pytorch/rl

23.2

Adjusted Score

23.2

Raw Score

100%

Time Factor

2026-08-29

Last Push

3.5K

Stars

Python

Language

532.7K

Lines of Code

1.3K

Files

9.8K

Pattern Hits

2026-08-29

Scan Date

0.22

HC Hit Rate

What These Metrics Mean

Adjusted Score: Primary synthetic code indicator. Raw score normalised per 1,000 lines of code and multiplied by the temporal discount factor. This is the definitive comparative metric — use it to rank repositories by AI authorship density.
Raw Score: The unmodified sum of all severity-weighted, context-multiplied pattern match scores before temporal discounting. Reflects the absolute signal strength independent of when the repository was last active.
Time Factor: The temporal discount multiplier (0–100%) applied to the raw score. Repositories last updated before ChatGPT's launch (Nov 2022) receive a 5% factor. Full signal is only assigned to repositories active in the post-adoption era (Jan 2024+).
Pattern Hits: Total count of individual pattern matches across all files and categories. A high hit count with a low score may indicate a very large codebase with isolated AI snippets; a low count with a high score indicates dense, concentrated AI signatures.
HC Hit Rate: High+Critical pattern hits per file, averaged across the repository. This orthogonal signal catches repositories where a few files are densely packed with high-severity AI tells — a strong indicator even when the normalised score appears moderate due to codebase size.
Lines of Code / Files: Total lines and files analysed. The scanner examines 94 file extensions. These denominators are used to normalise the score, enabling fair comparison between repositories of vastly different sizes.

Score History

Longitudinal tracking requires multiple scan runs. Once this repository is re-scanned after new commits land, this chart will visualise how the synthetic code signal evolves over time — enabling you to detect whether AI authorship is growing, stabilising, or being actively corrected by human engineers.

No multi-scan history yet — run the scanner again to build trend data.

Severity Breakdown

Classifies detected patterns by their diagnostic confidence and structural impact. CRITICAL patterns (coefficient 10) represent definitive synthetic signatures — hallucinated imports, explicit LLM attribution metadata — virtually never produced by human authors. HIGH (5) indicates strong structural tells such as cross-file repetition or cross-linguistic idioms. MEDIUM (2) covers recognisable conversational padding and AI-specific vocabulary. LOW (1) captures subtle indicators like tautological comments and generic boilerplate that require density to carry independent signal.

CRITICAL 3HIGH 280MEDIUM 1272LOW 8281

Directory Score Breakdown

This horizontal bar chart decomposes the repository's raw synthetic code score by top-level directory, allowing you to pinpoint precisely which modules or components carry the highest AI authorship density. Directories with disproportionately high scores relative to their size warrant targeted manual review: concentrated AI signatures often trace back to mass-generated configuration layers, auto-ported test suites, LLM-scaffolded boilerplate classes, or entire subsystems authored under heavy copilot assistance. Use this view to prioritise your human code-review effort.

Pattern Findings

The scanner identified 9836 distinct pattern matches across 25 syntactic categories. Each entry below represents a discrete location in the source code where the engine recorded a statistically significant AI authorship indicator. 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 diagnostic confidence level (CRITICAL / HIGH / MEDIUM / LOW). The Context column identifies whether the match occurred inside executable code, an inline comment, or a string literal — comment-context matches receive a ×1.5 weight because LLMs systematically over-annotate. The ⚡ bolt icon marks clustered matches: three or more patterns within a 10-line window, each receiving an additional ×1.5 density multiplier as dense clusters constitute far stronger evidence of synthetic authorship than isolated hits.

Hyper-Verbose Identifiers3705 hits · 3089 pts

Severity	File	Line	Snippet	Context
LOW	setup.py	145	def _check_and_clean_stale_builds():	CODE
LOW	AGENTS.md	120	def test_something_cuda_specific():	CODE
LOW	CLAUDE.md	120	def test_something_cuda_specific():	CODE
LOW	test/test_configs.py	301	def test_wrapped_class_kwargs_have_config_fields(self, config_name):	CODE
LOW	test/test_configs.py	388	def test_vmas_env_config_instantiation(self):	CODE
LOW	test/test_configs.py	424	def test_async_env_shared_exchange_config(self):	CODE
LOW	test/test_configs.py	447	def test_batched_env_config_omegaconf_schema(self):	CODE
LOW	test/test_configs.py	466	def test_batched_env_config_validation(self, field, value):	CODE
LOW	test/test_configs.py	497	def test_round_robin_writer_config(self):	CODE
LOW	test/test_configs.py	521	def test_random_sampler_config(self):	CODE
LOW	test/test_configs.py	556	def test_tensor_storage_config(self):	CODE
LOW	test/test_configs.py	577	def test_tensordict_replay_buffer_config(self):	CODE
LOW	test/test_configs.py	618	def test_replay_buffer_config(self):	CODE
LOW	test/test_configs.py	654	def test_tensordict_replay_buffer_config_optional_fields(self):	CODE
LOW	test/test_configs.py	670	def test_writer_ensemble_config(self):	CODE
LOW	test/test_configs.py	691	def test_tensor_dict_max_value_writer_config(self):	CODE
LOW	test/test_configs.py	708	def test_tensor_dict_round_robin_writer_config(self):	CODE
LOW	test/test_configs.py	727	def test_immutable_dataset_writer_config(self):	CODE
LOW	test/test_configs.py	741	def test_sampler_ensemble_config(self):	CODE
LOW	test/test_configs.py	762	def test_prioritized_slice_sampler_config(self):	CODE
LOW	test/test_configs.py	816	def test_slice_sampler_without_replacement_config(self):	CODE
LOW	test/test_configs.py	853	def test_slice_sampler_config(self):	CODE
LOW	test/test_configs.py	887	def test_prioritized_sampler_config(self):	CODE
LOW	test/test_configs.py	912	def test_sampler_without_replacement_config(self):	CODE
LOW	test/test_configs.py	930	def test_storage_ensemble_writer_config(self):	CODE
LOW	test/test_configs.py	950	def test_lazy_stack_storage_config(self):	CODE
LOW	test/test_configs.py	967	def test_storage_ensemble_config(self):	CODE
LOW	test/test_configs.py	992	def test_lazy_memmap_storage_config(self):	CODE
LOW	test/test_configs.py	1013	def test_lazy_tensor_storage_config(self):	CODE
LOW	test/test_configs.py	1042	def test_complex_replay_buffer_configuration(self):	CODE
LOW	test/test_configs.py	1194	def test_tensor_dict_module_config(self):	CODE
LOW	test/test_configs.py	1216	def test_tanh_normal_model_config(self):	CODE
LOW	test/test_configs.py	1250	def test_tanh_normal_model_config_defaults(self):	CODE
LOW	test/test_configs.py	1271	def test_tensordict_sequential_config(self):	CODE
LOW	test/test_configs.py	1401	def test_qmixer_network_config(self):	CODE
LOW	test/test_configs.py	1415	def test_additive_gaussian_module_config(self):	CODE
LOW	test/test_configs.py	1455	def test_generic_collector_backend_fields(self):	CODE
LOW	test/test_configs.py	1556	def test_collector_auto_configures_exploration_modules(self, factory, collector):	CODE
LOW	test/test_configs.py	1646	def test_gae_config_value_chunk_dim(self):	CODE
LOW	test/test_configs.py	1811	def test_reinforce_loss_config(self):	CODE
LOW	test/test_configs.py	1895	def test_wandb_logger_config_instantiation(self, monkeypatch):	CODE
LOW	test/test_configs.py	1922	def test_trackio_logger_config(self):	CODE
LOW	test/test_configs.py	1941	def test_trackio_logger_config_instantiation(self, monkeypatch):	CODE
LOW	test/test_configs.py	1993	def test_checkpoint_config_parity(self, config_name, field_name):	CODE
LOW	test/test_configs.py	1997	def test_nested_key_normalization_for_hydra_lists(self):	CODE
LOW	test/test_configs.py	2083	def test_reinforce_trainer_config(self):	CODE
LOW	test/test_configs.py	2113	def test_ppo_trainer_config_optional_fields(self):	CODE
LOW	test/test_configs.py	2276	def test_individual_hook_configs(self, config_cls, kwargs, hook_cls):	CODE
LOW	test/test_configs.py	4077	def test_weight_updater_config(self):	CODE
LOW	test/test_configs.py	4087	def test_vanilla_weight_updater_config(self):	CODE
LOW	test/test_configs.py	4099	def test_multiprocessed_weight_updater_config(self):	CODE
LOW	test/test_configs.py	4111	def test_ray_weight_updater_config(self):	CODE
LOW	test/test_configs.py	4124	def test_rpc_weight_updater_config(self):	CODE
LOW	test/test_configs.py	4139	def test_distributed_weight_updater_config(self):	CODE
LOW	test/test_configs.py	4153	def test_remote_module_weight_updater_config(self):	CODE
LOW	test/test_configs.py	2565	def test_batched_env_with_one_transform(self, tmpdir):	STRING
LOW	test/test_configs.py	2594	def test_batched_env_with_two_transforms(self, tmpdir):	STRING
LOW	test/test_configs.py	2635	def test_simple_config_instantiation(self, tmpdir):	STRING
LOW	test/test_configs.py	2685	def test_env_parsing_with_file(self, tmpdir):	STRING
LOW	test/test_configs.py	2705	def test_collector_parsing_with_file(self, tmpdir):	STRING
3645 more matches not shown…

Decorative Section Separators931 hits · 2952 pts

Severity	File	Line	Snippet	Context
MEDIUM⚡	test/test_inference_server.py	128	# =============================================================================	COMMENT
MEDIUM⚡	test/test_inference_server.py	130	# =============================================================================	COMMENT
MEDIUM⚡	test/test_inference_server.py	1813	# =============================================================================	COMMENT
MEDIUM⚡	test/test_inference_server.py	1815	# =============================================================================	COMMENT
MEDIUM⚡	test/test_inference_server.py	2539	# =============================================================================	COMMENT
MEDIUM⚡	test/test_inference_server.py	2541	# =============================================================================	COMMENT
MEDIUM⚡	test/test_inference_server.py	2582	# =============================================================================	COMMENT
MEDIUM⚡	test/test_inference_server.py	2584	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	77	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	79	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	1087	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	1089	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	1188	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	1190	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	1280	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	1282	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	1563	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	1565	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	1753	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	1755	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	1956	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	test/test_inference_server.py	1958	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	test/test_inference_server.py	2465	# =============================================================================	COMMENT
MEDIUM	test/test_inference_server.py	2467	# =============================================================================	COMMENT
MEDIUM⚡	test/test_collectors.py	7303	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	7305	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	7574	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	7576	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	7902	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	7904	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	8099	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	8101	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	8202	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	8204	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	8251	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	8253	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	8359	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	8361	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	8611	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_collectors.py	8613	# ------------------------------------------------------------------	COMMENT
MEDIUM	test/test_collectors.py	7287	# ------------------------------------------------------------------	COMMENT
MEDIUM	test/test_collectors.py	7289	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_rnd.py	98	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_rnd.py	100	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_rnd.py	230	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/test_rnd.py	232	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	test/test_rnd.py	20	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	test/test_rnd.py	22	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	test/test_rnd.py	35	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	test/test_rnd.py	37	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/llm/test_wrapper.py	1623	# ================================================	COMMENT
MEDIUM⚡	test/llm/test_wrapper.py	1625	# ================================================	COMMENT
MEDIUM⚡	test/llm/test_wrapper.py	1701	# ================================================	COMMENT
MEDIUM⚡	test/llm/test_wrapper.py	1703	# ================================================	COMMENT
MEDIUM⚡	test/llm/test_wrapper.py	3964	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/llm/test_wrapper.py	3966	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/llm/test_wrapper.py	4135	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/llm/test_wrapper.py	4137	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/llm/test_wrapper.py	4390	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	test/llm/test_wrapper.py	4392	# ------------------------------------------------------------------	COMMENT
871 more matches not shown…

Unused Imports2253 hits · 1814 pts

Severity	File	Line	Context
LOW	setup.py	1	CODE
LOW	test/test_configs.py	6	CODE
LOW	test/test_shared.py	5	CODE
LOW	test/test_exploration.py	5	CODE
LOW	test/smoke_test.py	5	CODE
LOW	test/test_utils.py	5	CODE
LOW	test/conftest.py	5	CODE
LOW	test/opengl_rendering.py	26	CODE
LOW	test/opengl_rendering.py	33	CODE
LOW	test/opengl_rendering.py	71	CODE
LOW	test/test_docstring_args_checker.py	1	CODE
LOW	test/test_offline_to_online.py	5	CODE
LOW	test/test_custom_envs.py	5	CODE
LOW	test/test_inference_server.py	5	CODE
LOW	test/test_loggers.py	5	CODE
LOW	test/test_rlhf.py	5	CODE
LOW	test/test_postprocs.py	5	CODE
LOW	test/test_specs.py	5	CODE
LOW	test/smoke_test_deps.py	5	CODE
LOW	test/smoke_test_deps.py	33	CODE
LOW	test/smoke_test_deps.py	34	CODE
LOW	test/smoke_test_deps.py	35	CODE
LOW	test/smoke_test_deps.py	36	CODE
LOW	test/smoke_test_deps.py	108	CODE
LOW	test/test_render.py	5	CODE
LOW	test/test_distributions.py	5	CODE
LOW	test/test_collectors.py	5	CODE
LOW	test/test_helpers.py	5	CODE
LOW	test/test_distributed.py	9	CODE
LOW	test/test_trainer.py	5	CODE
LOW	test/test_rnd.py	5	CODE
LOW	test/test_checkpoint.py	5	CODE
LOW	test/collectors/test_evaluator.py	5	CODE
LOW	test/llm/test_vllm.py	5	CODE
LOW	test/llm/test_llm_objectives.py	5	CODE
LOW	test/llm/test_wrapper.py	5	CODE
LOW	test/llm/smoke_test_llm.py	5	CODE
LOW	test/llm/test_sglang_updaters.py	6	CODE
LOW	test/llm/test_sglang.py	6	CODE
LOW	test/llm/test_vllm_plugin.py	5	CODE
LOW	test/llm/test_llm_envs.py	5	CODE
LOW	test/llm/smoke_test_llm_deps.py	5	CODE
LOW	test/llm/test_llm_transforms.py	6	CODE
LOW	test/llm/test_llm_collectors.py	5	CODE
LOW	test/llm/test_llm_updaters.py	5	CODE
LOW	test/llm/test_llm_updaters.py	30	CODE
LOW	test/llm/test_data.py	6	CODE
LOW	test/llm/libs/test_mlgym.py	5	CODE
LOW	test/objectives/test_controllers.py	6	CODE
LOW	test/objectives/test_loss_module.py	6	CODE
LOW	test/objectives/test_loss_module.py	79	CODE
LOW	test/objectives/test_loss_module.py	79	CODE
LOW	test/objectives/conftest.py	5	CODE
LOW	test/objectives/test_mappo.py	12	CODE
LOW	test/objectives/test_cql.py	6	CODE
LOW	test/objectives/test_cql.py	28	CODE
LOW	test/objectives/test_cql.py	28	CODE
LOW	test/objectives/test_cql.py	28	CODE
LOW	test/objectives/test_bc.py	5	CODE
LOW	test/objectives/_objectives_common.py	5	CODE
2193 more matches not shown…

Docstring Block Structure171 hits · 842 pts

Severity	File	Line	Snippet	Context
HIGH	…ta-implementations/vla_grpo/openvla_oft/train_utils.py	59	Find a specific checkpoint file matching a pattern. Args: pretrained_checkpoint: Path to the checkpoin	STRING
HIGH	sota-implementations/dreamer/dreamer_utils.py	85	Allocate CUDA devices for collectors, reserving cuda:0 for training. Device allocation strategy: - Training alw	STRING
HIGH	sota-implementations/grpo/grpo_utils.py	111	Creates and configures the training model with LoRA adapters. This function initializes the main training model wit	STRING
HIGH	sota-implementations/grpo/grpo_utils.py	196	Creates the inference model for fast generation. This function initializes a model server (vLLM or SGLang) for effi	STRING
HIGH	sota-implementations/grpo/grpo_utils.py	540	Creates and configures a HuggingFace model with optional optimizations. Args: model_name (str): HuggingFace	STRING
HIGH	sota-implementations/redq/utils.py	177	Creates a Trainer instance given its constituents. Args: collector (BaseCollector): A data collector to be	STRING
HIGH	sota-implementations/expert-iteration/ei_utils.py	107	Creates and configures the training model with LoRA adapters. This function initializes the main training model wit	STRING
HIGH	sota-implementations/expert-iteration/ei_utils.py	188	Creates the vLLM-based inference model for fast generation. This function initializes a vLLM model server for effic	STRING
HIGH	sota-implementations/expert-iteration/ei_utils.py	342	Creates and configures a HuggingFace model with optional optimizations. Args: model_name (str): HuggingFace	STRING
HIGH	torchrl/_utils.py	1119	Return the available accelerator device, or CPU if none is found. Checks for accelerator availability in the follow	STRING
HIGH	torchrl/_utils.py	1240	Compile a model with warm-up. This function wraps :func:`~torch.compile` to add a warm-up phase. During the warm-up	STRING
HIGH	torchrl/_utils.py	1417	Return current CUDA memory statistics for ``device`` in gigabytes. Wraps :func:`torch.cuda.memory_allocated`, :func	STRING
HIGH	torchrl/_utils.py	1803	Merge user-provided ray_init_config with default runtime_env excludes. This function ensures that the default Torch	STRING
HIGH	torchrl/weight_update/utils.py	32	Resolve an attribute path like 'policy' or 'env.value_net' to actual object. Also processes getitem notation like '	STRING
HIGH	torchrl/collectors/_base.py	651	Execute a method on a nested attribute of this collector. This method allows remote callers to invoke methods o	STRING
HIGH	torchrl/collectors/_base.py	684	Get a nested attribute of this collector. This method allows remote callers to retrieve attributes from nested	STRING
HIGH	torchrl/collectors/_base.py	715	Apply a method to each set of arguments. This method executes a method on the collector with different argument	STRING
HIGH	torchrl/collectors/_base.py	986	Update policy weights for the data collector. This method synchronizes the policy weights used by the collector	STRING
HIGH	torchrl/collectors/_base.py	1267	Receive and apply weights to the collector's policy. This method applies weights to the local policy. When rece	STRING
HIGH	torchrl/collectors/_multi_base.py	1196	Receive from a pipe while periodically checking worker health. This method prevents the main process from hangi	STRING
HIGH	torchrl/collectors/_multi_base.py	1967	Sets the seeds of the environments stored in the DataCollector. Args: seed: integer representing th	STRING
HIGH	torchrl/collectors/_multi_base.py	2112	Get an attribute from the policy of the first worker. Args: attr (str): The attribute name to retri	STRING
HIGH	torchrl/collectors/_multi_base.py	2138	Get an attribute from the environment of the first worker. Args: attr (str): The attribute name to	STRING
HIGH	torchrl/collectors/_multi_base.py	2168	Get model instance by ID (for weight sync schemes). Args: model_id: Model identifier (e.g., "policy	STRING
HIGH	torchrl/collectors/utils.py	98	A util function for trajectory separation. Takes a tensordict with a key traj_ids that indicates the id of each tra	STRING
HIGH	torchrl/collectors/_single.py	1738	Sets the seeds of the environments stored in the DataCollector. Args: seed (int): integer represent	STRING
HIGH	torchrl/collectors/_single.py	2527	Get model instance by ID (for weight sync schemes). Args: model_id: Model identifier (e.g., "policy	STRING
HIGH	torchrl/collectors/distributed/default_configs.py	61	Create a TCPStore with retry logic for handling port conflicts. This function attempts to create a TCPStore, and if	STRING
HIGH	torchrl/record/recorder.py	333	Convert recorded frames to a Matplotlib animation. This helper is intended for tutorials and notebooks where th	STRING
HIGH	torchrl/render/checkpoint.py	52	Writes a checkpoint in the layout expected by rlrender factories. The model weights are stored under the canonical	STRING
HIGH	torchrl/render/mujoco_wasm.py	57	Writes a local Vite viewer for MuJoCo WASM notebook rendering. Args: output_dir: Directory that will contai	STRING
HIGH	torchrl/render/mujoco_wasm.py	125	Starts and displays a generated MuJoCo WASM viewer in a notebook. Args: viewer_dir: Directory produced by :	STRING
HIGH	torchrl/render/mujoco_wasm.py	184	Sends one qpos vector to a live MuJoCo WASM notebook viewer. Args: qpos: Iterable of joint position values.	STRING
HIGH	torchrl/render/mujoco_wasm.py	235	Plays a qpos trajectory in a live MuJoCo WASM notebook viewer. Args: qpos: Iterable of waypoints, each cont	STRING
HIGH	torchrl/render/mujoco_wasm.py	312	Extracts a qpos trajectory from a rollout TensorDict. Args: rollout: Rollout TensorDict saved by ``rlrender	STRING
HIGH	torchrl/render/backends/mujoco.py	53	Returns a detached snapshot of the environment's MuJoCo state. Args: env: TorchRL-native MuJoCo env	STRING
HIGH	torchrl/objectives/dreamer_v3.py	73	Return DreamerV3 dynamics and representation KL losses. The dynamics term stops gradients through the posterior and	STRING
HIGH	torchrl/objectives/dreamer_v3.py	121	KL divergence with balancing between posterior and prior. Computes: loss = alpha * KL(sg(posterior) \|\| prio	STRING
HIGH	torchrl/objectives/dreamer_v3.py	1163	Compute the DreamerV3 critic loss on a replay sequence. The return of each replay state uses the reward of the	STRING
HIGH	torchrl/objectives/common.py	855	Value-function constructor. If the non-default value function is wanted, it must be built using this me	STRING
HIGH	torchrl/objectives/llm/distillation.py	27	Per-token k3 estimate of the KL divergence to a target distribution. Uses the k3 estimator ``exp(d) - 1 - d`` with	STRING
HIGH	torchrl/objectives/value/advantages.py	1099	Computes the TD(0) advantage given the data in tensordict. If a functional module is provided, a nested TensorD	STRING
HIGH	torchrl/objectives/value/advantages.py	1388	Computes the TD(1) advantage given the data in tensordict. If a functional module is provided, a nested TensorD	STRING
HIGH	torchrl/objectives/value/advantages.py	1701	Computes the TD(:math:`\lambda`) advantage given the data in tensordict. If a functional module is provided, a	STRING
HIGH	torchrl/objectives/value/advantages.py	2078	Computes the GAE given the data in tensordict. If a functional module is provided, a nested TensorDict containi	STRING
HIGH	torchrl/objectives/value/advantages.py	2691	Computes the V-Trace correction given the data in tensordict. If a functional module is provided, a nested Tens	STRING
HIGH	torchrl/objectives/value/functional.py	1393	Compute the discounted cumulative sum of rewards given multiple trajectories and the episode ends. Args: re	STRING
HIGH	torchrl/_comm/backends.py	79	Return the canonical spelling of a TorchRL service backend. The long ``threading`` and ``multiprocessing`` spelling	STRING
HIGH	torchrl/_comm/backends.py	104	Return the canonical spelling of a TorchRL payload transport. Args: backend: Transport selector to validate	STRING
HIGH	torchrl/trainers/trainers.py	2893	Resolve a module from a trainer using a string path. Args: trainer (Trainer): The trainer instance to resol	STRING
HIGH	torchrl/trainers/helpers/models.py	67	DQN constructor helper function. Args: proof_environment (EnvBase): a dummy environment to retrieve the obs	STRING
HIGH	torchrl/trainers/helpers/trainers.py	90	Creates a Trainer instance given its constituents. Args: collector (BaseCollector): A data collector to be	STRING
HIGH	torchrl/modules/functional.py	12	Apply the element-wise symmetric logarithm transform. The transform is defined as ``sign(value) * log(1 + abs(v	STRING
HIGH	torchrl/modules/functional.py	42	Apply the inverse symmetric exponential transform element-wise. Args: value (torch.Tensor): Input tensor in	STRING
HIGH	torchrl/modules/functional.py	69	Apply the signed hyperbolic value transform. This is the scale-compressing transform introduced by Pohlen et al. an	STRING
HIGH	torchrl/modules/functional.py	108	Apply the inverse of :func:`signed_hyperbolic` element-wise. Args: value (torch.Tensor): Input tensor in si	STRING
HIGH	torchrl/modules/llm/backends/vllm/vllm_async.py	1435	Create a load balancer for this AsyncVLLM service. Args: strategy: Load balancing strategy or seque	STRING
HIGH	torchrl/modules/llm/backends/vllm/vllm_async.py	1657	Select the optimal actor index based on the configured strategy hierarchy. Args: prompt: The input	STRING
HIGH	torchrl/modules/llm/backends/vllm/vllm_async.py	1776	Select actor based on prompt prefix for cache locality. Args: prompt: Input prompt as string or tok	STRING
HIGH	torchrl/modules/llm/backends/vllm/vllm_async.py	1826	Extract prefix tokens from prompt (string or token list). Args: prompt: Input prompt. Retu	STRING
111 more matches not shown…

Self-Referential Comments198 hits · 508 pts

Severity	File	Line	Snippet	Context
MEDIUM	test/test_configs.py	1051	# Create a complex configuration	COMMENT
MEDIUM	test/test_configs.py	2442	# Create a test script that follows the pattern	COMMENT
MEDIUM	test/test_configs.py	2463	# Create the config directory structure	STRING
MEDIUM	test/test_exploration.py	688	# Create a policy with exploration module that has spec=None	COMMENT
MEDIUM	test/test_exploration.py	1132	# Create a simple network with NoisyLinear using new behavior	COMMENT
MEDIUM	test/test_utils.py	624	# Create a tensor with batch dimension	COMMENT
MEDIUM	test/test_postprocs.py	230	# Create a sample TensorDict	COMMENT
MEDIUM	test/test_postprocs.py	244	# Create an instance of LastRewardToTraj	COMMENT
MEDIUM⚡	test/test_specs.py	4576	# Create a OneHot spec with a mask that varies across batch dimension	COMMENT
MEDIUM	test/smoke_test_deps.py	16	# This file is a smoke test for optional deps. All optional-dep imports must	COMMENT
MEDIUM	test/test_collectors.py	6313	# Create a ParallelEnv factory - this is the key component that was failing	COMMENT
MEDIUM	test/test_collectors.py	6331	# Create the multi-collector with ParallelEnv and replay_buffer	COMMENT
MEDIUM	test/test_collectors.py	6403	# Create a simple dummy environment	COMMENT
MEDIUM	test/test_collectors.py	6409	# Create a simple dummy policy	COMMENT
MEDIUM	test/test_collectors.py	6717	# Create a deterministic policy (linear, no bias) so zeroing weights → zero actions	COMMENT
MEDIUM⚡	test/llm/test_conversions.py	116	# Create a batch of ChatHistory objects	COMMENT
MEDIUM⚡	test/llm/test_conversions.py	427	# Create a 2x2 batch	COMMENT
MEDIUM⚡	test/llm/test_conversions.py	429	# Create a 2x2 batch of ChatHistory objects	COMMENT
MEDIUM	test/llm/test_conversions.py	62	# Create a batch of ChatHistory objects	COMMENT
MEDIUM	test/llm/test_conversions.py	181	# Create a batch of ChatHistory objects	COMMENT
MEDIUM	test/llm/test_conversions.py	233	# Create a batch of ChatHistory objects	COMMENT
MEDIUM	test/llm/test_conversions.py	310	# Create a batch of ChatHistory objects	COMMENT
MEDIUM	test/llm/test_conversions.py	408	# Create a batch of ChatHistory objects	COMMENT
MEDIUM	test/llm/test_vllm.py	118	# Create a transformer policy with the same model	COMMENT
MEDIUM	test/llm/test_vllm.py	153	# Create a proper collector mock that provides access to the AsyncVLLM service	COMMENT
MEDIUM	test/llm/test_wrapper.py	2734	# Create a custom mask matching the logits shape	COMMENT
MEDIUM	test/llm/test_wrapper.py	3385	# Create a simple mock function to test the decorator	COMMENT
MEDIUM	test/llm/test_wrapper.py	3393	# Create a mock self object with batching attributes	COMMENT
MEDIUM	test/llm/test_llm_envs.py	850	# Define a simple calculator tool	COMMENT
MEDIUM	test/llm/test_llm_envs.py	859	# Define the tool schema	COMMENT
MEDIUM	test/llm/test_llm_envs.py	970	# Define a tool that waits for a random amount of time	COMMENT
MEDIUM	test/llm/test_llm_envs.py	983	# Define the tool schema	COMMENT
MEDIUM	test/llm/test_data.py	577	# Create a simple history	COMMENT
MEDIUM	test/llm/test_data.py	726	# Create a custom template (trivially different from Qwen)	COMMENT
MEDIUM	test/llm/test_data.py	1000	# Create a batch of 2 prompt histories	COMMENT
MEDIUM⚡	test/objectives/test_loss_module.py	1371	# Create a value estimator instance	COMMENT
MEDIUM⚡	test/objectives/test_loss_module.py	1377	# Create a loss module that supports value estimation	COMMENT
MEDIUM⚡	test/objectives/test_loss_module.py	1470	# Create a GAE instance	COMMENT
MEDIUM⚡	test/objectives/test_loss_module.py	1477	# Create a PPO loss	COMMENT
MEDIUM	test/objectives/test_loss_module.py	1398	# Create a loss module	COMMENT
MEDIUM	test/objectives/test_loss_module.py	1424	# Create a loss module	COMMENT
MEDIUM	test/objectives/test_loss_module.py	1446	# Create a PPO loss which supports GAE	COMMENT
MEDIUM	test/libs/test_envpool.py	234	# Create the first env, set the seed, and perform a sequence of operations	COMMENT
MEDIUM	test/libs/test_envpool.py	248	# Create a new env, set the seed, and repeat same operations	COMMENT
MEDIUM	test/libs/test_isaac.py	937	# Create a fresh env with InitTracker (required for LSTM)	COMMENT
MEDIUM	test/libs/test_isaac.py	968	# Create a simple policy using the LSTM and move to correct device	COMMENT
MEDIUM⚡	test/transforms/test_reward_transforms.py	1208	# Create a simple env with multi-objective rewards	COMMENT
MEDIUM⚡	test/transforms/test_reward_transforms.py	1211	# Create a composed transform with LineariseRewards and RewardScaling	COMMENT
MEDIUM	test/transforms/test_reward_transforms.py	1250	# Create a dummy env that produces nested rewards	COMMENT
MEDIUM	test/transforms/test_compose_and_env.py	633	# Create a Compose transform that renames "action" to "action_1" and then to "action_2"	COMMENT
MEDIUM⚡	test/transforms/test_module_transforms.py	224	# Create a simple module that adds a learnable bias to observations	COMMENT
MEDIUM	test/rb/test_samplers.py	1512	# Create a new buffer with the loaded storage but NEW sampler	COMMENT
MEDIUM	test/rb/test_storages.py	414	# Create a list of tensordicts (like a collector would produce)	COMMENT
MEDIUM	test/rb/test_storages.py	752	# Create an instance of LazyStackStorage with given parameters	COMMENT
MEDIUM	test/rb/test_storages.py	754	# Create a ReplayBuffer using the created storage	COMMENT
MEDIUM	test/rb/test_storages.py	1548	# Create a script that creates a storage and exits normally	COMMENT
MEDIUM	test/rb/test_storages.py	1579	# Create a script that sleeps and can be interrupted	COMMENT
MEDIUM	…ta-implementations/vla_grpo/openvla_oft/train_utils.py	9	# Create a tensor marking positions of IGNORE_INDEX	COMMENT
MEDIUM	…ta-implementations/vla_grpo/openvla_oft/train_utils.py	15	# Create the mask	COMMENT
MEDIUM	…ta-implementations/vla_grpo/openvla_oft/train_utils.py	26	# Create a tensor marking positions of IGNORE_INDEX	COMMENT
138 more matches not shown…

Deep Nesting548 hits · 445 pts

Severity	File	Line	Context
LOW	setup.py	145	CODE
LOW	packaging/wheel/relocate.py	76	CODE
LOW	test/test_configs.py	1479	CODE
LOW	test/test_configs.py	2438	CODE
LOW	test/test_utils.py	379	CODE
LOW	test/opengl_rendering.py	87	CODE
LOW	test/test_specs.py	3482	CODE
LOW	test/test_specs.py	3528	CODE
LOW	test/test_specs.py	1910	CODE
LOW	test/test_specs.py	2245	CODE
LOW	test/test_specs.py	2530	CODE
LOW	test/test_specs.py	2826	CODE
LOW	test/test_collectors.py	1567	CODE
LOW	test/test_collectors.py	1649	CODE
LOW	test/test_collectors.py	2519	CODE
LOW	test/test_collectors.py	2871	CODE
LOW	test/test_collectors.py	3775	CODE
LOW	test/test_collectors.py	4214	CODE
LOW	test/test_collectors.py	4375	CODE
LOW	test/test_collectors.py	4590	CODE
LOW	test/test_collectors.py	4656	CODE
LOW	test/test_collectors.py	5503	CODE
LOW	test/test_collectors.py	5524	CODE
LOW	test/test_collectors.py	5554	CODE
LOW	test/test_collectors.py	6381	CODE
LOW	test/test_collectors.py	6543	CODE
LOW	test/test_collectors.py	6798	CODE
LOW	test/test_collectors.py	8616	CODE
LOW	test/test_collectors.py	8673	CODE
LOW	test/test_mcts.py	107	CODE
LOW	test/test_mcts.py	853	CODE
LOW	test/test_distributed.py	442	CODE
LOW	test/test_distributed.py	568	CODE
LOW	test/test_distributed.py	784	CODE
LOW	test/test_distributed.py	1524	CODE
LOW	test/test_distributed.py	1623	CODE
LOW	test/test_trainer.py	381	CODE
LOW	test/test_trainer.py	758	CODE
LOW	test/test_trainer.py	1147	CODE
LOW	test/llm/test_vllm.py	116	CODE
LOW	test/llm/test_wrapper.py	469	CODE
LOW	test/llm/test_wrapper.py	2337	CODE
LOW	test/llm/test_llm_collectors.py	242	CODE
LOW	test/llm/test_llm_collectors.py	349	CODE
LOW	test/llm/test_data.py	494	CODE
LOW	test/objectives/test_loss_module.py	182	CODE
LOW	test/objectives/test_cql.py	174	CODE
LOW	test/objectives/_objectives_common.py	208	CODE
LOW	test/objectives/_objectives_common.py	209	CODE
LOW	test/objectives/test_iql.py	289	CODE
LOW	test/objectives/test_iql.py	544	CODE
LOW	test/objectives/test_iql.py	1205	CODE
LOW	test/objectives/test_iql.py	1378	CODE
LOW	test/objectives/test_ddpg.py	375	CODE
LOW	test/objectives/test_ddpg.py	999	CODE
LOW	test/objectives/test_ddpg.py	1257	CODE
LOW	test/objectives/test_ddpg.py	1922	CODE
LOW	test/objectives/test_ddpg.py	2089	CODE
LOW	test/objectives/test_ppo.py	399	CODE
LOW	test/objectives/test_ppo.py	508	CODE
488 more matches not shown…

Excessive Try-Catch Wrapping390 hits · 400 pts

Severity	File	Line	Snippet	Context
LOW	setup.py	353	except Exception:	CODE
MEDIUM	setup.py	346	def _git_output(args) -> str \| None:	CODE
LOW	test/test_configs.py	2517	except Exception:	STRING
LOW	test/test_setup.py	216	except Exception as err:	CODE
MEDIUM	test/test_inference_server.py	1353	def run(idx):	CODE
MEDIUM	test/test_inference_server.py	2150	def _hammer(server, fn_name):	CODE
LOW	test/test_inference_server.py	1359	except Exception as exc: # noqa: BLE001	CODE
LOW	test/test_inference_server.py	2155	except Exception as exc:	CODE
LOW	test/test_inference_server.py	2603	except Exception:	CODE
LOW	test/test_loggers.py	757	except Exception:	CODE
LOW	test/smoke_test_deps.py	109	except Exception: # pragma: no cover	CODE
LOW	test/smoke_test_deps.py	113	except Exception as err: # gymnasium.error.NamespaceNotFound and similar	CODE
MEDIUM	test/smoke_test_deps.py	89	def test_gym():	CODE
MEDIUM	test/test_collectors.py	2464	def test_cudagraph_policy_stochastic_diversity(self):	CODE
LOW⚡	test/test_collectors.py	5261	except Exception:	CODE
LOW⚡	test/test_collectors.py	5276	except Exception:	CODE
LOW	test/test_collectors.py	5226	except Exception as e:	CODE
LOW	test/test_collectors.py	5307	except Exception:	CODE
MEDIUM	test/test_distributed.py	309	def _test_distributed_collector_basic(cls, queue, frames_per_batch):	CODE
MEDIUM	test/test_distributed.py	355	def _test_distributed_collector_mult(cls, queue, frames_per_batch):	CODE
MEDIUM	test/test_distributed.py	398	def _test_distributed_collector_sync(cls, queue, sync):	CODE
MEDIUM	test/test_distributed.py	517	def _test_distributed_collector_class(cls, queue, collector_class):	CODE
MEDIUM	test/test_distributed.py	568	def _test_distributed_collector_updatepolicy(	CODE
MEDIUM	test/test_distributed.py	895	def reset_process_group(self):	CODE
MEDIUM	test/test_distributed.py	1781	def reset_process_group(self):	CODE
LOW	test/test_distributed.py	329	except Exception as e:	CODE
LOW	test/test_distributed.py	374	except Exception as e:	CODE
LOW	test/test_distributed.py	418	except Exception as e:	CODE
LOW	test/test_distributed.py	479	except Exception as e:	CODE
LOW	test/test_distributed.py	537	except Exception as e:	CODE
LOW	test/test_distributed.py	630	except Exception as e:	CODE
LOW	test/test_distributed.py	698	except Exception as e:	CODE
LOW	test/test_distributed.py	830	except Exception as e:	CODE
LOW	test/test_distributed.py	898	except Exception:	CODE
LOW	test/test_distributed.py	1784	except Exception:	CODE
LOW	test/llm/test_llm_objectives.py	1395	except Exception as e:	CODE
LOW	test/llm/test_wrapper.py	109	except Exception as e:	CODE
LOW	test/llm/test_wrapper.py	142	except Exception as e:	CODE
LOW	test/llm/test_wrapper.py	588	except Exception as e:	CODE
LOW	test/llm/test_wrapper.py	2890	except Exception as e:	CODE
LOW	test/llm/test_wrapper.py	3155	except Exception as e:	CODE
LOW	test/llm/test_wrapper.py	3309	except Exception as e:	CODE
LOW	test/llm/test_wrapper.py	3536	except Exception:	CODE
LOW	test/llm/test_sglang_updaters.py	255	except Exception as e:	CODE
LOW	test/llm/test_llm_envs.py	1298	except Exception as e:	CODE
LOW	test/llm/test_llm_updaters.py	135	except Exception as e:	CODE
LOW	test/llm/test_llm_updaters.py	272	except Exception as e:	CODE
LOW	test/llm/test_llm_updaters.py	346	except Exception as e:	CODE
LOW	test/llm/test_llm_updaters.py	400	except Exception as e:	CODE
LOW	test/llm/test_data.py	691	except Exception as e:	CODE
LOW	test/libs/conftest.py	33	except Exception:	CODE
MEDIUM	test/libs/test_datasets.py	790	def test_correct_categorical_missions(self):	CODE
LOW	test/libs/test_datasets.py	445	except Exception as err:	CODE
LOW	test/libs/test_datasets.py	550	except Exception:	CODE
LOW	test/libs/test_datasets.py	555	except Exception:	CODE
LOW	test/libs/test_datasets.py	797	except Exception as e:	CODE
MEDIUM	test/libs/test_gym.py	119	def get_gym_pixel_wrapper():	CODE
LOW	test/libs/test_gym.py	68	except Exception:	CODE
LOW	test/libs/test_gym.py	125	except Exception:	CODE
LOW	test/libs/test_gym.py	139	except Exception:	CODE
330 more matches not shown…

Over-Commented Block503 hits · 372 pts

Severity	File	Line	Snippet	Context
LOW	build_nightly.sh	41	# PyPI rejects PEP 440 local version identifiers (e.g. "+g<sha>"). Pin the	COMMENT
LOW	pyproject.toml	261	# Auto-loaded by vLLM in every process (main, workers, and the registry	COMMENT
LOW	packaging/pkg_helpers.bash	1	# A set of useful bash functions for common functionality we need to do in	COMMENT
LOW	packaging/pkg_helpers.bash	21	# /usr/local/cuda-*	COMMENT
LOW	packaging/pkg_helpers.bash	201		COMMENT
LOW	test/opengl_rendering.py	1	# Copyright (c) Meta Platforms, Inc. and affiliates.	COMMENT
LOW	test/opengl_rendering.py	41		COMMENT
LOW	test/test_collectors.py	2041	def test_no_deepcopy_policy(self, collector_type):	COMMENT
LOW	test/test_collectors.py	2921	# Use -1 for padding to uniform with other preemption	COMMENT
LOW	test/test_collectors.py	3561	)	COMMENT
LOW	test/test_collectors.py	3581	# ) and not torch.cuda.is_available():	COMMENT
LOW	test/test_collectors.py	3601	# else:	COMMENT
LOW	test/test_collectors.py	3621	# policy=policy,	COMMENT
LOW	test/test_collectors.py	3641	# frames_per_batch=20,	COMMENT
LOW	test/test_collectors.py	3661	# @pytest.mark.parametrize("env_name", ["vec"])	COMMENT
LOW	test/test_collectors.py	3681	# create_env_kwargs=[{"seed": i} for i in range(seed, seed + num_env)],	COMMENT
LOW	test/test_collectors.py	3701	# assert (d["done"].sum(-2) >= 1).all()	COMMENT
LOW	test/test_collectors.py	3721	# create_env_fn=env_fn,	COMMENT
LOW	test/test_collectors.py	3821	# Debug: shutting down collector	COMMENT
LOW	test/test_collectors.py	3841	# assert collector.policy.out_keys == ["action"]	COMMENT
LOW	test/llm/test_wrapper.py	2841	# """Setup and teardown for each test.	COMMENT
LOW	test/objectives/test_loss_module.py	1181	class TestBuffer:	COMMENT
LOW	test/libs/test_isaac.py	401	# "AllegroKukaTwoArms",	COMMENT
LOW	test/modules/test_td_module.py	221		COMMENT
LOW	test/envs/test_env_base.py	661		COMMENT
LOW	test/rb/conftest.py	1	# Copyright (c) Meta Platforms, Inc. and affiliates.	COMMENT
LOW	test/rb/test_storage_map.py	561	node_id = vertex.node_id	COMMENT
LOW	sota-implementations/multiagent/utils/__init__.py	1	# Copyright (c) Meta Platforms, Inc. and affiliates.	COMMENT
LOW	sota-implementations/vla_grpo/openvla.py	61	PROMPT_TEMPLATE = "In: What action should the robot take to {instruction}?\nOut:"	COMMENT
LOW	sota-implementations/vla_grpo/openvla_oft/__init__.py	1	# Vendored from SimpleVLA-RL (https://github.com/PRIME-RL/SimpleVLA-RL),	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	521	dtype=labels.dtype,	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	541	# proprio=None,	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	561		COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	581	# # === Handle Unimodal Forward ===	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	601	# assert past_key_values is None, "Unexpected key `past_key_values` provided during multimodal forward!"	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	621	# # Add proprioceptive state if provided	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	641	# noisy_actions = noisy_actions.reshape(B, -1).unsqueeze(-1)	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	661	# # Build labels for multimodal sequence if needed	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	681		COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	821	output_hidden_states=False,	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	841		COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	861		COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	881	# #print("!!!!!!!!!!!!!!Entering forward!!!!!!!!!!")	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	901	# # Add stop token to sequence (needed in non-causal bi-directional self-attention, as it appears at train time)	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	921	# #test	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	941		COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	961	# newline_positions = labels != IGNORE_INDEX	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	981		COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	1001	# projected_patch_embeddings, proprio, proprio_projector	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	1021	# # Run diffusion-based prediction	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	1041	# # labels,	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	1061	# (projected_patch_embeddings.shape[0], projected_patch_embeddings.shape[1]),	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	1081		COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	1281	else:	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	1641	# if action_head is not None:	COMMENT
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	1661	device = language_model_output.logits.device	COMMENT
LOW	…a-implementations/vla_grpo/config/vla_grpo_libero.yaml	1	# SimpleVLA-RL (arXiv:2509.09674) on LIBERO: OpenVLA-OFT token variant (7B),	COMMENT
LOW	sota-implementations/cql/utils.py	261		COMMENT
LOW	sota-implementations/rnd/rnd_mujoco.py	81	# The env already normalizes and clips observations with VecNorm and	COMMENT
LOW	sota-implementations/dreamer/dreamer_utils.py	181		COMMENT
443 more matches not shown…

Modern Structural Boilerplate354 hits · 348 pts

Severity	File	Line	Snippet	Context
LOW	setup.py	26	logger = logging.getLogger(__name__)	CODE
LOW	packaging/prepare_nightly_build.py	7	logger = logging.getLogger(__name__)	CODE
LOW	test/conftest.py	81	def set_warnings() -> None:	STRING
LOW	test/test_render.py	162	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW	test/test_collectors.py	1754	def _set_seed(self, seed: Optional[int]) -> None:	CODE
LOW	test/test_collectors.py	3207	def _set_seed(self, seed: int \| None = None) -> None:	CODE
LOW	test/test_collectors.py	3264	def _set_seed(self, seed: int \| None = None) -> None:	CODE
LOW	test/test_collectors.py	3439	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW	test/test_collectors.py	4550	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW	test/collectors/test_evaluator.py	1041	def set_extra_state(self, state: OrderedDict) -> None:	CODE
LOW	test/objectives/_objectives_common.py	132	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW	test/libs/test_gym.py	195	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW⚡	test/transforms/test_action_transforms.py	135	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW⚡	test/transforms/test_key_transforms.py	95	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW⚡	test/transforms/test_key_transforms.py	333	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW⚡	test/transforms/test_key_transforms.py	934	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW	test/transforms/test_reward_transforms.py	1284	def _set_seed(self, seed: int \| None = None) -> None:	CODE
LOW	test/transforms/test_reward_transforms.py	1344	def _set_seed(self, seed: int \| None = None) -> None:	CODE
LOW	test/transforms/test_compose_and_env.py	174	def _set_seed(self, seed: int) -> None:	CODE
LOW	test/transforms/test_compose_and_env.py	1175	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW	test/transforms/test_env_transforms.py	439	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW	test/transforms/test_normalization.py	79	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW	test/envs/test_env_base.py	142	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW	test/envs/test_env_base.py	318	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW	…implementations/vla_grpo/compare_simplevla_rollouts.py	229	def _set_seed(seed: int) -> None:	CODE
LOW⚡	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	395	def set_input_embeddings(self, value: nn.Module) -> None:	CODE
LOW⚡	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	401	def set_output_embeddings(self, new_embeddings: nn.Module) -> None:	CODE
LOW⚡	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	407	def set_decoder(self, decoder: nn.Module) -> None:	CODE
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	43	logger = logging.getLogger(__name__)	CODE
LOW	…ementations/vla_grpo/openvla_oft/modeling_prismatic.py	199	def set_num_images_in_input(self, num_images_in_input: int) -> None:	CODE
LOW	sota-implementations/bandits/dqn.py	91	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW	examples/replay-buffers/filter-imcomplete-trajs.py	33	def _update_repertoire(self, tensordict: TensorDictBase) -> None:	CODE
LOW	examples/rlhf/models/actor_critic.py	15	__all__ = ["init_actor_critic"]	CODE
LOW	examples/rlhf/data/__init__.py	3	__all__ = ["get_prompt_dataloader_tldr"]	CODE
LOW	examples/envs/benchmark_compile_step_and_maybe_reset.py	110	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW	benchmarks/conftest.py	143	def set_warnings() -> None:	STRING
LOW	benchmarks/test_collectors_benchmark.py	51	def _set_seed(self, seed: int \| None) -> None:	CODE
LOW	benchmarks/bench_collectors.py	117	def _set_seed(self, seed: int) -> None:	CODE
LOW	torchrl/__init__.py	143	__all__ = [	CODE
LOW	torchrl/_utils.py	449	def set_profiling_enabled(enabled: bool) -> None:	CODE
LOW	torchrl/_utils.py	1168	def set_mode(self, type: Any \| None) -> None:	CODE
LOW	torchrl/weight_update/weight_sync_schemes.py	23	__all__ = [	CODE
LOW	torchrl/weight_update/__init__.py	20	__all__ = [	CODE
LOW	torchrl/weight_update/_shared.py	266	def _update_single_worker(self, worker_idx: int, weights: TensorDictBase) -> None:	CODE
LOW⚡	torchrl/weight_update/_ray.py	163	def set_model(self, model: Any) -> None:	CODE
LOW⚡	torchrl/weight_update/_ray.py	171	def set_process_group(self, process_group, store=None) -> None:	CODE
LOW⚡	torchrl/weight_update/_ray.py	176	def set_model_version(self, model_version: int \| None) -> None:	CODE
LOW	torchrl/weight_update/_ray.py	534	def _set_model_version(self, model_version: int) -> None:	CODE
LOW	torchrl/weight_update/_ray.py	971	def _set_transform(self, ray_transform) -> None:	CODE
LOW	torchrl/weight_update/llm/vllm_nccl.py	641	def update_weights(self, weights: Any \| None = None) -> None:	CODE
LOW	torchrl/weight_update/llm/__init__.py	26	__all__ = [	CODE
LOW	torchrl/weight_update/llm/vllm_double_buffer.py	260	def update_weights(self, weights: Any \| None = None) -> None:	CODE
LOW	torchrl/collectors/_constants.py	26	__all__ = [	CODE
LOW	torchrl/collectors/_base.py	1209	def _set_received_policy_version(self, model_version: int \| None) -> None:	CODE
LOW	torchrl/collectors/collectors.py	31	__all__ = [	CODE
LOW	torchrl/collectors/__init__.py	27	__all__ = [	CODE
LOW	torchrl/collectors/_multi_base.py	912	def _setup_preemptive_threshold(self, preemptive_threshold: float \| None) -> None:	CODE
LOW	torchrl/collectors/_multi_base.py	1735	def _set_worker_attr(self, attr_name: str, value: Any) -> None:	CODE
LOW⚡	torchrl/collectors/_single.py	1080	def _setup_policy_and_weights(self, policy: TensorDictModule \| Callable) -> None:	CODE
LOW⚡	torchrl/collectors/_single.py	1231	def _setup_max_frames_per_traj(self, max_frames_per_traj: int \| None) -> None:	CODE
294 more matches not shown…

Cross-File Repetition62 hits · 310 pts

Severity	File	Line	Snippet	Context
HIGH	test/llm/test_llm_objectives.py	0	create transformers model and tokenizer for testing.	STRING
HIGH	test/llm/test_wrapper.py	0	create transformers model and tokenizer for testing.	STRING
HIGH	test/llm/test_llm_envs.py	0	create transformers model and tokenizer for testing.	STRING
HIGH	test/objectives/test_cql.py	0	regression test for target_entropy='auto' should be -dim(a).	STRING
HIGH	test/objectives/test_dt.py	0	regression test for target_entropy='auto' should be -dim(a).	STRING
HIGH	test/objectives/test_sac.py	0	regression test for target_entropy='auto' should be -dim(a).	STRING
HIGH	test/libs/test_dm_control.py	0	smoke test that setting seed and reset works (seed forwarded into build).	STRING
HIGH	test/libs/test_gym.py	0	smoke test that setting seed and reset works (seed forwarded into build).	STRING
HIGH	test/libs/test_habitat.py	0	smoke test that setting seed and reset works (seed forwarded into build).	STRING
HIGH	sota-implementations/sac/sac.py	0	crossq example. this is a simple self-contained example of a crossq training script. it supports state environments like	STRING
HIGH	sota-implementations/discrete_sac/discrete_sac.py	0	crossq example. this is a simple self-contained example of a crossq training script. it supports state environments like	STRING
HIGH	sota-implementations/ddpg/ddpg.py	0	crossq example. this is a simple self-contained example of a crossq training script. it supports state environments like	STRING
HIGH	sota-implementations/td3/td3.py	0	crossq example. this is a simple self-contained example of a crossq training script. it supports state environments like	STRING
HIGH	sota-implementations/crossq/crossq.py	0	crossq example. this is a simple self-contained example of a crossq training script. it supports state environments like	STRING
HIGH	sota-implementations/cql/cql_offline.py	0	decision transformer example. this is a self-contained example of an offline decision transformer training script. the h	STRING
HIGH	sota-implementations/cql/discrete_cql_offline.py	0	decision transformer example. this is a self-contained example of an offline decision transformer training script. the h	STRING
HIGH	sota-implementations/iql/iql_offline.py	0	decision transformer example. this is a self-contained example of an offline decision transformer training script. the h	STRING
HIGH	sota-implementations/td3_bc/td3_bc.py	0	decision transformer example. this is a self-contained example of an offline decision transformer training script. the h	STRING
HIGH	sota-implementations/decision_transformer/online_dt.py	0	decision transformer example. this is a self-contained example of an offline decision transformer training script. the h	STRING
HIGH	sota-implementations/decision_transformer/dt.py	0	decision transformer example. this is a self-contained example of an offline decision transformer training script. the h	STRING
HIGH	sota-implementations/cql/cql_online.py	0	iql example. this is a self-contained example of an online iql training script. it works across gym and mujoco over a va	STRING
HIGH	sota-implementations/iql/discrete_iql.py	0	iql example. this is a self-contained example of an online iql training script. it works across gym and mujoco over a va	STRING
HIGH	sota-implementations/iql/iql_online.py	0	iql example. this is a self-contained example of an online iql training script. it works across gym and mujoco over a va	STRING
HIGH	sota-implementations/impala/impala_multi_node_ray.py	0	this script reproduces the impala algorithm results from espeholt et al. 2018 for the on atari environments.	STRING
HIGH	sota-implementations/impala/impala_single_node.py	0	this script reproduces the impala algorithm results from espeholt et al. 2018 for the on atari environments.	STRING
HIGH	…a-implementations/impala/impala_multi_node_submitit.py	0	this script reproduces the impala algorithm results from espeholt et al. 2018 for the on atari environments.	STRING
HIGH	sota-implementations/grpo/grpo-async.py	0	setup required environment variables and configurations.	STRING
HIGH	sota-implementations/grpo/grpo-sync.py	0	setup required environment variables and configurations.	STRING
HIGH	…lementations/expert-iteration/expert-iteration-sync.py	0	setup required environment variables and configurations.	STRING
HIGH	…ementations/expert-iteration/expert-iteration-async.py	0	setup required environment variables and configurations.	STRING
HIGH	examples/distributed/collectors/single_machine/sync.py	0	rpc data collection on a single node. the default configuration works fine on machines equipped with 4 gpus, but can be	STRING
HIGH	…mples/distributed/collectors/single_machine/generic.py	0	rpc data collection on a single node. the default configuration works fine on machines equipped with 4 gpus, but can be	STRING
HIGH	examples/distributed/collectors/single_machine/rpc.py	0	rpc data collection on a single node. the default configuration works fine on machines equipped with 4 gpus, but can be	STRING
HIGH	torchrl/record/loggers/mlflow.py	0	logs a scalar value to the tensorboard. args: name (str): the name of the scalar. value (float): the value of the scalar	STRING
HIGH	torchrl/record/loggers/trackio.py	0	logs a scalar value to the tensorboard. args: name (str): the name of the scalar. value (float): the value of the scalar	STRING
HIGH	torchrl/record/loggers/csv.py	0	logs a scalar value to the tensorboard. args: name (str): the name of the scalar. value (float): the value of the scalar	STRING
HIGH	torchrl/record/loggers/tensorboard.py	0	logs a scalar value to the tensorboard. args: name (str): the name of the scalar. value (float): the value of the scalar	STRING
HIGH	torchrl/record/loggers/mlflow.py	0	logs the hyperparameters of the experiment. args: cfg (dictconfig or dict): the configuration of the experiment.	STRING
HIGH	torchrl/record/loggers/wandb.py	0	logs the hyperparameters of the experiment. args: cfg (dictconfig or dict): the configuration of the experiment.	STRING
HIGH	torchrl/record/loggers/trackio.py	0	logs the hyperparameters of the experiment. args: cfg (dictconfig or dict): the configuration of the experiment.	STRING
HIGH	torchrl/record/loggers/csv.py	0	logs the hyperparameters of the experiment. args: cfg (dictconfig or dict): the configuration of the experiment.	STRING
HIGH	torchrl/record/loggers/tensorboard.py	0	logs the hyperparameters of the experiment. args: cfg (dictconfig or dict): the configuration of the experiment.	STRING
HIGH	torchrl/objectives/td3.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/objectives/td3_bc.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/objectives/iql.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/objectives/crossq.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/objectives/deprecated.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/objectives/redq.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/objectives/sac.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/objectives/ddpg.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/objectives/reinforce.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/objectives/ppo.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/objectives/a2c.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/objectives/cql.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/objectives/dqn.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/objectives/multiagent/qmixer.py	0	maintains default values for all configurable tensordict keys. this class defines which tensordict keys can be set using	STRING
HIGH	torchrl/_comm/distributed.py	0	return ``true`` when the result can be read without blocking.	STRING
HIGH	torchrl/_comm/mailbox.py	0	return ``true`` when the result can be read without blocking.	STRING
HIGH	torchrl/modules/inference_server/_shared_memory.py	0	return ``true`` when the result can be read without blocking.	STRING
HIGH	torchrl/modules/llm/policies/vllm_wrapper.py	0	returns a new version of the module with altered parameters. for instance, the generate parameter can be altered to enab	STRING
2 more matches not shown…

AI Structural Patterns275 hits · 242 pts

Severity	File	Line	Context
LOW	test/test_postprocs.py	300	CODE
LOW	test/llm/test_llm_objectives.py	660	CODE
LOW	test/llm/test_wrapper.py	636	CODE
LOW	test/objectives/test_iql.py	199	CODE
LOW	test/objectives/test_iql.py	1111	CODE
LOW	test/objectives/test_dqn.py	90	CODE
LOW	test/objectives/test_dqn.py	884	CODE
LOW	test/objectives/test_ddpg.py	138	CODE
LOW	test/objectives/test_ddpg.py	179	CODE
LOW	test/objectives/test_ddpg.py	905	CODE
LOW	test/objectives/test_ddpg.py	1827	CODE
LOW	test/objectives/test_ppo.py	74	CODE
LOW	test/objectives/test_ppo.py	262	CODE
LOW	test/objectives/test_ppo.py	314	CODE
LOW	test/objectives/test_ppo.py	1877	CODE
LOW	test/objectives/test_ppo.py	2029	CODE
LOW	test/objectives/test_sac.py	75	CODE
LOW	test/objectives/test_sac.py	257	CODE
LOW	test/objectives/test_sac.py	1585	CODE
LOW	test/objectives/test_sac.py	2420	CODE
LOW	test/objectives/test_sac.py	3269	CODE
LOW	test/libs/test_datasets.py	1153	CODE
LOW	test/libs/test_envpool.py	150	CODE
LOW	sota-implementations/vla_grpo/utils.py	529	CODE
LOW	sota-implementations/vla_grpo/utils.py	584	CODE
LOW	sota-implementations/vla_grpo/openvla.py	186	CODE
LOW	sota-implementations/vla_grpo/openvla.py	476	CODE
LOW	sota-implementations/vla_grpo/openvla.py	815	CODE
LOW	sota-implementations/vla_grpo/openvla.py	1101	CODE
LOW	…ations/vla_grpo/openvla_oft/configuration_prismatic.py	76	CODE
LOW	sota-implementations/grpo/grpo_utils.py	518	CODE
LOW	sota-implementations/decision_transformer/lamb.py	38	CODE
LOW	sota-implementations/ppo/utils_mujoco.py	43	CODE
LOW	sota-implementations/redq/utils.py	561	CODE
LOW	sota-implementations/expert-iteration/ei_utils.py	319	CODE
LOW	examples/satellite/_utils.py	443	CODE
LOW	examples/satellite/_utils.py	538	CODE
LOW	torchrl/weight_update/_shared.py	386	CODE
LOW	torchrl/weight_update/_shared.py	511	CODE
LOW	torchrl/weight_update/_mp.py	83	CODE
LOW	torchrl/collectors/_base.py	468	CODE
LOW	torchrl/collectors/_single_async.py	175	CODE
LOW	torchrl/collectors/_multi_base.py	423	CODE
LOW	torchrl/collectors/_async_batched.py	254	CODE
LOW	torchrl/collectors/_single.py	677	CODE
LOW	torchrl/collectors/_single.py	1762	CODE
LOW	torchrl/collectors/_evaluator.py	289	CODE
LOW	torchrl/collectors/_runner.py	36	CODE
LOW	torchrl/collectors/llm/ray_collector.py	82	CODE
LOW	torchrl/collectors/llm/base.py	146	CODE
LOW	torchrl/collectors/distributed/sync.py	288	CODE
LOW	torchrl/collectors/distributed/generic.py	542	CODE
LOW	torchrl/collectors/distributed/ray.py	348	CODE
LOW	torchrl/collectors/distributed/rpc.py	317	CODE
LOW	torchrl/record/recorder.py	142	CODE
LOW	torchrl/record/loggers/monitoring.py	219	CODE
LOW	torchrl/render/mujoco_wasm.py	220	CODE
LOW	torchrl/objectives/td3.py	224	CODE
LOW	torchrl/objectives/iql.py	269	CODE
LOW	torchrl/objectives/iql.py	826	CODE
215 more matches not shown…

Cross-Language Confusion44 hits · 230 pts

Severity	File	Line	Snippet	Context
HIGH	test/test_inference_server.py	1883	ws.push(new_weights)	CODE
HIGH	test/test_inference_server.py	1912	ws.push(new_weights)	CODE
HIGH	test/envs/test_special.py	254	board.push(move)	CODE
HIGH	.github/scripts/generate_dashboard.py	310	document.getElementById('new-flaky-count').textContent = report.summary.new_flaky_count \|\| 0;	CODE
HIGH	.github/scripts/generate_dashboard.py	311	document.getElementById('resolved-count').textContent = report.summary.resolved_count \|\| 0;	CODE
HIGH	.github/scripts/generate_dashboard.py	344	if (!flakyTests \|\| flakyTests.length === 0) {	CODE
HIGH	.github/scripts/generate_dashboard.py	370	const lastFailed = test.recent_failures && test.recent_failures.length > 0	CODE
HIGH	torchrl/_extension.py	59	" - make sure you ran `python setup.py clean && python setup.py develop` and that no error was raised\n"	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	833	let model = null;	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	834	let data = null;	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	835	let mujoco = null;	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	838	let threeScene = null;	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	839	let renderer = null;	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	840	let camera = null;	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	841	let controls = null;	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	844	let trajectoryPlayback = null;	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	1091	jointControls.push({name, jointId, qposAddress, slider, output});	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	1098	if (!trajectoryUrl \|\| params.get("autoplay") === "0") return;	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	1103	const result = applyTrajectoryMessage(payload, null);	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	1108	if (!value \|\| typeof value !== "object") return null;	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	1108	if (!value \|\| typeof value !== "object") return null;	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	1109	if (value.type !== "torchrl:mujoco_wasm:setQpos" && value.type !== "torchrl:mujoco_wasm:playTrajectory") return null;	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	1214	if (renderer && camera) renderer.render(rendererScene(), camera);	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	1222	if (!playback.loop && framePosition >= finalIndex) {	CODE
HIGH⚡	torchrl/render/mujoco_wasm.py	1224	trajectoryPlayback = null;	CODE
HIGH	torchrl/render/mujoco_wasm.py	613	if (requestId && callbackUrl) {{	CODE
HIGH	torchrl/render/mujoco_wasm.py	620	if (message.ok === false \|\| done \|\| cancelled) {{	CODE
HIGH	torchrl/render/mujoco_wasm.py	996	visuals.push({geomId, object: mesh});	CODE
HIGH	torchrl/render/mujoco_wasm.py	1017	return null;	CODE
HIGH	torchrl/render/mujoco_wasm.py	1057	if (jointType !== HINGE_JOINT_TYPE && jointType !== SLIDE_JOINT_TYPE) continue;	CODE
HIGH	torchrl/render/mujoco_wasm.py	1162	if (out.length !== model.nq && out.length !== jointControls.length) throw new Error(`qpos must have model.nq (${model.	CODE
HIGH	torchrl/render/mujoco_wasm.py	1181	if (message.dt !== undefined) {	CODE
HIGH	torchrl/render/mujoco_wasm.py	1186	if (message.fps !== undefined) {	CODE
HIGH	torchrl/render/mujoco_wasm.py	1253	trajectoryPlayback = null;	CODE
HIGH	torchrl/render/mujoco_wasm.py	1263	if (!data \|\| homeQpos.length < 3) return;	CODE
HIGH	torchrl/render/mujoco_wasm.py	1336	while (end < bytes.length && bytes[end] !== 0) end++;	CODE
HIGH	torchrl/objectives/dreamer_v3.py	125	+ (1 - alpha) * KL(posterior \|\| sg(prior))	STRING
HIGH	torchrl/objectives/llm/distillation.py	37	``KL(sampling distribution \|\| target distribution)``, and the direction	STRING
HIGH	torchrl/objectives/llm/distillation.py	118	- ``"reverse"`` (default): minimizes ``KL(student \|\| teacher)``. This is	CODE
HIGH	torchrl/objectives/llm/distillation.py	121	- ``"forward"``: minimizes ``KL(teacher \|\| student)``. Here the scored	CODE
HIGH	torchrl/envs/common.py	885	- The number of actions can be undefined, e.g., ``Categorical(n=-1)``;	STRING
HIGH	torchrl/envs/transforms/_reward.py	336	"""Maps the reward to a binary value (0 or 1) if the reward is null or non-null, respectively.	STRING
HIGH	torchrl/envs/custom/chess.py	498	board.push(move)	CODE
HIGH	torchrl/data/tensor_specs.py	3825	n (int): number of possible outcomes. If set to -1, the cardinality of the categorical spec is undefined,	STRING

Modern AI Meta-Vocabulary43 hits · 150 pts

Severity	File	Line	Snippet	Context
MEDIUM	README.md	143	# Nested keys make multi-agent, recurrent, and next-state data explicit.	COMMENT
MEDIUM	test/libs/test_mujoco_playground.py	67	# Cached environment dims for the multi-agent test env, avoiding O(n_tests)	COMMENT
MEDIUM	examples/multiagent/mappo_vmas.py	148	# multi-agent value estimator broadcasts the latter across agents.	COMMENT
MEDIUM	torchrl/envs/utils.py	641	# "multi-agent-emergence-environments": None,	COMMENT
MEDIUM	torchrl/envs/utils.py	642	# OpenAI, https://github.com/openai/multi-agent-emergence-environments	COMMENT
MEDIUM	torchrl/envs/libs/meltingpot.py	312	# Create multi-agent specs	COMMENT
MEDIUM	torchrl/envs/libs/vmas.py	429	# Create multi-agent specs	COMMENT
MEDIUM	torchrl/envs/libs/mujoco_playground.py	220	# Predefined multi-agent partitionings for common MuJoCo locomotion tasks.	COMMENT
MEDIUM	torchrl/envs/model_based/world_model_env.py	229	# done structure (e.g. multi-agent dones) consistent with the spec.	COMMENT
MEDIUM	torchrl/data/postprocs/postprocs.py	45	# Note: rewards could have a different shape than done (e.g. multi-agent with a single	COMMENT
MEDIUM⚡	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	936	# - How to create a competitive multi-group multi-agent environment in TorchRL, how its specs work, and how it integrate	COMMENT
MEDIUM⚡	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	937	# - How to create multi-agent network architectures in TorchRL for multiple groups;	COMMENT
MEDIUM⚡	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	938	# - How we can use :class:`tensordict.TensorDict` to carry multi-agent multi-group data;	COMMENT
MEDIUM⚡	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	939	# - How we can tie all the library components (collectors, modules, replay buffers, and losses) in a multi-agent multi-g	COMMENT
MEDIUM⚡	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	941	# Now that you are proficient with multi-agent DDPG, you can check out all the TorchRL multi-agent implementations in th	COMMENT
MEDIUM	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	63	# This approach has been extended to multi-agent learning in `Multi-Agent Actor-Critic for Mixed Cooperative-Competitive	COMMENT
MEDIUM	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	65	# In multi-agent settings, things are a bit different. We now have multiple policies :math:`\mathbf{\pi}`,	COMMENT
MEDIUM	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	88	# 2. Subsequently, we will construct a multi-agent environment, utilizing TorchRL's	COMMENT
MEDIUM	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	196	# TorchRL API allows integrating various types of multi-agent environment flavors.	COMMENT
MEDIUM	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	311	# (``"done", "terminated", "truncated"``), which do not have a leading multi-agent dimension.	COMMENT
MEDIUM	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	333	# We stress that, in multi-agent contexts, it is paramount to provide explicitly the keys to modify.	COMMENT
MEDIUM	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	612	# **From this point on, the multi-agent-specific components have been instantiated, and we will simply use the same	COMMENT
MEDIUM	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	677	), # We will store up to memory_size multi-agent transitions	CODE
MEDIUM	tutorials/sphinx-tutorials/getting-started-0.py	180	# multiple copies across various processes, a multi-agent environment, or even	COMMENT
MEDIUM	tutorials/sphinx-tutorials/getting-started-0.py	243	# :ref:`multi-agent environment API <MARL-environment-API>`	COMMENT
MEDIUM	tutorials/sphinx-tutorials/getting-started-4.py	20	# The final chapter of this series before we orchestrate everything in a	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	190	# TorchRL API allows integrating various types of multi-agent environment flavors.	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	192	# For more information on how the multi-agent environments API works in TorchRL, you can check out the dedicated	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	205	# The multi-agent task we will solve today is Navigation (see animated figure above).	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	525	# **From this point on, the multi-agent-specific components have been instantiated, and we will simply use the same	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	780	# - How to create a multi-agent environment in TorchRL, how its specs work, and how it integrates with the library;	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	782	# - How to create different multi-agent network architectures in TorchRL (e.g., using parameter sharing, centralised cri	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	783	# - How we can use :class:`tensordict.TensorDict` to carry multi-agent data;	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	784	# - How we can tie all the library components (collectors, modules, replay buffers, and losses) in a multi-agent MAPPO/I	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	786	# Now that you are proficient with multi-agent DDPG, you can check out all the TorchRL multi-agent implementations in th	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	791	# You can also check out our other multi-agent tutorial on how to train competitive	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	794	# If you are interested in creating or wrapping your own multi-agent environments in TorchRL,	COMMENT
MEDIUM	tutorials/sphinx-tutorials/multiagent_ppo.py	76	# In multi-agent settings, things are a bit different. We now have multiple policies :math:`\mathbf{\pi}`,	COMMENT
MEDIUM	tutorials/sphinx-tutorials/multiagent_ppo.py	98	# 2. Next, we will create a vectorized multi-agent environment, using TorchRL's	COMMENT
MEDIUM	tutorials/sphinx-tutorials/multiagent_ppo.py	301	# We stress that, in multi-agent contexts, it is paramount to provide explicitly the keys to modify.	COMMENT
MEDIUM	tutorials/sphinx-tutorials/getting-started-2.py	47	# module orchestrates the various components of your algorithm and	COMMENT
MEDIUM	tutorials/sphinx-tutorials/pendulum.py	289	# it (for example, in multi-agent settings we may want to indicate which agents need	COMMENT
MEDIUM	tutorials/sphinx-tutorials/coding_dqn.py	169	# the training script. The trainer is there to orchestrate everything for you!	COMMENT

Redundant / Tautological Comments83 hits · 120 pts

Severity	File	Line	Snippet	Context
LOW	build_nightly.sh	4	# Check if we're in a nightly build	COMMENT
LOW	packaging/verify_nightly_version.py	15	# Check if __version__ attribute exists	COMMENT
LOW	packaging/verify_nightly_version.py	53	# Check if it's the nightly version by looking at the version	COMMENT
LOW	packaging/verify_nightly_version.py	57	# Check if it's a nightly version (either date format or contains 'd' followed by date)	COMMENT
LOW	packaging/verify_nightly_version.py	68	# Check if it's a stable version that should not be used in nightly builds	COMMENT
LOW⚡	test/test_collectors.py	5270	# Check if there was an exception even with exitcode 0	COMMENT
LOW	test/test_mcts.py	451	): # Check if action_prob_key was also populated	CODE
LOW⚡	test/llm/test_conversions.py	109	# Check if it's a nested tensor by checking if it has the _values attribute	COMMENT
LOW⚡	test/llm/test_conversions.py	126	# Check if it's a nested tensor	COMMENT
LOW	test/llm/test_conversions.py	144	# Check if they're nested tensors	COMMENT
LOW	test/llm/conftest.py	78	# Check if test already has a timeout marker	COMMENT
LOW	test/llm/test_llm_envs.py	100	# Set list to stack for tensordict	COMMENT
LOW	test/objectives/test_loss_module.py	1529	# Assign float to entropy_coeff	COMMENT
LOW	test/transforms/test_module_transforms.py	150	# Check if ray is initialized	COMMENT
LOW	sota-implementations/vla_grpo/openvla_oft/constants.py	90	# Assign constants to global variables	COMMENT
LOW	sota-implementations/dreamer/dreamer.py	462	# Check if profiling is complete and we should exit	COMMENT
LOW	sota-implementations/dreamer/dreamer_utils.py	233	# Check if we should stop profiling	COMMENT
LOW	sota-implementations/grpo/grpo-async.py	272	# Check if num_devices is set	COMMENT
LOW	sota-implementations/grpo/grpo-sync.py	251	# Check if num_devices is set	COMMENT
LOW	…lementations/expert-iteration/expert-iteration-sync.py	407	# Check if num_devices is set	COMMENT
LOW	…ementations/expert-iteration/expert-iteration-async.py	408	# Check if num_devices is set	COMMENT
LOW	sota-check/submitit-release-check.sh	19	# Check if the script is called with --help or without any arguments	COMMENT
LOW⚡	examples/agents/recurrent_actor.py	126	# Print result:	COMMENT
LOW	.github/unittest/linux_libs/scripts_isaaclab/isaac.sh	42	# Check if isaaclab.sh exists	COMMENT
LOW	.github/workflows/auto-tag.yml	100	# Check if title starts with [...]	COMMENT
LOW	.github/workflows/release.yml	151	# Check if it's on a release branch matching the version	COMMENT
LOW	.github/workflows/release.yml	437	# Check if the version folder exists	COMMENT
LOW	.github/workflows/release.yml	480	# Check if changes were made	COMMENT
LOW	.github/workflows/release.yml	546	# Check if release already exists	COMMENT
LOW	.github/workflows/nightly_status_collector.yml	151	# Check if success rate dropped below 70% in last 7 days	COMMENT
LOW	.github/scripts/td_script.sh	16	# Check if ARCH is set to aarch64	COMMENT
LOW	torchrl/_utils.py	57	# Check if user has explicitly set a global start method	COMMENT
LOW	torchrl/weight_update/_rpc.py	256	# Check if all futures are complete	COMMENT
LOW	torchrl/weight_update/_shared.py	211	# Check if per-worker weights dict is provided	COMMENT
LOW	torchrl/weight_update/_shared.py	581	# Check if policy factories are uniform or distinct	COMMENT
LOW	torchrl/weight_update/_shared.py	936	# Check if per-worker weights dict is provided	COMMENT
LOW	torchrl/weight_update/_shared.py	1108	# Check if already shutdown	COMMENT
LOW	torchrl/weight_update/_distributed.py	478	# Check if we have weights to send	COMMENT
LOW	torchrl/weight_update/_distributed.py	549	# Check if already shutdown	COMMENT
LOW	torchrl/weight_update/_distributed.py	716	# Check if all futures are complete	STRING
LOW	torchrl/weight_update/llm/vllm_double_buffer.py	326	# Check if this is an AsyncVLLM instance (uses RPC to coordinate workers)	COMMENT
LOW	torchrl/collectors/_multi_base.py	1232	# Check if any worker has died	COMMENT
LOW	torchrl/collectors/_multi_base.py	1524	# Check if it's an error dict from worker	COMMENT
LOW⚡	torchrl/collectors/_single.py	1089	# Check if policy has meta-device parameters (sent from weight sync schemes)	COMMENT
LOW⚡	torchrl/collectors/_single.py	1226	# Check if we need to cast to env device	COMMENT
LOW	torchrl/collectors/_single.py	1430	# Check if policy has meta-device parameters (not yet initialized)	COMMENT
LOW	torchrl/collectors/llm/weight_update/vllm.py	311	# Check if it's a LoRA model	COMMENT
LOW	torchrl/collectors/llm/weight_update/vllm.py	339	# Check if the model has a LoRA adapter	COMMENT
LOW	torchrl/collectors/llm/weight_update/vllm_v2.py	119	# Check if it's a LoRA model	COMMENT
LOW	torchrl/collectors/llm/weight_update/vllm_v2.py	126	# Check if it's a LoRA model	COMMENT
LOW	torchrl/collectors/llm/weight_update/vllm_v2.py	290	# Check if it's a LoRA model	COMMENT
LOW	torchrl/collectors/llm/weight_update/vllm_v2.py	297	# Check if it's a LoRA model	COMMENT
LOW	torchrl/trainers/trainers.py	1861	# Check if we've reached the target frames	COMMENT
LOW	torchrl/trainers/trainers.py	1953	# Check if enough frames have passed since last logging for this key	COMMENT
LOW	torchrl/modules/llm/backends/vllm/vllm_async.py	254	# Check if it's a list of lists or a single list	COMMENT
LOW	torchrl/modules/llm/backends/vllm/vllm_async.py	832	# Check if this is actually a single prompt represented as token IDs	COMMENT
LOW	torchrl/modules/llm/backends/vllm/vllm_async.py	966	# Check if this is a batch request	COMMENT
LOW	torchrl/modules/llm/backends/vllm/vllm_async.py	1797	# Check if preferred actor is overloaded	COMMENT
LOW	torchrl/modules/llm/backends/vllm/vllm_async.py	2050	# Set tensor_parallel_size to num_devices if not set	COMMENT
LOW	torchrl/modules/llm/backends/sglang/sglang_utils.py	87	# Check if subprocess died	COMMENT
23 more matches not shown…

Structural Annotation Overuse60 hits · 108 pts

Severity	File	Line	Snippet	Context
LOW⚡	test/llm/test_wrapper.py	2389	# Step 1: Generate the canonical response tokens with vLLM.	COMMENT
LOW⚡	test/llm/test_wrapper.py	2395	# Step 2: Extract generated tokens and create new input for log-probs computation	COMMENT
LOW	test/llm/test_wrapper.py	2438	# Step 3: Create log-probs only wrappers	COMMENT
LOW	test/llm/test_wrapper.py	2457	# Step 4: Compute log-probs for the full sequence (original + generated)	COMMENT
LOW⚡	benchmarks/test_storage_write_benchmark.py	142	# Step 1: Stack to contiguous (this allocates new memory)	COMMENT
LOW⚡	benchmarks/test_storage_write_benchmark.py	144	# Step 2: Write to storage	COMMENT
LOW⚡	benchmarks/test_storage_write_benchmark.py	176	# Step 1: Create lazy stack (no allocation, just wrapping)	COMMENT
LOW⚡	benchmarks/test_storage_write_benchmark.py	178	# Step 2: Write to storage (storage handles the lazy stack)	COMMENT
LOW	.github/RELEASE_AGENT_PROMPT.md	60	## Step 1: Analyze Commits for Release Notes	COMMENT
LOW	.github/RELEASE_AGENT_PROMPT.md	167	## Step 2: Draft Release Notes	COMMENT
LOW	.github/RELEASE_AGENT_PROMPT.md	231	## Step 3: Update Version Files	COMMENT
LOW	.github/RELEASE_AGENT_PROMPT.md	270	## Step 4: Commit Version Changes on the main branch (Major releases only)	COMMENT
LOW	.github/RELEASE_AGENT_PROMPT.md	284	## Step 5: Create Release Branch	COMMENT
LOW	.github/RELEASE_AGENT_PROMPT.md	298	## Step 6: Commit Version Changes (Minor releases only)	COMMENT
LOW	.github/RELEASE_AGENT_PROMPT.md	309	## Step 7: Create and Push Tag	COMMENT
LOW	.github/RELEASE_AGENT_PROMPT.md	322	## Step 8: Trigger Release Workflow	COMMENT
LOW	.github/RELEASE_AGENT_PROMPT.md	337	## Step 9: Create Draft GitHub Release	COMMENT
LOW	.github/RELEASE_AGENT_PROMPT.md	350	## Step 10: Monitor Workflow	COMMENT
LOW⚡	torchrl/weight_update/_rpc.py	181	# Step 1: Signal the remote collector via RPC to start receiving (async)	COMMENT
LOW⚡	torchrl/weight_update/_rpc.py	185	# Step 2: Send weights via torch.distributed (blocks until receiver calls recv())	COMMENT
LOW⚡	torchrl/weight_update/_rpc.py	188	# Step 3: Wait for RPC to complete (receiver has applied weights)	COMMENT
LOW⚡	torchrl/weight_update/_rpc.py	207	# Step 1: Signal the remote collector via RPC to start receiving (async)	COMMENT
LOW⚡	torchrl/weight_update/_rpc.py	213	# Step 2: Send weights asynchronously via torch.distributed	COMMENT
LOW⚡	torchrl/weight_update/_ray.py	206	# Step 1: Signal the remote actor via Ray to start receiving (async)	COMMENT
LOW⚡	torchrl/weight_update/_ray.py	214	# Step 2: Send weights via torch.distributed (async)	COMMENT
LOW⚡	torchrl/weight_update/_ray.py	218	# Step 3: Wait for the Ray call to complete (receiver has applied weights)	COMMENT
LOW⚡	torchrl/weight_update/_ray.py	232	# Step 1: Signal the actor via Ray to start receiving (async)	COMMENT
LOW⚡	torchrl/weight_update/_ray.py	242	# Step 2: Send weights via torch.distributed (async)	COMMENT
LOW⚡	torchrl/weight_update/llm/sglang_nccl.py	604	# Step 1: Send a single HTTP request with all weight metadata.	COMMENT
LOW⚡	torchrl/weight_update/llm/sglang_nccl.py	614	# Step 2: Broadcast each weight tensor via NCCL in the same order.	COMMENT
LOW⚡	torchrl/weight_update/llm/sglang_nccl.py	623	# Step 3: Wait for the HTTP response confirming server received all weights	COMMENT
LOW⚡	torchrl/modules/llm/backends/vllm/vllm_async.py	1190	# Step 1: Start trainer NCCL group in a background thread — it blocks	COMMENT
LOW⚡	torchrl/modules/llm/backends/vllm/vllm_async.py	1204	# Step 2: Dispatch init_weight_transfer_engine to all vLLM actors	COMMENT
LOW⚡	torchrl/modules/llm/backends/vllm/vllm_async.py	1220	# Step 3: Wait for both sides to complete	COMMENT
LOW⚡	torchrl/modules/llm/backends/vllm/vllm_async.py	1293	# Step 1: Tell all actors to start receiving weights	COMMENT
LOW⚡	torchrl/modules/llm/backends/vllm/vllm_async.py	1298	# Step 2: Send weights from trainer side	COMMENT
LOW	torchrl/modules/llm/backends/vllm/vllm_async.py	1317	# Step 3: Wait for all actors to finish receiving	COMMENT
LOW⚡	torchrl/modules/llm/backends/sglang/sglang_server.py	707	# Step 1: Signal server to expect this weight	COMMENT
LOW⚡	torchrl/modules/llm/backends/sglang/sglang_server.py	712	# Step 2: Broadcast the weight via NCCL	COMMENT
LOW⚡	tutorials/sphinx-tutorials/llm_browser.py	81	# Step 1: Basic Environment Configuration	COMMENT
LOW⚡	tutorials/sphinx-tutorials/llm_browser.py	189	# Step 2: Tool Execution Helper	COMMENT
LOW⚡	tutorials/sphinx-tutorials/llm_browser.py	262	# Step 3: Starting the Interaction	COMMENT
LOW⚡	tutorials/sphinx-tutorials/llm_browser.py	292	# Step 4: Performing the Search	COMMENT
LOW⚡	tutorials/sphinx-tutorials/llm_browser.py	321	# Step 5: Providing the Answer	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	50	# Step 1: build and populate a replay buffer	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	77	# Step 2: wrap the buffer as an IterableDataset	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	109	# Step 3: key filtering	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	124	# Step 4: device placement	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	137	# Step 5: nested keys	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	181	# Step 1: define a stand-in reward model	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	216	# Step 2: wrap with default TorchRL token keys	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	234	# Step 3: run a batch through the wrapper	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	256	# Step 4: custom keys and nested reward key	COMMENT
LOW	knowledge_base/FLAKY_TESTS.md	42	## Step 1: Gather Flaky Test Information	COMMENT
LOW	knowledge_base/FLAKY_TESTS.md	96	## Step 2: Identify Non-Deterministic Failure Causes	COMMENT
LOW	knowledge_base/FLAKY_TESTS.md	131	## Step 3: Implement the Fix	COMMENT
LOW	knowledge_base/FLAKY_TESTS.md	184	## Step 4: Create a PR with the Fix AND Test Validation	COMMENT
LOW	knowledge_base/FLAKY_TESTS.md	274	## Step 5: Monitor the CI Run	COMMENT
LOW	knowledge_base/FLAKY_TESTS.md	302	## Step 6: Verify the Fix	COMMENT
LOW	knowledge_base/FLAKY_TESTS.md	328	## Step 7: Cleanup Before Merging	COMMENT

AI Slop Vocabulary44 hits · 101 pts

Severity	File	Line	Snippet	Context
LOW	test/llm/test_llm_collectors.py	547	# We have the tokens, let's just use them	COMMENT
LOW	sota-implementations/expert-iteration/ei_utils.py	88	# we can just use 0 here.	COMMENT
LOW	…lementations/expert-iteration/expert-iteration-sync.py	433	# we can just set a big number, the buffer will be emptied anyway	COMMENT
MEDIUM	examples/agents/recurrent_actor.py	182	# its recurrent connections to maintain state across time steps. This mode may utilize CuDNN to accelerate the processin	COMMENT
MEDIUM	.github/scripts/compare_pr_benchmarks.py	45	"""Return a rate based on the robust per-round duration statistic."""	STRING
LOW	torchrl/weight_update/_shared.py	909	# If no shared transport, just return the fresh weights	COMMENT
MEDIUM	torchrl/collectors/_single.py	1790	# this way of checking cuda is robust to lazy stacks with mismatching shapes	COMMENT
MEDIUM	torchrl/objectives/common.py	492	# To make it robust to device casting, we must register list of	COMMENT
LOW	torchrl/objectives/llm/grpo.py	605	# TODO: eventually, we want to always use `get_dist` and just pass the key of the mask	COMMENT
MEDIUM	torchrl/objectives/llm/grpo.py	1512	# Cat is the most robust way to combine the trajs	COMMENT
MEDIUM	torchrl/trainers/algorithms/on_policy.py	318	# Set up comprehensive logging for on-policy training	COMMENT
MEDIUM	torchrl/trainers/algorithms/sac.py	258	# Set up comprehensive logging for SAC training	COMMENT
LOW	torchrl/modules/llm/policies/common.py	747	# For log-prob distributions, we just return the pre-computed log-probs	COMMENT
MEDIUM	torchrl/modules/llm/policies/transformers_wrapper.py	1042	# We are going to map this tokens_in to a tensordict to facilitate the padding in case we need it	COMMENT
MEDIUM	torchrl/modules/llm/policies/transformers_wrapper.py	1465	# We are going to map this tokens_in to a tensordict to facilitate the padding in case we need it	COMMENT
LOW	torchrl/modules/mcts/scores.py	425	# Don't update weights for zero probability - just return	COMMENT
LOW	torchrl/envs/utils.py	1489	# case we just return the data	COMMENT
MEDIUM	torchrl/envs/utils.py	200	"""Represents the keys as a tree to facilitate iteration."""	STRING
LOW	torchrl/envs/libs/dm_control.py	318	# just use a common EGL_DEVICE_ID environment variable for all processes.	COMMENT
MEDIUM	torchrl/envs/transforms/_normalization.py	711	# to facilitate the computation of the stats using TD internals.	COMMENT
MEDIUM	torchrl/envs/transforms/_misc.py	75	"""Calls pin_memory on the tensordict to facilitate writing on CUDA devices."""	STRING
MEDIUM	torchrl/data/llm/topk.py	122	# Cat is the most robust way to combine the trajs	COMMENT
MEDIUM	torchrl/data/replay_buffers/utils.py	1200	# We currently ignore out in this call but we should leverage that	COMMENT
LOW	torchrl/data/replay_buffers/utils.py	1379	# if Tensor, we just create a MemoryMappedTensor of the desired shape, device and dtype	COMMENT
LOW	torchrl/data/replay_buffers/utils.py	1396	# if Tensor, we just create a MemoryMappedTensor of the desired shape, device and dtype	COMMENT
LOW	torchrl/data/replay_buffers/storages/tensor.py	1097	# if Tensor, we just create a MemoryMappedTensor of the desired shape, device and dtype	COMMENT
MEDIUM	tutorials/sphinx-tutorials/dqn_with_rnn.py	222	# values for these attributes to facilitate its construction.	COMMENT
MEDIUM	tutorials/sphinx-tutorials/torchrl_envs.py	652	# One can therefore think carefully about the transform design to leverage the	COMMENT
LOW	tutorials/sphinx-tutorials/torchrl_envs.py	341	# To compose transforms, simply use the ``Compose`` class:	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/getting-started-1.py	26	# Accommodating all these scenarios can be quite intricate. In this succinct	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/getting-started-1.py	27	# tutorial, we will delve into the core functionality of TorchRL in terms of	COMMENT
MEDIUM	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	337	# Essentially we just say that the	COMMENT
LOW	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	612	# **From this point on, the multi-agent-specific components have been instantiated, and we will simply use the same	COMMENT
MEDIUM	tutorials/sphinx-tutorials/getting-started-0.py	79	# Essentially, :class:`~tensordict.TensorDict` is a generic key-based data	COMMENT
MEDIUM	tutorials/sphinx-tutorials/getting-started-3.py	144	# Data collectors are very useful when it comes to coding state-of-the-art	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	200	# This allows leveraging the Single Instruction Multiple Data (SIMD) paradigm of GPUs and significantly	COMMENT
MEDIUM	tutorials/sphinx-tutorials/multiagent_ppo.py	224	# We will also use ``num_vmas_envs`` vectorized environments, to leverage batch simulation.	COMMENT
LOW⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	525	# **From this point on, the multi-agent-specific components have been instantiated, and we will simply use the same	COMMENT
LOW	tutorials/sphinx-tutorials/rb_tutorial.py	387	# To enable multithreaded sampling, just pass a positive integer to the	COMMENT
MEDIUM	tutorials/sphinx-tutorials/coding_ddpg.py	574	# To leverage the vectorization capabilities of PyTorch, we adopt	COMMENT
LOW	tutorials/sphinx-tutorials/coding_ddpg.py	181	# accessible but this will just return a detached version of the	COMMENT
LOW	tutorials/sphinx-tutorials/coding_ddpg.py	427	# are collected by :func:`env.step()`, simply pass the ``from_pixels=True``	COMMENT
MEDIUM	tutorials/sphinx-tutorials/pendulum.py	167	# To facilitate the reading and writing from that ``tensordict`` and to make sure	COMMENT
MEDIUM	tutorials/sphinx-tutorials/coding_dqn.py	67	# past, we can still pass it in conjunction with the various actions available	COMMENT

Fake / Example Data73 hits · 88 pts

Severity	File	Line	Snippet	Context
LOW	README.md	217	processes, weight updaters, evaluator loops, profiling hooks, and fake-data	CODE
LOW	test/llm/test_llm_objectives.py	43	"Lorem ipsum dolor sit amet,",	CODE
LOW	test/llm/test_llm_objectives.py	43	"Lorem ipsum dolor sit amet,",	CODE
LOW⚡	test/objectives/test_dreamer.py	421	loss_td, fake_data = loss_module(tensordict.reshape(-1))	CODE
LOW⚡	test/objectives/test_dreamer.py	422	assert not fake_data.requires_grad	CODE
LOW⚡	test/objectives/test_dreamer.py	423	assert fake_data.shape == torch.Size([tensordict.numel(), imagination_horizon])	CODE
LOW	test/objectives/test_dreamer.py	450	loss_td, fake_data = loss_module(tensordict)	CODE
LOW	test/objectives/test_dreamer.py	452	assert not fake_data.requires_grad	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	677	loss_td, fake_data = loss_module(	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	683	fake_data["discount_weight"][0, :, 0], expected_weight	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	686	fake_data["next", "continuation"],	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	687	torch.full_like(fake_data["next", "continuation"], 0.5),	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	689	assert not fake_data["discount_weight"].requires_grad	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	695	actor_inputs = fake_data.select(*actor_model.in_keys, strict=False).detach()	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	697	log_prob = distribution.log_prob(fake_data["action"].detach())	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	698	log_prob = _match_trailing_dim(log_prob, fake_data["lambda_target"])	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	699	baseline_td = fake_data.select(*value_model.in_keys, strict=False)	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	701	advantage = (fake_data["lambda_target"] - baseline_td["state_value"]).detach()	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	702	expected_loss = -(fake_data["discount_weight"] * log_prob * advantage).mean()	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	712	value_loss(fake_data.detach())	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	1481	loss_td, fake_data = loss_module(	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	1484	baseline_td = fake_data.select(*value_model.in_keys, strict=False)	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	1486	advantage = (fake_data["lambda_target"] - baseline_td["state_value"]).detach()	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	1488	fake_data["action_log_prob"], fake_data["lambda_target"]	CODE
LOW⚡	test/objectives/test_dreamer_v3.py	1498	compiled_scale(fake_data["lambda_target"]),	CODE
LOW	test/objectives/test_dreamer_v3.py	633	loss_td, fake_data = loss_module(tensordict.reshape(-1))	CODE
LOW	test/objectives/test_dreamer_v3.py	851	loss_td, fake_data = actor_loss(	CODE
LOW	test/objectives/test_dreamer_v3.py	855	assert fake_data["lambda_target"].shape[-1] == 1	CODE
LOW	test/objectives/test_dreamer_v3.py	1521	loss_td, fake_data = loss_module(	CODE
LOW	test/objectives/test_dreamer_v3.py	1524	expected = -(fake_data["lambda_target"] / 10.0).mean()	CODE
LOW	test/objectives/test_dreamer_v3.py	1543	loss_td, fake_data = loss_module(	CODE
LOW	test/objectives/test_dreamer_v3.py	1547	fake_data["lambda_target"].detach(),	CODE
LOW	test/objectives/test_dreamer_v3.py	1568	loss_td, fake_data = loss_module(	CODE
LOW	test/objectives/test_dreamer_v3.py	1572	fake_data["lambda_target"].detach(),	CODE
LOW⚡	test/transforms/test_module_transforms.py	210	dummy_data = TensorDict(observation=torch.zeros(2, 3), batch_size=[2])	CODE
LOW⚡	test/transforms/test_module_transforms.py	213	assert (module(dummy_data)["action"] == 2).all()	CODE
LOW⚡	test/transforms/test_module_transforms.py	221	assert (module(dummy_data)["action"] == 1).all()	CODE
LOW⚡	sota-implementations/dreamer_v3/train.py	542	actor_loss_td, fake_data = actor_loss(actor_input)	CODE
LOW⚡	sota-implementations/dreamer_v3/train.py	543	value_loss_td, _ = value_loss(fake_data.detach())	CODE
LOW⚡	sota-implementations/dreamer_v3/train.py	549	"bootstrap": fake_data.get("lambda_target")[..., 0, 0].reshape(	CODE
LOW⚡	benchmarks/test_llm.py	64	"Lorem ipsum dolor sit amet",	CODE
LOW⚡	benchmarks/test_llm.py	64	"Lorem ipsum dolor sit amet",	CODE
LOW⚡	benchmarks/test_llm.py	72	"Lorem ipsum dolor sit amet",	CODE
LOW⚡	benchmarks/test_llm.py	72	"Lorem ipsum dolor sit amet",	CODE
LOW⚡	torchrl/objectives/dreamer.py	299	fake_data = self.model_based_env.rollout(	CODE
LOW⚡	torchrl/objectives/dreamer.py	305	next_tensordict = step_mdp(fake_data, keep_other=True)	CODE
LOW⚡	torchrl/objectives/dreamer.py	309	reward = fake_data.get(("next", self.tensor_keys.reward))	CODE
LOW⚡	torchrl/objectives/dreamer.py	312	fake_data.set("lambda_target", lambda_target)	CODE
LOW	torchrl/objectives/dreamer.py	325	return loss_tensordict, fake_data.data	CODE
LOW⚡	torchrl/objectives/dreamer.py	427	def forward(self, fake_data) -> torch.Tensor:	CODE
LOW⚡	torchrl/objectives/dreamer.py	428	lambda_target = fake_data.get("lambda_target")	CODE
LOW⚡	torchrl/objectives/dreamer.py	430	tensordict_select = fake_data.select(*self.value_model.in_keys, strict=False)	CODE
LOW	torchrl/objectives/dreamer.py	463	self._clear_weakrefs(fake_data, loss_tensordict)	CODE
LOW	torchrl/objectives/dreamer.py	465	return loss_tensordict, fake_data	CODE
LOW⚡	torchrl/objectives/dreamer_v3.py	755	fake_data = self.model_based_env.rollout(	CODE
LOW⚡	torchrl/objectives/dreamer_v3.py	761	next_tensordict = step_mdp(fake_data, keep_other=True)	CODE
LOW⚡	torchrl/objectives/dreamer_v3.py	766	reward = fake_data.get(("next", self.tensor_keys.reward))	CODE
LOW⚡	torchrl/objectives/dreamer_v3.py	772	first_td = fake_data[..., :1].select(	CODE
LOW⚡	torchrl/objectives/dreamer_v3.py	789	fake_data.set(self.tensor_keys.continuation, root_continuation)	CODE
LOW⚡	torchrl/objectives/dreamer_v3.py	790	fake_data.set(("next", self.tensor_keys.continuation), continuation)	CODE
13 more matches not shown…

Verbosity Indicators44 hits · 85 pts

Severity	File	Line	Snippet	Context
LOW⚡	test/llm/test_wrapper.py	2389	# Step 1: Generate the canonical response tokens with vLLM.	COMMENT
LOW⚡	test/llm/test_wrapper.py	2395	# Step 2: Extract generated tokens and create new input for log-probs computation	COMMENT
LOW	test/llm/test_wrapper.py	2438	# Step 3: Create log-probs only wrappers	COMMENT
LOW	test/llm/test_wrapper.py	2457	# Step 4: Compute log-probs for the full sequence (original + generated)	COMMENT
LOW⚡	benchmarks/test_storage_write_benchmark.py	142	# Step 1: Stack to contiguous (this allocates new memory)	COMMENT
LOW⚡	benchmarks/test_storage_write_benchmark.py	144	# Step 2: Write to storage	COMMENT
LOW⚡	benchmarks/test_storage_write_benchmark.py	176	# Step 1: Create lazy stack (no allocation, just wrapping)	COMMENT
LOW⚡	benchmarks/test_storage_write_benchmark.py	178	# Step 2: Write to storage (storage handles the lazy stack)	COMMENT
LOW⚡	torchrl/weight_update/_rpc.py	181	# Step 1: Signal the remote collector via RPC to start receiving (async)	COMMENT
LOW⚡	torchrl/weight_update/_rpc.py	185	# Step 2: Send weights via torch.distributed (blocks until receiver calls recv())	COMMENT
LOW⚡	torchrl/weight_update/_rpc.py	188	# Step 3: Wait for RPC to complete (receiver has applied weights)	COMMENT
LOW⚡	torchrl/weight_update/_rpc.py	207	# Step 1: Signal the remote collector via RPC to start receiving (async)	COMMENT
LOW⚡	torchrl/weight_update/_rpc.py	213	# Step 2: Send weights asynchronously via torch.distributed	COMMENT
LOW⚡	torchrl/weight_update/_ray.py	206	# Step 1: Signal the remote actor via Ray to start receiving (async)	COMMENT
LOW⚡	torchrl/weight_update/_ray.py	214	# Step 2: Send weights via torch.distributed (async)	COMMENT
LOW⚡	torchrl/weight_update/_ray.py	218	# Step 3: Wait for the Ray call to complete (receiver has applied weights)	COMMENT
LOW⚡	torchrl/weight_update/_ray.py	232	# Step 1: Signal the actor via Ray to start receiving (async)	COMMENT
LOW⚡	torchrl/weight_update/_ray.py	242	# Step 2: Send weights via torch.distributed (async)	COMMENT
LOW⚡	torchrl/weight_update/llm/sglang_nccl.py	604	# Step 1: Send a single HTTP request with all weight metadata.	COMMENT
LOW⚡	torchrl/weight_update/llm/sglang_nccl.py	614	# Step 2: Broadcast each weight tensor via NCCL in the same order.	COMMENT
LOW⚡	torchrl/weight_update/llm/sglang_nccl.py	623	# Step 3: Wait for the HTTP response confirming server received all weights	COMMENT
LOW⚡	torchrl/modules/llm/backends/vllm/vllm_async.py	1190	# Step 1: Start trainer NCCL group in a background thread — it blocks	COMMENT
LOW⚡	torchrl/modules/llm/backends/vllm/vllm_async.py	1204	# Step 2: Dispatch init_weight_transfer_engine to all vLLM actors	COMMENT
LOW⚡	torchrl/modules/llm/backends/vllm/vllm_async.py	1220	# Step 3: Wait for both sides to complete	COMMENT
LOW⚡	torchrl/modules/llm/backends/vllm/vllm_async.py	1293	# Step 1: Tell all actors to start receiving weights	COMMENT
LOW⚡	torchrl/modules/llm/backends/vllm/vllm_async.py	1298	# Step 2: Send weights from trainer side	COMMENT
LOW	torchrl/modules/llm/backends/vllm/vllm_async.py	1317	# Step 3: Wait for all actors to finish receiving	COMMENT
LOW⚡	torchrl/modules/llm/backends/sglang/sglang_server.py	707	# Step 1: Signal server to expect this weight	COMMENT
LOW⚡	torchrl/modules/llm/backends/sglang/sglang_server.py	712	# Step 2: Broadcast the weight via NCCL	COMMENT
LOW	torchrl/modules/distributions/discrete.py	844	# For token-level masking, we need to check if specific tokens are masked	COMMENT
LOW⚡	tutorials/sphinx-tutorials/llm_browser.py	81	# Step 1: Basic Environment Configuration	COMMENT
LOW⚡	tutorials/sphinx-tutorials/llm_browser.py	189	# Step 2: Tool Execution Helper	COMMENT
LOW⚡	tutorials/sphinx-tutorials/llm_browser.py	262	# Step 3: Starting the Interaction	COMMENT
LOW⚡	tutorials/sphinx-tutorials/llm_browser.py	292	# Step 4: Performing the Search	COMMENT
LOW⚡	tutorials/sphinx-tutorials/llm_browser.py	321	# Step 5: Providing the Answer	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	50	# Step 1: build and populate a replay buffer	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	77	# Step 2: wrap the buffer as an IterableDataset	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	109	# Step 3: key filtering	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	124	# Step 4: device placement	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	137	# Step 5: nested keys	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	181	# Step 1: define a stand-in reward model	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	216	# Step 2: wrap with default TorchRL token keys	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	234	# Step 3: run a batch through the wrapper	COMMENT
LOW⚡	tutorials/sphinx-tutorials/trl_interop.py	256	# Step 4: custom keys and nested reward key	COMMENT

Slop Phrases18 hits · 43 pts

Severity	File	Line	Snippet	Context
LOW	torchrl/envs/transforms/_base.py	211	- Don't forget to edits the specs if needed: top level: :meth:`~torchrl.envs.Transform.transform_output_spec`,	STRING
MEDIUM	torchrl/data/llm/history.py	316	>>> # Now you can use it with auto-detection	CODE
MEDIUM	tutorials/sphinx-tutorials/dqn_with_rnn.py	259	# As mentioned earlier, we have one more optional transform to add to our	COMMENT
LOW	tutorials/sphinx-tutorials/dqn_with_rnn.py	61	# If you are running this in Google Colab, make sure you install the following dependencies:	COMMENT
LOW	tutorials/sphinx-tutorials/torchrl_envs.py	277	# To run this part of the tutorial, make sure you have installed dm_control:	COMMENT
MEDIUM	tutorials/sphinx-tutorials/getting-started-1.py	310	# - Have a look at how you can use an RNN within the policy (a	COMMENT
MEDIUM⚡	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	948	# Finally, you can modify the parameters of this tutorial to try many other configurations and scenarios	COMMENT
MEDIUM⚡	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	952	# Here are a few videos of some possible scenarios you can try in VMAS.	COMMENT
MEDIUM	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	663	# To store your buffer on the GPU, you can use :class:`~torchrl.data.replay_buffers.LazyTensorStorage`, passing the desi	COMMENT
LOW	…orials/sphinx-tutorials/multiagent_competitive_ddpg.py	43	# If you are running this in Google Colab, make sure you install the following dependencies:	COMMENT
MEDIUM	tutorials/sphinx-tutorials/getting-started-4.py	31	# portable CSV logger (:class:`~torchrl.record.loggers.csv.CSVLogger`) that you can use	COMMENT
MEDIUM	tutorials/sphinx-tutorials/export.py	335	# Here's an example of how you can use AOTInductor to export your policy, inspired by the	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	798	# Finally, you can modify the parameters of this tutorial to try many other configurations and scenarios	COMMENT
MEDIUM⚡	tutorials/sphinx-tutorials/multiagent_ppo.py	800	# Here are a few videos of some possible scenarios you can try in VMAS.	COMMENT
LOW	tutorials/sphinx-tutorials/multiagent_ppo.py	43	# If you are running this in Google Colab, make sure you install the following dependencies:	COMMENT
LOW	tutorials/sphinx-tutorials/coding_ddpg.py	283	# When computing this value, we must make sure to take the value parameters out	COMMENT
LOW	tutorials/sphinx-tutorials/getting-started-2.py	111	# loss. If you want to make sure you're fully in control of what is happening,	COMMENT
LOW	tutorials/sphinx-tutorials/coding_ppo.py	39	# If you are running this in Google Colab, make sure you install the following dependencies:	COMMENT

Hallucination Indicators3 hits · 40 pts

Severity	File	Line	Snippet	Context
CRITICAL⚡	test/test_render.py	1304	env.base_env.unwrapped.data.qpos.copy(),	CODE
CRITICAL⚡	test/test_render.py	1310	env.base_env.unwrapped.data.qpos.copy(),	CODE
CRITICAL	torchrl/envs/libs/gym.py	800	return gym.envs.registration.registry.env_specs.keys()	CODE

Dead Code18 hits · 36 pts

Severity	File	Line	Context
MEDIUM	test/llm/test_data.py	889	CODE
MEDIUM	test/llm/test_data.py	892	CODE
MEDIUM	test/objectives/test_bc.py	41	CODE
MEDIUM	test/objectives/test_bc.py	42	CODE
MEDIUM	torchrl/envs/common.py	1069	CODE
MEDIUM	torchrl/envs/transforms/_normalization.py	912	CODE
MEDIUM	torchrl/envs/transforms/_normalization.py	913	CODE
MEDIUM	torchrl/envs/transforms/_normalization.py	915	CODE
MEDIUM	torchrl/envs/transforms/_normalization.py	916	CODE
MEDIUM	torchrl/envs/transforms/_normalization.py	917	CODE
MEDIUM	torchrl/envs/transforms/_normalization.py	918	CODE
MEDIUM	torchrl/envs/transforms/_normalization.py	919	CODE
MEDIUM	torchrl/envs/transforms/_normalization.py	924	CODE
MEDIUM	torchrl/envs/transforms/_normalization.py	925	CODE
MEDIUM	torchrl/envs/transforms/_normalization.py	937	CODE
MEDIUM	torchrl/envs/transforms/_normalization.py	938	CODE
MEDIUM	torchrl/envs/transforms/_normalization.py	939	CODE
MEDIUM	torchrl/envs/transforms/_normalization.py	941	CODE

AI Response Leakage3 hits · 22 pts

Severity	File	Line	Snippet	Context
HIGH	tutorials/sphinx-tutorials/rb_tutorial.py	807	# ``("collector", "traj_ids")``. In this example, we construct data	COMMENT
HIGH	tutorials/sphinx-tutorials/pendulum.py	304	# In this example, we do not pass a done state as this is not mandatory	COMMENT
HIGH	tutorials/sphinx-tutorials/pendulum.py	850	# In this example, we will train a simple policy using the reward as a	COMMENT

Example Usage Blocks5 hits · 8 pts

Severity	File	Line	Snippet	Context
LOW	docs/build_local.sh	4	# Usage:	COMMENT
LOW	…b/unittest/helpers/assert_torch_tensordict_versions.sh	4	# Usage:	COMMENT
LOW	.github/unittest/helpers/assert_torch_version.sh	4	# Usage:	COMMENT
LOW	.github/workflows/auto-tag.yml	3	# Usage:	COMMENT
LOW	.github/workflows/assign.yml	3	# Usage:	COMMENT

TODO Padding3 hits · 4 pts

Severity	File	Line	Snippet	Context
LOW	torchrl/envs/llm/libs/mlgym.py	228	# TODO: implement add commands method in environment	COMMENT
LOW	torchrl/data/map/tree.py	228	# TODO: implement specs	COMMENT
LOW	torchrl/data/map/tree.py	245	# TODO: implement specs	COMMENT

Overly Generic Function Names5 hits · 2 pts

Severity	File	Line	Snippet	Context
LOW	test/test_docstring_args_checker.py	31	def helper(first, second):	STRING
LOW	test/test_docstring_args_checker.py	71	def helper(first, second, *, mode="strict", retries=0):	STRING
LOW	test/test_docstring_args_checker.py	93	def helper(first, second, *, mode="strict", retries=0):	STRING
LOW	torchrl/_utils.py	246	>>> def my_function():	STRING
LOW	torchrl/_utils.py	1319	... def my_function():	STRING

Analysis Overview

What These Metrics Mean

Score History

Severity Breakdown

Directory Score Breakdown

Pattern Findings