How to use feedback method in tox
Best Python code snippet using tox_python
nudging_loss.py
Source:nudging_loss.py 
...211 reduction = 'sum'212 # ===== Get good action values above the best bad action213 loss_feedback = regression_loss(qs_a_bad, min_no_feedback_per_line, reduction=reduction)214 return loss_feedback215def feedback_frontier_margin_learnt_feedback(qs, action=None, feedback=None, margin=None, regression_loss=None, testing=False, feedback_logits=None, ceil=None):216 """217 Compute the expert loss218 ab is the "bad" action219 m is a margin function220 """221 if feedback_logits is None :222 return 0223 n_actions = qs.size(1)224 almost_sure_feedback = torch.zeros(*feedback_logits.size()).to(TORCH_DEVICE)225 almost_sure_no_feedback = torch.zeros(*feedback_logits.size()).to(TORCH_DEVICE)226 # Feedback (or no feedback) that is almost sure (using a classification network), use them.227 almost_sure_feedback[feedback_logits > 0.5 + ceil] = 1228 almost_sure_no_feedback[feedback_logits < 0.5 - ceil] = 1229 # check if at least there are "sure" predicted feedback per line230 at_least_one_feedback_per_line = almost_sure_feedback.sum(dim=1).type(torch.uint8)231 at_least_one_nofeedback_per_line = almost_sure_no_feedback.sum(dim=1).type(torch.uint8)232 # Statistics about the number of action that are considered233 certainty_percentage = ((at_least_one_nofeedback_per_line + at_least_one_feedback_per_line).float() / n_actions).mean()234 certainty_percentage_feed = (at_least_one_feedback_per_line.float() / n_actions).mean()235 certainty_percentage_no_feed = (at_least_one_nofeedback_per_line.float() / n_actions).mean()236 both_per_line = at_least_one_feedback_per_line + at_least_one_nofeedback_per_line > 1237 qs = qs[both_per_line, :]238 if qs.size(0) == 0:239 return 0240 almost_sure_no_feedback = almost_sure_no_feedback[both_per_line, :]241 almost_sure_feedback = almost_sure_feedback[both_per_line, :]242 qs_feedback = qs.clone().detach() # Q(s,a) for action flagged as "gives feedback" aka bad actions by classification algorithm243 qs_no_feedback = qs.clone().detach() #Q(s,a) for action flagged as "don't give feedback" aka good actions by classification algorithm244 # Don't know if it's a feedback yet ? Temporarily set to the minimum qs so the optim doesn't touch it245 qs_feedback[almost_sure_feedback == 0] = torch.min(qs).item() - margin246 # Don't know if it's NOT a feedback yet ? Temporarily set to the maximum qs so the optim doesn't touch it247 qs_no_feedback[almost_sure_no_feedback == 0] = torch.max(qs).item() + margin248 min_no_feedback_per_line = qs_no_feedback.min(dim=1)[0].repeat([n_actions, 1]).t() - margin249 max_feedback_per_line = qs_feedback.max(dim=1)[0].repeat([n_actions, 1]).t() + margin250 sure_feedback_and_above_min = almost_sure_feedback.byte() * (qs > min_no_feedback_per_line) # '*' is logical and251 sure_no_feedback_and_below_max = almost_sure_no_feedback.byte() * (qs < max_feedback_per_line)252 qs_target = qs.clone().detach()253 qs_feedback_target = torch.where(sure_feedback_and_above_min, min_no_feedback_per_line, qs_target)254 qs_no_feedback_target = torch.where(sure_no_feedback_and_below_max, max_feedback_per_line, qs_target)255 if not testing:256 assert qs_no_feedback_target.requires_grad == False257 assert qs_feedback_target.requires_grad == False258 assert qs.requires_grad == True259 reduction = 'mean'260 else:261 reduction = 'sum'262 # ===== Get good action values above the best bad action263 loss_no_feedback = regression_loss(qs, qs_no_feedback_target, reduction=reduction)264 loss_feedback = regression_loss(qs, qs_feedback_target, reduction=reduction)265 return loss_no_feedback + loss_feedback266if __name__ == "__main__":267 import torch268 TORCH_DEVICE = 'cpu'269 regr_loss = torch.nn.functional.smooth_l1_loss270 margin = 0.1271 # Test 1272 qs = torch.arange(12).view(4,3).float()273 actions = torch.Tensor([0,0,0,0]).long()274 feedback = torch.Tensor([1,1,1,0])275 assert feedback_bad_to_min_when_max(qs, actions, testing=True, feedback=feedback, margin=margin, regression_loss=regr_loss) == 0276 # Test 2277 qs = torch.arange(12).view(4,3).float()278 actions = torch.Tensor([0,0,0,0]).long()279 feedback = torch.Tensor([1,1,1,0])280 loss1 = feedback_bad_to_min_when_max(qs, actions, testing=True, feedback=feedback, margin=margin, regression_loss=regr_loss)281 qs = torch.arange(12).view(4,3).float()282 actions = torch.Tensor([0,1,2,0]).long()283 feedback = torch.Tensor([1,1,1,0])284 loss2 = feedback_bad_to_min_when_max(qs, actions, testing=True, feedback=feedback, margin=margin, regression_loss=regr_loss)285 assert loss1 < loss2286 # # Test 3287 # qs = torch.arange(12).view(4, 3).float()288 #289 # max_margin = 0.50 # If bad action is 50% of the max : Put it down niggah290 # actions = torch.Tensor([0, 1, 2, 0]).long()291 # feedback = torch.Tensor([1, 1, 1, 0])292 # loss1 = feedback_bad_to_percent_max(qs, actions, feedback, regr_loss, max_margin)293 #294 # max_margin = 0.90 # If bad action is 90% of the max : Put it down niggah295 # qs = torch.arange(12).view(4, 3).float()296 # actions = torch.Tensor([0, 1, 2, 0]).long()297 # feedback = torch.Tensor([1, 1, 1, 0])298 # loss2 = feedback_bad_to_percent_max(qs, actions, feedback, regr_loss, max_margin)299 #300 # # assert loss1 > loss2, "loss1 {}, loss2 {}".format(loss1, loss2)301 #302 # max_margin = 0.90 # If bad action is 90% of the max : Put it down niggah303 # qs = - torch.arange(12).view(4, 3).float()304 # actions = torch.Tensor([0, 1, 2, 0]).long()305 # feedback = torch.Tensor([1, 1, 1, 0])306 # loss3 = feedback_bad_to_percent_max(qs, actions, feedback, regr_loss, max_margin)307 #308 # #=================================================309 #310 # qs = torch.arange(12).view(4,3).float()311 # actions = torch.Tensor([1,2,1,0]).long()312 # feedback = torch.Tensor([1,1,1,0])313 # loss1 = feedback_bad_to_min(qs, actions, feedback, margin, regr_loss)314 #==================================================315 #316 # qs = torch.arange(21).view(7, 3).float()317 # actions = torch.Tensor([1, 2, 1, 0, 2 , 1, 0]).long().view(-1, 1)318 # feedback = torch.Tensor([1, 0, 1, 0, 1, 0, 1])319 # loss1 = feedback_frontier_margin(qs, actions, feedback, margin, regr_loss, testing=True)320 #321 # actions = torch.Tensor([[1, 2, 1, 0, 2, 1, 0]]).long().view(-1, 1)322 # feedback = torch.Tensor([1, 0, 1, 0, 1, 0, 1])323 #324 # qs = - torch.arange(21).view(7, 3).float()325 # loss2s = feedback_frontier_margin(qs, actions, feedback, margin, regr_loss, testing=True)326 #=======================================================327 ceil = 0.4328 #=======================================================329 #330 # qs = torch.arange(21).view(7, 3).float()331 # logits = torch.ones(7,3)332 # logits[:, 2] *= -1333 # logits[:, 1] = 0334 #335 #336 # loss1 = feedback_frontier_margin_learnt_feedback(qs, margin=margin, regression_loss=regr_loss, feedback_logits=logits,337 # testing=True, ceil=ceil)338 # assert loss1 == 0339 #340 # # =======================================================341 # qs = torch.arange(21).view(7, 3).float()342 # logits = torch.ones(7, 3)343 # logits[:, 0] = 0.5344 # logits[:, 1] = 0.5345 #346 # loss2 = feedback_frontier_margin_learnt_feedback(qs, margin=margin, regression_loss=regr_loss,347 # feedback_logits=logits,348 # testing=True, ceil=ceil)349 # assert loss2 == 0350 #351 # #=======================================================352 # qs = -torch.arange(21).view(7, 3).float()353 # logits = torch.ones(7, 3)354 # logits[:, 2] *= 0355 # logits[:, 1] = 0.5356 #357 # loss3 = feedback_frontier_margin_learnt_feedback(qs, margin=margin, regression_loss=regr_loss,358 # feedback_logits=logits,359 # testing=True, ceil=ceil)360 #361 # #========================================================362 #363 # qs = torch.arange(21).view(7, 3).float()364 # logits = torch.ones(7, 3)365 # logits[:, 0] *= 0366 # logits[:, 1] = 0.5367 #368 # loss4 = feedback_frontier_margin_learnt_feedback(qs, margin=margin, regression_loss=regr_loss,369 # feedback_logits=logits,370 # testing=True, ceil=ceil)371 #372 # assert loss3 == loss4373 #374 # # ========================================================375 # qs = torch.arange(21).view(7, 3).float()376 # logits = torch.ones(7, 3)377 # logits[:, 0] *= -1378 # logits[:, 1] = 0379 # logits[-1, :] = 0380 #381 # loss5 = feedback_frontier_margin_learnt_feedback(qs, margin=margin, regression_loss=regr_loss,382 # feedback_logits=logits,383 # testing=True, ceil=ceil)384 #385 # assert loss5 != 0386 # assert loss5 < loss4387 #388 # # =========================================================389 # qs = torch.arange(21).view(7, 3).float()390 # logits = torch.zeros(7, 3)391 # logits[:,0] = 1392 #393 # loss6 = feedback_frontier_margin_learnt_feedback(qs, margin=margin, regression_loss=regr_loss,394 # feedback_logits=logits,395 # testing=True, ceil=ceil)396 #397 # assert loss6 == 0398 #===========================================================399 qs = torch.tensor([[0,0,0],[1,0,0.5],[0,1,0.5]]).float()400 actions = torch.Tensor([1,1,1]).long()401 feedback = torch.Tensor([0,1,1])402 logits = torch.zeros_like(qs)403 logits[1, 1] = 1404 logits[2, 1] = 1405 loss7 = feedback_ponctual_negative_only(qs,406 action=actions,407 feedback=feedback,...
feedback_trigger.py
Source:feedback_trigger.py
...151 triggers = { dt[0]: dt[1] for dt in triggers }152 return triggers153def get_context(doc):154 return { "doc": doc }155def delete_feedback_request_and_feedback(reference_doctype, reference_name):156 """ delete all the feedback request and feedback communication """157 if not all([reference_doctype, reference_name]):158 return159 feedback_requests = frappe.get_all("Feedback Request", filters={160 "is_feedback_submitted": 0,161 "reference_doctype": reference_doctype,162 "reference_name": reference_name163 })164 communications = frappe.get_all("Communication", {165 "communication_type": "Feedback",166 "reference_doctype": reference_doctype,167 "reference_name": reference_name168 })169 for request in feedback_requests:...
test_feedback_trigger.py
Source:test_feedback_trigger.py
1# -*- coding: utf-8 -*-2# Copyright (c) 2015, Frappe Technologies and Contributors3# See license.txt4from __future__ import unicode_literals5import frappe6import unittest7# test_records = frappe.get_test_records('Feedback Trigger')8def get_feedback_request(todo, feedback_trigger):9 return frappe.db.get_value("Feedback Request", {10 "is_sent": 1,11 "is_feedback_submitted": 0,12 "reference_doctype": "ToDo",13 "reference_name": todo,14 "feedback_trigger": feedback_trigger15 }, ["name", "key"])16class TestFeedbackTrigger(unittest.TestCase):17 def setUp(self):18 new_user = frappe.get_doc(dict(doctype='User', email='test-feedback@example.com',19 first_name='Tester')).insert(ignore_permissions=True)20 new_user.add_roles("System Manager")21 def tearDown(self):22 frappe.db.sql("delete from tabContact where email_id='test-feedback@example.com'")23 frappe.delete_doc("User", "test-feedback@example.com")24 frappe.delete_doc("Feedback Trigger", "ToDo")25 frappe.db.sql('delete from `tabEmail Queue`')26 frappe.db.sql('delete from `tabFeedback Request`')27 def test_feedback_trigger(self):28 """ Test feedback trigger """29 from frappe.www.feedback import accept30 frappe.delete_doc("Feedback Trigger", "ToDo")31 frappe.db.sql('delete from `tabEmail Queue`')32 frappe.db.sql('delete from `tabFeedback Request`')33 feedback_trigger = frappe.get_doc({34 "enabled": 1,35 "doctype": "Feedback Trigger",36 "document_type": "ToDo",37 "email_field": "assigned_by",38 "email_fieldname": "assigned_by",39 "subject": "{{ doc.name }} Task Completed",40 "condition": "doc.status == 'Closed'",41 "message": """Task {{ doc.name }} is Completed by {{ doc.owner }}.42 regarding the Task {{ doc.name }}"""43 }).insert(ignore_permissions=True)44 # create a todo45 todo = frappe.get_doc({46 "doctype": "ToDo",47 "owner": "test-feedback@example.com",48 "assigned_by": "test-feedback@example.com",49 "description": "Unable To Submit Sales Order #SO-00001"50 }).insert(ignore_permissions=True)51 # feedback alert mail should be sent only on 'Closed' status52 email_queue = frappe.db.sql("""select name from `tabEmail Queue` where53 reference_doctype='ToDo' and reference_name='{0}'""".format(todo.name))54 self.assertFalse(email_queue)55 # add a communication56 frappe.get_doc({57 "reference_doctype": "ToDo",58 "reference_name": todo.name,59 "communication_type": "Communication",60 "content": "Test Communication",61 "subject": "Test Communication",62 "doctype": "Communication"63 }).insert(ignore_permissions=True)64 # check if feedback mail alert is triggered65 todo.reload()66 todo.status = "Closed"67 todo.save(ignore_permissions=True)68 email_queue = frappe.db.sql("""select name from `tabEmail Queue` where69 reference_doctype='ToDo' and reference_name='{0}'""".format(todo.name))70 self.assertTrue(email_queue)71 # test if feedback is submitted for the todo72 feedback_request, request_key = get_feedback_request(todo.name, feedback_trigger.name)73 self.assertTrue(feedback_request)74 # test if mail alerts are triggered multiple times for same document75 todo.save(ignore_permissions=True)76 email_queue = frappe.db.sql("""select name from `tabEmail Queue` where77 reference_doctype='ToDo' and reference_name='{0}'""".format(todo.name))78 self.assertTrue(len(email_queue) == 1)79 frappe.db.sql('delete from `tabEmail Queue`')80 # Test if feedback is submitted sucessfully81 result = accept(request_key, "test-feedback@example.com", "ToDo", todo.name, "Great Work !!", 4, fullname="Test User")82 self.assertTrue(result)83 # test if feedback is saved in Communication84 docname = frappe.db.get_value("Communication", {85 "reference_doctype": "ToDo",86 "reference_name": todo.name,87 "communication_type": "Feedback",88 "feedback_request": feedback_request89 })90 communication = frappe.get_doc("Communication", docname)91 self.assertEqual(communication.rating, 4)92 self.assertEqual(communication.content, "Great Work !!")93 # test if link expired after feedback submission94 self.assertRaises(Exception, accept, key=request_key, sender="test-feedback@example.com",95 reference_doctype="ToDo", reference_name=todo.name, feedback="Thank You !!", rating=4, fullname="Test User")96 # auto feedback request should trigger only once97 todo.reload()98 todo.save(ignore_permissions=True)99 email_queue = frappe.db.sql("""select name from `tabEmail Queue` where100 reference_doctype='ToDo' and reference_name='{0}'""".format(todo.name))101 self.assertFalse(email_queue)102 frappe.delete_doc("ToDo", todo.name)103 # test if feedback requests and feedback communications are deleted?104 communications = frappe.get_all("Communication", {105 "reference_doctype": "ToDo",106 "reference_name": todo.name,107 "communication_type": "Feedback"108 })109 self.assertFalse(communications)110 feedback_requests = frappe.get_all("Feedback Request", {111 "reference_doctype": "ToDo",112 "reference_name": todo.name,113 "is_feedback_submitted": 0114 })...
Automation Testing Tutorials
Learn to execute automation testing from scratch with LambdaTest Learning Hub. Right from setting up the prerequisites to run your first automation test, to following best practices and diving deeper into advanced test scenarios. LambdaTest Learning Hubs compile a list of step-by-step guides to help you be proficient with different test automation frameworks i.e. Selenium, Cypress, TestNG etc.
LambdaTest Learning Hubs:
- JUnit Tutorial
- TestNG Tutorial
- Webdriver Tutorial
- WebDriverIO Tutorial
- Protractor Tutorial
- Selenium 4 Tutorial
- Jenkins Tutorial
- NUnit Tutorial
- Jest Tutorial
- Playwright Tutorial
- Cypress Tutorial
- PyTest Tutorial
YouTube
You could also refer to video tutorials over LambdaTest YouTube channel to get step by step demonstration from industry experts.
Run tox automation tests on LambdaTest cloud grid
Perform automation testing on 3000+ real desktop and mobile devices online.
Try LambdaTest Now !!
Get 100 minutes of automation test minutes FREE!!
Was this article helpful?
