mirror of
https://github.com/gryf/coach.git
synced 2025-12-17 19:20:19 +01:00
6d40ad1650
* updating the documentation website * adding the built docs * update of api docstrings across coach and tutorials 0-2 * added some missing api documentation * New Sphinx based documentation
442 lines
35 KiB
HTML
442 lines
35 KiB
HTML
|
|
|
|
<!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.architectures.architecture — Reinforcement Learning Coach 0.11.0 documentation</title>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
<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">
|
|
|
|
|
|
|
|
<script src="../../../_static/js/modernizr.min.js"></script>
|
|
|
|
</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="../../../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>
|
|
</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/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/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> »</li>
|
|
|
|
<li><a href="../../index.html">Module code</a> »</li>
|
|
|
|
<li>rl_coach.architectures.architecture</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.architectures.architecture</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 "License");</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 "AS IS" 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">Any</span><span class="p">,</span> <span class="n">Dict</span><span class="p">,</span> <span class="n">List</span><span class="p">,</span> <span class="n">Tuple</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.base_parameters</span> <span class="k">import</span> <span class="n">AgentParameters</span>
|
|
<span class="kn">from</span> <span class="nn">rl_coach.spaces</span> <span class="k">import</span> <span class="n">SpacesDefinition</span>
|
|
|
|
|
|
<div class="viewcode-block" id="Architecture"><a class="viewcode-back" href="../../../components/architectures/index.html#rl_coach.architectures.architecture.Architecture">[docs]</a><span class="k">class</span> <span class="nc">Architecture</span><span class="p">(</span><span class="nb">object</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">AgentParameters</span><span class="p">,</span> <span class="n">spaces</span><span class="p">:</span> <span class="n">SpacesDefinition</span><span class="p">,</span> <span class="n">name</span><span class="p">:</span> <span class="nb">str</span><span class="o">=</span> <span class="s2">""</span><span class="p">):</span>
|
|
<span class="sd">"""</span>
|
|
<span class="sd"> Creates a neural network 'architecture', that can be trained and used for inference.</span>
|
|
|
|
<span class="sd"> :param agent_parameters: the agent parameters</span>
|
|
<span class="sd"> :param spaces: the spaces (observation, action, etc.) definition of the agent</span>
|
|
<span class="sd"> :param name: the name of the network</span>
|
|
<span class="sd"> """</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">spaces</span> <span class="o">=</span> <span class="n">spaces</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">name</span> <span class="o">=</span> <span class="n">name</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">network_wrapper_name</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">name</span><span class="o">.</span><span class="n">split</span><span class="p">(</span><span class="s1">'/'</span><span class="p">)[</span><span class="mi">0</span><span class="p">]</span> <span class="c1"># e.g. 'main/online' --> 'main'</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">full_name</span> <span class="o">=</span> <span class="s2">"</span><span class="si">{}</span><span class="s2">/</span><span class="si">{}</span><span class="s2">"</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">agent_parameters</span><span class="o">.</span><span class="n">full_name_id</span><span class="p">,</span> <span class="n">name</span><span class="p">)</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">network_parameters</span> <span class="o">=</span> <span class="n">agent_parameters</span><span class="o">.</span><span class="n">network_wrappers</span><span class="p">[</span><span class="bp">self</span><span class="o">.</span><span class="n">network_wrapper_name</span><span class="p">]</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">batch_size</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">network_parameters</span><span class="o">.</span><span class="n">batch_size</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">learning_rate</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">network_parameters</span><span class="o">.</span><span class="n">learning_rate</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">optimizer</span> <span class="o">=</span> <span class="kc">None</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">ap</span> <span class="o">=</span> <span class="n">agent_parameters</span>
|
|
|
|
<div class="viewcode-block" id="Architecture.predict"><a class="viewcode-back" href="../../../components/architectures/index.html#rl_coach.architectures.architecture.Architecture.predict">[docs]</a> <span class="k">def</span> <span class="nf">predict</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span>
|
|
<span class="n">inputs</span><span class="p">:</span> <span class="n">Dict</span><span class="p">[</span><span class="nb">str</span><span class="p">,</span> <span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">],</span>
|
|
<span class="n">outputs</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">Any</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
|
|
<span class="n">squeeze_output</span><span class="p">:</span> <span class="nb">bool</span> <span class="o">=</span> <span class="kc">True</span><span class="p">,</span>
|
|
<span class="n">initial_feed_dict</span><span class="p">:</span> <span class="n">Dict</span><span class="p">[</span><span class="n">Any</span><span class="p">,</span> <span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">)</span> <span class="o">-></span> <span class="n">Tuple</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">,</span> <span class="o">...</span><span class="p">]:</span>
|
|
<span class="sd">"""</span>
|
|
<span class="sd"> Given input observations, use the model to make predictions (e.g. action or value).</span>
|
|
|
|
<span class="sd"> :param inputs: current state (i.e. observations, measurements, goals, etc.)</span>
|
|
<span class="sd"> (e.g. `{'observation': numpy.ndarray}` of shape (batch_size, observation_space_size))</span>
|
|
<span class="sd"> :param outputs: list of outputs to return. Return all outputs if unspecified. Type of the list elements</span>
|
|
<span class="sd"> depends on the framework backend.</span>
|
|
<span class="sd"> :param squeeze_output: call squeeze_list on output before returning if True</span>
|
|
<span class="sd"> :param initial_feed_dict: a dictionary of extra inputs for forward pass.</span>
|
|
<span class="sd"> :return: predictions of action or value of shape (batch_size, action_space_size) for action predictions)</span>
|
|
<span class="sd"> """</span>
|
|
<span class="k">raise</span> <span class="ne">NotImplementedError</span></div>
|
|
|
|
<div class="viewcode-block" id="Architecture.parallel_predict"><a class="viewcode-back" href="../../../components/architectures/index.html#rl_coach.architectures.architecture.Architecture.parallel_predict">[docs]</a> <span class="nd">@staticmethod</span>
|
|
<span class="k">def</span> <span class="nf">parallel_predict</span><span class="p">(</span><span class="n">sess</span><span class="p">:</span> <span class="n">Any</span><span class="p">,</span>
|
|
<span class="n">network_input_tuples</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">Tuple</span><span class="p">[</span><span class="s1">'Architecture'</span><span class="p">,</span> <span class="n">Dict</span><span class="p">[</span><span class="nb">str</span><span class="p">,</span> <span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">]]])</span> <span class="o">-></span> \
|
|
<span class="n">Tuple</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">,</span> <span class="o">...</span><span class="p">]:</span>
|
|
<span class="sd">"""</span>
|
|
<span class="sd"> :param sess: active session to use for prediction</span>
|
|
<span class="sd"> :param network_input_tuples: tuple of network and corresponding input</span>
|
|
<span class="sd"> :return: list or tuple of outputs from all networks</span>
|
|
<span class="sd"> """</span>
|
|
<span class="k">raise</span> <span class="ne">NotImplementedError</span></div>
|
|
|
|
<div class="viewcode-block" id="Architecture.train_on_batch"><a class="viewcode-back" href="../../../components/architectures/index.html#rl_coach.architectures.architecture.Architecture.train_on_batch">[docs]</a> <span class="k">def</span> <span class="nf">train_on_batch</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span>
|
|
<span class="n">inputs</span><span class="p">:</span> <span class="n">Dict</span><span class="p">[</span><span class="nb">str</span><span class="p">,</span> <span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">],</span>
|
|
<span class="n">targets</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">],</span>
|
|
<span class="n">scaler</span><span class="p">:</span> <span class="nb">float</span><span class="o">=</span><span class="mf">1.</span><span class="p">,</span>
|
|
<span class="n">additional_fetches</span><span class="p">:</span> <span class="nb">list</span><span class="o">=</span><span class="kc">None</span><span class="p">,</span>
|
|
<span class="n">importance_weights</span><span class="p">:</span> <span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="o">=</span><span class="kc">None</span><span class="p">)</span> <span class="o">-></span> <span class="n">Tuple</span><span class="p">[</span><span class="nb">float</span><span class="p">,</span> <span class="n">List</span><span class="p">[</span><span class="nb">float</span><span class="p">],</span> <span class="nb">float</span><span class="p">,</span> <span class="nb">list</span><span class="p">]:</span>
|
|
<span class="sd">"""</span>
|
|
<span class="sd"> Given a batch of inputs (e.g. states) and targets (e.g. discounted rewards), takes a training step: i.e. runs a</span>
|
|
<span class="sd"> forward pass and backward pass of the network, accumulates the gradients and applies an optimization step to</span>
|
|
<span class="sd"> update the weights.</span>
|
|
<span class="sd"> Calls `accumulate_gradients` followed by `apply_and_reset_gradients`.</span>
|
|
<span class="sd"> Note: Currently an unused method.</span>
|
|
|
|
<span class="sd"> :param inputs: typically the environment states (but can also contain other data necessary for loss).</span>
|
|
<span class="sd"> (e.g. `{'observation': numpy.ndarray}` with `observation` of shape (batch_size, observation_space_size) or</span>
|
|
<span class="sd"> (batch_size, observation_space_size, stack_size) or</span>
|
|
<span class="sd"> `{'observation': numpy.ndarray, 'output_0_0': numpy.ndarray}` with `output_0_0` of shape (batch_size,))</span>
|
|
<span class="sd"> :param targets: target values of shape (batch_size, ). For example discounted rewards for value network</span>
|
|
<span class="sd"> for calculating the value-network loss would be a target. Length of list and order of arrays in</span>
|
|
<span class="sd"> the list matches that of network losses which are defined by network parameters</span>
|
|
<span class="sd"> :param scaler: value to scale gradients by before optimizing network weights</span>
|
|
<span class="sd"> :param additional_fetches: list of additional values to fetch and return. The type of each list</span>
|
|
<span class="sd"> element is framework dependent.</span>
|
|
<span class="sd"> :param importance_weights: ndarray of shape (batch_size,) to multiply with batch loss.</span>
|
|
<span class="sd"> :return: tuple of total_loss, losses, norm_unclipped_grads, fetched_tensors</span>
|
|
<span class="sd"> total_loss (float): sum of all head losses</span>
|
|
<span class="sd"> losses (list of float): list of all losses. The order is list of target losses followed by list</span>
|
|
<span class="sd"> of regularization losses. The specifics of losses is dependant on the network parameters</span>
|
|
<span class="sd"> (number of heads, etc.)</span>
|
|
<span class="sd"> norm_unclippsed_grads (float): global norm of all gradients before any gradient clipping is applied</span>
|
|
<span class="sd"> fetched_tensors: all values for additional_fetches</span>
|
|
<span class="sd"> """</span>
|
|
<span class="k">raise</span> <span class="ne">NotImplementedError</span></div>
|
|
|
|
<div class="viewcode-block" id="Architecture.get_weights"><a class="viewcode-back" href="../../../components/architectures/index.html#rl_coach.architectures.architecture.Architecture.get_weights">[docs]</a> <span class="k">def</span> <span class="nf">get_weights</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-></span> <span class="n">List</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">]:</span>
|
|
<span class="sd">"""</span>
|
|
<span class="sd"> Gets model weights as a list of ndarrays. It is used for synchronizing weight between two identical networks.</span>
|
|
|
|
<span class="sd"> :return: list weights as ndarray</span>
|
|
<span class="sd"> """</span>
|
|
<span class="k">raise</span> <span class="ne">NotImplementedError</span></div>
|
|
|
|
<div class="viewcode-block" id="Architecture.set_weights"><a class="viewcode-back" href="../../../components/architectures/index.html#rl_coach.architectures.architecture.Architecture.set_weights">[docs]</a> <span class="k">def</span> <span class="nf">set_weights</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">weights</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">],</span> <span class="n">rate</span><span class="p">:</span> <span class="nb">float</span><span class="o">=</span><span class="mf">1.0</span><span class="p">)</span> <span class="o">-></span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="sd">"""</span>
|
|
<span class="sd"> Sets model weights for provided layer parameters.</span>
|
|
|
|
<span class="sd"> :param weights: list of model weights in the same order as received in get_weights</span>
|
|
<span class="sd"> :param rate: controls the mixture of given weight values versus old weight values.</span>
|
|
<span class="sd"> i.e. new_weight = rate * given_weight + (1 - rate) * old_weight</span>
|
|
<span class="sd"> :return: None</span>
|
|
<span class="sd"> """</span>
|
|
<span class="k">raise</span> <span class="ne">NotImplementedError</span></div>
|
|
|
|
<div class="viewcode-block" id="Architecture.reset_accumulated_gradients"><a class="viewcode-back" href="../../../components/architectures/index.html#rl_coach.architectures.architecture.Architecture.reset_accumulated_gradients">[docs]</a> <span class="k">def</span> <span class="nf">reset_accumulated_gradients</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-></span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="sd">"""</span>
|
|
<span class="sd"> Sets gradient of all parameters to 0.</span>
|
|
|
|
<span class="sd"> Once gradients are reset, they must be accessible by `accumulated_gradients` property of this class,</span>
|
|
<span class="sd"> which must return a list of numpy ndarrays. Child class must ensure that `accumulated_gradients` is set.</span>
|
|
<span class="sd"> """</span>
|
|
<span class="k">raise</span> <span class="ne">NotImplementedError</span></div>
|
|
|
|
<div class="viewcode-block" id="Architecture.accumulate_gradients"><a class="viewcode-back" href="../../../components/architectures/index.html#rl_coach.architectures.architecture.Architecture.accumulate_gradients">[docs]</a> <span class="k">def</span> <span class="nf">accumulate_gradients</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span>
|
|
<span class="n">inputs</span><span class="p">:</span> <span class="n">Dict</span><span class="p">[</span><span class="nb">str</span><span class="p">,</span> <span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">],</span>
|
|
<span class="n">targets</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">],</span>
|
|
<span class="n">additional_fetches</span><span class="p">:</span> <span class="nb">list</span><span class="o">=</span><span class="kc">None</span><span class="p">,</span>
|
|
<span class="n">importance_weights</span><span class="p">:</span> <span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="o">=</span><span class="kc">None</span><span class="p">,</span>
|
|
<span class="n">no_accumulation</span><span class="p">:</span> <span class="nb">bool</span><span class="o">=</span><span class="kc">False</span><span class="p">)</span> <span class="o">-></span> <span class="n">Tuple</span><span class="p">[</span><span class="nb">float</span><span class="p">,</span> <span class="n">List</span><span class="p">[</span><span class="nb">float</span><span class="p">],</span> <span class="nb">float</span><span class="p">,</span> <span class="nb">list</span><span class="p">]:</span>
|
|
<span class="sd">"""</span>
|
|
<span class="sd"> Given a batch of inputs (i.e. states) and targets (e.g. discounted rewards), computes and accumulates the</span>
|
|
<span class="sd"> gradients for model parameters. Will run forward and backward pass to compute gradients, clip the gradient</span>
|
|
<span class="sd"> values if required and then accumulate gradients from all learners. It does not update the model weights,</span>
|
|
<span class="sd"> that's performed in `apply_and_reset_gradients` method.</span>
|
|
|
|
<span class="sd"> Once gradients are accumulated, they are accessed by `accumulated_gradients` property of this class.å</span>
|
|
|
|
<span class="sd"> :param inputs: typically the environment states (but can also contain other data for loss)</span>
|
|
<span class="sd"> (e.g. `{'observation': numpy.ndarray}` with `observation` of shape (batch_size, observation_space_size) or</span>
|
|
<span class="sd"> (batch_size, observation_space_size, stack_size) or</span>
|
|
<span class="sd"> `{'observation': numpy.ndarray, 'output_0_0': numpy.ndarray}` with `output_0_0` of shape (batch_size,))</span>
|
|
<span class="sd"> :param targets: targets for calculating loss. For example discounted rewards for value network</span>
|
|
<span class="sd"> for calculating the value-network loss would be a target. Length of list and order of arrays in</span>
|
|
<span class="sd"> the list matches that of network losses which are defined by network parameters</span>
|
|
<span class="sd"> :param additional_fetches: list of additional values to fetch and return. The type of each list</span>
|
|
<span class="sd"> element is framework dependent.</span>
|
|
<span class="sd"> :param importance_weights: ndarray of shape (batch_size,) to multiply with batch loss.</span>
|
|
<span class="sd"> :param no_accumulation: if True, set gradient values to the new gradients, otherwise sum with previously</span>
|
|
<span class="sd"> calculated gradients</span>
|
|
<span class="sd"> :return: tuple of total_loss, losses, norm_unclipped_grads, fetched_tensors</span>
|
|
<span class="sd"> total_loss (float): sum of all head losses</span>
|
|
<span class="sd"> losses (list of float): list of all losses. The order is list of target losses followed by list of</span>
|
|
<span class="sd"> regularization losses. The specifics of losses is dependant on the network parameters</span>
|
|
<span class="sd"> (number of heads, etc.)</span>
|
|
<span class="sd"> norm_unclippsed_grads (float): global norm of all gradients before any gradient clipping is applied</span>
|
|
<span class="sd"> fetched_tensors: all values for additional_fetches</span>
|
|
<span class="sd"> """</span>
|
|
<span class="k">raise</span> <span class="ne">NotImplementedError</span></div>
|
|
|
|
<div class="viewcode-block" id="Architecture.apply_and_reset_gradients"><a class="viewcode-back" href="../../../components/architectures/index.html#rl_coach.architectures.architecture.Architecture.apply_and_reset_gradients">[docs]</a> <span class="k">def</span> <span class="nf">apply_and_reset_gradients</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">gradients</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">],</span> <span class="n">scaler</span><span class="p">:</span> <span class="nb">float</span><span class="o">=</span><span class="mf">1.</span><span class="p">)</span> <span class="o">-></span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="sd">"""</span>
|
|
<span class="sd"> Applies the given gradients to the network weights and resets the gradient accumulations.</span>
|
|
<span class="sd"> Has the same impact as calling `apply_gradients`, then `reset_accumulated_gradients`.</span>
|
|
|
|
<span class="sd"> :param gradients: gradients for the parameter weights, taken from `accumulated_gradients` property</span>
|
|
<span class="sd"> of an identical network (either self or another identical network)</span>
|
|
<span class="sd"> :param scaler: A scaling factor that allows rescaling the gradients before applying them</span>
|
|
<span class="sd"> """</span>
|
|
<span class="k">raise</span> <span class="ne">NotImplementedError</span></div>
|
|
|
|
<div class="viewcode-block" id="Architecture.apply_gradients"><a class="viewcode-back" href="../../../components/architectures/index.html#rl_coach.architectures.architecture.Architecture.apply_gradients">[docs]</a> <span class="k">def</span> <span class="nf">apply_gradients</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">gradients</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">],</span> <span class="n">scaler</span><span class="p">:</span> <span class="nb">float</span><span class="o">=</span><span class="mf">1.</span><span class="p">)</span> <span class="o">-></span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="sd">"""</span>
|
|
<span class="sd"> Applies the given gradients to the network weights.</span>
|
|
<span class="sd"> Will be performed sync or async depending on `network_parameters.async_training`</span>
|
|
|
|
<span class="sd"> :param gradients: gradients for the parameter weights, taken from `accumulated_gradients` property</span>
|
|
<span class="sd"> of an identical network (either self or another identical network)</span>
|
|
<span class="sd"> :param scaler: A scaling factor that allows rescaling the gradients before applying them</span>
|
|
<span class="sd"> """</span>
|
|
<span class="k">raise</span> <span class="ne">NotImplementedError</span></div>
|
|
|
|
<div class="viewcode-block" id="Architecture.get_variable_value"><a class="viewcode-back" href="../../../components/architectures/index.html#rl_coach.architectures.architecture.Architecture.get_variable_value">[docs]</a> <span class="k">def</span> <span class="nf">get_variable_value</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">variable</span><span class="p">:</span> <span class="n">Any</span><span class="p">)</span> <span class="o">-></span> <span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">:</span>
|
|
<span class="sd">"""</span>
|
|
<span class="sd"> Gets value of a specified variable. Type of variable is dependant on the framework.</span>
|
|
<span class="sd"> Example of a variable is head.kl_coefficient, which could be a symbol for evaluation</span>
|
|
<span class="sd"> or could be a string representing the value.</span>
|
|
|
|
<span class="sd"> :param variable: variable of interest</span>
|
|
<span class="sd"> :return: value of the specified variable</span>
|
|
<span class="sd"> """</span>
|
|
<span class="k">raise</span> <span class="ne">NotImplementedError</span></div>
|
|
|
|
<div class="viewcode-block" id="Architecture.set_variable_value"><a class="viewcode-back" href="../../../components/architectures/index.html#rl_coach.architectures.architecture.Architecture.set_variable_value">[docs]</a> <span class="k">def</span> <span class="nf">set_variable_value</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">assign_op</span><span class="p">:</span> <span class="n">Any</span><span class="p">,</span> <span class="n">value</span><span class="p">:</span> <span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">,</span> <span class="n">placeholder</span><span class="p">:</span> <span class="n">Any</span><span class="p">):</span>
|
|
<span class="sd">"""</span>
|
|
<span class="sd"> Updates the value of a specified variable. Type of assign_op is dependant on the framework</span>
|
|
<span class="sd"> and is a unique identifier for assigning value to a variable. For example an agent may use</span>
|
|
<span class="sd"> head.assign_kl_coefficient. There is a one to one mapping between assign_op and placeholder</span>
|
|
<span class="sd"> (in the example above, placeholder would be head.kl_coefficient_ph).</span>
|
|
|
|
<span class="sd"> :param assign_op: a parameter representing the operation for assigning value to a specific variable</span>
|
|
<span class="sd"> :param value: value of the specified variable used for update</span>
|
|
<span class="sd"> :param placeholder: a placeholder for binding the value to assign_op.</span>
|
|
<span class="sd"> """</span>
|
|
<span class="k">raise</span> <span class="ne">NotImplementedError</span></div></div>
|
|
</pre></div>
|
|
|
|
</div>
|
|
|
|
</div>
|
|
<footer>
|
|
|
|
|
|
<hr/>
|
|
|
|
<div role="contentinfo">
|
|
<p>
|
|
© Copyright 2018, 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" 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 async="async" type="text/javascript" src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.1/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script>
|
|
|
|
|
|
|
|
|
|
<script type="text/javascript" src="../../../_static/js/theme.js"></script>
|
|
|
|
<script type="text/javascript">
|
|
jQuery(function () {
|
|
SphinxRtdTheme.Navigation.enable(true);
|
|
});
|
|
</script>
|
|
|
|
</body>
|
|
</html> |