gryf/coach
1
0
Fork 0
mirror of https://github.com/gryf/coach.git synced 2025-12-17 19:20:19 +01:00
Code Issues Projects Releases Wiki Activity
Files
master
coach/docs/_modules/rl_coach/agents/acer_agent.html
anabwan ddffac8570 fixed release version (#333) 
* fixed release version

* update docs
2019-05-28 11:11:15 +03:00

423 lines
40 KiB
HTML
Raw Permalink Blame History

<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>rl_coach.agents.acer_agent &mdash; Reinforcement Learning Coach 0.12.0 documentation</title>
<script type="text/javascript" src="../../../_static/js/modernizr.min.js"></script>
<script type="text/javascript" id="documentation_options" data-url_root="../../../" src="../../../_static/documentation_options.js"></script>
<script type="text/javascript" src="../../../_static/jquery.js"></script>
<script type="text/javascript" src="../../../_static/underscore.js"></script>
<script type="text/javascript" src="../../../_static/doctools.js"></script>
<script type="text/javascript" src="../../../_static/language_data.js"></script>
<script async="async" type="text/javascript" src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/latest.js?config=TeX-AMS-MML_HTMLorMML"></script>
<script type="text/javascript" src="../../../_static/js/theme.js"></script>
<link rel="stylesheet" href="../../../_static/css/theme.css" type="text/css" />
<link rel="stylesheet" href="../../../_static/pygments.css" type="text/css" />
<link rel="stylesheet" href="../../../_static/css/custom.css" type="text/css" />
<link rel="index" title="Index" href="../../../genindex.html" />
<link rel="search" title="Search" href="../../../search.html" />
<link href="../../../_static/css/custom.css" rel="stylesheet" type="text/css">
</head>
<body class="wy-body-for-nav">
<div class="wy-grid-for-nav">
<nav data-toggle="wy-nav-shift" class="wy-nav-side">
<div class="wy-side-scroll">
<div class="wy-side-nav-search" >
<a href="../../../index.html" class="icon icon-home"> Reinforcement Learning Coach
<img src="../../../_static/dark_logo.png" class="logo" alt="Logo"/>
</a>
<div role="search">
<form id="rtd-search-form" class="wy-form" action="../../../search.html" method="get">
<input type="text" name="q" placeholder="Search docs" />
<input type="hidden" name="check_keywords" value="yes" />
<input type="hidden" name="area" value="default" />
</form>
</div>
</div>
<div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation">
<p class="caption"><span class="caption-text">Intro</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../../usage.html">Usage</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../dist_usage.html">Usage - Distributed Coach</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../features/index.html">Features</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../selecting_an_algorithm.html">Selecting an Algorithm</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../dashboard.html">Coach Dashboard</a></li>
</ul>
<p class="caption"><span class="caption-text">Design</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../../design/control_flow.html">Control Flow</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../design/network.html">Network Design</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../design/horizontal_scaling.html">Distributed Coach - Horizontal Scale-Out</a></li>
</ul>
<p class="caption"><span class="caption-text">Contributing</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../../contributing/add_agent.html">Adding a New Agent</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../contributing/add_env.html">Adding a New Environment</a></li>
</ul>
<p class="caption"><span class="caption-text">Components</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../../components/agents/index.html">Agents</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../components/architectures/index.html">Architectures</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../components/data_stores/index.html">Data Stores</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../components/environments/index.html">Environments</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../components/exploration_policies/index.html">Exploration Policies</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../components/filters/index.html">Filters</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../components/memories/index.html">Memories</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../components/memory_backends/index.html">Memory Backends</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../components/orchestrators/index.html">Orchestrators</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../components/core_types.html">Core Types</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../components/spaces.html">Spaces</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../components/additional_parameters.html">Additional Parameters</a></li>
</ul>
</div>
</div>
</nav>
<section data-toggle="wy-nav-shift" class="wy-nav-content-wrap">
<nav class="wy-nav-top" aria-label="top navigation">
<i data-toggle="wy-nav-top" class="fa fa-bars"></i>
<a href="../../../index.html">Reinforcement Learning Coach</a>
</nav>
<div class="wy-nav-content">
<div class="rst-content">
<div role="navigation" aria-label="breadcrumbs navigation">
<ul class="wy-breadcrumbs">
<li><a href="../../../index.html">Docs</a> &raquo;</li>
<li><a href="../../index.html">Module code</a> &raquo;</li>
<li>rl_coach.agents.acer_agent</li>
<li class="wy-breadcrumbs-aside">
</li>
</ul>
<hr/>
</div>
<div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
<div itemprop="articleBody">
<h1>Source code for rl_coach.agents.acer_agent</h1><div class="highlight"><pre>
<span></span><span class="c1">#</span>
<span class="c1"># Copyright (c) 2017 Intel Corporation</span>
<span class="c1">#</span>
<span class="c1"># Licensed under the Apache License, Version 2.0 (the &quot;License&quot;);</span>
<span class="c1"># you may not use this file except in compliance with the License.</span>
<span class="c1"># You may obtain a copy of the License at</span>
<span class="c1">#</span>
<span class="c1"># http://www.apache.org/licenses/LICENSE-2.0</span>
<span class="c1">#</span>
<span class="c1"># Unless required by applicable law or agreed to in writing, software</span>
<span class="c1"># distributed under the License is distributed on an &quot;AS IS&quot; BASIS,</span>
<span class="c1"># WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
<span class="c1"># See the License for the specific language governing permissions and</span>
<span class="c1"># limitations under the License.</span>
<span class="c1">#</span>
<span class="kn">from</span> <span class="nn">typing</span> <span class="k">import</span> <span class="n">Union</span>
<span class="kn">import</span> <span class="nn">numpy</span> <span class="k">as</span> <span class="nn">np</span>
<span class="kn">from</span> <span class="nn">rl_coach.agents.policy_optimization_agent</span> <span class="k">import</span> <span class="n">PolicyOptimizationAgent</span>
<span class="kn">from</span> <span class="nn">rl_coach.architectures.embedder_parameters</span> <span class="k">import</span> <span class="n">InputEmbedderParameters</span>
<span class="kn">from</span> <span class="nn">rl_coach.architectures.head_parameters</span> <span class="k">import</span> <span class="n">ACERPolicyHeadParameters</span><span class="p">,</span> <span class="n">QHeadParameters</span>
<span class="kn">from</span> <span class="nn">rl_coach.architectures.middleware_parameters</span> <span class="k">import</span> <span class="n">FCMiddlewareParameters</span>
<span class="kn">from</span> <span class="nn">rl_coach.base_parameters</span> <span class="k">import</span> <span class="n">AlgorithmParameters</span><span class="p">,</span> <span class="n">NetworkParameters</span><span class="p">,</span> <span class="n">AgentParameters</span>
<span class="kn">from</span> <span class="nn">rl_coach.core_types</span> <span class="k">import</span> <span class="n">Batch</span>
<span class="kn">from</span> <span class="nn">rl_coach.exploration_policies.categorical</span> <span class="k">import</span> <span class="n">CategoricalParameters</span>
<span class="kn">from</span> <span class="nn">rl_coach.memories.episodic.episodic_experience_replay</span> <span class="k">import</span> <span class="n">EpisodicExperienceReplayParameters</span>
<span class="kn">from</span> <span class="nn">rl_coach.spaces</span> <span class="k">import</span> <span class="n">DiscreteActionSpace</span>
<span class="kn">from</span> <span class="nn">rl_coach.utils</span> <span class="k">import</span> <span class="n">eps</span><span class="p">,</span> <span class="n">last_sample</span>
<div class="viewcode-block" id="ACERAlgorithmParameters"><a class="viewcode-back" href="../../../components/agents/policy_optimization/acer.html#rl_coach.agents.acer_agent.ACERAlgorithmParameters">[docs]</a><span class="k">class</span> <span class="nc">ACERAlgorithmParameters</span><span class="p">(</span><span class="n">AlgorithmParameters</span><span class="p">):</span>
<span class="sd">&quot;&quot;&quot;</span>
<span class="sd"> :param num_steps_between_gradient_updates: (int)</span>
<span class="sd"> Every num_steps_between_gradient_updates transitions will be considered as a single batch and use for</span>
<span class="sd"> accumulating gradients. This is also the number of steps used for bootstrapping according to the n-step formulation.</span>
<span class="sd"> :param ratio_of_replay: (int)</span>
<span class="sd"> The number of off-policy training iterations in each ACER iteration.</span>
<span class="sd"> :param num_transitions_to_start_replay: (int)</span>
<span class="sd"> Number of environment steps until ACER starts to train off-policy from the experience replay.</span>
<span class="sd"> This emulates a heat-up phase where the agents learns only on-policy until there are enough transitions in</span>
<span class="sd"> the experience replay to start the off-policy training.</span>
<span class="sd"> :param rate_for_copying_weights_to_target: (float)</span>
<span class="sd"> The rate of the exponential moving average for the average policy which is used for the trust region optimization.</span>
<span class="sd"> The target network in this algorithm is used as the average policy.</span>
<span class="sd"> :param importance_weight_truncation: (float)</span>
<span class="sd"> The clipping constant for the importance weight truncation (not used in the Q-retrace calculation).</span>
<span class="sd"> :param use_trust_region_optimization: (bool)</span>
<span class="sd"> If set to True, the gradients of the network will be modified with a term dependant on the KL divergence between</span>
<span class="sd"> the average policy and the current one, to bound the change of the policy during the network update.</span>
<span class="sd"> :param max_KL_divergence: (float)</span>
<span class="sd"> The upper bound parameter for the trust region optimization, use_trust_region_optimization needs to be set true</span>
<span class="sd"> for this parameter to have an effect.</span>
<span class="sd"> :param beta_entropy: (float)</span>
<span class="sd"> An entropy regulaization term can be added to the loss function in order to control exploration. This term</span>
<span class="sd"> is weighted using the beta value defined by beta_entropy.</span>
<span class="sd"> &quot;&quot;&quot;</span>
<span class="k">def</span> <span class="nf">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
<span class="nb">super</span><span class="p">()</span><span class="o">.</span><span class="fm">__init__</span><span class="p">()</span>
<span class="bp">self</span><span class="o">.</span><span class="n">apply_gradients_every_x_episodes</span> <span class="o">=</span> <span class="mi">5</span>
<span class="bp">self</span><span class="o">.</span><span class="n">num_steps_between_gradient_updates</span> <span class="o">=</span> <span class="mi">5000</span>
<span class="bp">self</span><span class="o">.</span><span class="n">ratio_of_replay</span> <span class="o">=</span> <span class="mi">4</span>
<span class="bp">self</span><span class="o">.</span><span class="n">num_transitions_to_start_replay</span> <span class="o">=</span> <span class="mi">10000</span>
<span class="bp">self</span><span class="o">.</span><span class="n">rate_for_copying_weights_to_target</span> <span class="o">=</span> <span class="mf">0.01</span>
<span class="bp">self</span><span class="o">.</span><span class="n">importance_weight_truncation</span> <span class="o">=</span> <span class="mf">10.0</span>
<span class="bp">self</span><span class="o">.</span><span class="n">use_trust_region_optimization</span> <span class="o">=</span> <span class="kc">True</span>
<span class="bp">self</span><span class="o">.</span><span class="n">max_KL_divergence</span> <span class="o">=</span> <span class="mf">1.0</span>
<span class="bp">self</span><span class="o">.</span><span class="n">beta_entropy</span> <span class="o">=</span> <span class="mi">0</span></div>
<span class="k">class</span> <span class="nc">ACERNetworkParameters</span><span class="p">(</span><span class="n">NetworkParameters</span><span class="p">):</span>
<span class="k">def</span> <span class="nf">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
<span class="nb">super</span><span class="p">()</span><span class="o">.</span><span class="fm">__init__</span><span class="p">()</span>
<span class="bp">self</span><span class="o">.</span><span class="n">input_embedders_parameters</span> <span class="o">=</span> <span class="p">{</span><span class="s1">&#39;observation&#39;</span><span class="p">:</span> <span class="n">InputEmbedderParameters</span><span class="p">()}</span>
<span class="bp">self</span><span class="o">.</span><span class="n">middleware_parameters</span> <span class="o">=</span> <span class="n">FCMiddlewareParameters</span><span class="p">()</span>
<span class="bp">self</span><span class="o">.</span><span class="n">heads_parameters</span> <span class="o">=</span> <span class="p">[</span><span class="n">QHeadParameters</span><span class="p">(</span><span class="n">loss_weight</span><span class="o">=</span><span class="mf">0.5</span><span class="p">),</span> <span class="n">ACERPolicyHeadParameters</span><span class="p">(</span><span class="n">loss_weight</span><span class="o">=</span><span class="mf">1.0</span><span class="p">)]</span>
<span class="bp">self</span><span class="o">.</span><span class="n">optimizer_type</span> <span class="o">=</span> <span class="s1">&#39;Adam&#39;</span>
<span class="bp">self</span><span class="o">.</span><span class="n">async_training</span> <span class="o">=</span> <span class="kc">True</span>
<span class="bp">self</span><span class="o">.</span><span class="n">clip_gradients</span> <span class="o">=</span> <span class="mf">40.0</span>
<span class="bp">self</span><span class="o">.</span><span class="n">create_target_network</span> <span class="o">=</span> <span class="kc">True</span>
<span class="k">class</span> <span class="nc">ACERAgentParameters</span><span class="p">(</span><span class="n">AgentParameters</span><span class="p">):</span>
<span class="k">def</span> <span class="nf">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
<span class="nb">super</span><span class="p">()</span><span class="o">.</span><span class="fm">__init__</span><span class="p">(</span><span class="n">algorithm</span><span class="o">=</span><span class="n">ACERAlgorithmParameters</span><span class="p">(),</span>
<span class="n">exploration</span><span class="o">=</span><span class="p">{</span><span class="n">DiscreteActionSpace</span><span class="p">:</span> <span class="n">CategoricalParameters</span><span class="p">()},</span>
<span class="n">memory</span><span class="o">=</span><span class="n">EpisodicExperienceReplayParameters</span><span class="p">(),</span>
<span class="n">networks</span><span class="o">=</span><span class="p">{</span><span class="s2">&quot;main&quot;</span><span class="p">:</span> <span class="n">ACERNetworkParameters</span><span class="p">()})</span>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">path</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
<span class="k">return</span> <span class="s1">&#39;rl_coach.agents.acer_agent:ACERAgent&#39;</span>
<span class="c1"># Actor-Critic with Experience Replay - https://arxiv.org/abs/1611.01224</span>
<span class="k">class</span> <span class="nc">ACERAgent</span><span class="p">(</span><span class="n">PolicyOptimizationAgent</span><span class="p">):</span>
<span class="k">def</span> <span class="nf">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">agent_parameters</span><span class="p">,</span> <span class="n">parent</span><span class="p">:</span> <span class="n">Union</span><span class="p">[</span><span class="s1">&#39;LevelManager&#39;</span><span class="p">,</span> <span class="s1">&#39;CompositeAgent&#39;</span><span class="p">]</span><span class="o">=</span><span class="kc">None</span><span class="p">):</span>
<span class="nb">super</span><span class="p">()</span><span class="o">.</span><span class="fm">__init__</span><span class="p">(</span><span class="n">agent_parameters</span><span class="p">,</span> <span class="n">parent</span><span class="p">)</span>
<span class="c1"># signals definition</span>
<span class="bp">self</span><span class="o">.</span><span class="n">q_loss</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">register_signal</span><span class="p">(</span><span class="s1">&#39;Q Loss&#39;</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">policy_loss</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">register_signal</span><span class="p">(</span><span class="s1">&#39;Policy Loss&#39;</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">probability_loss</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">register_signal</span><span class="p">(</span><span class="s1">&#39;Probability Loss&#39;</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">bias_correction_loss</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">register_signal</span><span class="p">(</span><span class="s1">&#39;Bias Correction Loss&#39;</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">unclipped_grads</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">register_signal</span><span class="p">(</span><span class="s1">&#39;Grads (unclipped)&#39;</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">V_Values</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">register_signal</span><span class="p">(</span><span class="s1">&#39;Values&#39;</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">kl_divergence</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">register_signal</span><span class="p">(</span><span class="s1">&#39;KL Divergence&#39;</span><span class="p">)</span>
<span class="k">def</span> <span class="nf">_learn_from_batch</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">batch</span><span class="p">):</span>
<span class="n">fetches</span> <span class="o">=</span> <span class="p">[</span><span class="bp">self</span><span class="o">.</span><span class="n">networks</span><span class="p">[</span><span class="s1">&#39;main&#39;</span><span class="p">]</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">output_heads</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">.</span><span class="n">probability_loss</span><span class="p">,</span>
<span class="bp">self</span><span class="o">.</span><span class="n">networks</span><span class="p">[</span><span class="s1">&#39;main&#39;</span><span class="p">]</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">output_heads</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">.</span><span class="n">bias_correction_loss</span><span class="p">,</span>
<span class="bp">self</span><span class="o">.</span><span class="n">networks</span><span class="p">[</span><span class="s1">&#39;main&#39;</span><span class="p">]</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">output_heads</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">.</span><span class="n">kl_divergence</span><span class="p">]</span>
<span class="c1"># batch contains a list of transitions to learn from</span>
<span class="n">network_keys</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">ap</span><span class="o">.</span><span class="n">network_wrappers</span><span class="p">[</span><span class="s1">&#39;main&#39;</span><span class="p">]</span><span class="o">.</span><span class="n">input_embedders_parameters</span><span class="o">.</span><span class="n">keys</span><span class="p">()</span>
<span class="c1"># get the values for the current states</span>
<span class="n">Q_values</span><span class="p">,</span> <span class="n">policy_prob</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">networks</span><span class="p">[</span><span class="s1">&#39;main&#39;</span><span class="p">]</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">batch</span><span class="o">.</span><span class="n">states</span><span class="p">(</span><span class="n">network_keys</span><span class="p">))</span>
<span class="n">avg_policy_prob</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">networks</span><span class="p">[</span><span class="s1">&#39;main&#39;</span><span class="p">]</span><span class="o">.</span><span class="n">target_network</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">batch</span><span class="o">.</span><span class="n">states</span><span class="p">(</span><span class="n">network_keys</span><span class="p">))[</span><span class="mi">1</span><span class="p">]</span>
<span class="n">current_state_values</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">sum</span><span class="p">(</span><span class="n">policy_prob</span> <span class="o">*</span> <span class="n">Q_values</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span>
<span class="n">actions</span> <span class="o">=</span> <span class="n">batch</span><span class="o">.</span><span class="n">actions</span><span class="p">()</span>
<span class="n">num_transitions</span> <span class="o">=</span> <span class="n">batch</span><span class="o">.</span><span class="n">size</span>
<span class="n">Q_head_targets</span> <span class="o">=</span> <span class="n">Q_values</span>
<span class="n">Q_i</span> <span class="o">=</span> <span class="n">Q_values</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">arange</span><span class="p">(</span><span class="n">num_transitions</span><span class="p">),</span> <span class="n">actions</span><span class="p">]</span>
<span class="n">mu</span> <span class="o">=</span> <span class="n">batch</span><span class="o">.</span><span class="n">info</span><span class="p">(</span><span class="s1">&#39;all_action_probabilities&#39;</span><span class="p">)</span>
<span class="n">rho</span> <span class="o">=</span> <span class="n">policy_prob</span> <span class="o">/</span> <span class="p">(</span><span class="n">mu</span> <span class="o">+</span> <span class="n">eps</span><span class="p">)</span>
<span class="n">rho_i</span> <span class="o">=</span> <span class="n">rho</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">arange</span><span class="p">(</span><span class="n">batch</span><span class="o">.</span><span class="n">size</span><span class="p">),</span> <span class="n">actions</span><span class="p">]</span>
<span class="n">rho_bar</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">minimum</span><span class="p">(</span><span class="mf">1.0</span><span class="p">,</span> <span class="n">rho_i</span><span class="p">)</span>
<span class="k">if</span> <span class="n">batch</span><span class="o">.</span><span class="n">game_overs</span><span class="p">()[</span><span class="o">-</span><span class="mi">1</span><span class="p">]:</span>
<span class="n">Qret</span> <span class="o">=</span> <span class="mi">0</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">result</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">networks</span><span class="p">[</span><span class="s1">&#39;main&#39;</span><span class="p">]</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">last_sample</span><span class="p">(</span><span class="n">batch</span><span class="o">.</span><span class="n">next_states</span><span class="p">(</span><span class="n">network_keys</span><span class="p">)))</span>
<span class="n">Qret</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">sum</span><span class="p">(</span><span class="n">result</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span> <span class="o">*</span> <span class="n">result</span><span class="p">[</span><span class="mi">1</span><span class="p">],</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)[</span><span class="mi">0</span><span class="p">]</span>
<span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="nb">reversed</span><span class="p">(</span><span class="nb">range</span><span class="p">(</span><span class="n">num_transitions</span><span class="p">)):</span>
<span class="n">Qret</span> <span class="o">=</span> <span class="n">batch</span><span class="o">.</span><span class="n">rewards</span><span class="p">()[</span><span class="n">i</span><span class="p">]</span> <span class="o">+</span> <span class="bp">self</span><span class="o">.</span><span class="n">ap</span><span class="o">.</span><span class="n">algorithm</span><span class="o">.</span><span class="n">discount</span> <span class="o">*</span> <span class="n">Qret</span>
<span class="n">Q_head_targets</span><span class="p">[</span><span class="n">i</span><span class="p">,</span> <span class="n">actions</span><span class="p">[</span><span class="n">i</span><span class="p">]]</span> <span class="o">=</span> <span class="n">Qret</span>
<span class="n">Qret</span> <span class="o">=</span> <span class="n">rho_bar</span><span class="p">[</span><span class="n">i</span><span class="p">]</span> <span class="o">*</span> <span class="p">(</span><span class="n">Qret</span> <span class="o">-</span> <span class="n">Q_i</span><span class="p">[</span><span class="n">i</span><span class="p">])</span> <span class="o">+</span> <span class="n">current_state_values</span><span class="p">[</span><span class="n">i</span><span class="p">]</span>
<span class="n">Q_retrace</span> <span class="o">=</span> <span class="n">Q_head_targets</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">arange</span><span class="p">(</span><span class="n">num_transitions</span><span class="p">),</span> <span class="n">actions</span><span class="p">]</span>
<span class="c1"># train</span>
<span class="n">result</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">networks</span><span class="p">[</span><span class="s1">&#39;main&#39;</span><span class="p">]</span><span class="o">.</span><span class="n">train_and_sync_networks</span><span class="p">({</span><span class="o">**</span><span class="n">batch</span><span class="o">.</span><span class="n">states</span><span class="p">(</span><span class="n">network_keys</span><span class="p">),</span>
<span class="s1">&#39;output_1_0&#39;</span><span class="p">:</span> <span class="n">actions</span><span class="p">,</span>
<span class="s1">&#39;output_1_1&#39;</span><span class="p">:</span> <span class="n">rho</span><span class="p">,</span>
<span class="s1">&#39;output_1_2&#39;</span><span class="p">:</span> <span class="n">rho_i</span><span class="p">,</span>
<span class="s1">&#39;output_1_3&#39;</span><span class="p">:</span> <span class="n">Q_values</span><span class="p">,</span>
<span class="s1">&#39;output_1_4&#39;</span><span class="p">:</span> <span class="n">Q_retrace</span><span class="p">,</span>
<span class="s1">&#39;output_1_5&#39;</span><span class="p">:</span> <span class="n">avg_policy_prob</span><span class="p">},</span>
<span class="p">[</span><span class="n">Q_head_targets</span><span class="p">,</span> <span class="n">current_state_values</span><span class="p">],</span>
<span class="n">additional_fetches</span><span class="o">=</span><span class="n">fetches</span><span class="p">)</span>
<span class="k">for</span> <span class="n">network</span> <span class="ow">in</span> <span class="bp">self</span><span class="o">.</span><span class="n">networks</span><span class="o">.</span><span class="n">values</span><span class="p">():</span>
<span class="n">network</span><span class="o">.</span><span class="n">update_target_network</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">ap</span><span class="o">.</span><span class="n">algorithm</span><span class="o">.</span><span class="n">rate_for_copying_weights_to_target</span><span class="p">)</span>
<span class="c1"># logging</span>
<span class="n">total_loss</span><span class="p">,</span> <span class="n">losses</span><span class="p">,</span> <span class="n">unclipped_grads</span><span class="p">,</span> <span class="n">fetch_result</span> <span class="o">=</span> <span class="n">result</span><span class="p">[:</span><span class="mi">4</span><span class="p">]</span>
<span class="bp">self</span><span class="o">.</span><span class="n">q_loss</span><span class="o">.</span><span class="n">add_sample</span><span class="p">(</span><span class="n">losses</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span>
<span class="bp">self</span><span class="o">.</span><span class="n">policy_loss</span><span class="o">.</span><span class="n">add_sample</span><span class="p">(</span><span class="n">losses</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span>
<span class="bp">self</span><span class="o">.</span><span class="n">probability_loss</span><span class="o">.</span><span class="n">add_sample</span><span class="p">(</span><span class="n">fetch_result</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span>
<span class="bp">self</span><span class="o">.</span><span class="n">bias_correction_loss</span><span class="o">.</span><span class="n">add_sample</span><span class="p">(</span><span class="n">fetch_result</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span>
<span class="bp">self</span><span class="o">.</span><span class="n">unclipped_grads</span><span class="o">.</span><span class="n">add_sample</span><span class="p">(</span><span class="n">unclipped_grads</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">V_Values</span><span class="o">.</span><span class="n">add_sample</span><span class="p">(</span><span class="n">current_state_values</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">kl_divergence</span><span class="o">.</span><span class="n">add_sample</span><span class="p">(</span><span class="n">fetch_result</span><span class="p">[</span><span class="mi">2</span><span class="p">])</span>
<span class="k">return</span> <span class="n">total_loss</span><span class="p">,</span> <span class="n">losses</span><span class="p">,</span> <span class="n">unclipped_grads</span>
<span class="k">def</span> <span class="nf">learn_from_batch</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">batch</span><span class="p">):</span>
<span class="c1"># perform on-policy training iteration</span>
<span class="n">total_loss</span><span class="p">,</span> <span class="n">losses</span><span class="p">,</span> <span class="n">unclipped_grads</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_learn_from_batch</span><span class="p">(</span><span class="n">batch</span><span class="p">)</span>
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">ap</span><span class="o">.</span><span class="n">algorithm</span><span class="o">.</span><span class="n">ratio_of_replay</span> <span class="o">&gt;</span> <span class="mi">0</span> \
<span class="ow">and</span> <span class="bp">self</span><span class="o">.</span><span class="n">memory</span><span class="o">.</span><span class="n">num_transitions</span><span class="p">()</span> <span class="o">&gt;</span> <span class="bp">self</span><span class="o">.</span><span class="n">ap</span><span class="o">.</span><span class="n">algorithm</span><span class="o">.</span><span class="n">num_transitions_to_start_replay</span><span class="p">:</span>
<span class="n">n</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">poisson</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">ap</span><span class="o">.</span><span class="n">algorithm</span><span class="o">.</span><span class="n">ratio_of_replay</span><span class="p">)</span>
<span class="c1"># perform n off-policy training iterations</span>
<span class="k">for</span> <span class="n">_</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="n">n</span><span class="p">):</span>
<span class="n">new_batch</span> <span class="o">=</span> <span class="n">Batch</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">call_memory</span><span class="p">(</span><span class="s1">&#39;sample&#39;</span><span class="p">,</span> <span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">ap</span><span class="o">.</span><span class="n">algorithm</span><span class="o">.</span><span class="n">num_steps_between_gradient_updates</span><span class="p">,</span> <span class="kc">True</span><span class="p">)))</span>
<span class="n">result</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_learn_from_batch</span><span class="p">(</span><span class="n">new_batch</span><span class="p">)</span>
<span class="n">total_loss</span> <span class="o">+=</span> <span class="n">result</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span>
<span class="n">losses</span> <span class="o">+=</span> <span class="n">result</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span>
<span class="n">unclipped_grads</span> <span class="o">+=</span> <span class="n">result</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span>
<span class="k">return</span> <span class="n">total_loss</span><span class="p">,</span> <span class="n">losses</span><span class="p">,</span> <span class="n">unclipped_grads</span>
<span class="k">def</span> <span class="nf">get_prediction</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">states</span><span class="p">):</span>
<span class="n">tf_input_state</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">prepare_batch_for_inference</span><span class="p">(</span><span class="n">states</span><span class="p">,</span> <span class="s2">&quot;main&quot;</span><span class="p">)</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">networks</span><span class="p">[</span><span class="s1">&#39;main&#39;</span><span class="p">]</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">tf_input_state</span><span class="p">)[</span><span class="mi">1</span><span class="p">:]</span> <span class="c1"># index 0 is the state value</span>
</pre></div>
</div>
</div>
<footer>
<hr/>
<div role="contentinfo">
<p>
&copy; Copyright 2018-2019, Intel AI Lab
</p>
</div>
Built with <a href="http://sphinx-doc.org/">Sphinx</a> using a <a href="https://github.com/rtfd/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>.
</footer>
</div>
</div>
</section>
</div>
<script type="text/javascript">
jQuery(function () {
SphinxRtdTheme.Navigation.enable(true);
});
</script>
</body>
</html>
Reference in New Issue View Git Blame Copy Permalink