traiSTRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_large.py0utility function for checkpointing model + optimizer dictionaries the main purpose for this is to be able to resume traiSTRING
HIGHtraining/bing_bert/run_glue_classifier_bert_base.py0a single training/test example for simple sequence classification.STRING
HIGHtraining/bing_bert/run_glue_classifier_bert_large.py0a single training/test example for simple sequence classification.STRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_base.py0a single training/test example for simple sequence classification.STRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_large.py0a single training/test example for simple sequence classification.STRING
HIGHtraining/bing_bert/run_glue_classifier_bert_base.py0constructs a inputexample. args: guid: unique id for the example. text_a: string. the untokenized text of the first sequSTRING
HIGHtraining/bing_bert/run_glue_classifier_bert_large.py0constructs a inputexample. args: guid: unique id for the example. text_a: string. the untokenized text of the first sequSTRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_base.py0constructs a inputexample. args: guid: unique id for the example. text_a: string. the untokenized text of the first sequSTRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_large.py0constructs a inputexample. args: guid: unique id for the example. text_a: string. the untokenized text of the first sequSTRING
HIGHtraining/bing_bert/run_glue_classifier_bert_base.py0base class for data converters for sequence classification data sets.STRING
HIGHtraining/bing_bert/run_glue_classifier_bert_large.py0base class for data converters for sequence classification data sets.STRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_base.py0base class for data converters for sequence classification data sets.STRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_large.py0base class for data converters for sequence classification data sets.STRING
HIGHtraining/bing_bert/run_glue_classifier_bert_base.py0gets a collection of `inputexample`s for the train set.STRING
HIGHtraining/bing_bert/run_glue_classifier_bert_large.py0gets a collection of `inputexample`s for the train set.STRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_base.py0gets a collection of `inputexample`s for the train set.STRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_large.py0gets a collection of `inputexample`s for the train set.STRING
HIGHtraining/bing_bert/run_glue_classifier_bert_base.py0gets a collection of `inputexample`s for the dev set.STRING
HIGHtraining/bing_bert/run_glue_classifier_bert_large.py0gets a collection of `inputexample`s for the dev set.STRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_base.py0gets a collection of `inputexample`s for the dev set.STRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_large.py0gets a collection of `inputexample`s for the dev set.STRING
HIGHtraining/bing_bert/run_glue_classifier_bert_base.py0processor for the multinli mismatched data set (glue version).STRING
HIGHtraining/bing_bert/run_glue_classifier_bert_large.py0processor for the multinli mismatched data set (glue version).STRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_base.py0processor for the multinli mismatched data set (glue version).STRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_large.py0processor for the multinli mismatched data set (glue version).STRING
HIGHtraining/bing_bert/run_glue_classifier_bert_base.py0truncates a sequence pair in place to the maximum length.STRING
HIGHtraining/bing_bert/run_glue_classifier_bert_large.py0truncates a sequence pair in place to the maximum length.STRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_base.py0truncates a sequence pair in place to the maximum length.STRING
HIGHtraining/BingBertGlue/run_glue_classifier_bert_large.py0truncates a sequence pair in place to the maximum length.STRING
HIGH…ning/bing_bert/pytorch_pretrained_bert/optimization.py0implements bert version of adam algorithm with weight decay fix. params: lr: learning rate warmup: portion of t_total foSTRING
HIGH…g/BingBertGlue/pytorch_pretrained_bert/optimization.py0implements bert version of adam algorithm with weight decay fix. params: lr: learning rate warmup: portion of t_total foSTRING
HIGH…/BingBertSquad/pytorch_pretrained_bert/optimization.py0implements bert version of adam algorithm with weight decay fix. params: lr: learning rate warmup: portion of t_total foSTRING
HIGH…ning/bing_bert/pytorch_pretrained_bert/optimization.py0performs a single optimization step. arguments: closure (callable, optional): a closure that reevaluates the model and rSTRING
HIGH…g/BingBertGlue/pytorch_pretrained_bert/optimization.py0performs a single optimization step. arguments: closure (callable, optional): a closure that reevaluates the model and rSTRING
HIGH…/BingBertSquad/pytorch_pretrained_bert/optimization.py0performs a single optimization step. arguments: closure (callable, optional): a closure that reevaluates the model and rSTRING
HIGH…ning/bing_bert/pytorch_pretrained_bert/tokenization.py0runs basic whitespace cleaning and splitting on a peice of text.STRING
HIGH…g/BingBertGlue/pytorch_pretrained_bert/tokenization.py0runs basic whitespace cleaning and splitting on a peice of text.STRING
HIGH…/BingBertSquad/pytorch_pretrained_bert/tokenization.py0runs basic whitespace cleaning and splitting on a peice of text.STRING
HIGH…ning/bing_bert/pytorch_pretrained_bert/tokenization.py0runs end-to-end tokenization: punctuation splitting + wordpieceSTRING
HIGH…g/BingBertGlue/pytorch_pretrained_bert/tokenization.py0runs end-to-end tokenization: punctuation splitting + wordpieceSTRING
HIGH…/BingBertSquad/pytorch_pretrained_bert/tokenization.py0runs end-to-end tokenization: punctuation splitting + wordpieceSTRING
HIGH…ning/bing_bert/pytorch_pretrained_bert/tokenization.py0converts a sequence of tokens into ids using the vocab.STRING
HIGH…g/BingBertGlue/pytorch_pretrained_bert/tokenization.py0converts a sequence of tokens into ids using the vocab.STRING
HIGH…/BingBertSquad/pytorch_pretrained_bert/tokenization.py0converts a sequence of tokens into ids using the vocab.STRING
HIGH…ning/bing_bert/pytorch_pretrained_bert/tokenization.py0converts a sequence of ids in wordpiece tokens using the vocab.STRING
HIGH…g/BingBertGlue/pytorch_pretrained_bert/tokenization.py0converts a sequence of ids in wordpiece tokens using the vocab.STRING
HIGH…/BingBertSquad/pytorch_pretrained_bert/tokenization.py0converts a sequence of ids in wordpiece tokens using the vocab.STRING
HIGH…ning/bing_bert/pytorch_pretrained_bert/tokenization.py0instantiate a pretrainedbertmodel from a pre-trained model file. download and cache the pre-trained model file if neededSTRING
HIGH…g/BingBertGlue/pytorch_pretrained_bert/tokenization.py0instantiate a pretrainedbertmodel from a pre-trained model file. download and cache the pre-trained model file if neededSTRING
HIGH…/BingBertSquad/pytorch_pretrained_bert/tokenization.py0instantiate a pretrainedbertmodel from a pre-trained model file. download and cache the pre-trained model file if neededSTRING
HIGH…ning/bing_bert/pytorch_pretrained_bert/tokenization.py0runs basic tokenization (punctuation splitting, lower casing, etc.).STRING
HIGH…g/BingBertGlue/pytorch_pretrained_bert/tokenization.py0runs basic tokenization (punctuation splitting, lower casing, etc.).STRING
HIGH…/BingBertSquad/pytorch_pretrained_bert/tokenization.py0runs basic tokenization (punctuation splitting, lower casing, etc.).STRING
HIGH…ning/bing_bert/pytorch_pretrained_bert/tokenization.py0constructs a basictokenizer. args: do_lower_case: whether to lower case the input.STRING
HIGH…g/BingBertGlue/pytorch_pretrained_bert/tokenization.py0constructs a basictokenizer. args: do_lower_case: whether to lower case the input.STRING
189 more matches not shown…
Unused Imports366 hits · 366 pts
SeverityFileLineSnippetContext
LOW…aining/data_efficiency/vit_finetuning/main_imagenet.py28CODE
LOW…aining/data_efficiency/vit_finetuning/main_imagenet.py30CODE
LOW…aining/data_efficiency/vit_finetuning/main_imagenet.py30CODE
LOWtraining/data_efficiency/vit_finetuning/main_cifar.py16CODE
LOWtraining/data_efficiency/vit_finetuning/main_cifar.py20CODE
LOWtraining/data_efficiency/vit_finetuning/main_cifar.py21CODE
LOWtraining/data_efficiency/vit_finetuning/main_cifar.py23CODE
LOWtraining/data_efficiency/vit_finetuning/main_cifar.py29CODE
LOWtraining/data_efficiency/vit_finetuning/main_cifar.py30CODE
LOWtraining/data_efficiency/vit_finetuning/main_cifar.py33CODE
LOW…ining/data_efficiency/vit_finetuning/utils/__init__.py15CODE
LOW…ining/data_efficiency/vit_finetuning/utils/__init__.py15CODE
LOW…ining/data_efficiency/vit_finetuning/utils/__init__.py15CODE
LOW…ining/data_efficiency/vit_finetuning/utils/__init__.py15CODE
LOW…ining/data_efficiency/vit_finetuning/utils/__init__.py15CODE
LOW…ining/data_efficiency/vit_finetuning/utils/__init__.py15CODE
LOW…ining/data_efficiency/vit_finetuning/utils/__init__.py15CODE
LOW…ining/data_efficiency/vit_finetuning/utils/__init__.py15CODE
LOW…ining/data_efficiency/vit_finetuning/utils/__init__.py16CODE
LOWtraining/data_efficiency/vit_finetuning/utils/utils.py15CODE
LOWtraining/data_efficiency/vit_finetuning/utils/utils.py17CODE
LOWtraining/data_efficiency/vit_finetuning/utils/utils.py21CODE
LOWtraining/data_efficiency/vit_finetuning/utils/utils.py23CODE
LOWtraining/data_efficiency/vit_finetuning/utils/utils.py24CODE
LOWtraining/data_efficiency/vit_finetuning/utils/utils.py30CODE
LOWtraining/data_efficiency/vit_finetuning/utils/utils.py32CODE
LOW…ning/data_efficiency/vit_finetuning/models/__init__.py15CODE
LOW…ning/data_efficiency/vit_finetuning/models/__init__.py16CODE
LOW…ning/data_efficiency/vit_finetuning/models/__init__.py17CODE
LOWtraining/data_efficiency/vit_finetuning/models/vit.py26CODE
LOWtraining/data_efficiency/vit_finetuning/models/vit.py33CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py27CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py29CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py31CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py32CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py34CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py37CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py37CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py37CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py38CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py39CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py41CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py41CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py41CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py41CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py41CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py55CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py65CODE
LOW…g/data_efficiency/gpt_finetuning/run_clm_no_trainer.py31CODE
LOW…g/data_efficiency/gpt_finetuning/run_clm_no_trainer.py32CODE
LOW…g/data_efficiency/gpt_finetuning/run_clm_no_trainer.py34CODE
LOW…g/data_efficiency/gpt_finetuning/run_clm_no_trainer.py39CODE
LOWtraining/pipeline_parallelism/train.py12CODE
LOWtraining/bing_bert/deepspeed_train.py11CODE
LOWtraining/bing_bert/deepspeed_train.py18CODE
LOWtraining/bing_bert/deepspeed_train.py19CODE
LOWtraining/bing_bert/deepspeed_train.py19CODE
LOWtraining/bing_bert/deepspeed_train.py19CODE
LOWtraining/bing_bert/deepspeed_train.py21CODE
LOWtraining/bing_bert/deepspeed_train.py21CODE
306 more matches not shown…
Hyper-Verbose Identifiers164 hits · 176 pts
SeverityFileLineSnippetContext
LOWtraining/data_efficiency/vit_finetuning/models/vit.py662def _create_vision_transformer(variant, pretrained=False, **kwargs):CODE
LOW…g/tensor_parallel/hf_integration/train_bench_length.py67def smart_tokenizer_and_embedding_resize(CODE
LOW…g/tensor_parallel/hf_integration/train_bench_length.py193def make_supervised_data_module(tokenizer: transformers.PreTrainedTokenizer, data_args) -> Dict:CODE
LOWtraining/tensor_parallel/hf_integration/train.py65def smart_tokenizer_and_embedding_resize(CODE
LOWtraining/tensor_parallel/hf_integration/train.py188def make_supervised_data_module(tokenizer: transformers.PreTrainedTokenizer, data_args) -> Dict:CODE
LOW…aining/tensor_parallel/basic_example/autotp_example.py27 def get_data_parallel_world_size(self):CODE
LOW…aining/tensor_parallel/basic_example/autotp_example.py30 def get_model_parallel_world_size(self):CODE
LOW…tensor_parallel/basic_example/autotp_memory_compare.py28 def get_data_parallel_world_size(self):CODE
LOW…tensor_parallel/basic_example/autotp_memory_compare.py31 def get_model_parallel_world_size(self):CODE
LOW…sor_parallel/custom_patterns/autotp_custom_patterns.py32 def get_data_parallel_world_size(self):CODE
LOW…sor_parallel/custom_patterns/autotp_custom_patterns.py35 def get_model_parallel_world_size(self):CODE
LOWtraining/opsd/losses.py64def chunked_distillation_loss(CODE
LOWtraining/opsd/losses.py122def streamed_distillation_loss(CODE
LOW⚡training/opsd/tests/test_losses.py28def test_positive_when_different(loss_type):CODE
LOW⚡training/opsd/tests/test_losses.py38def test_chunking_equivalent_to_unchunked(loss_type):CODE
LOW⚡training/opsd/tests/test_losses.py48def test_mask_excludes_tokens():CODE
LOW⚡training/opsd/tests/test_losses.py58def test_gradient_flows_to_student():CODE
LOW⚡training/opsd/tests/test_losses.py69def test_gradient_does_not_flow_to_teacher_when_detached():CODE
LOW⚡training/opsd/tests/test_losses.py79def test_unknown_loss_type_raises():CODE
LOW⚡training/opsd/tests/test_losses.py87def test_shape_mismatch_raises():CODE
LOW⚡training/opsd/tests/test_losses.py95def test_mask_shape_mismatch_raises():CODE
LOW⚡training/opsd/tests/test_losses.py104def test_temperature_changes_loss_but_stays_finite(temperature):CODE
LOW⚡training/opsd/tests/test_losses.py123def test_all_zero_mask_returns_zero():CODE
LOW⚡training/opsd/tests/test_losses.py132def test_build_response_mask_basic():CODE
LOW⚡training/opsd/tests/test_losses.py140def test_build_response_mask_validates_shapes():CODE
LOWtraining/opsd/tests/test_teacher_caching.py17def test_round_trip_preserves_values_within_dtype():CODE
LOWtraining/opsd/tests/test_teacher_caching.py29def test_chunk_slicing_is_correct():CODE
LOWtraining/opsd/tests/test_teacher_caching.py39def test_invalid_chunk_bounds_raise():CODE
LOW⚡training/opsd/tests/test_teacher_caching.py54def test_rejects_gpu_resident_logits():CODE
LOW⚡training/opsd/tests/test_teacher_caching.py61def test_dtype_override_in_chunk_to_device():CODE
LOW⚡training/opsd/tests/test_teacher_caching.py68def test_free_releases_buffer():CODE
LOWtraining/opsd/tests/test_teacher_caching.py82def test_streamed_chunked_loss_matches_full_loss():CODE
LOWtraining/opsd/tests/test_data.py36def test_collator_left_pads_real_tokens_to_right_edge():CODE
LOW⚡training/opsd/tests/test_data.py55def test_collator_truncates_to_max_prompt_length():CODE
LOW⚡training/opsd/tests/test_data.py64def test_collator_falls_back_to_eos_when_no_pad_id():CODE
LOW⚡training/opsd/tests/test_data.py74def test_collator_raises_without_pad_and_eos():CODE
LOW⚡training/opsd/tests/test_data.py80def test_prompt_dataset_reads_jsonl(tmp_path):CODE
LOW⚡training/opsd/tests/test_data.py89def test_prompt_dataset_missing_field_raises(tmp_path):CODE
LOWtraining/bing_bert/deepspeed_train.py366def prepare_optimizer_parameters(args, model):CODE
LOWtraining/bing_bert/nvidia_bert_dataset_provider.py30def create_pretraining_dataset(input_file, max_predictions_per_seq,CODE
LOWtraining/bing_bert/run_glue_classifier_bert_base.py82class DataProcessor(object):CODE
LOWtraining/bing_bert/run_glue_classifier_bert_base.py411def convert_examples_to_features(examples, label_list, max_seq_length,CODE
LOWtraining/bing_bert/run_glue_classifier_bert_large.py109class DataProcessor(object):CODE
LOWtraining/bing_bert/run_glue_classifier_bert_large.py453def convert_examples_to_features(examples, label_list, max_seq_length,CODE
LOW…ch_pretrained_bert/convert_tf_checkpoint_to_pytorch.py31def convert_tf_checkpoint_to_pytorch(tf_checkpoint_path, bert_config_file,CODE
LOWtraining/bing_bert/turing/dataset.py344 def create_masked_lm_predictions(self, tokens):CODE
LOWtraining/BingBertGlue/nvidia_bert_dataset_provider.py30def create_pretraining_dataset(input_file, max_predictions_per_seq,CODE
LOWtraining/BingBertGlue/run_glue_classifier_bert_base.py81class DataProcessor(object):CODE
LOWtraining/BingBertGlue/run_glue_classifier_bert_base.py410def convert_examples_to_features(examples, label_list, max_seq_length,CODE
LOWtraining/BingBertGlue/run_glue_classifier_bert_large.py110class DataProcessor(object):CODE
LOWtraining/BingBertGlue/run_glue_classifier_bert_large.py454def convert_examples_to_features(examples, label_list, max_seq_length,CODE
LOW…ch_pretrained_bert/convert_tf_checkpoint_to_pytorch.py31def convert_tf_checkpoint_to_pytorch(tf_checkpoint_path, bert_config_file,CODE
LOWtraining/BingBertGlue/turing/dataset.py344 def create_masked_lm_predictions(self, tokens):CODE
LOWtraining/DeepSpeed-Domino/domino/language_model.py258 def state_dict_for_save_checkpoint(self, prefix='', keep_vars=False):CODE
LOWtraining/DeepSpeed-Domino/domino/language_model.py516 def state_dict_for_save_checkpoint(self, prefix='', keep_vars=False):CODE
LOWtraining/DeepSpeed-Domino/domino/gpt_model.py16def post_language_model_processing(lm_output, labels, logit_weights,CODE
LOWtraining/DeepSpeed-Domino/domino/gpt_model.py100 def state_dict_for_save_checkpoint(self, prefix='', keep_vars=False):CODE
LOWtraining/DeepSpeed-Domino/domino/training.py155def setup_model_and_optimizer(model_builder,CODE
LOWtraining/DeepSpeed-Domino/domino/training.py248def get_optimizer_param_scheduler(optimizer):CODE
LOWtraining/DeepSpeed-Domino/domino/training.py803def evaluate_and_print_results(prefix, forward_step_func,CODE
104 more matches not shown…
Deep Nesting129 hits · 128 pts
SeverityFileLineSnippetContext
LOWtraining/gan/gan_baseline_train.py14CODE
LOWtraining/gan/gan_deepspeed_train.py16CODE
LOW…aining/data_efficiency/vit_finetuning/main_imagenet.py109CODE
LOW…aining/data_efficiency/vit_finetuning/main_imagenet.py513CODE
LOWtraining/data_efficiency/vit_finetuning/utils/utils.py63CODE
LOWtraining/data_efficiency/vit_finetuning/utils/utils.py75CODE
LOWtraining/data_efficiency/vit_finetuning/models/vit.py525CODE
LOWtraining/data_efficiency/vit_finetuning/models/vit.py631CODE
LOWtraining/data_efficiency/vit_finetuning/models/vit.py530CODE
LOW…g/data_efficiency/gpt_finetuning/run_clm_no_trainer.py241CODE
LOW…g/data_efficiency/gpt_finetuning/run_clm_no_trainer.py497CODE
LOWtraining/tensor_parallel/hf_integration/utils.py39CODE
LOWtraining/bing_bert/deepspeed_train.py132CODE
LOWtraining/bing_bert/run_glue_classifier_bert_base.py546CODE
LOWtraining/bing_bert/run_glue_classifier_bert_base.py572CODE
LOWtraining/bing_bert/run_glue_classifier_bert_large.py587CODE
LOWtraining/bing_bert/run_glue_classifier_bert_large.py613CODE
LOW…ch_pretrained_bert/convert_tf_checkpoint_to_pytorch.py31CODE
LOW…ning/bing_bert/pytorch_pretrained_bert/tokenization.py297CODE
LOWtraining/bing_bert/nvidia/modelingpreln.py78CODE
LOWtraining/bing_bert/turing/sources.py167CODE
LOWtraining/bing_bert/turing/sources.py258CODE
LOWtraining/BingBertGlue/run_glue_classifier_bert_base.py545CODE
LOWtraining/BingBertGlue/run_glue_classifier_bert_base.py571CODE
LOWtraining/BingBertGlue/run_glue_classifier_bert_large.py588CODE
LOWtraining/BingBertGlue/run_glue_classifier_bert_large.py614CODE
LOW…ch_pretrained_bert/convert_tf_checkpoint_to_pytorch.py31CODE
LOW…g/BingBertGlue/pytorch_pretrained_bert/tokenization.py297CODE
LOWtraining/BingBertGlue/nvidia/modeling.py70CODE
LOWtraining/BingBertGlue/nvidia/modelingpreln.py78CODE
LOWtraining/BingBertGlue/turing/sources.py167CODE
LOWtraining/BingBertGlue/turing/sources.py258CODE
LOWtraining/DeepSpeed-Domino/domino/language_model.py276CODE
LOWtraining/DeepSpeed-Domino/domino/language_model.py544CODE
LOWtraining/DeepSpeed-Domino/domino/training.py476CODE
LOWtraining/DeepSpeed-Domino/domino/training.py739CODE
LOWtraining/imagenet/main.py141CODE
LOWtraining/imagenet/main.py453CODE
LOWtraining/deepspeed_finetune_demo/convert_ds_to_hf.py55CODE
LOWtraining/deepspeed_finetune_demo/finetune_llama.py331CODE
LOW…g/deepspeed_finetune_demo/evaluate/gsm8k/eval_gsm8k.py13CODE
LOW…ning/deepspeed_finetune_demo/evaluate/mmlu/gen_mmlu.py31CODE
LOW…ing/deepspeed_finetune_demo/evaluate/mmlu/eval_mmlu.py13CODE
LOWtraining/stable_diffusion/train_sd_distil_lora.py707CODE
LOW…ng/stable_diffusion/local_pipeline_stable_diffusion.py279CODE
LOW…ng/stable_diffusion/local_pipeline_stable_diffusion.py518CODE
LOW…aining/BingBertSquad/convert_bert_ckpt_to_deepspeed.py25CODE
LOW…aining/BingBertSquad/convert_bert_ckpt_to_deepspeed.py173CODE
LOW…aining/BingBertSquad/convert_bert_ckpt_to_deepspeed.py282CODE
LOWtraining/BingBertSquad/nvidia_run_squad_baseline.py112CODE
LOWtraining/BingBertSquad/nvidia_run_squad_baseline.py183CODE
LOWtraining/BingBertSquad/nvidia_run_squad_baseline.py420CODE
LOWtraining/BingBertSquad/nvidia_run_squad_baseline.py767CODE
LOWtraining/BingBertSquad/evaluate.py54CODE
LOWtraining/BingBertSquad/nvidia_run_squad_deepspeed.py116CODE
LOWtraining/BingBertSquad/nvidia_run_squad_deepspeed.py187CODE
LOWtraining/BingBertSquad/nvidia_run_squad_deepspeed.py424CODE
LOWtraining/BingBertSquad/nvidia_run_squad_deepspeed.py742CODE
LOW…/BingBertSquad/pytorch_pretrained_bert/tokenization.py297CODE
LOWtraining/BingBertSquad/turing/nvidia_modeling.py70CODE
69 more matches not shown…
Self-Referential Comments29 hits · 85 pts
SeverityFileLineSnippetContext
MEDIUMtraining/HelloDeepSpeed/train_bert_ds.py149 # Create the labels firstCOMMENT
MEDIUMtraining/HelloDeepSpeed/train_bert.py119 # Create the labels firstCOMMENT
MEDIUMtraining/DeepSpeed-Domino/pretrain_gpt3_13b.sh1# This file is adapted from pretrain_gpt.sh in Megatron-LMCOMMENT
MEDIUMtraining/DeepSpeed-Domino/pretrain_gpt.py2# This file is adapted from pretrain_gpt.py in Megatron-LMCOMMENT
MEDIUMtraining/DeepSpeed-Domino/pretrain_gpt3_6.7b.sh1# This file is adapted from pretrain_gpt.sh in Megatron-LMCOMMENT
MEDIUMtraining/DeepSpeed-Domino/domino/language_model.py1# This file is adapted from language_model.py in Megatron-LMCOMMENT
MEDIUMtraining/DeepSpeed-Domino/domino/gpt_model.py1# This file is adapted from gpt_model.py in Megatron-LMCOMMENT
MEDIUMtraining/DeepSpeed-Domino/domino/training.py2# This file is adapted from training.py in Megatron-LMCOMMENT
MEDIUMtraining/stable_diffusion/train_sd_distil_lora.py1235 # Create the pipeline using using the trained modules and save it.STRING
MEDIUMtraining/cifar/cifar10_deepspeed.py237 # Define the test dataloader.COMMENT
MEDIUMtraining/cifar/cifar10_deepspeed.py356 # Define the Classification Cross-Entropy loss function.COMMENT
MEDIUM…chmarks/inference/deepspeedometer/tests/test_config.py22 # Create a config that would generate 6 benchmark settingsCOMMENT
MEDIUM…ce/deepspeedometer/src/deepspeedometer/sample_input.py95 # Define the dataset and data collatorCOMMENT
MEDIUM…ce/deepspeedometer/src/deepspeedometer/sample_input.py99 # Define the training argumentsCOMMENT
MEDIUMbenchmarks/inference/mii/src/sample_input.py95 # Define the dataset and data collatorCOMMENT
MEDIUMbenchmarks/inference/mii/src/sample_input.py99 # Define the training argumentsCOMMENT
MEDIUMbenchmarks/inference/mii/src/utils.py150 # Create the parser, inheriting from the server and/or client parsersCOMMENT
MEDIUMbenchmarks/autosp/distributed_attention.py65 # Create the attention mask from padding mask + causal maskCOMMENT
MEDIUM…lications/DeepSpeed-VisualChat/utils/model/vis_proj.py28# The following code is adopted from QWen-ClipCOMMENT
MEDIUM…ons/DeepSpeed-VisualChat/utils/data/ocr_vqa_dataset.py1# This file is adapted from https://github.com/open-mmlab/Multimodal-GPTCOMMENT
MEDIUM…eepSpeed-VisualChat/utils/data/cc_sbu_align_dataset.py1# This file is adapted from https://github.com/open-mmlab/Multimodal-GPTCOMMENT
MEDIUM…ations/DeepSpeed-VisualChat/utils/data/dial_dataset.py1# This file is adapted from https://github.com/open-mmlab/Multimodal-GPTCOMMENT
MEDIUM…plications/DeepSpeed-VisualChat/utils/data/__init__.py1# This file is adapted from https://github.com/open-mmlab/Multimodal-GPTCOMMENT
MEDIUMapplications/DeepSpeed-VisualChat/utils/data/builder.py1# This file is adapted from https://github.com/open-mmlab/Multimodal-GPTCOMMENT
MEDIUM…cations/DeepSpeed-VisualChat/utils/data/vqa_dataset.py1# This file is adapted from https://github.com/open-mmlab/Multimodal-GPTCOMMENT
MEDIUM…ions/DeepSpeed-VisualChat/utils/data/aokvqa_dataset.py1# This file is adapted from https://github.com/open-mmlab/Multimodal-GPTCOMMENT
MEDIUM…tions/DeepSpeed-VisualChat/utils/data/llava_dataset.py1# This file is adapted from https://github.com/open-mmlab/Multimodal-GPTCOMMENT
MEDIUM…eepSpeed-VisualChat/utils/data/coco_caption_dataset.py1# This file is adapted from https://github.com/open-mmlab/Multimodal-GPTCOMMENT
MEDIUMapplications/DeepSpeed-Chat/dschat/utils/utils.py152# This function is a modified version of code available in the from_pretrained API of HuggingFace TransformersCOMMENT
Over-Commented Block82 hits · 82 pts
SeverityFileLineSnippetContext
LOW…atch_size_and_lr/variable_batch_size_and_lr_example.py141 "data_sampling": {COMMENT
LOW…g/data_efficiency/variable_batch_size_and_lr/README.md81 "dynamic_batching": {COMMENT
LOWtraining/data_efficiency/vit_finetuning/main_cifar.py1# Copyright 2021 Google LLCCOMMENT
LOW…ncy/vit_finetuning/bash_script/run_cifar_random_ltd.sh1#!/bin/bashCOMMENT
LOW…/vit_finetuning/bash_script/run_imagenet_random_ltd.sh21# --deepspeed \COMMENT
LOW…ining/data_efficiency/vit_finetuning/utils/__init__.py1# Copyright 2021 Google LLCCOMMENT
LOW…ining/data_efficiency/vit_finetuning/utils/get_data.py1# Copyright 2021 Google LLCCOMMENT
LOWtraining/data_efficiency/vit_finetuning/utils/utils.py1# Copyright 2021 Google LLCCOMMENT
LOW…ning/data_efficiency/vit_finetuning/models/__init__.py1# Copyright 2021 Google LLCCOMMENT
LOW…ining/data_efficiency/gpt_finetuning/learning_rates.py1# This code is adopted from https://github.com/deepspeedai/Megatron-DeepSpeed/blob/main/megatron/learning_rates.pyCOMMENT
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py1#!/usr/bin/env pythonCOMMENT
LOW…g/data_efficiency/gpt_finetuning/run_clm_no_trainer.py1#!/usr/bin/env pythonCOMMENT
LOW…ency/gpt_finetuning/bash_script/run_base_random_ltd.sh21# --random_ltd \COMMENT
LOW…g/tensor_parallel/hf_integration/train_bench_length.py1# Copyright 2023 Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen LiCOMMENT
LOWtraining/tensor_parallel/hf_integration/train.py1# Copyright 2023 Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen LiCOMMENT
LOWtraining/bing_bert/run_glue_classifier_bert_base.py1# coding=utf-8COMMENT
LOWtraining/bing_bert/run_glue_classifier_bert_base.py441 # tokens: [CLS] the dog is hairy . [SEP]COMMENT
LOWtraining/bing_bert/NOTICE.txt21Open Source License/Copyright Notice.COMMENT
LOWtraining/bing_bert/run_glue_classifier_bert_large.py1# coding=utf-8COMMENT
LOWtraining/bing_bert/run_glue_classifier_bert_large.py481 # type_ids: 0 0 0 0 0 0 0 0 1 1 1 1 1 1COMMENT
LOW…ch_pretrained_bert/convert_tf_checkpoint_to_pytorch.py1# coding=utf-8COMMENT
LOW…ning/bing_bert/pytorch_pretrained_bert/optimization.py1# coding=utf-8COMMENT
LOW…ning/bing_bert/pytorch_pretrained_bert/tokenization.py1# coding=utf-8COMMENT
LOWtraining/bing_bert/pytorch_pretrained_bert/modeling.py1# coding=utf-8COMMENT
LOWtraining/bing_bert/nvidia/modelingpreln.py1# DeepSpeed note, code taken from commit 3d59216cec89a363649b4fe3d15295ba936ced0fCOMMENT
LOWtraining/bing_bert/nvidia/modelingpreln.py601 if output_all_encoded_layers:COMMENT
LOWtraining/BingBertGlue/run_glue_classifier_bert_base.py1# coding=utf-8COMMENT
LOWtraining/BingBertGlue/run_glue_classifier_bert_base.py441 # type_ids: 0 0 0 0 0 0 0COMMENT
LOWtraining/BingBertGlue/run_glue_classifier_bert_large.py1# coding=utf-8COMMENT
LOWtraining/BingBertGlue/run_glue_classifier_bert_large.py481 # tokens: [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]COMMENT
LOW…ch_pretrained_bert/convert_tf_checkpoint_to_pytorch.py1# coding=utf-8COMMENT
LOW…g/BingBertGlue/pytorch_pretrained_bert/optimization.py1# coding=utf-8COMMENT
LOW…g/BingBertGlue/pytorch_pretrained_bert/tokenization.py1# coding=utf-8COMMENT
LOW…ining/BingBertGlue/pytorch_pretrained_bert/modeling.py1# coding=utf-8COMMENT
LOWtraining/BingBertGlue/nvidia/modeling.py1# DeepSpeed note, code taken from commit 3d59216cec89a363649b4fe3d15295ba936ced0fCOMMENT
LOWtraining/BingBertGlue/nvidia/modeling.py581#class BertEncoder(nn.Module):COMMENT
LOWtraining/BingBertGlue/nvidia/modelingpreln.py1# DeepSpeed note, code taken from commit 3d59216cec89a363649b4fe3d15295ba936ced0fCOMMENT
LOWtraining/imagenet/extract_ILSVRC.sh1#!/bin/bashCOMMENT
LOWtraining/imagenet/extract_ILSVRC.sh21# │ ├── ILSVRC2012_val_00002138.JPEGCOMMENT
LOWtraining/imagenet/extract_ILSVRC.sh41# 2. extract and copy contents of .tar file into directoryCOMMENT
LOWtraining/imagenet/extract_ILSVRC.sh61# get script from soumith and run; this script creates all class directories and moves images into corresponding directoCOMMENT
LOWtraining/deepspeed_finetune_demo/run_and_evaluate.sh1#!/bin/bashCOMMENT
LOW…une_demo/run_autoep_affine_ir_checkpoint_experiment.sh1#!/usr/bin/env bashCOMMENT
LOWtraining/stable_diffusion/train_sd_distil_lora.py1#!/usr/bin/env pythonCOMMENT
LOW…ng/stable_diffusion/local_pipeline_stable_diffusion.py1# Copyright 2023 The HuggingFace Team. All rights reserved.COMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_baseline.py1# coding=utf-8COMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_baseline.py341def _improve_answer_span(doc_tokens, input_start, input_end, tokenizer,COMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_baseline.py381 # Because of the sliding window approach taken to scoring documents, a singleCOMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_baseline.py561 # (whitespace tokenized) tokens and our WordPiece tokenized tokens. SoCOMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_deepspeed.py1# coding=utf-8COMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_deepspeed.py341COMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_deepspeed.py381COMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_deepspeed.py561def get_final_text(pred_text, orig_text, do_lower_case, verbose_logging=False):COMMENT
LOW…/BingBertSquad/pytorch_pretrained_bert/optimization.py1# coding=utf-8COMMENT
LOW…/BingBertSquad/pytorch_pretrained_bert/tokenization.py1# coding=utf-8COMMENT
LOW…ning/BingBertSquad/pytorch_pretrained_bert/modeling.py1# coding=utf-8COMMENT
LOWtraining/BingBertSquad/turing/nvidia_modeling.py1# DeepSpeed note, code taken from commit 3d59216cec89a363649b4fe3d15295ba936ced0fCOMMENT
LOWtraining/BingBertSquad/turing/nvidia_modeling.py581COMMENT
LOWtraining/BingBertSquad/turing/nvidia_modelingpreln.py1# DeepSpeed note, code taken from commit 3d59216cec89a363649b4fe3d15295ba936ced0fCOMMENT
LOWtraining/BingBertSquad/turing/nvidia_modelingpreln.py601COMMENT
22 more matches not shown…
Empty Error Handlers13 hits · 65 pts
SeverityFileLineSnippetContext
HIGHtraining/gan/utils.py46except OSError:CODE
HIGHtraining/bing_bert/turing/utils.py166except (AttributeError, ValueError):CODE
HIGHtraining/BingBertGlue/turing/utils.py166except (AttributeError, ValueError):CODE
HIGH…me/model_checkpoint/torch/serialization_orig_v2.6.0.py1527except KeyError:CODE
HIGH…me/model_checkpoint/torch/serialization_orig_v2.6.0.py1950except KeyError:CODE
HIGH…me/model_checkpoint/torch/serialization_fast_v2.6.0.py1531except KeyError:CODE
HIGH…me/model_checkpoint/torch/serialization_fast_v2.6.0.py1954except KeyError:CODE
HIGH…eepspeedometer/src/deepspeedometer/benchmark_runner.py153except queue.Empty:CODE
HIGHbenchmarks/autosp/correctness/correctness_run.py123except FileNotFoundError:CODE
HIGH…pression/reasoning_aware_compression/rac/sequential.py116except _StopForward:CODE
HIGH…ed-Chat/training/step2_reward_model_finetuning/main.py334except:CODE
HIGH…s/DeepSpeed-Chat/training/step2_dpo_finetuning/main.py400except:CODE
HIGH…peed-Chat/training/step1_supervised_finetuning/main.py309except:CODE
AI Structural Patterns54 hits · 54 pts
SeverityFileLineSnippetContext
LOWtraining/data_efficiency/vit_finetuning/models/vit.py229CODE
LOWtraining/data_efficiency/vit_finetuning/models/vit.py252CODE
LOWtraining/data_efficiency/vit_finetuning/models/vit.py282CODE
LOWtraining/data_efficiency/vit_finetuning/models/vit.py328CODE
LOWtraining/opsd/teacher.py90CODE
LOW…ning/bing_bert/pytorch_pretrained_bert/optimization.py108CODE
LOWtraining/bing_bert/pytorch_pretrained_bert/modeling.py80CODE
LOWtraining/bing_bert/pytorch_pretrained_bert/modeling.py162CODE
LOWtraining/bing_bert/nvidia/modelingpreln.py228CODE
LOWtraining/bing_bert/nvidia/modelingpreln.py310CODE
LOWtraining/HelloDeepSpeed/train_bert_ds.py595CODE
LOWtraining/HelloDeepSpeed/train_bert.py554CODE
LOW…g/BingBertGlue/pytorch_pretrained_bert/optimization.py108CODE
LOW…ining/BingBertGlue/pytorch_pretrained_bert/modeling.py80CODE
LOW…ining/BingBertGlue/pytorch_pretrained_bert/modeling.py162CODE
LOWtraining/BingBertGlue/nvidia/modeling.py217CODE
LOWtraining/BingBertGlue/nvidia/modeling.py298CODE
LOWtraining/BingBertGlue/nvidia/modelingpreln.py228CODE
LOWtraining/BingBertGlue/nvidia/modelingpreln.py310CODE
LOWtraining/DeepSpeed-Domino/domino/language_model.py433CODE
LOWtraining/stable_diffusion/train_sd_distil_lora.py545CODE
LOW…ng/stable_diffusion/local_pipeline_stable_diffusion.py518CODE
LOW…/BingBertSquad/pytorch_pretrained_bert/optimization.py80CODE
LOW…ning/BingBertSquad/pytorch_pretrained_bert/modeling.py78CODE
LOW…ning/BingBertSquad/pytorch_pretrained_bert/modeling.py158CODE
LOWtraining/BingBertSquad/turing/nvidia_modeling.py217CODE
LOWtraining/BingBertSquad/turing/nvidia_modeling.py299CODE
LOWtraining/BingBertSquad/turing/nvidia_modelingpreln.py220CODE
LOWtraining/BingBertSquad/turing/nvidia_modelingpreln.py304CODE
LOW…eedometer/src/deepspeedometer/clients/openai_client.py54CODE
LOW…speedometer/src/deepspeedometer/clients/vllm_client.py85CODE
LOWbenchmarks/inference/mii/src/client.py115CODE
LOWbenchmarks/inference/mii/src/client.py184CODE
LOWbenchmarks/autosp/distributed_attention.py93CODE
LOWbenchmarks/autosp/ring_attention.py391CODE
LOWbenchmarks/autosp/ring_attention.py417CODE
LOWbenchmarks/autosp/ring_attention.py444CODE
LOW…pression/reasoning_aware_compression/rac/sequential.py129CODE
LOW…ression/reasoning_aware_compression/rac/calibration.py182CODE
LOW…ions/DeepSpeed-VisualChat/utils/model/modeling_dsvl.py277CODE
LOW…t/utils/model/third_party_model/qwen_clip/qwen_clip.py243CODE
LOW…odel/third_party_model/hf_model/configuration_llama.py106CODE
LOW…ils/model/third_party_model/hf_model/modeling_llama.py683CODE
LOW…ils/model/third_party_model/hf_model/modeling_llama.py843CODE
LOW…ils/model/third_party_model/hf_model/modeling_llama.py1009CODE
LOW…ations/DeepSpeed-VisualChat/utils/data/dial_dataset.py78CODE
LOW…cations/DeepSpeed-VisualChat/utils/data/vqa_dataset.py225CODE
LOW…at/training/step1_supervised_finetuning/prompt_eval.py102CODE
LOW…at/training/step1_supervised_finetuning/prompt_eval.py122CODE
LOWapplications/DeepSpeed-Chat/inference/chatbot.py70CODE
LOWapplications/DeepSpeed-Chat/dschat/utils/ds_utils.py13CODE
LOW…tions/DeepSpeed-Chat/dschat/utils/model/model_utils.py32CODE
LOW…ions/DeepSpeed-Chat/dschat/utils/model/reward_model.py127CODE
LOW…ce/stable-diffusion/local_pipeline_stable_diffusion.py519CODE
Modern Structural Boilerplate38 hits · 38 pts
SeverityFileLineSnippetContext
LOW…ining/data_efficiency/vit_finetuning/utils/__init__.py18__all__ = [ "get_dataset", "ImageMemFolder", "AverageMeter", "ProgressMeter", "accuracy", "get_optimizer", "get_scheduleCODE
LOWtraining/data_efficiency/vit_finetuning/models/vit.py1097__all__ = [ 'vits16r224', 'vits16r384', 'lvits16r224','lvits16r384', 'Block']CODE
LOWtraining/data_efficiency/gpt_finetuning/analyze_data.py68logger = logging.getLogger(__name__)CODE
LOW…g/data_efficiency/gpt_finetuning/run_clm_no_trainer.py65logger = logging.getLogger(__name__)CODE
LOWtraining/bing_bert/run_glue_classifier_bert_base.py48logger = logging.getLogger(__name__)CODE
LOWtraining/bing_bert/run_glue_classifier_bert_large.py48logger = logging.getLogger(__name__)CODE
LOW…ning/bing_bert/pytorch_pretrained_bert/tokenization.py28logger = logging.getLogger(__name__)CODE
LOW…aining/bing_bert/pytorch_pretrained_bert/file_utils.py24logger = logging.getLogger(__name__) # pylint: disable=invalid-nameCODE
LOWtraining/bing_bert/pytorch_pretrained_bert/modeling.py37logger = logging.getLogger(__name__)CODE
LOWtraining/bing_bert/nvidia/modelingpreln.py47logger = logging.getLogger(__name__)CODE
LOWtraining/bing_bert/turing/logger.py8logger = logging.getLogger(__name__)CODE
LOWtraining/bing_bert/turing/file_utils.py39logger = logging.getLogger(__name__) # pylint: disable=invalid-nameCODE
LOWtraining/BingBertGlue/run_glue_classifier_bert_base.py47logger = logging.getLogger(__name__)CODE
LOWtraining/BingBertGlue/run_glue_classifier_bert_large.py49logger = logging.getLogger(__name__)CODE
LOW…g/BingBertGlue/pytorch_pretrained_bert/tokenization.py28logger = logging.getLogger(__name__)CODE
LOW…ing/BingBertGlue/pytorch_pretrained_bert/file_utils.py24logger = logging.getLogger(__name__) # pylint: disable=invalid-nameCODE
LOW…ining/BingBertGlue/pytorch_pretrained_bert/modeling.py37logger = logging.getLogger(__name__)CODE
LOWtraining/BingBertGlue/nvidia/modeling.py47logger = logging.getLogger(__name__)CODE
LOWtraining/BingBertGlue/nvidia/modelingpreln.py47logger = logging.getLogger(__name__)CODE
LOWtraining/BingBertGlue/turing/logger.py8logger = logging.getLogger(__name__)CODE
LOWtraining/BingBertGlue/turing/file_utils.py39logger = logging.getLogger(__name__) # pylint: disable=invalid-nameCODE
LOW…aining/BingBertSquad/convert_bert_ckpt_to_deepspeed.py15logger = logging.getLogger(__name__)CODE
LOWtraining/BingBertSquad/nvidia_run_squad_baseline.py51logger = logging.getLogger(__name__)CODE
LOWtraining/BingBertSquad/nvidia_run_squad_deepspeed.py54logger = logging.getLogger(__name__)CODE
LOW…/BingBertSquad/pytorch_pretrained_bert/tokenization.py28logger = logging.getLogger(__name__)CODE
LOW…ng/BingBertSquad/pytorch_pretrained_bert/file_utils.py24logger = logging.getLogger(__name__) # pylint: disable=invalid-nameCODE
LOW…ning/BingBertSquad/pytorch_pretrained_bert/modeling.py38logger = logging.getLogger(__name__)CODE
LOWtraining/BingBertSquad/turing/nvidia_modeling.py47logger = logging.getLogger(__name__)CODE
LOWtraining/BingBertSquad/turing/file_utils.py39logger = logging.getLogger(__name__) # pylint: disable=invalid-nameCODE
LOWtraining/BingBertSquad/turing/nvidia_modelingpreln.py47logger = logging.getLogger(__name__)CODE
LOWtraining/expert_parallel/train.py42logger = logging.getLogger(__name__)CODE
LOWtraining/expert_parallel/utils/prepare_init_weights.py25logger = logging.getLogger(__name__)CODE
LOW…me/model_checkpoint/torch/serialization_orig_v2.6.0.py41__all__ = [CODE
LOW…me/model_checkpoint/torch/serialization_fast_v2.6.0.py41__all__ = [CODE
LOWbenchmarks/autosp/ring_attention.py14__all__ = ["update_out_and_lse", "RingComm", "get_default_args"]CODE
LOWcompression/reasoning_aware_compression/rac/__init__.py24__all__ = [CODE
LOW…at/training/step1_supervised_finetuning/prompt_eval.py16logger = logging.getLogger(__name__)CODE
LOW…eneration/run-generation-script/test-run-generation.py51logger = logging.getLogger(__name__)CODE
Excessive Try-Catch Wrapping27 hits · 34 pts
SeverityFileLineSnippetContext
LOWtraining/DeepSpeed-SuperOffload/finetune_zero3.py220 except Exception as e:CODE
LOWtraining/DeepSpeed-SuperOffload/finetune_zero3.py370 except Exception as e:CODE
LOWtraining/DeepSpeed-SuperOffload/finetune_zero3.py377 except Exception as e:CODE
MEDIUMtraining/bf16_master_weight/gather_memory.py65 print(f"Error: Log directory '{log_dir}' does not exist")CODE
LOWtraining/expert_parallel/init_weights.py95 except Exception:CODE
MEDIUMtraining/expert_parallel/init_weights.py92def _transformers_version() -> str:CODE
LOWtraining/expert_parallel/data_utils.py142 except Exception as e:CODE
MEDIUMtraining/expert_parallel/data_utils.py144 print(f"Error loading {dataset_name!r}: {e}; falling back to wikitext.")CODE
LOWtraining/expert_parallel/train.py267 except Exception as exc:CODE
LOWtraining/expert_parallel/train.py357 except Exception as exc:CODE
LOWtraining/expert_parallel/train.py482 except Exception as exc:CODE
MEDIUM…me/model_checkpoint/torch/serialization_orig_v2.6.0.py1531def _check_container_source(container_type, source_file, original_source):CODE
LOW…me/model_checkpoint/torch/serialization_orig_v2.6.0.py1534 except Exception: # saving the source is optional, so we can ignore any errorsCODE
LOW…me/model_checkpoint/torch/serialization_orig_v2.6.0.py513 except Exception as e:STRING
MEDIUM…me/model_checkpoint/torch/serialization_fast_v2.6.0.py1535def _check_container_source(container_type, source_file, original_source):CODE
LOW…me/model_checkpoint/torch/serialization_fast_v2.6.0.py1538 except Exception: # saving the source is optional, so we can ignore any errorsCODE
LOW…me/model_checkpoint/torch/serialization_fast_v2.6.0.py513 except Exception as e:STRING
LOWbenchmarks/pin_memory/model_tensor_offload/bench.py141 except Exception:CODE
MEDIUMbenchmarks/pin_memory/model_tensor_offload/bench.py138def _mem(fn):CODE
LOWbenchmarks/pin_memory/cpu_pin/bench.py42 except Exception as exc:CODE
LOWbenchmarks/pin_memory/activation_offload/bench.py102 except Exception:CODE
MEDIUMbenchmarks/pin_memory/activation_offload/bench.py99def _peak():CODE
MEDIUMbenchmarks/inference/collect_results.py121 print(f"Error found in {file_path}, collecting error info...")CODE
LOW…dometer/src/deepspeedometer/clients/azure_ml_client.py72 except Exception as e:CODE
LOWbenchmarks/inference/mii/src/client.py254 except Exception as e:CODE
LOWbenchmarks/autosp/correctness/correctness_run.py32 except Exception:CODE
MEDIUMbenchmarks/autosp/correctness/correctness_run.py26def get_host_ip():CODE
Decorative Section Separators9 hits · 30 pts
SeverityFileLineSnippetContext
MEDIUMtraining/opsd/trainer.py77 # ------------------------------------------------------------------COMMENT
MEDIUMtraining/opsd/trainer.py79 # ------------------------------------------------------------------COMMENT
MEDIUMtraining/opsd/trainer.py97 # ------------------------------------------------------------------COMMENT
MEDIUMtraining/opsd/trainer.py99 # ------------------------------------------------------------------COMMENT
MEDIUMtraining/opsd/trainer.py187 # ------------------------------------------------------------------COMMENT
MEDIUMtraining/opsd/trainer.py189 # ------------------------------------------------------------------COMMENT
MEDIUMtraining/cifar/cifar10_tutorial.py298# ----------------COMMENT
MEDIUM⚡training/cifar/cifar10_tutorial.py342# -------------------------COMMENT
MEDIUM⚡training/cifar/cifar10_tutorial.py347# -------------------COMMENT
Docstring Block Structure4 hits · 20 pts
SeverityFileLineSnippetContext
HIGH…ng/stable_diffusion/local_pipeline_stable_diffusion.py538 Function invoked when calling the pipeline for generation. Args: prompt (`str` or `List[stSTRING
HIGH…me/model_checkpoint/torch/serialization_orig_v2.6.0.py673 Restores `storage` using a deserializer function registered for the `location`. This function looks in the regSTRING
HIGH…me/model_checkpoint/torch/serialization_fast_v2.6.0.py673 Restores `storage` using a deserializer function registered for the `location`. This function looks in the regSTRING
HIGH…ce/stable-diffusion/local_pipeline_stable_diffusion.py540 Function invoked when calling the pipeline for generation. Args: prompt (`str` or `List[stSTRING
AI Slop Vocabulary9 hits · 20 pts
SeverityFileLineSnippetContext
LOW…ining/data_efficiency/gpt_finetuning/learning_rates.py66 # If the learning rate is constant, just return the initial value.COMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_baseline.py365 # "Japanese", we just use "Japanese" as the annotation. This is fairly rareCOMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_baseline.py523 # just create a nonce prediction in this case to avoid failure.COMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_baseline.py583 # can fail in certain cases in which case we just return `orig_text`.COMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_deepspeed.py369 # "Japanese", we just use "Japanese" as the annotation. This is fairly rareCOMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_deepspeed.py527 # just create a nonce prediction in this case to avoid failure.COMMENT
LOWtraining/BingBertSquad/nvidia_run_squad_deepspeed.py587 # can fail in certain cases in which case we just return `orig_text`.COMMENT
MEDIUM⚡training/cifar/cifar10_tutorial.py350# - `Train a state-of-the-art ResNet network on imagenet`_COMMENT
MEDIUM⚡training/cifar/cifar10_tutorial.py358# .. _Train a state-of-the-art ResNet network on imagenet: https://github.com/pytorch/examples/tree/master/imagenetCOMMENT
Cross-Language Confusion4 hits · 14 pts
SeverityFileLineSnippetContext
HIGH⚡training/opsd/losses.py12 * ``forward_kl`` = D_KL(teacher || student) — mode-covering for studentSTRING
HIGH⚡training/opsd/losses.py13 * ``reverse_kl`` = D_KL(student || teacher) — mode-seeking for studentSTRING
HIGH⚡training/opsd/losses.py14 * ``jsd`` = 0.5 * D_KL(P || M) + 0.5 * D_KL(Q || M), M = (P+Q)/2STRING
HIGHtraining/opsd/tests/test_losses.py8 cd examples/opsd && python -m pytest tests/test_losses.py -vSTRING
Redundant / Tautological Comments7 hits · 10 pts
SeverityFileLineSnippetContext
LOW…g/tensor_parallel/hf_integration/train_bench_length.py243 # Print message except when distributed but not rank 0COMMENT
LOWtraining/tensor_parallel/hf_integration/train.py238 # Print message except when distributed but not rank 0COMMENT
LOWtraining/DeepSpeed-Domino/domino/training.py660 # Set grad to zero.COMMENT
LOW…mpression/bash_script/collect_traces_zero_inference.sh25# Set NVME_OFFLOAD_DIR to spill parameters to NVMe instead of CPU memory.COMMENT
LOW…ions/DeepSpeed-Chat/dschat/utils/model/reward_model.py99 # Check if there is any padding otherwise take length of sequenceCOMMENT
LOW…cations/DeepSpeed-Chat/dschat/utils/data/data_utils.py364 if sft_train_datasets: # Check if sft_train_datasets is not emptyCODE
LOW…cations/DeepSpeed-Chat/dschat/utils/data/data_utils.py370 if sft_eval_datasets: # Check if sft_eval_datasets is not emptyCODE
AI Response Leakage1 hit · 8 pts
SeverityFileLineSnippetContext
HIGH…/huggingface/stable-diffusion/test-stable-diffusion.py8# In this example the SD inference pipeline is optimized based on recommendations in the research paperCOMMENT
Structural Annotation Overuse3 hits · 4 pts
SeverityFileLineSnippetContext
LOWapplications/DeepSpeed-Chat/training/README.md7### Step 1: Supervised FinetuningCOMMENT
LOWapplications/DeepSpeed-Chat/training/README.md16### Step 2: Reward Model FinetuningCOMMENT
LOWapplications/DeepSpeed-Chat/training/README.md34### Step 3: RLHF finetuningCOMMENT
Example Usage Blocks2 hits · 3 pts
SeverityFileLineSnippetContext
LOWtraining/bf16_master_weight/run_comparison.sh4# Usage:COMMENT
LOWbenchmarks/autosp/correctness/correctness.sh10# Usage:COMMENT
Modern AI Meta-Vocabulary1 hit · 3 pts
SeverityFileLineSnippetContext
MEDIUM…re_compression/bash_script/run_calibration_ablation.sh11# rac those prompts plus the dense model's own chain-of-thoughtCOMMENT
Slop Phrases2 hits · 2 pts
SeverityFileLineSnippetContext
LOWtraining/stable_diffusion/train_sd_distil_lora.py850 # make sure to pop weight so that corresponding model is not saved againSTRING
LOWbenchmarks/inference/mii/src/utils.py200 # If we are not running the benchmark, we need to make sure to only have oneCOMMENT
Verbosity Indicators1 hit · 2 pts
SeverityFileLineSnippetContext
LOWtraining/cifar/cifar10_tutorial.py213# But we need to check if the network has learnt anything at all.COMMENT