Sourcery refactored master branch #1
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
![sourcery-ai[bot]](https://avatars.githubusercontent.com/in/48477?s=60&v=4){kind=small}
| if inp.sum() > 0.: | ||
| output = self.weight + inp | ||
| else: | ||
| output = self.weight - inp | ||
| return output | ||
| return self.weight + inp if inp.sum() > 0. else self.weight - inp |
There was a problem hiding this comment.
Function SimpleIf.forward refactored with the following changes:
- Inline variable that is immediately returned (
inline-immediately-returned-variable) - Replace if statement with if expression (
assign-if-exp)
Comment on lines
-28
to
+26
| if inp.mean() > 0.: | ||
| output = self.weight + inp | ||
| else: | ||
| output = self.weight - inp | ||
| return self.weight + inp if inp.mean() > 0. else self.weight - inp | ||
| else: | ||
| if inp.mean() > 0.: | ||
| output = self.weight * inp | ||
| else: | ||
| output = self.weight / inp | ||
|
|
||
| return output | ||
| return self.weight * inp if inp.mean() > 0. else self.weight / inp |
There was a problem hiding this comment.
Function NestedIf.forward refactored with the following changes:
- Lift return into if (
lift-return-into-if) - Replace if statement with if expression [×2] (
assign-if-exp)
Comment on lines
-54
to
+42
| for i in range(inp.size(0)): | ||
| for _ in range(a.size(0)): |
There was a problem hiding this comment.
Function SimpleLoop.forward refactored with the following changes:
- Use previously assigned local variable (
use-assigned-variable) - Replace unused for index with underscore (
for-index-underscore)
Comment on lines
-64
to
+52
| for i in range(inp.size(0)): | ||
| for _ in range(a.size(0)): |
There was a problem hiding this comment.
Function LoopWithIf.forward refactored with the following changes:
- Use previously assigned local variable (
use-assigned-variable) - Replace unused for index with underscore (
for-index-underscore)
Comment on lines
-98
to
+86
| while i < inp.size(0): | ||
| while i < a.size(0): |
There was a problem hiding this comment.
Function SimpleWhileLoop.forward refactored with the following changes:
- Use previously assigned local variable (
use-assigned-variable)
Comment on lines
-173
to
-189
| if True: | ||
| with auto_scheduler.ApplyHistoryBest("logs/maskrcnn_rtx3070.log"): | ||
| with tvm.transform.PassContext(opt_level=3, config={"relay.backend.use_auto_scheduler": True}): | ||
| desired_layouts = {'nn.conv2d': ['NHWC', 'default'], "vision.roi_align": ["NHWC", "default"]} | ||
| seq = tvm.transform.Sequential([relay.transform.ConvertLayout(desired_layouts)]) | ||
| mod = seq(mod) | ||
| vm_exec = relay.vm.compile(mod, target=target, params=params) | ||
| else: | ||
| # with auto_scheduler.ApplyHistoryBest("logs/maskrcnn_nvptx.log"): | ||
| # # with auto_scheduler.ApplyHistoryBest("logs/maskrcnn_cuda.log"): | ||
| # with tvm.transform.PassContext(opt_level=3, config={"relay.backend.use_auto_scheduler": True}): | ||
| with tvm.transform.PassContext(opt_level=3): | ||
| # desired_layouts = {'nn.conv2d': ['NHWC', 'default'], "vision.roi_align": ["NHWC", "default"]} | ||
| # seq = tvm.transform.Sequential([relay.transform.ConvertLayout(desired_layouts)]) | ||
| # mod = seq(mod) | ||
| with auto_scheduler.ApplyHistoryBest("logs/maskrcnn_rtx3070.log"): | ||
| with tvm.transform.PassContext(opt_level=3, config={"relay.backend.use_auto_scheduler": True}): | ||
| desired_layouts = {'nn.conv2d': ['NHWC', 'default'], "vision.roi_align": ["NHWC", "default"]} | ||
| seq = tvm.transform.Sequential([relay.transform.ConvertLayout(desired_layouts)]) | ||
| mod = seq(mod) | ||
| vm_exec = relay.vm.compile(mod, target=target, params=params) | ||
|
|
There was a problem hiding this comment.
Function bench_tvm refactored with the following changes:
- Remove redundant conditional (
remove-redundant-if)
This removes the following comments ( why? ):
# desired_layouts = {'nn.conv2d': ['NHWC', 'default'], "vision.roi_align": ["NHWC", "default"]}
# seq = tvm.transform.Sequential([relay.transform.ConvertLayout(desired_layouts)])
# with auto_scheduler.ApplyHistoryBest("logs/maskrcnn_nvptx.log"):
# # with auto_scheduler.ApplyHistoryBest("logs/maskrcnn_cuda.log"):
# mod = seq(mod)
# with tvm.transform.PassContext(opt_level=3, config={"relay.backend.use_auto_scheduler": True}):
| rois = torch.cat([ids, concat_boxes], dim=1) | ||
| return rois | ||
| return torch.cat([ids, concat_boxes], dim=1) |
There was a problem hiding this comment.
Function MultiScaleRoIAlign.convert_to_roi_format refactored with the following changes:
- Inline variable that is immediately returned (
inline-immediately-returned-variable)
Comment on lines
-144
to
+143
| assert len(image_shapes) != 0 | ||
| assert image_shapes |
There was a problem hiding this comment.
Function MultiScaleRoIAlign.setup_scales refactored with the following changes:
- Simplify sequence length comparison (
simplify-len-comparison)
Comment on lines
-179
to
+178
| x_filtered = [] | ||
| for k, v in x.items(): | ||
| if k in self.featmap_names: | ||
| x_filtered.append(v) | ||
| x_filtered = [v for k, v in x.items() if k in self.featmap_names] |
There was a problem hiding this comment.
Function MultiScaleRoIAlign.forward refactored with the following changes:
- Convert for loop into list comprehension (
list-comprehension)
Comment on lines
-128
to
-131
| mask_loss = F.binary_cross_entropy_with_logits( | ||
| mask_logits[torch.arange(labels.shape[0], device=labels.device), labels], mask_targets | ||
| return F.binary_cross_entropy_with_logits( | ||
| mask_logits[ | ||
| torch.arange(labels.shape[0], device=labels.device), labels | ||
| ], | ||
| mask_targets, | ||
| ) | ||
| return mask_loss |
There was a problem hiding this comment.
Function maskrcnn_loss refactored with the following changes:
- Inline variable that is immediately returned (
inline-immediately-returned-variable)
Comment on lines
-308
to
+310
| keypoint_loss = F.cross_entropy(keypoint_logits[valid], keypoint_targets[valid]) | ||
| return keypoint_loss | ||
| return F.cross_entropy(keypoint_logits[valid], keypoint_targets[valid]) |
There was a problem hiding this comment.
Function keypointrcnn_loss refactored with the following changes:
- Inline variable that is immediately returned (
inline-immediately-returned-variable)
Comment on lines
-342
to
+343
| boxes_exp = torch.stack((boxes_exp0, boxes_exp1, boxes_exp2, boxes_exp3), 1) | ||
| return boxes_exp | ||
| return torch.stack((boxes_exp0, boxes_exp1, boxes_exp2, boxes_exp3), 1) |
There was a problem hiding this comment.
Function _onnx_expand_boxes refactored with the following changes:
- Inline variable that is immediately returned (
inline-immediately-returned-variable)
| im_mask = torch.cat((zeros_x0, | ||
| concat_0, | ||
| zeros_x1), 1)[:, :im_w] | ||
| return im_mask | ||
| return torch.cat((zeros_x0, concat_0, zeros_x1), 1)[:, :im_w] |
There was a problem hiding this comment.
Function _onnx_paste_mask_in_image refactored with the following changes:
- Inline variable that is immediately returned (
inline-immediately-returned-variable)
Comment on lines
-483
to
+484
| if len(res) > 0: | ||
| ret = torch.stack(res, dim=0)[:, None] | ||
| else: | ||
| ret = masks.new_empty((0, 1, im_h, im_w)) | ||
| return ret | ||
| return ( | ||
| torch.stack(res, dim=0)[:, None] | ||
| if res | ||
| else masks.new_empty((0, 1, im_h, im_w)) | ||
| ) |
There was a problem hiding this comment.
Function paste_masks_in_image refactored with the following changes:
- Inline variable that is immediately returned (
inline-immediately-returned-variable) - Replace if statement with if expression (
assign-if-exp) - Simplify sequence length comparison (
simplify-len-comparison)
Comment on lines
-609
to
+606
| for img_idx, (pos_inds_img, neg_inds_img) in enumerate( | ||
| zip(sampled_pos_inds, sampled_neg_inds) | ||
| ): | ||
| for pos_inds_img, neg_inds_img in zip(sampled_pos_inds, sampled_neg_inds): |
There was a problem hiding this comment.
Function RoIHeads.subsample refactored with the following changes:
- Remove unnecessary calls to
enumeratewhen the index is not used (remove-unused-enumerate)
| for i in range(3): | ||
| for _ in range(3): | ||
| pt_model(**inputs) | ||
|
|
||
| t1 = time.time() | ||
| for i in range(n_repeat): | ||
| for _ in range(n_repeat): |
There was a problem hiding this comment.
Function perf_bench_torch refactored with the following changes:
- Replace unused for index with underscore [×2] (
for-index-underscore)
Comment on lines
-298
to
+294
| quantized_output_dir = configs.output_dir + "quantized/" | ||
| quantized_output_dir = f"{configs.output_dir}quantized/" |
There was a problem hiding this comment.
Lines 298-298 refactored with the following changes:
- Use f-string instead of string concatenation (
use-fstring-for-concatenation)
| (args.output_dir, args.output_dir + "-MM") | ||
| (args.output_dir, f"{args.output_dir}-MM") | ||
| if args.task_name == "mnli" | ||
| else (args.output_dir,) | ||
| ) | ||
|
|
There was a problem hiding this comment.
Function evaluate_tvm refactored with the following changes:
- Use f-string instead of string concatenation [×2] (
use-fstring-for-concatenation) - Replace call to format with f-string [×2] (
use-fstring-for-formatting) - Merge dictionary updates via the union operator (
dict-assign-update-to-union)
Comment on lines
-437
to
+440
| # PyTorch eval | ||
| if True: | ||
| # # Evaluate the original FP32 BERT model | ||
| # print("Evaluating PyTorch full precision accuracy and performance:") | ||
| # time_model_evaluation(model, configs, tokenizer) | ||
| # # Evaluate the original FP32 BERT model | ||
| # print("Evaluating PyTorch full precision accuracy and performance:") | ||
| # time_model_evaluation(model, configs, tokenizer) | ||
|
|
||
| # Evaluate the INT8 BERT model after the dynamic quantization | ||
| print("Evaluating PyTorch quantization accuracy and performance:") | ||
| time_model_evaluation(quantized_model, configs, tokenizer) | ||
| # Evaluate the INT8 BERT model after the dynamic quantization | ||
| print("Evaluating PyTorch quantization accuracy and performance:") | ||
| time_model_evaluation(quantized_model, configs, tokenizer) |
There was a problem hiding this comment.
Lines 437-445 refactored with the following changes:
- Remove redundant conditional (
remove-redundant-if)
This removes the following comments ( why? ):
# PyTorch eval
Comment on lines
67
to
-82
|
|
||
| for idx, task in enumerate(tasks): | ||
| print("========== Task %d (workload key: %s) ==========" % (idx, task.workload_key)) | ||
| print(task.compute_dag) | ||
|
|
||
| measure_ctx = auto_scheduler.LocalRPCMeasureContext(repeat=1, min_repeat_ms=300, timeout=100) | ||
|
|
||
| tuner = auto_scheduler.TaskScheduler(tasks, task_weights) | ||
| # tuner = auto_scheduler.TaskScheduler(tasks, task_weights, load_log_file=log_file) | ||
| tune_option = auto_scheduler.TuningOptions( | ||
| num_measure_trials=50000, # change this to 20000 to achieve the best performance | ||
| runner=measure_ctx.runner, | ||
| measure_callbacks=[auto_scheduler.RecordToFile(log_file)], | ||
| ) | ||
|
|
||
| tuner.tune(tune_option) |
There was a problem hiding this comment.
Function auto_schedule refactored with the following changes:
- Remove unreachable code (
remove-unreachable-code)
This removes the following comments ( why? ):
# tuner = auto_scheduler.TaskScheduler(tasks, task_weights, load_log_file=log_file)
# change this to 20000 to achieve the best performance
Sourcery Code Quality Report✅ Merging this PR will increase code quality in the affected files by 0.43%.
Here are some functions in these files that still need a tune-up:
Legend and ExplanationThe emojis denote the absolute quality of the code:
The 👍 and 👎 indicate whether the quality has improved or gotten worse with this pull request. Please see our documentation here for details on how these metrics are calculated. We are actively working on this report - lots more documentation and extra metrics to come! Help us improve this quality report! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Branch
masterrefactored by Sourcery.If you're happy with these changes, merge this Pull Request using the Squash and merge strategy.
See our documentation here.
Run Sourcery locally
Reduce the feedback loop during development by using the Sourcery editor plugin:
Review changes via command line
To manually merge these changes, make sure you're on the
masterbranch, then run:Help us improve this pull request!