I'm playing around with OpenAI's Universe module and am having trouble understanding what the observation_n is exactly and how it works. On their website it just says, "Universe allows an AI agent to use a computer like a human does: by looking at screen pixels..."
I modified some code that was meant for a gym game called "CartPole-v1" and tried applying it to "flashgames.DriftRunners-v0". However I keep getting an index error, so I assume observation_n is not an index.
def get_action(self, observation_n):
turn_angle = observation_n[2]
action = [('KeyEvent', 'ArrowUp', True), ('KeyEvent', 'ArrowRight', True)] if turn_angle > 0 else [('KeyEvent', 'ArrowUp', True)]
return action
agent = Agent(env)
observation_n = env.reset()
I'm getting this error,
line 19, in get_action
turn_angle = observation_n[2]
IndexError: list index out of range
Related
Pdict = {"KobeBryant":0,"JoeJohnson":1,"LeBronJames":2,"CarmeloAnthony":3,"DwightHoward":4,"ChrisBosh":5,"ChrisPaul":6,"KevinDurant":7,"DerrickRose":8,"DwayneWade":9}
Salary = np.array([KobeBryant_Salary, JoeJohnson_Salary, LeBronJames_Salary, CarmeloAnthony_Salary, DwightHoward_Salary, ChrisBosh_Salary, ChrisPaul_Salary, KevinDurant_Salary, DerrickRose_Salary, DwayneWade_Salary])
Games = np.array([KobeBryant_G, JoeJohnson_G, LeBronJames_G, CarmeloAnthony_G, DwightHoward_G, ChrisBosh_G, ChrisPaul_G, KevinDurant_G, DerrickRose_G, DwayneWade_G])
plr =[]
for v in Pdict.values():
for s in Salary:
for g in Games:
if np.isnan(g[v]) == True: continue
z = np.round(np.divide(s[v], Games[v]), 2)
plr.append(z)
print(plr)
print(type(z))
Im trying to make a new matrix called plr, there is a zero in Games[] and Im trying to make it skip it instead of giving me that error. I found the np.isnan() but it seems to do nothing here. If I run the code with or without that line, it still gives me the runtime warning. Im not sure what Im doing wrong with it or is there a better way to do this?
I've been playing around with BERT and TensorFlow following the example here and have a trained working model.
I then wanted to save and deploy the model, so used the export_saved_model function, which requires you build a serving_input_fn to handle any incoming requests when the model is reloaded.
I wanted to be able to pass a single string for sentiment analysis to the deployed model, rather than having a theoretical client side application do the tokenisation and feature generation etc, so tried to write an input function that would handle that and pass the constructed features to the model. Is this possible? I wrote the following which I feel should do what I want:
import json
import base64
def plain_text_serving_input_fn():
input_string = tf.placeholder(dtype=tf.string, shape=None, name='input_string_text')
# What format to expect input in.
receiver_tensors = {'input_text': input_string}
input_examples = [run_classifier.InputExample(guid="", text_a = str(input_string), text_b = None, label = 0)] # here, "" is just a dummy label
input_features = run_classifier.convert_examples_to_features(input_examples, label_list, MAX_SEQ_LENGTH, tokenizer)
variables = {}
for i in input_features:
variables["input_ids"] = i.input_ids
variables["input_mask"] = i.input_mask
variables["segment_ids"] = i.segment_ids
variables["label_id"] = i.label_id
feature_spec = {
"input_ids" : tf.FixedLenFeature([MAX_SEQ_LENGTH], tf.int64),
"input_mask" : tf.FixedLenFeature([MAX_SEQ_LENGTH], tf.int64),
"segment_ids" : tf.FixedLenFeature([MAX_SEQ_LENGTH], tf.int64),
"label_ids" : tf.FixedLenFeature([], tf.int64)
}
string_variables = json.dumps(variables)
encode_input = base64.b64encode(string_variables.encode('utf-8'))
encode_string = base64.decodestring(encode_input)
features_to_input = tf.parse_example([encode_string], feature_spec)
return tf.estimator.export.ServingInputReceiver(features_to_input, receiver_tensors)
I would expect that this would allow me to call predict on my deployed model with
variables = {"input_text" : "This is some test input"}
predictor.predict(variables)
I've tried a range of variations of this (putting it in an array, converting to base 64 etc), but I get a range of errors either telling me
"error": "Failed to process element: 0 of 'instances' list. Error: Invalid argument: JSON Value: {\n \"input_text\": \"This is some test input\"\n} not formatted correctly for base64 data" }"
or
Object of type 'bytes' is not JSON serializable
I suspect I'm formatting my requests incorrectly, but I also can't find any examples of something similar being done in a serving_input_fn, so has anyone ever done something similar?
I am creating a platform game. I was trying to do a platform that went constantly up and down like an elevator, when i got this error: TypeError: 'Sprite' object is not iterable.
Here's the code related to the elevator:
IN SET UP:
self.platform_list = arcade.SpriteList()
self.elevator = arcade.Sprite()
for i in range(600,800,32):
self.elevator =
arcade.Sprite("imagenes/platformer/Ground/dirt_grass.png", SCALE_PLATFORMS)
self.elevator.center_x = i
self.elevator.center_y = 420
self.elevator.change_y = 2
self.platform_list.append(self.elevator)
IN UPDATE:
self.platform_list.update()
for plat in self.elevator :
if plat.center_y > 680 :
plat.change_y = - 2
if plat.center_y < 400 :
plat.change_y = 2
if i change the update by replacing plat for self.elevator, and get rid of the for, only one block of the ones that conform the platform goes up and down, the rest goes up without stoping, and gets out of the screen. I have tried also changing the for by "for self.elevator in self.platform_list" and then changing plat for self.elevator, but when i execute it, all platforms move up and down.
¿How can i fix this? Please help me.
I am trying to Check Tree items from a function after the Tree is initialized. Check marks never appear. I'm not sure if I need to 'refresh' the Tree or that I am not interacting with the right object.
I've read about this widget a lot and I see the same question more or less on several websites but the answers are 8+ years old, two pages of code or I just don't understand them.
What am I doing wrong here? Why are there no check marks showing up in the tree when I call checkActiveTreeItems()?
After a REST call, I want to update the checked items in the tree. This seems to be ok but the items never get checked.
def checkActiveTreeItems(self):
for label in self.transactie['transactie']['labels']:
test = QtWidgets.QTreeWidgetItem([label])
test.setCheckState(0, QtCore.Qt.Checked)
This is the function that creates the tree, I do not call it again after setting the QTreeWidgetItem to Checked in the function above.
def initTree(self):
print("initTree")
self.treeWidget = QtWidgets.QTreeWidget(self.gridLayoutWidget)
self.treeWidget.setSelectionMode(QtWidgets.QAbstractItemView.ContiguousSelection)
self.treeWidget.setObjectName("treeWidget")
inkomstenstructuur = requests.get(url="http://localhost:32769/_db/piekenpijp/incoming/inkomstenstructuur")
iscontent = json.loads(inkomstenstructuur.content)
self.inkomsten = iscontent
inkomstentree = QtWidgets.QTreeWidgetItem(["Inkomsten"])
for groep in iscontent:
child = QtWidgets.QTreeWidgetItem([groep['groep']])
inkomstentree.addChild(child)
for categorie in groep['categorien']:
child2 = QtWidgets.QTreeWidgetItem([categorie])
child2.setCheckState(0, QtCore.Qt.Unchecked)
child.addChild(child2)
uitgavenstructuur = requests.get(url="http://localhost:32769/_db/piekenpijp/incoming/uitgavenstructuur")
uscontent = json.loads(uitgavenstructuur.content)
self.uitgaven = uscontent
uitgaventree = QtWidgets.QTreeWidgetItem(["Uitgaven"])
for groep in uscontent:
child = QtWidgets.QTreeWidgetItem([groep['groep']])
uitgaventree.addChild(child)
for categorie in groep['categorien']:
child2 = QtWidgets.QTreeWidgetItem([categorie])
child2.setCheckState(0, QtCore.Qt.Unchecked)
child.addChild(child2)
self.treeWidget.itemChanged.connect(self.vinkje)
self.treeWidget.resize(500, 200)
self.treeWidget.setColumnCount(1)
self.treeWidget.setHeaderLabels(["Categorie"])
self.treeWidget.addTopLevelItem(inkomstentree)
self.treeWidget.addTopLevelItem(uitgaventree)
self.treeWidget.expandItem(uitgaventree)
self.gridLayout.addWidget(self.treeWidget, 0, 2, 1, 1)
Your current method creates a new item, sets its check-state, and then throws it away without using it. Instead, you need to find the items in the tree that you have already created:
def checkActiveTreeItems(self):
for label in self.transactie['transactie']['labels']:
for item in self.treeWidget.findItems(label, Qt.MatchExactly):
item.setCheckState(0, QtCore.Qt.Checked)
I'm not sure if the title accurately describes what I'm trying to do. I have a Python3.x script that I wrote that will issue flood warning to my facebook page when the river near my home has reached it's lowest flood stage. Right now the script works, however it only reports data from one measuring station. I would like to be able to process the data from all of the stations in my county (total of 5), so I was thinking that maybe a class method may do the trick but I'm not sure how to implement it. I've been teaching myself Python since January and feel pretty comfortable with the language for the most part, and while I have a good idea of how to build a class object I'm not sure how my flow chart should look. Here is the code now:
#!/usr/bin/env python3
'''
Facebook Flood Warning Alert System - this script will post a notification to
to Facebook whenever the Sabine River # Hawkins reaches flood stage (22.3')
'''
import requests
import facebook
from lxml import html
graph = facebook.GraphAPI(access_token='My_Access_Token')
river_url = 'http://water.weather.gov/ahps2/river.php?wfo=SHV&wfoid=18715&riverid=203413&pt%5B%5D=147710&allpoints=143204%2C147710%2C141425%2C144668%2C141750%2C141658%2C141942%2C143491%2C144810%2C143165%2C145368&data%5B%5D=obs'
ref_url = 'http://water.weather.gov/ahps2/river.php?wfo=SHV&wfoid=18715&riverid=203413&pt%5B%5D=147710&allpoints=143204%2C147710%2C141425%2C144668%2C141750%2C141658%2C141942%2C143491%2C144810%2C143165%2C145368&data%5B%5D=all'
def checkflood():
r = requests.get(river_url)
tree = html.fromstring(r.content)
stage = ''.join(tree.xpath('//div[#class="stage_stage_flow"]//text()'))
warn = ''.join(tree.xpath('//div[#class="current_warns_statmnts_ads"]/text()'))
stage_l = stage.split()
level = float(stage_l[2])
#check if we're at flood level
if level < 22.5:
pass
elif level == 37:
major_diff = level - 23.0
major_r = ('The Sabine River near Hawkins, Tx has reached [Major Flood Stage]: #', stage_l[2], 'Ft. ', str(round(major_diff, 2)), ' Ft. \n Please click the link for more information.\n\n Current Warnings and Alerts:\n ', warn)
major_p = ''.join(major_r)
graph.put_object(parent_object='me', connection_name='feed', message = major_p, link = ref_url)
<--snip-->
checkflood()
Each station has different 5 different catagories for flood stage: Action, Flood, Moderate, Major, each different depths per station. So for Sabine river in Hawkins it will be Action - 22', Flood - 24', Moderate - 28', Major - 32'. For the other statinos those depths are different. So I know that I'll have to start out with something like:
class River:
def __init__(self, id, stage):
self.id = id #station ID
self.stage = stage #river level'
#staticmethod
def check_flood(stage):
if stage < 22.5:
pass
elif stage.....
but from there I'm not sure what to do. Where should it be added in(to?) the code, should I write a class to handle the Facebook postings as well, is this even something that needs a class method to handle, is there any way to clean this up for efficiency? I'm not looking for anyone to write this up for me, but some tips and pointers would sure be helpful. Thanks everyone!
EDIT Here is what I figured out and is working:
class River:
name = ""
stage = ""
action = ""
flood = ""
mod = ""
major = ""
warn = ""
def checkflood(self):
if float(self.stage) < float(self.action):
pass
elif float(self.stage) >= float(self.major):
<--snip-->
mineola = River()
mineola.name = stations[0]
mineola.stage = stages[0]
mineola.action = "13.5"
mineola.flood = "14.0"
mineola.mod = "18.0"
mineola.major = "21.0"
mineola.alert = warn[0]
hawkins = River()
hawkins.name = stations[1]
hawkins.stage = stages[1]
hawkins.action = "22.5"
hawkins.flood = "23.0"
hawkins.mod = "32.0"
hawkins.major = "37.0"
hawkins.alert = warn[1]
<--snip-->
So from here I'm tring to stick all the individual river blocks into one block. What I have tried so far is this:
class River:
... name = ""
... stage = ""
... def testcheck(self):
... return self.name, self.stage
...
>>> for n in range(num_river):
... stations[n] = River()
... stations[n].name = stations[n]
... stations[n].stage = stages[n]
...
>>> for n in range(num_river):
... stations[n].testcheck()
...
<__main__.River object at 0x7fbea469bc50> 4.13
<__main__.River object at 0x7fbea46b4748> 20.76
<__main__.River object at 0x7fbea46b4320> 22.13
<__main__.River object at 0x7fbea46b4898> 16.08
So this doesn't give me the printed results that I was expecting. How can I return the string instead of the object? Will I be able to define the Class variables in this manner or will I have to list them out individually? Thanks again!
After reading many, many, many articles and tutorials on class objects I was able to come up with a solution for creating the objects using list elements.
class River():
def __init__(self, river, stage, flood, action):
self.river = river
self.stage = stage
self.action = action
self.flood = flood
self.action = action
def alerts(self):
if float(self.stage < self.flood):
#alert = "The %s is below Flood Stage (%sFt) # %s Ft. \n" % (self.river, self.flood, self.stage)
pass
elif float(self.stage > self.flood):
alert = "The %s has reached Flood Stage(%sFt) # %sFt. Warnings: %s \n" % (self.river, self.flood, self.stage, self.action)
return alert
'''this is the function that I was trying to create
to build the class objects automagically'''
def riverlist():
river_list = []
for n in range(len(rivers)):
station = River(river[n], stages[n], floods[n], warns[n])
river_list.append(station)
return river_list
if __name__ == '__main__':
for x in riverlist():
print(x.alerts())