Add Lookahead+RAdam optimizer

slm_lab/agent/net/net_util.py

@@ -10,6 +10,7 @@
 # register custom torch.optim
 setattr(torch.optim, 'GlobalAdam', optimizer.GlobalAdam)
 setattr(torch.optim, 'GlobalRMSprop', optimizer.GlobalRMSprop)
+setattr(torch.optim, 'Lookahead', optimizer.Lookahead)
 setattr(torch.optim, 'RAdam', optimizer.RAdam)
 
 
@@ -329,9 +330,11 @@ def init_global_nets(algorithm):
             global_nets[f'global_{net_name}'] = g_net
         # if optim is Global, set to override the local optim and its scheduler
         optim = getattr(algorithm, optim_name)
-        if 'Global' in util.get_class_name(optim):
+        if hasattr(optim, 'share_memory'):
             optim.share_memory()  # make optim global
             global_nets[optim_name] = optim
+            if hasattr(optim, 'optimizer'):  # for Lookahead with an inner optimizer
+                global_nets[f'{optim_name}_optimizer'] = optim.optimizer
             lr_scheduler_name = net_name.replace('net', 'lr_scheduler')
             lr_scheduler = getattr(algorithm, lr_scheduler_name)
             global_nets[lr_scheduler_name] = lr_scheduler
@@ -346,6 +349,12 @@ def set_global_nets(algorithm, global_nets):
         setattr(algorithm, f'global_{net_name}', None)
     # set attr created in init_global_nets
     if global_nets is not None:
+        # handle inner-optimizer recovery
+        inner_opt_keys = [k for k in global_nets if k.endswith('_optimizer')]
+        for inner_opt_key in inner_opt_keys:
+            opt = global_nets[inner_opt_key.replace('_optimizer', '')]  # optimizer which has a inner optimizer
+            setattr(opt, 'optimizer', global_nets.pop(inner_opt_key))
+        # set global nets and optims
         util.set_attr(algorithm, global_nets)
         logger.info(f'Set global_nets attr {list(global_nets.keys())} for Hogwild')
 

slm_lab/lib/optimizer.py

@@ -1,5 +1,6 @@
 # Custom PyTorch optimizer classes, to be registered in net_util.py
 from torch.optim.optimizer import Optimizer
+import itertools as it
 import math
 import torch
 
@@ -103,6 +104,56 @@ def step(self, closure=None):
         return loss
 
 
+class Lookahead(Optimizer):
+    '''
+    Lookahead Optimizer: k steps forward, 1 step back
+    https://arxiv.org/abs/1907.08610
+    Implementation modified from https://github.com/lonePatient/lookahead_pytorch; reference from https://medium.com/@lessw/new-deep-learning-optimizer-ranger-synergistic-combination-of-radam-lookahead-for-the-best-of-2dc83f79a48d
+    '''
+
+    def __init__(self, params, alpha=0.5, k=5, optimizer='RAdam', **optimizer_kwargs):
+        if not 0.0 <= alpha <= 1.0:
+            raise ValueError(f'Invalid slow update rate: {alpha}')
+        if not 1 <= k:
+            raise ValueError(f'Invalid lookahead steps: {k}')
+        # construct base optimizer
+        OptimClass = getattr(torch.optim, optimizer)
+        self.optimizer = OptimClass(params, **optimizer_kwargs)
+        self.param_groups = self.optimizer.param_groups
+        self.state = self.optimizer.state
+        # create and use defaults to track params to retain them in multiprocessing spawn
+        self.defaults = self.optimizer.defaults
+        self.defaults['alpha'] = alpha
+        self.defaults['k'] = k
+        for group in self.param_groups:
+            group['step_counter'] = 0
+        self.defaults['slow_weights'] = [[
+            p.clone().detach() for p in group['params']]
+            for group in self.param_groups]
+
+        for w in it.chain(*self.defaults['slow_weights']):
+            w.requires_grad = False
+
+    def share_memory(self):
+        self.optimizer.share_memory()
+
+    def step(self, closure=None):
+        loss = None
+        if closure is not None:
+            loss = closure()
+        loss = self.optimizer.step()
+        for group, slow_weights in zip(self.param_groups, self.defaults['slow_weights']):
+            group['step_counter'] += 1
+            if group['step_counter'] % self.defaults['k'] != 0:
+                continue
+            for p, q in zip(group['params'], slow_weights):
+                if p.grad is None:
+                    continue
+                q.data.add_(self.defaults['alpha'], p.data - q.data)
+                p.data.copy_(q.data)
+        return loss
+
+
 class RAdam(Optimizer):
     '''
     RAdam optimizer which stabilizes training vs. different learning rates.
@@ -111,21 +162,33 @@ class RAdam(Optimizer):
     '''
 
     def __init__(self, params, lr=1e-3, betas=(0.9, 0.999), eps=1e-8, weight_decay=0):
-        defaults = dict(lr=lr, betas=betas, eps=eps, weight_decay=weight_decay)
-        self.buffer = [[None, None, None] for ind in range(10)]
+        defaults = dict(lr=lr, betas=betas, eps=eps, weight_decay=weight_decay, buffer=torch.zeros((10, 3)))
         super(RAdam, self).__init__(params, defaults)
 
+        for group in self.param_groups:
+            for p in group['params']:
+                state = self.state[p]
+                state['step'] = torch.zeros(1)
+                state['exp_avg'] = p.data.new().resize_as_(p.data).zero_()
+                state['exp_avg_sq'] = p.data.new().resize_as_(p.data).zero_()
+
     def __setstate__(self, state):
         super(RAdam, self).__setstate__(state)
 
-    def step(self, closure=None):
+    def share_memory(self):
+        for group in self.param_groups:
+            for p in group['params']:
+                state = self.state[p]
+                state['step'].share_memory_()
+                state['exp_avg'].share_memory_()
+                state['exp_avg_sq'].share_memory_()
 
+    def step(self, closure=None):
         loss = None
         if closure is not None:
             loss = closure()
 
         for group in self.param_groups:
-
             for p in group['params']:
                 if p.grad is None:
                     continue
@@ -134,16 +197,9 @@ def step(self, closure=None):
                     raise RuntimeError('RAdam does not support sparse gradients')
 
                 p_data_fp32 = p.data.float()
-
                 state = self.state[p]
-
-                if len(state) == 0:
-                    state['step'] = 0
-                    state['exp_avg'] = torch.zeros_like(p_data_fp32)
-                    state['exp_avg_sq'] = torch.zeros_like(p_data_fp32)
-                else:
-                    state['exp_avg'] = state['exp_avg'].type_as(p_data_fp32)
-                    state['exp_avg_sq'] = state['exp_avg_sq'].type_as(p_data_fp32)
+                state['exp_avg'] = state['exp_avg'].type_as(p_data_fp32)
+                state['exp_avg_sq'] = state['exp_avg_sq'].type_as(p_data_fp32)
 
                 exp_avg, exp_avg_sq = state['exp_avg'], state['exp_avg_sq']
                 beta1, beta2 = group['betas']
@@ -152,7 +208,7 @@ def step(self, closure=None):
                 exp_avg.mul_(beta1).add_(1 - beta1, grad)
 
                 state['step'] += 1
-                buffered = self.buffer[int(state['step'] % 10)]
+                buffered = self.defaults['buffer'][int(state['step'] % 10)]
                 if state['step'] == buffered[0]:
                     N_sma, step_size = buffered[1], buffered[2]
                 else:
@@ -164,20 +220,21 @@ def step(self, closure=None):
 
                     # more conservative since it's an approximated value
                     if N_sma >= 5:
-                        step_size = group['lr'] * math.sqrt((1 - beta2_t) * (N_sma - 4) / (N_sma_max - 4) * (N_sma - 2) / N_sma * N_sma_max / (N_sma_max - 2)) / (1 - beta1 ** state['step'])
+                        step_size = math.sqrt((1 - beta2_t) * (N_sma - 4) / (N_sma_max - 4) * (N_sma - 2) / N_sma * N_sma_max / (N_sma_max - 2)) / (1 - beta1 ** state['step'])
                     else:
-                        step_size = group['lr'] / (1 - beta1 ** state['step'])
+                        step_size = 1.0 / (1 - beta1 ** state['step'])
                     buffered[2] = step_size
 
                 if group['weight_decay'] != 0:
                     p_data_fp32.add_(-group['weight_decay'] * group['lr'], p_data_fp32)
 
                 # more conservative since it's an approximated value
+                adap_lr = (-step_size * group['lr']).squeeze(dim=0).item()
                 if N_sma >= 5:
                     denom = exp_avg_sq.sqrt().add_(group['eps'])
-                    p_data_fp32.addcdiv_(-step_size, exp_avg, denom)
+                    p_data_fp32.addcdiv_(adap_lr, exp_avg, denom)
                 else:
-                    p_data_fp32.add_(-step_size, exp_avg)
+                    p_data_fp32.add_(adap_lr, exp_avg)
 
                 p.data.copy_(p_data_fp32)
 

slm_lab/spec/benchmark/a2c/a2c_gae_roboschool.json

@@ -35,11 +35,13 @@
           "name": "MSELoss"
         },
         "actor_optim_spec": {
-          "name": "Adam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "critic_optim_spec": {
-          "name": "Adam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "lr_scheduler_spec": null,
@@ -66,7 +68,7 @@
     },
     "spec_params": {
       "env": [
-        "RoboschoolAnt-v1", "RoboschoolAtlasForwardWalk-v1", "RoboschoolHalfCheetah-v1", "RoboschoolHopper-v1", "RoboschoolInvertedDoublePendulum-v1", "RoboschoolInvertedPendulum-v1", "RoboschoolInvertedPendulumSwingup-v1",  "RoboschoolReacher-v1", "RoboschoolWalker2d-v1"
+        "RoboschoolAnt-v1", "RoboschoolAtlasForwardWalk-v1", "RoboschoolHalfCheetah-v1", "RoboschoolHopper-v1", "RoboschoolInvertedDoublePendulum-v1", "RoboschoolInvertedPendulum-v1",  "RoboschoolReacher-v1", "RoboschoolWalker2d-v1"
       ]
     }
   },
@@ -106,11 +108,13 @@
           "name": "MSELoss"
         },
         "actor_optim_spec": {
-          "name": "RAdam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "critic_optim_spec": {
-          "name": "RAdam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "lr_scheduler_spec": null,
@@ -172,11 +176,13 @@
           "name": "MSELoss"
         },
         "actor_optim_spec": {
-          "name": "RAdam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "critic_optim_spec": {
-          "name": "RAdam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "lr_scheduler_spec": null,
@@ -238,11 +244,13 @@
           "name": "MSELoss"
         },
         "actor_optim_spec": {
-          "name": "RAdam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "critic_optim_spec": {
-          "name": "RAdam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "lr_scheduler_spec": null,

slm_lab/spec/benchmark/a2c/a2c_nstep_roboschool.json

@@ -35,11 +35,13 @@
           "name": "MSELoss"
         },
         "actor_optim_spec": {
-          "name": "Adam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "critic_optim_spec": {
-          "name": "Adam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "lr_scheduler_spec": null,
@@ -66,7 +68,7 @@
     },
     "spec_params": {
       "env": [
-        "RoboschoolAnt-v1", "RoboschoolAtlasForwardWalk-v1", "RoboschoolHalfCheetah-v1", "RoboschoolHopper-v1", "RoboschoolInvertedDoublePendulum-v1", "RoboschoolInvertedPendulum-v1", "RoboschoolInvertedPendulumSwingup-v1", "RoboschoolReacher-v1", "RoboschoolWalker2d-v1"
+        "RoboschoolAnt-v1", "RoboschoolAtlasForwardWalk-v1", "RoboschoolHalfCheetah-v1", "RoboschoolHopper-v1", "RoboschoolInvertedDoublePendulum-v1", "RoboschoolInvertedPendulum-v1", "RoboschoolReacher-v1", "RoboschoolWalker2d-v1"
       ]
     }
   },
@@ -106,11 +108,13 @@
           "name": "MSELoss"
         },
         "actor_optim_spec": {
-          "name": "RAdam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "critic_optim_spec": {
-          "name": "RAdam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "lr_scheduler_spec": null,
@@ -172,11 +176,13 @@
           "name": "MSELoss"
         },
         "actor_optim_spec": {
-          "name": "RAdam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "critic_optim_spec": {
-          "name": "RAdam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "lr_scheduler_spec": null,
@@ -238,11 +244,13 @@
           "name": "MSELoss"
         },
         "actor_optim_spec": {
-          "name": "RAdam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "critic_optim_spec": {
-          "name": "RAdam",
+          "name": "Lookahead",
+          "optimizer": "RAdam",
           "lr": 3e-4,
         },
         "lr_scheduler_spec": null,