mirror of
https://github.com/gryf/coach.git
synced 2025-12-18 03:30:19 +01:00
TD3 (#338)
This commit is contained in:
Gal Leibovich
@@ -278,19 +278,6 @@
|
||||
<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">memory</span><span class="o">.</span><span class="n">memory_backend_params</span><span class="o">.</span><span class="n">run_type</span> <span class="o">!=</span> <span class="s1">'trainer'</span><span class="p">:</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">memory</span><span class="o">.</span><span class="n">set_memory_backend</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">memory_backend</span><span class="p">)</span>
|
||||
|
||||
<span class="k">if</span> <span class="n">agent_parameters</span><span class="o">.</span><span class="n">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="p">:</span>
|
||||
<span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">agent_parameters</span><span class="o">.</span><span class="n">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="p">,</span> <span class="n">PickledReplayBuffer</span><span class="p">):</span>
|
||||
<span class="n">screen</span><span class="o">.</span><span class="n">log_title</span><span class="p">(</span><span class="s2">"Loading a pickled replay buffer. Pickled file path: </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">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="o">.</span><span class="n">filepath</span><span class="p">))</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">memory</span><span class="o">.</span><span class="n">load_pickled</span><span class="p">(</span><span class="n">agent_parameters</span><span class="o">.</span><span class="n">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="o">.</span><span class="n">filepath</span><span class="p">)</span>
|
||||
<span class="k">elif</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">agent_parameters</span><span class="o">.</span><span class="n">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="p">,</span> <span class="n">CsvDataset</span><span class="p">):</span>
|
||||
<span class="n">screen</span><span class="o">.</span><span class="n">log_title</span><span class="p">(</span><span class="s2">"Loading a replay buffer from a CSV file. CSV file path: </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">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="o">.</span><span class="n">filepath</span><span class="p">))</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">memory</span><span class="o">.</span><span class="n">load_csv</span><span class="p">(</span><span class="n">agent_parameters</span><span class="o">.</span><span class="n">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="p">)</span>
|
||||
<span class="k">else</span><span class="p">:</span>
|
||||
<span class="k">raise</span> <span class="ne">ValueError</span><span class="p">(</span><span class="s1">'Trying to load a replay buffer using an unsupported method - </span><span class="si">{}</span><span class="s1">. '</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">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="p">))</span>
|
||||
|
||||
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">shared_memory</span> <span class="ow">and</span> <span class="bp">self</span><span class="o">.</span><span class="n">is_chief</span><span class="p">:</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">shared_memory_scratchpad</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">memory_lookup_name</span><span class="p">,</span> <span class="bp">self</span><span class="o">.</span><span class="n">memory</span><span class="p">)</span>
|
||||
|
||||
@@ -444,7 +431,39 @@
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">input_filter</span><span class="o">.</span><span class="n">set_session</span><span class="p">(</span><span class="n">sess</span><span class="p">)</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">output_filter</span><span class="o">.</span><span class="n">set_session</span><span class="p">(</span><span class="n">sess</span><span class="p">)</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">pre_network_filter</span><span class="o">.</span><span class="n">set_session</span><span class="p">(</span><span class="n">sess</span><span class="p">)</span>
|
||||
<span class="p">[</span><span class="n">network</span><span class="o">.</span><span class="n">set_session</span><span class="p">(</span><span class="n">sess</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></div>
|
||||
<span class="p">[</span><span class="n">network</span><span class="o">.</span><span class="n">set_session</span><span class="p">(</span><span class="n">sess</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="bp">self</span><span class="o">.</span><span class="n">initialize_session_dependent_components</span><span class="p">()</span></div>
|
||||
|
||||
<div class="viewcode-block" id="Agent.initialize_session_dependent_components"><a class="viewcode-back" href="../../../components/agents/index.html#rl_coach.agents.agent.Agent.initialize_session_dependent_components">[docs]</a> <span class="k">def</span> <span class="nf">initialize_session_dependent_components</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
|
||||
<span class="sd">"""</span>
|
||||
<span class="sd"> Initialize components which require a session as part of their initialization.</span>
|
||||
|
||||
<span class="sd"> :return: None</span>
|
||||
<span class="sd"> """</span>
|
||||
|
||||
<span class="c1"># Loading a memory from a CSV file, requires an input filter to filter through the data.</span>
|
||||
<span class="c1"># The filter needs a session before it can be used.</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">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="p">:</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">load_memory_from_file</span><span class="p">()</span></div>
|
||||
|
||||
<div class="viewcode-block" id="Agent.load_memory_from_file"><a class="viewcode-back" href="../../../components/agents/index.html#rl_coach.agents.agent.Agent.load_memory_from_file">[docs]</a> <span class="k">def</span> <span class="nf">load_memory_from_file</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
|
||||
<span class="sd">"""</span>
|
||||
<span class="sd"> Load memory transitions from a file.</span>
|
||||
|
||||
<span class="sd"> :return: None</span>
|
||||
<span class="sd"> """</span>
|
||||
|
||||
<span class="k">if</span> <span class="nb">isinstance</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">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="p">,</span> <span class="n">PickledReplayBuffer</span><span class="p">):</span>
|
||||
<span class="n">screen</span><span class="o">.</span><span class="n">log_title</span><span class="p">(</span><span class="s2">"Loading a pickled replay buffer. Pickled file path: </span><span class="si">{}</span><span class="s2">"</span>
|
||||
<span class="o">.</span><span class="n">format</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">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="o">.</span><span class="n">filepath</span><span class="p">))</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">memory</span><span class="o">.</span><span class="n">load_pickled</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">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="o">.</span><span class="n">filepath</span><span class="p">)</span>
|
||||
<span class="k">elif</span> <span class="nb">isinstance</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">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="p">,</span> <span class="n">CsvDataset</span><span class="p">):</span>
|
||||
<span class="n">screen</span><span class="o">.</span><span class="n">log_title</span><span class="p">(</span><span class="s2">"Loading a replay buffer from a CSV file. CSV file path: </span><span class="si">{}</span><span class="s2">"</span>
|
||||
<span class="o">.</span><span class="n">format</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">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="o">.</span><span class="n">filepath</span><span class="p">))</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">memory</span><span class="o">.</span><span class="n">load_csv</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">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="p">,</span> <span class="bp">self</span><span class="o">.</span><span class="n">input_filter</span><span class="p">)</span>
|
||||
<span class="k">else</span><span class="p">:</span>
|
||||
<span class="k">raise</span> <span class="ne">ValueError</span><span class="p">(</span><span class="s1">'Trying to load a replay buffer using an unsupported method - </span><span class="si">{}</span><span class="s1">. '</span>
|
||||
<span class="o">.</span><span class="n">format</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">memory</span><span class="o">.</span><span class="n">load_memory_from_file_path</span><span class="p">))</span></div>
|
||||
|
||||
<div class="viewcode-block" id="Agent.register_signal"><a class="viewcode-back" href="../../../components/agents/index.html#rl_coach.agents.agent.Agent.register_signal">[docs]</a> <span class="k">def</span> <span class="nf">register_signal</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">signal_name</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span> <span class="n">dump_one_value_per_episode</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">dump_one_value_per_step</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">Signal</span><span class="p">:</span>
|
||||
@@ -868,7 +887,10 @@
|
||||
<span class="sd"> """</span>
|
||||
<span class="n">loss</span> <span class="o">=</span> <span class="mi">0</span>
|
||||
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">_should_train</span><span class="p">():</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">training_epoch</span> <span class="o">+=</span> <span class="mi">1</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">is_batch_rl_training</span><span class="p">:</span>
|
||||
<span class="c1"># when training an agent for generating a dataset in batch-rl, we don't want it to be counted as part of</span>
|
||||
<span class="c1"># the training epochs. we only care for training epochs in batch-rl anyway.</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">training_epoch</span> <span class="o">+=</span> <span class="mi">1</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">set_is_training</span><span class="p">(</span><span class="kc">True</span><span class="p">)</span>
|
||||
|
||||
@@ -1229,7 +1251,15 @@
|
||||
<span class="n">TimeTypes</span><span class="o">.</span><span class="n">TrainingIteration</span><span class="p">:</span> <span class="bp">self</span><span class="o">.</span><span class="n">training_iteration</span><span class="p">,</span>
|
||||
<span class="n">TimeTypes</span><span class="o">.</span><span class="n">EnvironmentSteps</span><span class="p">:</span> <span class="bp">self</span><span class="o">.</span><span class="n">total_steps_counter</span><span class="p">,</span>
|
||||
<span class="n">TimeTypes</span><span class="o">.</span><span class="n">WallClockTime</span><span class="p">:</span> <span class="bp">self</span><span class="o">.</span><span class="n">agent_logger</span><span class="o">.</span><span class="n">get_current_wall_clock_time</span><span class="p">(),</span>
|
||||
<span class="n">TimeTypes</span><span class="o">.</span><span class="n">Epoch</span><span class="p">:</span> <span class="bp">self</span><span class="o">.</span><span class="n">training_epoch</span><span class="p">}[</span><span class="bp">self</span><span class="o">.</span><span class="n">parent_level_manager</span><span class="o">.</span><span class="n">parent_graph_manager</span><span class="o">.</span><span class="n">time_metric</span><span class="p">]</span></div>
|
||||
<span class="n">TimeTypes</span><span class="o">.</span><span class="n">Epoch</span><span class="p">:</span> <span class="bp">self</span><span class="o">.</span><span class="n">training_epoch</span><span class="p">}[</span><span class="bp">self</span><span class="o">.</span><span class="n">parent_level_manager</span><span class="o">.</span><span class="n">parent_graph_manager</span><span class="o">.</span><span class="n">time_metric</span><span class="p">]</span>
|
||||
|
||||
<div class="viewcode-block" id="Agent.freeze_memory"><a class="viewcode-back" href="../../../components/agents/index.html#rl_coach.agents.agent.Agent.freeze_memory">[docs]</a> <span class="k">def</span> <span class="nf">freeze_memory</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
|
||||
<span class="sd">"""</span>
|
||||
<span class="sd"> Shuffle episodes in the memory and freeze it to make sure that no extra data is being pushed anymore.</span>
|
||||
<span class="sd"> :return: None</span>
|
||||
<span class="sd"> """</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">call_memory</span><span class="p">(</span><span class="s1">'shuffle_episodes'</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">'freeze'</span><span class="p">)</span></div></div>
|
||||
</pre></div>
|
||||
|
||||
</div>
|
||||
|
||||
@@ -196,7 +196,6 @@
|
||||
<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>
|
||||
@@ -266,13 +265,22 @@
|
||||
|
||||
<span class="c1"># prediction's format is (batch,actions,atoms)</span>
|
||||
<span class="k">def</span> <span class="nf">get_all_q_values_for_states</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">StateType</span><span class="p">):</span>
|
||||
<span class="n">q_values</span> <span class="o">=</span> <span class="kc">None</span>
|
||||
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">exploration_policy</span><span class="o">.</span><span class="n">requires_action_values</span><span class="p">():</span>
|
||||
<span class="n">q_values</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">get_prediction</span><span class="p">(</span><span class="n">states</span><span class="p">,</span>
|
||||
<span class="n">outputs</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">'main'</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">0</span><span class="p">]</span><span class="o">.</span><span class="n">q_values</span><span class="p">])</span>
|
||||
<span class="k">else</span><span class="p">:</span>
|
||||
<span class="n">q_values</span> <span class="o">=</span> <span class="kc">None</span>
|
||||
|
||||
<span class="k">return</span> <span class="n">q_values</span>
|
||||
|
||||
<span class="k">def</span> <span class="nf">get_all_q_values_for_states_and_softmax_probabilities</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">StateType</span><span class="p">):</span>
|
||||
<span class="n">actions_q_values</span><span class="p">,</span> <span class="n">softmax_probabilities</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span> <span class="kc">None</span>
|
||||
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">exploration_policy</span><span class="o">.</span><span class="n">requires_action_values</span><span class="p">():</span>
|
||||
<span class="n">outputs</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">'main'</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">0</span><span class="p">]</span><span class="o">.</span><span class="n">q_values</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">'main'</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">0</span><span class="p">]</span><span class="o">.</span><span class="n">softmax</span><span class="p">]</span>
|
||||
<span class="n">actions_q_values</span><span class="p">,</span> <span class="n">softmax_probabilities</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">get_prediction</span><span class="p">(</span><span class="n">states</span><span class="p">,</span> <span class="n">outputs</span><span class="o">=</span><span class="n">outputs</span><span class="p">)</span>
|
||||
|
||||
<span class="k">return</span> <span class="n">actions_q_values</span><span class="p">,</span> <span class="n">softmax_probabilities</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">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">'main'</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>
|
||||
|
||||
|
||||
@@ -206,7 +206,7 @@
|
||||
<span class="kn">from</span> <span class="nn">rl_coach.agents.actor_critic_agent</span> <span class="k">import</span> <span class="n">ActorCriticAgent</span>
|
||||
<span class="kn">from</span> <span class="nn">rl_coach.agents.agent</span> <span class="k">import</span> <span class="n">Agent</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">DDPGActorHeadParameters</span><span class="p">,</span> <span class="n">VHeadParameters</span>
|
||||
<span class="kn">from</span> <span class="nn">rl_coach.architectures.head_parameters</span> <span class="k">import</span> <span class="n">DDPGActorHeadParameters</span><span class="p">,</span> <span class="n">DDPGVHeadParameters</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">NetworkParameters</span><span class="p">,</span> <span class="n">AlgorithmParameters</span><span class="p">,</span> \
|
||||
<span class="n">AgentParameters</span><span class="p">,</span> <span class="n">EmbedderScheme</span>
|
||||
@@ -222,14 +222,17 @@
|
||||
<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">'observation'</span><span class="p">:</span> <span class="n">InputEmbedderParameters</span><span class="p">(</span><span class="n">batchnorm</span><span class="o">=</span><span class="kc">True</span><span class="p">),</span>
|
||||
<span class="s1">'action'</span><span class="p">:</span> <span class="n">InputEmbedderParameters</span><span class="p">(</span><span class="n">scheme</span><span class="o">=</span><span class="n">EmbedderScheme</span><span class="o">.</span><span class="n">Shallow</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">VHeadParameters</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">DDPGVHeadParameters</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">'Adam'</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">batch_size</span> <span class="o">=</span> <span class="mi">64</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">async_training</span> <span class="o">=</span> <span class="kc">False</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">learning_rate</span> <span class="o">=</span> <span class="mf">0.001</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">adam_optimizer_beta2</span> <span class="o">=</span> <span class="mf">0.999</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">optimizer_epsilon</span> <span class="o">=</span> <span class="mf">1e-8</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="bp">self</span><span class="o">.</span><span class="n">shared_optimizer</span> <span class="o">=</span> <span class="kc">True</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">scale_down_gradients_by_number_of_workers_for_sync_training</span> <span class="o">=</span> <span class="kc">False</span>
|
||||
<span class="c1"># self.l2_regularization = 1e-2</span>
|
||||
|
||||
|
||||
<span class="k">class</span> <span class="nc">DDPGActorNetworkParameters</span><span class="p">(</span><span class="n">NetworkParameters</span><span class="p">):</span>
|
||||
@@ -240,6 +243,8 @@
|
||||
<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">DDPGActorHeadParameters</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">'Adam'</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">batch_size</span> <span class="o">=</span> <span class="mi">64</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">adam_optimizer_beta2</span> <span class="o">=</span> <span class="mf">0.999</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">optimizer_epsilon</span> <span class="o">=</span> <span class="mf">1e-8</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">async_training</span> <span class="o">=</span> <span class="kc">False</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">learning_rate</span> <span class="o">=</span> <span class="mf">0.0001</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>
|
||||
@@ -323,7 +328,7 @@
|
||||
|
||||
<span class="n">critic_inputs</span> <span class="o">=</span> <span class="n">copy</span><span class="o">.</span><span class="n">copy</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">critic_keys</span><span class="p">))</span>
|
||||
<span class="n">critic_inputs</span><span class="p">[</span><span class="s1">'action'</span><span class="p">]</span> <span class="o">=</span> <span class="n">next_actions</span>
|
||||
<span class="n">q_st_plus_1</span> <span class="o">=</span> <span class="n">critic</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">critic_inputs</span><span class="p">)</span>
|
||||
<span class="n">q_st_plus_1</span> <span class="o">=</span> <span class="n">critic</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">critic_inputs</span><span class="p">)[</span><span class="mi">0</span><span class="p">]</span>
|
||||
|
||||
<span class="c1"># calculate the bootstrapped TD targets while discounting terminal states according to</span>
|
||||
<span class="c1"># use_non_zero_discount_for_terminal_states</span>
|
||||
@@ -343,7 +348,7 @@
|
||||
<span class="n">critic_inputs</span> <span class="o">=</span> <span class="n">copy</span><span class="o">.</span><span class="n">copy</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">critic_keys</span><span class="p">))</span>
|
||||
<span class="n">critic_inputs</span><span class="p">[</span><span class="s1">'action'</span><span class="p">]</span> <span class="o">=</span> <span class="n">actions_mean</span>
|
||||
<span class="n">action_gradients</span> <span class="o">=</span> <span class="n">critic</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">critic_inputs</span><span class="p">,</span>
|
||||
<span class="n">outputs</span><span class="o">=</span><span class="n">critic</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">gradients_wrt_inputs</span><span class="p">[</span><span class="mi">0</span><span class="p">][</span><span class="s1">'action'</span><span class="p">])</span>
|
||||
<span class="n">outputs</span><span class="o">=</span><span class="n">critic</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">gradients_wrt_inputs</span><span class="p">[</span><span class="mi">1</span><span class="p">][</span><span class="s1">'action'</span><span class="p">])</span>
|
||||
|
||||
<span class="c1"># train the critic</span>
|
||||
<span class="n">critic_inputs</span> <span class="o">=</span> <span class="n">copy</span><span class="o">.</span><span class="n">copy</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">critic_keys</span><span class="p">))</span>
|
||||
|
||||
@@ -365,7 +365,7 @@
|
||||
<span class="n">action_values</span> <span class="o">=</span> <span class="kc">None</span>
|
||||
|
||||
<span class="c1"># choose action according to the exploration policy and the current phase (evaluating or training the agent)</span>
|
||||
<span class="n">action</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">exploration_policy</span><span class="o">.</span><span class="n">get_action</span><span class="p">(</span><span class="n">action_values</span><span class="p">)</span>
|
||||
<span class="n">action</span><span class="p">,</span> <span class="n">_</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">exploration_policy</span><span class="o">.</span><span class="n">get_action</span><span class="p">(</span><span class="n">action_values</span><span class="p">)</span>
|
||||
|
||||
<span class="k">if</span> <span class="n">action_values</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
||||
<span class="n">action_values</span> <span class="o">=</span> <span class="n">action_values</span><span class="o">.</span><span class="n">squeeze</span><span class="p">()</span>
|
||||
|
||||
@@ -232,6 +232,7 @@
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">batch_size</span> <span class="o">=</span> <span class="mi">32</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">replace_mse_with_huber_loss</span> <span class="o">=</span> <span class="kc">True</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="bp">self</span><span class="o">.</span><span class="n">should_get_softmax_probabilities</span> <span class="o">=</span> <span class="kc">False</span>
|
||||
|
||||
|
||||
<span class="k">class</span> <span class="nc">DQNAgentParameters</span><span class="p">(</span><span class="n">AgentParameters</span><span class="p">):</span>
|
||||
|
||||
@@ -199,7 +199,7 @@
|
||||
|
||||
<span class="kn">import</span> <span class="nn">os</span>
|
||||
<span class="kn">import</span> <span class="nn">pickle</span>
|
||||
<span class="kn">from</span> <span class="nn">typing</span> <span class="k">import</span> <span class="n">Union</span>
|
||||
<span class="kn">from</span> <span class="nn">typing</span> <span class="k">import</span> <span class="n">Union</span><span class="p">,</span> <span class="n">List</span>
|
||||
|
||||
<span class="kn">import</span> <span class="nn">numpy</span> <span class="k">as</span> <span class="nn">np</span>
|
||||
|
||||
@@ -223,6 +223,7 @@
|
||||
<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">DNDQHeadParameters</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">'Adam'</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">should_get_softmax_probabilities</span> <span class="o">=</span> <span class="kc">False</span>
|
||||
|
||||
|
||||
<div class="viewcode-block" id="NECAlgorithmParameters"><a class="viewcode-back" href="../../../components/agents/value_optimization/nec.html#rl_coach.agents.nec_agent.NECAlgorithmParameters">[docs]</a><span class="k">class</span> <span class="nc">NECAlgorithmParameters</span><span class="p">(</span><span class="n">AlgorithmParameters</span><span class="p">):</span>
|
||||
@@ -349,11 +350,25 @@
|
||||
|
||||
<span class="k">return</span> <span class="nb">super</span><span class="p">()</span><span class="o">.</span><span class="n">act</span><span class="p">()</span>
|
||||
|
||||
<span class="k">def</span> <span class="nf">get_all_q_values_for_states</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">StateType</span><span class="p">):</span>
|
||||
<span class="k">def</span> <span class="nf">get_all_q_values_for_states</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">StateType</span><span class="p">,</span> <span class="n">additional_outputs</span><span class="p">:</span> <span class="n">List</span> <span class="o">=</span> <span class="kc">None</span><span class="p">):</span>
|
||||
<span class="c1"># we need to store the state embeddings regardless if the action is random or not</span>
|
||||
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">get_prediction</span><span class="p">(</span><span class="n">states</span><span class="p">)</span>
|
||||
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">get_prediction_and_update_embeddings</span><span class="p">(</span><span class="n">states</span><span class="p">)</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="k">def</span> <span class="nf">get_all_q_values_for_states_and_softmax_probabilities</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">StateType</span><span class="p">):</span>
|
||||
<span class="c1"># get the actions q values and the state embedding</span>
|
||||
<span class="n">embedding</span><span class="p">,</span> <span class="n">actions_q_values</span><span class="p">,</span> <span class="n">softmax_probabilities</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">'main'</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="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="s1">'main'</span><span class="p">),</span>
|
||||
<span class="n">outputs</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">'main'</span><span class="p">]</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">state_embedding</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">'main'</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">0</span><span class="p">]</span><span class="o">.</span><span class="n">output</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">'main'</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">0</span><span class="p">]</span><span class="o">.</span><span class="n">softmax</span><span class="p">]</span>
|
||||
<span class="p">)</span>
|
||||
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">phase</span> <span class="o">!=</span> <span class="n">RunPhase</span><span class="o">.</span><span class="n">TEST</span><span class="p">:</span>
|
||||
<span class="c1"># store the state embedding for inserting it to the DND later</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">current_episode_state_embeddings</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">embedding</span><span class="o">.</span><span class="n">squeeze</span><span class="p">())</span>
|
||||
<span class="n">actions_q_values</span> <span class="o">=</span> <span class="n">actions_q_values</span><span class="p">[</span><span class="mi">0</span><span class="p">][</span><span class="mi">0</span><span class="p">]</span>
|
||||
<span class="k">return</span> <span class="n">actions_q_values</span><span class="p">,</span> <span class="n">softmax_probabilities</span>
|
||||
|
||||
<span class="k">def</span> <span class="nf">get_prediction_and_update_embeddings</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="c1"># get the actions q values and the state embedding</span>
|
||||
<span class="n">embedding</span><span class="p">,</span> <span class="n">actions_q_values</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">'main'</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="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="s1">'main'</span><span class="p">),</span>
|
||||
@@ -362,7 +377,7 @@
|
||||
<span class="p">)</span>
|
||||
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">phase</span> <span class="o">!=</span> <span class="n">RunPhase</span><span class="o">.</span><span class="n">TEST</span><span class="p">:</span>
|
||||
<span class="c1"># store the state embedding for inserting it to the DND later</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">current_episode_state_embeddings</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">embedding</span><span class="o">.</span><span class="n">squeeze</span><span class="p">())</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">current_episode_state_embeddings</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">embedding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">.</span><span class="n">squeeze</span><span class="p">())</span>
|
||||
<span class="n">actions_q_values</span> <span class="o">=</span> <span class="n">actions_q_values</span><span class="p">[</span><span class="mi">0</span><span class="p">][</span><span class="mi">0</span><span class="p">]</span>
|
||||
<span class="k">return</span> <span class="n">actions_q_values</span>
|
||||
|
||||
|
||||
@@ -196,7 +196,7 @@
|
||||
<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">copy</span> <span class="k">import</span> <span class="n">copy</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>
|
||||
@@ -262,6 +262,17 @@
|
||||
<span class="n">actions_q_values</span> <span class="o">=</span> <span class="kc">None</span>
|
||||
<span class="k">return</span> <span class="n">actions_q_values</span>
|
||||
|
||||
<span class="c1"># prediction's format is (batch,actions,atoms)</span>
|
||||
<span class="k">def</span> <span class="nf">get_all_q_values_for_states_and_softmax_probabilities</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">StateType</span><span class="p">):</span>
|
||||
<span class="n">actions_q_values</span><span class="p">,</span> <span class="n">softmax_probabilities</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span> <span class="kc">None</span>
|
||||
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">exploration_policy</span><span class="o">.</span><span class="n">requires_action_values</span><span class="p">():</span>
|
||||
<span class="n">outputs</span> <span class="o">=</span> <span class="n">copy</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">'main'</span><span class="p">]</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">outputs</span><span class="p">)</span>
|
||||
<span class="n">outputs</span><span class="o">.</span><span class="n">append</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">'main'</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">0</span><span class="p">]</span><span class="o">.</span><span class="n">softmax</span><span class="p">)</span>
|
||||
<span class="n">quantile_values</span><span class="p">,</span> <span class="n">softmax_probabilities</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">get_prediction</span><span class="p">(</span><span class="n">states</span><span class="p">,</span> <span class="n">outputs</span><span class="p">)</span>
|
||||
<span class="n">actions_q_values</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">get_q_values</span><span class="p">(</span><span class="n">quantile_values</span><span class="p">)</span>
|
||||
|
||||
<span class="k">return</span> <span class="n">actions_q_values</span><span class="p">,</span> <span class="n">softmax_probabilities</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">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">'main'</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>
|
||||
|
||||
|
||||
448
docs/_modules/rl_coach/agents/td3_agent.html
Normal file
448
docs/_modules/rl_coach/agents/td3_agent.html
Normal file
@@ -0,0 +1,448 @@
|
||||
|
||||
|
||||
<!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.td3_agent — 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> »</li>
|
||||
|
||||
<li><a href="../../index.html">Module code</a> »</li>
|
||||
|
||||
<li>rl_coach.agents.td3_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.td3_agent</h1><div class="highlight"><pre>
|
||||
<span></span><span class="c1">#</span>
|
||||
<span class="c1"># Copyright (c) 2019 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">import</span> <span class="nn">copy</span>
|
||||
<span class="kn">from</span> <span class="nn">typing</span> <span class="k">import</span> <span class="n">Union</span>
|
||||
<span class="kn">from</span> <span class="nn">collections</span> <span class="k">import</span> <span class="n">OrderedDict</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.agent</span> <span class="k">import</span> <span class="n">Agent</span>
|
||||
<span class="kn">from</span> <span class="nn">rl_coach.agents.ddpg_agent</span> <span class="k">import</span> <span class="n">DDPGAgent</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">DDPGActorHeadParameters</span><span class="p">,</span> <span class="n">TD3VHeadParameters</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">NetworkParameters</span><span class="p">,</span> <span class="n">AlgorithmParameters</span><span class="p">,</span> \
|
||||
<span class="n">AgentParameters</span><span class="p">,</span> <span class="n">EmbedderScheme</span>
|
||||
<span class="kn">from</span> <span class="nn">rl_coach.core_types</span> <span class="k">import</span> <span class="n">ActionInfo</span><span class="p">,</span> <span class="n">TrainingSteps</span><span class="p">,</span> <span class="n">Transition</span>
|
||||
<span class="kn">from</span> <span class="nn">rl_coach.exploration_policies.additive_noise</span> <span class="k">import</span> <span class="n">AdditiveNoiseParameters</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">BoxActionSpace</span><span class="p">,</span> <span class="n">GoalsSpace</span>
|
||||
|
||||
|
||||
<span class="k">class</span> <span class="nc">TD3CriticNetworkParameters</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="n">num_q_networks</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">'observation'</span><span class="p">:</span> <span class="n">InputEmbedderParameters</span><span class="p">(),</span>
|
||||
<span class="s1">'action'</span><span class="p">:</span> <span class="n">InputEmbedderParameters</span><span class="p">(</span><span class="n">scheme</span><span class="o">=</span><span class="n">EmbedderScheme</span><span class="o">.</span><span class="n">Shallow</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="n">num_streams</span><span class="o">=</span><span class="n">num_q_networks</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">TD3VHeadParameters</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">'Adam'</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">adam_optimizer_beta2</span> <span class="o">=</span> <span class="mf">0.999</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">optimizer_epsilon</span> <span class="o">=</span> <span class="mf">1e-8</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">batch_size</span> <span class="o">=</span> <span class="mi">100</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">async_training</span> <span class="o">=</span> <span class="kc">False</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">learning_rate</span> <span class="o">=</span> <span class="mf">0.001</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="bp">self</span><span class="o">.</span><span class="n">shared_optimizer</span> <span class="o">=</span> <span class="kc">True</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">scale_down_gradients_by_number_of_workers_for_sync_training</span> <span class="o">=</span> <span class="kc">False</span>
|
||||
|
||||
|
||||
<span class="k">class</span> <span class="nc">TD3ActorNetworkParameters</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">'observation'</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">DDPGActorHeadParameters</span><span class="p">(</span><span class="n">batchnorm</span><span class="o">=</span><span class="kc">False</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">'Adam'</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">adam_optimizer_beta2</span> <span class="o">=</span> <span class="mf">0.999</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">optimizer_epsilon</span> <span class="o">=</span> <span class="mf">1e-8</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">batch_size</span> <span class="o">=</span> <span class="mi">100</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">async_training</span> <span class="o">=</span> <span class="kc">False</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">learning_rate</span> <span class="o">=</span> <span class="mf">0.001</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="bp">self</span><span class="o">.</span><span class="n">shared_optimizer</span> <span class="o">=</span> <span class="kc">True</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">scale_down_gradients_by_number_of_workers_for_sync_training</span> <span class="o">=</span> <span class="kc">False</span>
|
||||
|
||||
|
||||
<div class="viewcode-block" id="TD3AlgorithmParameters"><a class="viewcode-back" href="../../../components/agents/policy_optimization/td3.html#rl_coach.agents.td3_agent.TD3AlgorithmParameters">[docs]</a><span class="k">class</span> <span class="nc">TD3AlgorithmParameters</span><span class="p">(</span><span class="n">AlgorithmParameters</span><span class="p">):</span>
|
||||
<span class="sd">"""</span>
|
||||
<span class="sd"> :param num_steps_between_copying_online_weights_to_target: (StepMethod)</span>
|
||||
<span class="sd"> The number of steps between copying the online network weights to the target network weights.</span>
|
||||
|
||||
<span class="sd"> :param rate_for_copying_weights_to_target: (float)</span>
|
||||
<span class="sd"> When copying the online network weights to the target network weights, a soft update will be used, which</span>
|
||||
<span class="sd"> weight the new online network weights by rate_for_copying_weights_to_target</span>
|
||||
|
||||
<span class="sd"> :param num_consecutive_playing_steps: (StepMethod)</span>
|
||||
<span class="sd"> The number of consecutive steps to act between every two training iterations</span>
|
||||
|
||||
<span class="sd"> :param use_target_network_for_evaluation: (bool)</span>
|
||||
<span class="sd"> If set to True, the target network will be used for predicting the actions when choosing actions to act.</span>
|
||||
<span class="sd"> Since the target network weights change more slowly, the predicted actions will be more consistent.</span>
|
||||
|
||||
<span class="sd"> :param action_penalty: (float)</span>
|
||||
<span class="sd"> The amount by which to penalize the network on high action feature (pre-activation) values.</span>
|
||||
<span class="sd"> This can prevent the actions features from saturating the TanH activation function, and therefore prevent the</span>
|
||||
<span class="sd"> gradients from becoming very low.</span>
|
||||
|
||||
<span class="sd"> :param clip_critic_targets: (Tuple[float, float] or None)</span>
|
||||
<span class="sd"> The range to clip the critic target to in order to prevent overestimation of the action values.</span>
|
||||
|
||||
<span class="sd"> :param use_non_zero_discount_for_terminal_states: (bool)</span>
|
||||
<span class="sd"> If set to True, the discount factor will be used for terminal states to bootstrap the next predicted state</span>
|
||||
<span class="sd"> values. If set to False, the terminal states reward will be taken as the target return for the network.</span>
|
||||
<span class="sd"> """</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">rate_for_copying_weights_to_target</span> <span class="o">=</span> <span class="mf">0.005</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">use_target_network_for_evaluation</span> <span class="o">=</span> <span class="kc">False</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">action_penalty</span> <span class="o">=</span> <span class="mi">0</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">clip_critic_targets</span> <span class="o">=</span> <span class="kc">None</span> <span class="c1"># expected to be a tuple of the form (min_clip_value, max_clip_value) or None</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">use_non_zero_discount_for_terminal_states</span> <span class="o">=</span> <span class="kc">False</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">act_for_full_episodes</span> <span class="o">=</span> <span class="kc">True</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">update_policy_every_x_episode_steps</span> <span class="o">=</span> <span class="mi">2</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">num_steps_between_copying_online_weights_to_target</span> <span class="o">=</span> <span class="n">TrainingSteps</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">update_policy_every_x_episode_steps</span><span class="p">)</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">policy_noise</span> <span class="o">=</span> <span class="mf">0.2</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">noise_clipping</span> <span class="o">=</span> <span class="mf">0.5</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">num_q_networks</span> <span class="o">=</span> <span class="mi">2</span></div>
|
||||
|
||||
|
||||
<span class="k">class</span> <span class="nc">TD3AgentExplorationParameters</span><span class="p">(</span><span class="n">AdditiveNoiseParameters</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">noise_as_percentage_from_action_space</span> <span class="o">=</span> <span class="kc">False</span>
|
||||
|
||||
|
||||
<span class="k">class</span> <span class="nc">TD3AgentParameters</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="n">td3_algorithm_params</span> <span class="o">=</span> <span class="n">TD3AlgorithmParameters</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">td3_algorithm_params</span><span class="p">,</span>
|
||||
<span class="n">exploration</span><span class="o">=</span><span class="n">TD3AgentExplorationParameters</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="n">OrderedDict</span><span class="p">([(</span><span class="s2">"actor"</span><span class="p">,</span> <span class="n">TD3ActorNetworkParameters</span><span class="p">()),</span>
|
||||
<span class="p">(</span><span class="s2">"critic"</span><span class="p">,</span>
|
||||
<span class="n">TD3CriticNetworkParameters</span><span class="p">(</span><span class="n">td3_algorithm_params</span><span class="o">.</span><span class="n">num_q_networks</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">'rl_coach.agents.td3_agent:TD3Agent'</span>
|
||||
|
||||
|
||||
<span class="c1"># Twin Delayed DDPG - https://arxiv.org/pdf/1802.09477.pdf</span>
|
||||
<span class="k">class</span> <span class="nc">TD3Agent</span><span class="p">(</span><span class="n">DDPGAgent</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">'LevelManager'</span><span class="p">,</span> <span class="s1">'CompositeAgent'</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="bp">self</span><span class="o">.</span><span class="n">q_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="s2">"Q"</span><span class="p">)</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">TD_targets_signal</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="s2">"TD targets"</span><span class="p">)</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">action_signal</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="s2">"actions"</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">actor</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">'actor'</span><span class="p">]</span>
|
||||
<span class="n">critic</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">'critic'</span><span class="p">]</span>
|
||||
|
||||
<span class="n">actor_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">'actor'</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="n">critic_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">'critic'</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"># TD error = r + discount*max(q_st_plus_1) - q_st</span>
|
||||
<span class="n">next_actions</span><span class="p">,</span> <span class="n">actions_mean</span> <span class="o">=</span> <span class="n">actor</span><span class="o">.</span><span class="n">parallel_prediction</span><span class="p">([</span>
|
||||
<span class="p">(</span><span class="n">actor</span><span class="o">.</span><span class="n">target_network</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">actor_keys</span><span class="p">)),</span>
|
||||
<span class="p">(</span><span class="n">actor</span><span class="o">.</span><span class="n">online_network</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">actor_keys</span><span class="p">))</span>
|
||||
<span class="p">])</span>
|
||||
|
||||
<span class="c1"># add noise to the next_actions</span>
|
||||
<span class="n">noise</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">normal</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">ap</span><span class="o">.</span><span class="n">algorithm</span><span class="o">.</span><span class="n">policy_noise</span><span class="p">,</span> <span class="n">next_actions</span><span class="o">.</span><span class="n">shape</span><span class="p">)</span><span class="o">.</span><span class="n">clip</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">noise_clipping</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">noise_clipping</span><span class="p">)</span>
|
||||
<span class="n">next_actions</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">spaces</span><span class="o">.</span><span class="n">action</span><span class="o">.</span><span class="n">clip_action_to_space</span><span class="p">(</span><span class="n">next_actions</span> <span class="o">+</span> <span class="n">noise</span><span class="p">)</span>
|
||||
|
||||
<span class="n">critic_inputs</span> <span class="o">=</span> <span class="n">copy</span><span class="o">.</span><span class="n">copy</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">critic_keys</span><span class="p">))</span>
|
||||
<span class="n">critic_inputs</span><span class="p">[</span><span class="s1">'action'</span><span class="p">]</span> <span class="o">=</span> <span class="n">next_actions</span>
|
||||
<span class="n">q_st_plus_1</span> <span class="o">=</span> <span class="n">critic</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">critic_inputs</span><span class="p">)[</span><span class="mi">2</span><span class="p">]</span> <span class="c1"># output #2 is the min (Q1, Q2)</span>
|
||||
|
||||
<span class="c1"># calculate the bootstrapped TD targets while discounting terminal states according to</span>
|
||||
<span class="c1"># use_non_zero_discount_for_terminal_states</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">use_non_zero_discount_for_terminal_states</span><span class="p">:</span>
|
||||
<span class="n">TD_targets</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">expand_dims</span><span class="o">=</span><span class="kc">True</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">q_st_plus_1</span>
|
||||
<span class="k">else</span><span class="p">:</span>
|
||||
<span class="n">TD_targets</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">expand_dims</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> <span class="o">+</span> \
|
||||
<span class="p">(</span><span class="mf">1.0</span> <span class="o">-</span> <span class="n">batch</span><span class="o">.</span><span class="n">game_overs</span><span class="p">(</span><span class="n">expand_dims</span><span class="o">=</span><span class="kc">True</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">q_st_plus_1</span>
|
||||
|
||||
<span class="c1"># clip the TD targets to prevent overestimation errors</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">clip_critic_targets</span><span class="p">:</span>
|
||||
<span class="n">TD_targets</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">clip</span><span class="p">(</span><span class="n">TD_targets</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">clip_critic_targets</span><span class="p">)</span>
|
||||
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">TD_targets_signal</span><span class="o">.</span><span class="n">add_sample</span><span class="p">(</span><span class="n">TD_targets</span><span class="p">)</span>
|
||||
|
||||
<span class="c1"># train the critic</span>
|
||||
<span class="n">critic_inputs</span> <span class="o">=</span> <span class="n">copy</span><span class="o">.</span><span class="n">copy</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">critic_keys</span><span class="p">))</span>
|
||||
<span class="n">critic_inputs</span><span class="p">[</span><span class="s1">'action'</span><span class="p">]</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="nb">len</span><span class="p">(</span><span class="n">batch</span><span class="o">.</span><span class="n">actions</span><span class="p">()</span><span class="o">.</span><span class="n">shape</span><span class="p">)</span> <span class="o">==</span> <span class="mi">1</span><span class="p">)</span>
|
||||
<span class="n">result</span> <span class="o">=</span> <span class="n">critic</span><span class="o">.</span><span class="n">train_and_sync_networks</span><span class="p">(</span><span class="n">critic_inputs</span><span class="p">,</span> <span class="n">TD_targets</span><span class="p">)</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="n">result</span><span class="p">[:</span><span class="mi">3</span><span class="p">]</span>
|
||||
|
||||
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">training_iteration</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">update_policy_every_x_episode_steps</span> <span class="o">==</span> <span class="mi">0</span><span class="p">:</span>
|
||||
<span class="c1"># get the gradients of output #3 (=mean of Q1 network) w.r.t the action</span>
|
||||
<span class="n">critic_inputs</span> <span class="o">=</span> <span class="n">copy</span><span class="o">.</span><span class="n">copy</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">critic_keys</span><span class="p">))</span>
|
||||
<span class="n">critic_inputs</span><span class="p">[</span><span class="s1">'action'</span><span class="p">]</span> <span class="o">=</span> <span class="n">actions_mean</span>
|
||||
<span class="n">action_gradients</span> <span class="o">=</span> <span class="n">critic</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">critic_inputs</span><span class="p">,</span>
|
||||
<span class="n">outputs</span><span class="o">=</span><span class="n">critic</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">gradients_wrt_inputs</span><span class="p">[</span><span class="mi">3</span><span class="p">][</span><span class="s1">'action'</span><span class="p">])</span>
|
||||
|
||||
<span class="c1"># apply the gradients from the critic to the actor</span>
|
||||
<span class="n">initial_feed_dict</span> <span class="o">=</span> <span class="p">{</span><span class="n">actor</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">gradients_weights_ph</span><span class="p">[</span><span class="mi">0</span><span class="p">]:</span> <span class="o">-</span><span class="n">action_gradients</span><span class="p">}</span>
|
||||
<span class="n">gradients</span> <span class="o">=</span> <span class="n">actor</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">actor_keys</span><span class="p">),</span>
|
||||
<span class="n">outputs</span><span class="o">=</span><span class="n">actor</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">weighted_gradients</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span>
|
||||
<span class="n">initial_feed_dict</span><span class="o">=</span><span class="n">initial_feed_dict</span><span class="p">)</span>
|
||||
|
||||
<span class="k">if</span> <span class="n">actor</span><span class="o">.</span><span class="n">has_global</span><span class="p">:</span>
|
||||
<span class="n">actor</span><span class="o">.</span><span class="n">apply_gradients_to_global_network</span><span class="p">(</span><span class="n">gradients</span><span class="p">)</span>
|
||||
<span class="n">actor</span><span class="o">.</span><span class="n">update_online_network</span><span class="p">()</span>
|
||||
<span class="k">else</span><span class="p">:</span>
|
||||
<span class="n">actor</span><span class="o">.</span><span class="n">apply_gradients_to_online_network</span><span class="p">(</span><span class="n">gradients</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">train</span><span class="p">(</span><span class="bp">self</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_consecutive_training_steps</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">current_episode_steps_counter</span>
|
||||
<span class="k">return</span> <span class="n">Agent</span><span class="o">.</span><span class="n">train</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span>
|
||||
|
||||
<span class="k">def</span> <span class="nf">update_transition_before_adding_to_replay_buffer</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">transition</span><span class="p">:</span> <span class="n">Transition</span><span class="p">)</span> <span class="o">-></span> <span class="n">Transition</span><span class="p">:</span>
|
||||
<span class="sd">"""</span>
|
||||
<span class="sd"> Allows agents to update the transition just before adding it to the replay buffer.</span>
|
||||
<span class="sd"> Can be useful for agents that want to tweak the reward, termination signal, etc.</span>
|
||||
|
||||
<span class="sd"> :param transition: the transition to update</span>
|
||||
<span class="sd"> :return: the updated transition</span>
|
||||
<span class="sd"> """</span>
|
||||
<span class="n">transition</span><span class="o">.</span><span class="n">game_over</span> <span class="o">=</span> <span class="kc">False</span> <span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">current_episode_steps_counter</span> <span class="o">==</span>\
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">parent_level_manager</span><span class="o">.</span><span class="n">environment</span><span class="o">.</span><span class="n">env</span><span class="o">.</span><span class="n">_max_episode_steps</span>\
|
||||
<span class="k">else</span> <span class="n">transition</span><span class="o">.</span><span class="n">game_over</span>
|
||||
|
||||
<span class="k">return</span> <span class="n">transition</span>
|
||||
</pre></div>
|
||||
|
||||
</div>
|
||||
|
||||
</div>
|
||||
<footer>
|
||||
|
||||
|
||||
<hr/>
|
||||
|
||||
<div role="contentinfo">
|
||||
<p>
|
||||
© 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>
|
||||
@@ -197,7 +197,7 @@
|
||||
<span class="c1"># limitations under the License.</span>
|
||||
<span class="c1">#</span>
|
||||
<span class="kn">from</span> <span class="nn">collections</span> <span class="k">import</span> <span class="n">OrderedDict</span>
|
||||
<span class="kn">from</span> <span class="nn">typing</span> <span class="k">import</span> <span class="n">Union</span>
|
||||
<span class="kn">from</span> <span class="nn">typing</span> <span class="k">import</span> <span class="n">Union</span><span class="p">,</span> <span class="n">List</span>
|
||||
|
||||
<span class="kn">import</span> <span class="nn">numpy</span> <span class="k">as</span> <span class="nn">np</span>
|
||||
|
||||
@@ -207,7 +207,8 @@
|
||||
<span class="kn">from</span> <span class="nn">rl_coach.logger</span> <span class="k">import</span> <span class="n">screen</span>
|
||||
<span class="kn">from</span> <span class="nn">rl_coach.memories.non_episodic.prioritized_experience_replay</span> <span class="k">import</span> <span class="n">PrioritizedExperienceReplay</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">copy</span> <span class="k">import</span> <span class="n">deepcopy</span>
|
||||
<span class="kn">from</span> <span class="nn">copy</span> <span class="k">import</span> <span class="n">deepcopy</span><span class="p">,</span> <span class="n">copy</span>
|
||||
|
||||
|
||||
<span class="c1">## This is an abstract agent - there is no learn_from_batch method ##</span>
|
||||
|
||||
@@ -218,6 +219,12 @@
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">q_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="s2">"Q"</span><span class="p">)</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">q_value_for_action</span> <span class="o">=</span> <span class="p">{}</span>
|
||||
|
||||
<span class="c1"># currently we use softmax action probabilities only in batch-rl,</span>
|
||||
<span class="c1"># but we might want to extend this later at some point.</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">should_get_softmax_probabilities</span> <span class="o">=</span> \
|
||||
<span class="nb">hasattr</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">network_wrappers</span><span class="p">[</span><span class="s1">'main'</span><span class="p">],</span> <span class="s1">'should_get_softmax_probabilities'</span><span class="p">)</span> <span class="ow">and</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">'main'</span><span class="p">]</span><span class="o">.</span><span class="n">should_get_softmax_probabilities</span>
|
||||
|
||||
<span class="k">def</span> <span class="nf">init_environment_dependent_modules</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="n">init_environment_dependent_modules</span><span class="p">()</span>
|
||||
<span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">spaces</span><span class="o">.</span><span class="n">action</span><span class="p">,</span> <span class="n">DiscreteActionSpace</span><span class="p">):</span>
|
||||
@@ -228,12 +235,21 @@
|
||||
|
||||
<span class="c1"># Algorithms for which q_values are calculated from predictions will override this function</span>
|
||||
<span class="k">def</span> <span class="nf">get_all_q_values_for_states</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">StateType</span><span class="p">):</span>
|
||||
<span class="n">actions_q_values</span> <span class="o">=</span> <span class="kc">None</span>
|
||||
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">exploration_policy</span><span class="o">.</span><span class="n">requires_action_values</span><span class="p">():</span>
|
||||
<span class="n">actions_q_values</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">get_prediction</span><span class="p">(</span><span class="n">states</span><span class="p">)</span>
|
||||
<span class="k">else</span><span class="p">:</span>
|
||||
<span class="n">actions_q_values</span> <span class="o">=</span> <span class="kc">None</span>
|
||||
|
||||
<span class="k">return</span> <span class="n">actions_q_values</span>
|
||||
|
||||
<span class="k">def</span> <span class="nf">get_all_q_values_for_states_and_softmax_probabilities</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">StateType</span><span class="p">):</span>
|
||||
<span class="n">actions_q_values</span><span class="p">,</span> <span class="n">softmax_probabilities</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span> <span class="kc">None</span>
|
||||
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">exploration_policy</span><span class="o">.</span><span class="n">requires_action_values</span><span class="p">():</span>
|
||||
<span class="n">outputs</span> <span class="o">=</span> <span class="n">copy</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">'main'</span><span class="p">]</span><span class="o">.</span><span class="n">online_network</span><span class="o">.</span><span class="n">outputs</span><span class="p">)</span>
|
||||
<span class="n">outputs</span><span class="o">.</span><span class="n">append</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">'main'</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">0</span><span class="p">]</span><span class="o">.</span><span class="n">softmax</span><span class="p">)</span>
|
||||
|
||||
<span class="n">actions_q_values</span><span class="p">,</span> <span class="n">softmax_probabilities</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">get_prediction</span><span class="p">(</span><span class="n">states</span><span class="p">,</span> <span class="n">outputs</span><span class="o">=</span><span class="n">outputs</span><span class="p">)</span>
|
||||
<span class="k">return</span> <span class="n">actions_q_values</span><span class="p">,</span> <span class="n">softmax_probabilities</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">outputs</span><span class="o">=</span><span class="kc">None</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">'main'</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="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="s1">'main'</span><span class="p">),</span>
|
||||
<span class="n">outputs</span><span class="o">=</span><span class="n">outputs</span><span class="p">)</span>
|
||||
@@ -255,10 +271,19 @@
|
||||
<span class="p">)</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">policy</span><span class="o">.</span><span class="vm">__class__</span><span class="o">.</span><span class="vm">__name__</span><span class="p">))</span>
|
||||
|
||||
<span class="k">def</span> <span class="nf">choose_action</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">curr_state</span><span class="p">):</span>
|
||||
<span class="n">actions_q_values</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">get_all_q_values_for_states</span><span class="p">(</span><span class="n">curr_state</span><span class="p">)</span>
|
||||
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">should_get_softmax_probabilities</span><span class="p">:</span>
|
||||
<span class="n">actions_q_values</span><span class="p">,</span> <span class="n">softmax_probabilities</span> <span class="o">=</span> \
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">get_all_q_values_for_states_and_softmax_probabilities</span><span class="p">(</span><span class="n">curr_state</span><span class="p">)</span>
|
||||
<span class="k">else</span><span class="p">:</span>
|
||||
<span class="n">actions_q_values</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">get_all_q_values_for_states</span><span class="p">(</span><span class="n">curr_state</span><span class="p">)</span>
|
||||
|
||||
<span class="c1"># choose action according to the exploration policy and the current phase (evaluating or training the agent)</span>
|
||||
<span class="n">action</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">exploration_policy</span><span class="o">.</span><span class="n">get_action</span><span class="p">(</span><span class="n">actions_q_values</span><span class="p">)</span>
|
||||
<span class="n">action</span><span class="p">,</span> <span class="n">action_probabilities</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">exploration_policy</span><span class="o">.</span><span class="n">get_action</span><span class="p">(</span><span class="n">actions_q_values</span><span class="p">)</span>
|
||||
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">should_get_softmax_probabilities</span> <span class="ow">and</span> <span class="n">softmax_probabilities</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
||||
<span class="c1"># override the exploration policy's generated probabilities when an action was taken</span>
|
||||
<span class="c1"># with the agent's actual policy</span>
|
||||
<span class="n">action_probabilities</span> <span class="o">=</span> <span class="n">softmax_probabilities</span>
|
||||
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">_validate_action</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">exploration_policy</span><span class="p">,</span> <span class="n">action</span><span class="p">)</span>
|
||||
|
||||
<span class="k">if</span> <span class="n">actions_q_values</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
||||
@@ -270,15 +295,18 @@
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">q_values</span><span class="o">.</span><span class="n">add_sample</span><span class="p">(</span><span class="n">actions_q_values</span><span class="p">)</span>
|
||||
|
||||
<span class="n">actions_q_values</span> <span class="o">=</span> <span class="n">actions_q_values</span><span class="o">.</span><span class="n">squeeze</span><span class="p">()</span>
|
||||
<span class="n">action_probabilities</span> <span class="o">=</span> <span class="n">action_probabilities</span><span class="o">.</span><span class="n">squeeze</span><span class="p">()</span>
|
||||
|
||||
<span class="k">for</span> <span class="n">i</span><span class="p">,</span> <span class="n">q_value</span> <span class="ow">in</span> <span class="nb">enumerate</span><span class="p">(</span><span class="n">actions_q_values</span><span class="p">):</span>
|
||||
<span class="bp">self</span><span class="o">.</span><span class="n">q_value_for_action</span><span class="p">[</span><span class="n">i</span><span class="p">]</span><span class="o">.</span><span class="n">add_sample</span><span class="p">(</span><span class="n">q_value</span><span class="p">)</span>
|
||||
|
||||
<span class="n">action_info</span> <span class="o">=</span> <span class="n">ActionInfo</span><span class="p">(</span><span class="n">action</span><span class="o">=</span><span class="n">action</span><span class="p">,</span>
|
||||
<span class="n">action_value</span><span class="o">=</span><span class="n">actions_q_values</span><span class="p">[</span><span class="n">action</span><span class="p">],</span>
|
||||
<span class="n">max_action_value</span><span class="o">=</span><span class="n">np</span><span class="o">.</span><span class="n">max</span><span class="p">(</span><span class="n">actions_q_values</span><span class="p">))</span>
|
||||
<span class="n">max_action_value</span><span class="o">=</span><span class="n">np</span><span class="o">.</span><span class="n">max</span><span class="p">(</span><span class="n">actions_q_values</span><span class="p">),</span>
|
||||
<span class="n">all_action_probabilities</span><span class="o">=</span><span class="n">action_probabilities</span><span class="p">)</span>
|
||||
|
||||
<span class="k">else</span><span class="p">:</span>
|
||||
<span class="n">action_info</span> <span class="o">=</span> <span class="n">ActionInfo</span><span class="p">(</span><span class="n">action</span><span class="o">=</span><span class="n">action</span><span class="p">)</span>
|
||||
<span class="n">action_info</span> <span class="o">=</span> <span class="n">ActionInfo</span><span class="p">(</span><span class="n">action</span><span class="o">=</span><span class="n">action</span><span class="p">,</span> <span class="n">all_action_probabilities</span><span class="o">=</span><span class="n">action_probabilities</span><span class="p">)</span>
|
||||
|
||||
<span class="k">return</span> <span class="n">action_info</span>
|
||||
|
||||
|
||||
Reference in New Issue
Block a user