Week 3 notes and notebooks
This commit is contained in:
levdoescode
Submodule CM3020 Artificial Intelligence/Week 3/3.8 Lab Files/gs-gym added at 23a1bed2cd
@ -0,0 +1,32 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": []
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3 (ipykernel)",
|
||||
"language": "python",
|
||||
"name": "python3"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3",
|
||||
"version": "3.11.0"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 4
|
||||
}
|
||||
@ -0,0 +1,320 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 3,
|
||||
"id": "0bbb53e7-84e7-4d65-baab-05421b7e7bc3",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"2.10.0\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"import keras\n",
|
||||
"print(keras.__version__)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "9d23266a-38fb-430f-9502-85526a86a5d1",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from keras import layers\n",
|
||||
"inputs = layers.Input(shape=(1,))\n",
|
||||
"layer1 = layers.Dense(512, activation=\"relu\")(inputs)\n",
|
||||
"outputs = layers.Dense(3, activation=\"linear\")(layer1)\n",
|
||||
"nn_model = keras.Model(inputs=inputs, outputs=outputs)\n",
|
||||
"nn_model.summary()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "397f97d8-18de-4855-ad94-89e461f85b6e",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"output = nn_model(0.5) #throws erros"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "bded47b5-69fe-40cf-9cb4-78a360ee4244",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"input = keras.backend.constant([[0.5]])\n",
|
||||
"output = nn_model(input)\n",
|
||||
"print(output)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "d10e5212-c663-49c2-bb0f-34956c23070a",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import numpy as np\n",
|
||||
"inputs = np.arange(0, 1.1, 0.1)\n",
|
||||
"input = keras.backend.constant([[0.5]])\n",
|
||||
"input2 = keras.backend.constant([inputs])\n",
|
||||
"output = nn_model(input)\n",
|
||||
"print(input)\n",
|
||||
"print(input2)\n",
|
||||
"print(output)\n",
|
||||
"print(inputs)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "0111c2c1-6c00-4859-8715-e74fcc3883bd",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import gym\n",
|
||||
"e = gym.make('MountainCarContinuous-v0')\n",
|
||||
"e.reset()\n",
|
||||
"e.render()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "2dcfd9d1-43b6-48f0-9435-9ef75158152c",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import gym\n",
|
||||
"e = gym.make('MountainCarContinuous-v0')\n",
|
||||
"s1 = e.reset()\n",
|
||||
"print(s1)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "1d4a9391-4af2-47c8-b310-06c5b3016f59",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"e.action_space.sample()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "77781c43-bb1e-41ce-b878-2245edf9a094",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from keras import layers\n",
|
||||
"inputs = layers.Input(shape=(2, ))\n",
|
||||
"layer1 = layers.Dense(512, activation=\"relu\")(inputs)\n",
|
||||
"outputs = layers.Dense(1, activation=\"linear\")(layer1)\n",
|
||||
"nn_model = keras.Model(inputs=inputs, outputs=outputs)\n",
|
||||
"s = e.reset()\n",
|
||||
"input = keras.backend.constant([s])\n",
|
||||
"a = nn_model(input)\n",
|
||||
"s,r,d,i = e.step(a)\n",
|
||||
"print(s)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "e94051cc-7498-4c85-b296-a79669d4bc64",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# Control the agent with the network\n",
|
||||
"from keras import layers\n",
|
||||
"inputs = layers.Input(shape=(2, ))\n",
|
||||
"layer1 = layers.Dense(512, activation=\"relu\")(inputs)\n",
|
||||
"outputs = layers.Dense(1, activation=\"linear\")(layer1)\n",
|
||||
"nn_model = keras.Model(inputs=inputs, outputs=outputs)\n",
|
||||
"for item in range(10):\n",
|
||||
" s = e.reset()\n",
|
||||
" input = keras.backend.constant([s])\n",
|
||||
" a = nn_model(input)\n",
|
||||
" s,r,d,i = e.step(a)\n",
|
||||
" e.render()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "0cba086b-4a65-4ee0-92d3-4ee752bf2161",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"s = e.reset()\n",
|
||||
"input = keras.backend.constant([s]) # s is an array\n",
|
||||
"a = nn_model(input)\n",
|
||||
"s,r,d,i = e.step(a)\n",
|
||||
"print(s)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 1,
|
||||
"id": "fb686ff6-aa76-4446-b81f-1b26db98022d",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"pygame 2.1.2 (SDL 2.0.18, Python 3.10.8)\n",
|
||||
"Hello from the pygame community. https://www.pygame.org/contribute.html\n",
|
||||
"Discrete(4)\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"# Breakwall enviroment again\n",
|
||||
"import keras\n",
|
||||
"import gym\n",
|
||||
"# We get a discrete (4) action space\n",
|
||||
"env_name = \"gym_gs:BreakwallNoFrameskip-v1\"\n",
|
||||
"e = gym.make(env_name)\n",
|
||||
"s = e.reset()\n",
|
||||
"print(e.action_space)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 2,
|
||||
"id": "585e3e61-4745-45f1-beec-d25bf6a78c11",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"2\n",
|
||||
"Observation/ state shape: (600, 780, 3)\n",
|
||||
"0\n",
|
||||
"Observation/ state shape: (600, 780, 3)\n",
|
||||
"3\n",
|
||||
"Observation/ state shape: (600, 780, 3)\n",
|
||||
"0\n",
|
||||
"Observation/ state shape: (600, 780, 3)\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"for i in range(4):\n",
|
||||
" print(e.action_space.sample())\n",
|
||||
" print(\"Observation/ state shape:\", s.shape)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 3,
|
||||
"id": "2a45d350-cf9b-459d-a1f7-1470809e14cf",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"tf.Tensor([[0. 0. 0. 0.]], shape=(1, 4), dtype=float32)\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"# We'll update the imput layer\n",
|
||||
"from keras import layers\n",
|
||||
"inputs = layers.Input(shape=(600,780,3,))\n",
|
||||
"layer1 = layers.Flatten()(inputs) # We flatten the the 3D input layer\n",
|
||||
"layer2 = layers.Dense(512, activation=\"relu\")(layer1)\n",
|
||||
"outputs = layers.Dense(4, activation=\"linear\")(layer2)\n",
|
||||
"nn_model = keras.Model(inputs=inputs, outputs=outputs)\n",
|
||||
"state = e.reset()\n",
|
||||
"state_tensor = keras.backend.constant(state)\n",
|
||||
"state_tensor = keras.backend.expand_dims(state_tensor, 0)\n",
|
||||
"action_values = nn_model(state_tensor)\n",
|
||||
"print(action_values)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "c27e9935-2a2c-4dd4-82e0-43b86d3e0de5",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"action = keras.backend.argmax(action_values[0]).numpy()\n",
|
||||
"action = tf.argmax(action_probs[0]).numpy() # if renamed\n",
|
||||
"print(keras.backend.argmax([0.1,0.5,0.01,0.2]).numpy())"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 4,
|
||||
"id": "053e95dd-48af-4c18-949b-12d5a13e1a3e",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"tf.Tensor([[-135.33052 -34.96535 -2.6175923 104.20433 ]], shape=(1, 4), dtype=float32) 3\n",
|
||||
"tf.Tensor([[-133.47585 -40.308247 -9.9967165 100.298096 ]], shape=(1, 4), dtype=float32) 3\n",
|
||||
"tf.Tensor([[-133.52586 -34.83712 -9.089392 106.48525 ]], shape=(1, 4), dtype=float32) 3\n",
|
||||
"tf.Tensor([[-142.65623 -37.653713 -12.290092 100.456726]], shape=(1, 4), dtype=float32) 3\n",
|
||||
"tf.Tensor([[-139.46967 -29.941523 -15.024807 102.14433 ]], shape=(1, 4), dtype=float32) 3\n",
|
||||
"tf.Tensor([[-137.86325 -28.1824 -18.672073 102.72069 ]], shape=(1, 4), dtype=float32) 3\n",
|
||||
"tf.Tensor([[-134.41286 -29.641228 -12.233908 105.396324]], shape=(1, 4), dtype=float32) 3\n",
|
||||
"tf.Tensor([[-139.85345 -27.45109 -11.1273575 93.42384 ]], shape=(1, 4), dtype=float32) 3\n",
|
||||
"tf.Tensor([[-134.20053 -29.01193 -11.014997 94.585106]], shape=(1, 4), dtype=float32) 3\n",
|
||||
"tf.Tensor([[-136.16206 -31.467659 -6.581274 96.69547 ]], shape=(1, 4), dtype=float32) 3\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"state = e.reset()\n",
|
||||
"state,r,d,i = e.step(1)\n",
|
||||
"for i in range(10):\n",
|
||||
" state_tensor = keras.backend.constant(state)\n",
|
||||
" state_tensor = keras.backend.expand_dims(state_tensor, 0)\n",
|
||||
" action_values = nn_model(state_tensor)\n",
|
||||
" action = keras.backend.argmax(action_values[0]).numpy()\n",
|
||||
" print(action_values, action)\n",
|
||||
" state,r,d,i = e.step(1)\n",
|
||||
" e.render()"
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3 (ipykernel)",
|
||||
"language": "python",
|
||||
"name": "python3"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3",
|
||||
"version": "3.10.8"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 5
|
||||
}
|
||||
@ -0,0 +1,108 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "ec755efa-2052-422b-b49d-fe46a343d568",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import gym\n",
|
||||
"for n in gym.envs.registry.all():\n",
|
||||
" print(n)\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 2,
|
||||
"id": "01167b07-819c-40c8-a5c8-5463c43c8925",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"data": {
|
||||
"text/plain": [
|
||||
"True"
|
||||
]
|
||||
},
|
||||
"execution_count": 2,
|
||||
"metadata": {},
|
||||
"output_type": "execute_result"
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"import gym\n",
|
||||
"import matplotlib.pyplot as plt\n",
|
||||
"#e = gym.make('gym_gs:BreakwallNoFrameskip-v1')\n",
|
||||
"env_name = \"MountainCarContinuous-v0\"\n",
|
||||
"e = gym.make(env_name)\n",
|
||||
"e.reset()\n",
|
||||
"e.render()\n",
|
||||
"# img = e.render('rgb_array')\n",
|
||||
"# plt.imshow(e.render(mode='rgb_array'))b\n",
|
||||
"# img = e.render('rgb_array')\n",
|
||||
"# plt.imshow(e.render(mode='rgb_array'))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 7,
|
||||
"id": "9328b324-7f9a-4178-94c4-d1f8cec07542",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"data": {
|
||||
"text/plain": [
|
||||
"array([-0.51512, 0. ])"
|
||||
]
|
||||
},
|
||||
"execution_count": 7,
|
||||
"metadata": {},
|
||||
"output_type": "execute_result"
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"e.reset()\n",
|
||||
"e.step(e.action_space.sample())\n",
|
||||
"e.render()\n",
|
||||
"e.reset()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "547ccc71-8e9e-4d56-8cea-bfeddd002845",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": []
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "77a096e6-3fa9-48be-8bf0-06367ac9f9aa",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": []
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3 (ipykernel)",
|
||||
"language": "python",
|
||||
"name": "python3"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3",
|
||||
"version": "3.10.8"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 5
|
||||
}
|
||||
@ -0,0 +1,75 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "9d9f64ee-0186-4475-ba63-b5a569662227",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import gym\n",
|
||||
"for n in gym.envs.registry.all():\n",
|
||||
" print(n)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 1,
|
||||
"id": "ace720b1-7121-4a81-8369-f5cfc40cbe36",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"ename": "Exception",
|
||||
"evalue": "Please add mujoco library to your PATH:\nset PATH=C:\\Users\\gofor\\.mujoco\\mujoco210\\bin;%PATH%",
|
||||
"output_type": "error",
|
||||
"traceback": [
|
||||
"\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
|
||||
"\u001b[1;31mException\u001b[0m Traceback (most recent call last)",
|
||||
"Cell \u001b[1;32mIn [1], line 1\u001b[0m\n\u001b[1;32m----> 1\u001b[0m \u001b[38;5;28;01mimport\u001b[39;00m \u001b[38;5;21;01mmujoco_py\u001b[39;00m\n\u001b[0;32m 2\u001b[0m \u001b[38;5;28;01mimport\u001b[39;00m \u001b[38;5;21;01mos\u001b[39;00m\n\u001b[0;32m 3\u001b[0m mj_path \u001b[38;5;241m=\u001b[39m mujoco_py\u001b[38;5;241m.\u001b[39mutils\u001b[38;5;241m.\u001b[39mdiscover_mujoco()\n",
|
||||
"File \u001b[1;32m~\\venv310\\lib\\site-packages\\mujoco_py\\__init__.py:2\u001b[0m\n\u001b[0;32m 1\u001b[0m \u001b[38;5;66;03m#!/usr/bin/env python\u001b[39;00m\n\u001b[1;32m----> 2\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mmujoco_py\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mbuilder\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m cymj, ignore_mujoco_warnings, functions, MujocoException\n\u001b[0;32m 3\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mmujoco_py\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mgenerated\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m const\n\u001b[0;32m 4\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mmujoco_py\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mmjrenderpool\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m MjRenderPool\n",
|
||||
"File \u001b[1;32m~\\venv310\\lib\\site-packages\\mujoco_py\\builder.py:504\u001b[0m\n\u001b[0;32m 500\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m module\u001b[38;5;241m.\u001b[39mlib\u001b[38;5;241m.\u001b[39m__fun\n\u001b[0;32m 503\u001b[0m mujoco_path \u001b[38;5;241m=\u001b[39m discover_mujoco()\n\u001b[1;32m--> 504\u001b[0m cymj \u001b[38;5;241m=\u001b[39m \u001b[43mload_cython_ext\u001b[49m\u001b[43m(\u001b[49m\u001b[43mmujoco_path\u001b[49m\u001b[43m)\u001b[49m\n\u001b[0;32m 507\u001b[0m \u001b[38;5;66;03m# Trick to expose all mj* functions from mujoco in mujoco_py.*\u001b[39;00m\n\u001b[0;32m 508\u001b[0m \u001b[38;5;28;01mclass\u001b[39;00m \u001b[38;5;21;01mdict2\u001b[39;00m(\u001b[38;5;28mobject\u001b[39m):\n",
|
||||
"File \u001b[1;32m~\\venv310\\lib\\site-packages\\mujoco_py\\builder.py:83\u001b[0m, in \u001b[0;36mload_cython_ext\u001b[1;34m(mujoco_path)\u001b[0m\n\u001b[0;32m 81\u001b[0m var \u001b[38;5;241m=\u001b[39m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mPATH\u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[0;32m 82\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m var \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;129;01min\u001b[39;00m os\u001b[38;5;241m.\u001b[39menviron \u001b[38;5;129;01mor\u001b[39;00m lib_path \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;129;01min\u001b[39;00m os\u001b[38;5;241m.\u001b[39menviron[var]\u001b[38;5;241m.\u001b[39msplit(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m;\u001b[39m\u001b[38;5;124m\"\u001b[39m):\n\u001b[1;32m---> 83\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mException\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mPlease add mujoco library to your PATH:\u001b[39m\u001b[38;5;130;01m\\n\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\n\u001b[0;32m 84\u001b[0m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mset \u001b[39m\u001b[38;5;132;01m%s\u001b[39;00m\u001b[38;5;124m=\u001b[39m\u001b[38;5;132;01m%s\u001b[39;00m\u001b[38;5;124m;\u001b[39m\u001b[38;5;132;01m%%\u001b[39;00m\u001b[38;5;132;01m%s\u001b[39;00m\u001b[38;5;132;01m%%\u001b[39;00m\u001b[38;5;124m\"\u001b[39m \u001b[38;5;241m%\u001b[39m (var, lib_path, var))\n\u001b[0;32m 85\u001b[0m Builder \u001b[38;5;241m=\u001b[39m WindowsExtensionBuilder\n\u001b[0;32m 86\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n",
|
||||
"\u001b[1;31mException\u001b[0m: Please add mujoco library to your PATH:\nset PATH=C:\\Users\\gofor\\.mujoco\\mujoco210\\bin;%PATH%"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"import mujoco_py\n",
|
||||
"import os\n",
|
||||
"mj_path = mujoco_py.utils.discover_mujoco()\n",
|
||||
"xml_path = os.path.join(mj_path, 'model', 'humanoid.xml')\n",
|
||||
"model = mujoco_py.load_model_from_path(xml_path)\n",
|
||||
"sim = mujoco_py.MjSim(model)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "aa696af6-b6d2-4916-be15-93052eb43b3c",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": []
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3 (ipykernel)",
|
||||
"language": "python",
|
||||
"name": "python3"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3",
|
||||
"version": "3.10.8"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 5
|
||||
}
|
||||
File diff suppressed because one or more lines are too long
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
@ -0,0 +1,189 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# Run a pre-trained model\n",
|
||||
"\n",
|
||||
"This notebook loads a pre-trained model and uses it to play games. \n",
|
||||
"Note that it does not render the image of the game, it just prints out the episodic score. "
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 3,
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"ename": "ModuleNotFoundError",
|
||||
"evalue": "No module named 'gym'",
|
||||
"output_type": "error",
|
||||
"traceback": [
|
||||
"\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
|
||||
"\u001b[1;31mModuleNotFoundError\u001b[0m Traceback (most recent call last)",
|
||||
"Cell \u001b[1;32mIn [3], line 2\u001b[0m\n\u001b[0;32m 1\u001b[0m \u001b[38;5;66;03m# sanity check: can we create breakwall?\u001b[39;00m\n\u001b[1;32m----> 2\u001b[0m \u001b[38;5;28;01mimport\u001b[39;00m \u001b[38;5;21;01mgym\u001b[39;00m\n\u001b[0;32m 3\u001b[0m e \u001b[38;5;241m=\u001b[39m gym\u001b[38;5;241m.\u001b[39mmake(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mgym_gs:BreakwallNoFrameskip-v1\u001b[39m\u001b[38;5;124m'\u001b[39m)\n",
|
||||
"\u001b[1;31mModuleNotFoundError\u001b[0m: No module named 'gym'"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"# sanity check: can we create breakwall?\n",
|
||||
"import gym\n",
|
||||
"e = gym.make('gym_gs:BreakwallNoFrameskip-v1')"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": []
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": []
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": []
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# install baselines and other stuff\n",
|
||||
"!pip install git+https://github.com/openai/baselines.git"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"## full check - can we use the full opencv/ openai version \n",
|
||||
"## of the gym?\n",
|
||||
"\n",
|
||||
"# Script to test a pre-trained model\n",
|
||||
"# Written by Matthew Yee-King\n",
|
||||
"# MIT license \n",
|
||||
"# https://mit-license.org/\n",
|
||||
"\n",
|
||||
"import sys\n",
|
||||
"import os\n",
|
||||
"from baselines.common.atari_wrappers import make_atari, wrap_deepmind\n",
|
||||
"import numpy as np\n",
|
||||
"import tensorflow as tf\n",
|
||||
"from tensorflow import keras\n",
|
||||
"from tensorflow.keras import layers\n",
|
||||
"import datetime\n",
|
||||
"import random\n",
|
||||
"import time \n",
|
||||
"\n",
|
||||
"env_name = \"gym_gs:BreakwallNoFrameskip-v1\" \n",
|
||||
"# for notebook users - make sure you have uploaded your pre-trained\n",
|
||||
"# models... then adjust this to reflect the file path\n",
|
||||
"model_file = \"./pre-trained/mac_hard_breakwall/gym_gs:BreakwallNoFrameskip-v1_20211018-114642_5424\"\n",
|
||||
"\n",
|
||||
"def create_q_model(num_actions):\n",
|
||||
" # Network defined by the Deepmind paper\n",
|
||||
" inputs = layers.Input(shape=(84, 84, 4,))\n",
|
||||
" # Convolutions on the frames on the screen\n",
|
||||
" layer1 = layers.Conv2D(32, 8, strides=4, activation=\"relu\")(inputs) \n",
|
||||
" layer2 = layers.Conv2D(64, 4, strides=2, activation=\"relu\")(layer1)\n",
|
||||
" layer3 = layers.Conv2D(64, 3, strides=1, activation=\"relu\")(layer2)\n",
|
||||
" layer4 = layers.Flatten()(layer3)\n",
|
||||
" layer5 = layers.Dense(512, activation=\"relu\")(layer4) \n",
|
||||
" action = layers.Dense(num_actions, activation=\"linear\")(layer5) \n",
|
||||
" return keras.Model(inputs=inputs, outputs=action)\n",
|
||||
"\n",
|
||||
"def create_env(env_name, seed=42):\n",
|
||||
" try:\n",
|
||||
" # Use the Baseline Atari environment because of Deepmind helper functions\n",
|
||||
" env = make_atari(env_name)\n",
|
||||
" # Warp the frames, grey scale, stake four frame and scale to smaller ratio\n",
|
||||
" env = wrap_deepmind(env, frame_stack=True, scale=True)\n",
|
||||
" print(\"Loaded gym\")\n",
|
||||
" env.seed(seed)\n",
|
||||
" return env\n",
|
||||
" except:\n",
|
||||
" print(\"Failed to make gym env\", env_name)\n",
|
||||
" return None\n",
|
||||
"\n",
|
||||
"def run_sim(env, model, frame_count):\n",
|
||||
" state = np.array(env.reset())\n",
|
||||
" total_reward = 0\n",
|
||||
" for i in range(frame_count):\n",
|
||||
" # in the notebook version we cannot really \n",
|
||||
" # render in realtime, so you just have\n",
|
||||
" # to check the score :( \n",
|
||||
" #env.render('human')\n",
|
||||
" state_tensor = keras.backend.constant(state)\n",
|
||||
" state_tensor = keras.backend.expand_dims(state_tensor, 0)\n",
|
||||
" action_values = model(state_tensor, training=False)\n",
|
||||
" # Take best action\n",
|
||||
" action = keras.backend.argmax(action_values[0]).numpy()\n",
|
||||
" state, reward, done, _ = env.step(action)\n",
|
||||
" state = np.array(state)\n",
|
||||
" total_reward += reward\n",
|
||||
" if done:\n",
|
||||
" print(\"Game over at frame\", i, \"rew\", total_reward)\n",
|
||||
" env.reset()\n",
|
||||
" #break\n",
|
||||
" #time.sleep(0.1)\n",
|
||||
" print(\"Sim ended : rew is \", total_reward)\n",
|
||||
"\n",
|
||||
"def main(env_name, model_file,frame_count=1000, seed=42):\n",
|
||||
" env = create_env(env_name=env_name)\n",
|
||||
" assert env is not None, \"Failed to make env \" + env_name\n",
|
||||
" model = create_q_model(num_actions=env.action_space.n)\n",
|
||||
" model_testfile = model_file + \".data-00000-of-00001\"\n",
|
||||
" assert os.path.exists(model_testfile), \"Failed to load model: \" + model_testfile\n",
|
||||
" print(\"Model weights look loadable\", model_testfile)\n",
|
||||
" model.load_weights(model_file)\n",
|
||||
" print(\"Model loaded weights - starting sim\")\n",
|
||||
" run_sim(env, model, frame_count)\n",
|
||||
" \n",
|
||||
"main(env_name, model_file, frame_count=1000)\n",
|
||||
"\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": []
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3 (ipykernel)",
|
||||
"language": "python",
|
||||
"name": "python3"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3",
|
||||
"version": "3.11.0"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 4
|
||||
}
|
||||
49
CM3020 Artificial Intelligence/Week 3/Week 3 Notes.md
Normal file
49
CM3020 Artificial Intelligence/Week 3/Week 3 Notes.md
Normal file
@ -0,0 +1,49 @@
|
||||
# What is a gym?
|
||||
It's a way of providing standarized environments (API) for reinforcement learnign algorithms to operate in.
|
||||
|
||||
Allows the developer to focus on the agent instead of the simulation.
|
||||
|
||||
Includes versioned, standard set of enviroments enabling easy comparison.
|
||||
|
||||
---
|
||||
There are various gym-like systems such as:
|
||||
* Arcade Learning Enviroment (2013)
|
||||
* Vizdoom (2016)
|
||||
* OpenAI gym (2016)
|
||||
|
||||
# Open AI gym
|
||||
Aims to combine benchmarks collections in a convenient and accessible software package.
|
||||
|
||||
# Keras
|
||||
A Neural network is a set of processing units which are interconnected, they are fed with values and also output values.
|
||||
|
||||
Keras is a high level neural network API sitting atop tensorflow.
|
||||
|
||||
The DQN we'll look at uses Keras.
|
||||
|
||||
# Layers
|
||||
Neural networks are made of layers of which there are different types with different nodes in them which process signals in different ways.
|
||||
|
||||
Keras provides many different types of layers
|
||||
https://keras.io/api/layers
|
||||
|
||||
Convolution layers are great for image processing.
|
||||
|
||||
Recurrent layers for sequential data.
|
||||
|
||||
Attention layers for sequential data (better).
|
||||
|
||||
DQN uses some common layers plus some convolutional layers.
|
||||
|
||||
The nodes are fed data, the nodes create signals with value which are sent to other nodes. The results will be back propagated to the network again for processing.
|
||||
|
||||
# Convolution layers
|
||||
Convolution is a filtering technique often used for images but also other signals. This is done by adding together scaled values of the original pixel and its surrounding pixels.
|
||||
|
||||
Example of convolution filters are edge detection and blur filter.
|
||||
|
||||
We can visualize what convolutional layers are doing to image data as it passes through the neural network.
|
||||
|
||||
https://github.com/gabrielpierobon/cnnshapes/
|
||||
|
||||
|
||||
Reference in New Issue
Block a user