evolutionary-policy-optimization 0.0.11__py3-none-any.whl → 0.0.14__py3-none-any.whl

evolutionary_policy_optimization/__init__.py

@@ -2,6 +2,7 @@ from evolutionary_policy_optimization.epo import (
     MLP,
     Actor,
     Critic,
+    create_agent,
     Agent,
     LatentGenePool
 )

evolutionary_policy_optimization/epo.py

@@ -502,6 +502,38 @@ class LatentGenePool(Module):
 
 # agent contains the actor, critic, and the latent genetic pool
 
+def create_agent(
+    dim_state,
+    num_latents,
+    dim_latent,
+    actor_num_actions,
+    actor_dim_hiddens: int | tuple[int, ...],
+    critic_dim_hiddens: int | tuple[int, ...],
+    num_latent_sets = 1
+) -> Agent:
+
+    actor = Actor(
+        num_actions = actor_num_actions,
+        dim_state = dim_state,
+        dim_latent = dim_latent,
+        dim_hiddens = actor_dim_hiddens
+    )
+
+    critic = Critic(
+        dim_state = dim_state,
+        dim_latent = dim_latent,
+        dim_hiddens = critic_dim_hiddens
+    )
+
+    latent_gene_pool = LatentGenePool(
+        dim_state = dim_state,
+        num_latents = num_latents,
+        dim_latent = dim_latent,
+        num_latent_sets = num_latent_sets
+    )
+
+    return Agent(actor = actor, critic = critic, latent_gene_pool = latent_gene_pool)
+
 class Agent(Module):
     def __init__(
         self,
@@ -516,6 +548,28 @@ class Agent(Module):
 
         self.latent_gene_pool = latent_gene_pool
 
+    def get_actor_actions(
+        self,
+        state,
+        latent_id
+    ):
+        latent = self.latent_gene_pool(latent_id = latent_id, state = state)
+        return self.actor(state, latent)
+
+    def get_critic_values(
+        self,
+        state,
+        latent_id
+    ):
+        latent = self.latent_gene_pool(latent_id = latent_id, state = state)
+        return self.critic(state, latent)
+
+    def update_latent_gene_pool_(
+        self,
+        fitnesses
+    ):
+        return self.latent_gene_pool.genetic_algorithm_step(fitnesses)
+
     def forward(
         self,
         memories: list[Memory]
@@ -539,13 +593,10 @@ class EPO(Module):
 
     def __init__(
         self,
-        agent: Agent,
-        latent_gene_pool: LatentGenePool
+        agent: Agent
     ):
         super().__init__()
-
         self.agent = agent
-        self.latent_gene_pool = latent_gene_pool
 
     def forward(
         self,

{evolutionary_policy_optimization-0.0.11.dist-info → evolutionary_policy_optimization-0.0.14.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: evolutionary-policy-optimization
-Version: 0.0.11
+Version: 0.0.14
 Summary: EPO - Pytorch
 Project-URL: Homepage, https://pypi.org/project/evolutionary-policy-optimization/
 Project-URL: Repository, https://github.com/lucidrains/evolutionary-policy-optimization

evolutionary_policy_optimization-0.0.14.dist-info/RECORD

@@ -0,0 +1,7 @@
+evolutionary_policy_optimization/__init__.py,sha256=Qavcia0n13jjaWIS_LPW7QrxSLT_BBeKujCjF9kQjbA,133
+evolutionary_policy_optimization/epo.py,sha256=vGSq_KCHxyJClzU23kWRFjwBv511JZN43bBd7KOpWGo,16842
+evolutionary_policy_optimization/experimental.py,sha256=ktBKxRF27Qsj7WIgBpYlWXqMVxO9zOx2oD1JuDYRAwM,548
+evolutionary_policy_optimization-0.0.14.dist-info/METADATA,sha256=KV5cnqxeUqEAwNYj03sHivKo-cpd12RL_I2Wfu3KLaM,4357
+evolutionary_policy_optimization-0.0.14.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+evolutionary_policy_optimization-0.0.14.dist-info/licenses/LICENSE,sha256=1yCiA9b5nhslTavxPjsQAO-wpOnwJR9-l8LTVi7GJuk,1066
+evolutionary_policy_optimization-0.0.14.dist-info/RECORD,,

evolutionary_policy_optimization-0.0.11.dist-info/RECORD

@@ -1,7 +0,0 @@
-evolutionary_policy_optimization/__init__.py,sha256=A07bhbBI_p-GlSTkI15pioQ1XgtJ0V4tBN6v3vs2nuU,115
-evolutionary_policy_optimization/epo.py,sha256=JGow9ofx7IgFy7QNL0dL0K_SCL_bVkBUznMG8aSGM9Q,15591
-evolutionary_policy_optimization/experimental.py,sha256=ktBKxRF27Qsj7WIgBpYlWXqMVxO9zOx2oD1JuDYRAwM,548
-evolutionary_policy_optimization-0.0.11.dist-info/METADATA,sha256=fkouRBZU5nrPgHt0eT5izSHdOiYGAg67N5Gn3t039mQ,4357
-evolutionary_policy_optimization-0.0.11.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-evolutionary_policy_optimization-0.0.11.dist-info/licenses/LICENSE,sha256=1yCiA9b5nhslTavxPjsQAO-wpOnwJR9-l8LTVi7GJuk,1066
-evolutionary_policy_optimization-0.0.11.dist-info/RECORD,,