您最多选择25个主题
主题必须以中文或者字母或数字开头，可以包含连字符 (-)，并且长度不得超过35个字符
132 行
15 KiB
132 行
15 KiB
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 1,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"import csv\n",
|
|
"import numpy as np\n",
|
|
"import matplotlib.pyplot as plt"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 2,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"def read_reward(dirname):\n",
|
|
" reward = []\n",
|
|
" with open(dirname+'/Simple.csv', newline='') as csvfile:\n",
|
|
" reader = csv.DictReader(csvfile)\n",
|
|
" for row in reader:\n",
|
|
" reward.append(float(row['Environment/Cumulative Reward']))\n",
|
|
" return reward"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 3,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"def smooth(scalars, weight: float): # Weight between 0 and 1\n",
|
|
" last = scalars[0] # First value in the plot (first timestep)\n",
|
|
" smoothed = list()\n",
|
|
" for point in scalars:\n",
|
|
" smoothed_val = last * weight + (1 - weight) * point # Calculate smoothed value\n",
|
|
" smoothed.append(smoothed_val) # Save it\n",
|
|
" last = smoothed_val # Anchor the last smoothed value\n",
|
|
"\n",
|
|
" return smoothed"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 8,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"seeds = [0,1,2,3,4]\n",
|
|
"obs_types = [\"long\", \"longpre\"]"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 9,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"rewards_ppo = {} \n",
|
|
"for obs in obs_types:\n",
|
|
" rewards_ppo[obs] = []\n",
|
|
" for s in seeds:\n",
|
|
" rewards_ppo[obs].append(read_reward(\"transfer_results/ppo_{}_s{}\".format(obs, s)))\n",
|
|
" rewards_ppo[obs] = np.array(rewards_ppo[obs])"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 15,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"image/png": "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\n",
|
|
"text/plain": [
|
|
"<Figure size 432x288 with 1 Axes>"
|
|
]
|
|
},
|
|
"metadata": {
|
|
"needs_background": "light"
|
|
},
|
|
"output_type": "display_data"
|
|
}
|
|
],
|
|
"source": [
|
|
"xs = list(range(len(rewards_ppo[\"long\"])))\n",
|
|
"for obs in obs_types:\n",
|
|
" plt.plot(xs, smooth(np.mean(rewards_ppo[obs], axis=1), 0.8), label=obs)\n",
|
|
"plt.legend()\n",
|
|
"plt.show()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"rewards_model = {} \n",
|
|
"for obs in obs_types:\n",
|
|
" rewards_ppo[obs] = []\n",
|
|
" for s in seeds:\n",
|
|
" rewards_ppo[obs].append(read_reward(\"transfer_results/ppo_{}_s{}\".format(obs, s)))\n",
|
|
" rewards_ppo[obs] = np.array(rewards_ppo[obs])"
|
|
]
|
|
}
|
|
],
|
|
"metadata": {
|
|
"kernelspec": {
|
|
"display_name": "mlagents-env",
|
|
"language": "python",
|
|
"name": "mlagents-env"
|
|
},
|
|
"language_info": {
|
|
"codemirror_mode": {
|
|
"name": "ipython",
|
|
"version": 3
|
|
},
|
|
"file_extension": ".py",
|
|
"mimetype": "text/x-python",
|
|
"name": "python",
|
|
"nbconvert_exporter": "python",
|
|
"pygments_lexer": "ipython3",
|
|
"version": "3.7.6"
|
|
}
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 4
|
|
}
|