# Copyright 2025 The RLinf Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# https://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import torch.nn as nn
[docs]
class ValueHead(nn.Module):
def __init__(
self,
input_dim: int,
hidden_sizes=(512, 128),
output_dim: int = 1,
activation: str = "gelu", # 'relu' or 'gelu'
bias_last: bool = False,
):
super().__init__()
layers = []
in_dim = input_dim
if activation.lower() == "relu":
act = nn.ReLU
elif activation.lower() == "gelu":
act = nn.GELU
elif activation.lower() == "tanh":
act = nn.Tanh
else:
raise ValueError(f"Unsupported activation: {activation}")
for h in hidden_sizes:
layers.append(nn.Linear(in_dim, h))
layers.append(act())
in_dim = h
layers.append(nn.Linear(in_dim, output_dim, bias=bias_last))
self.mlp = nn.Sequential(*layers)
self._init_weights(activation.lower())
def _init_weights(self, nonlinearity="relu"):
for m in self.mlp:
if isinstance(m, nn.Linear):
if m is self.mlp[-1]:
nn.init.normal_(m.weight, mean=0.0, std=0.02)
if m.bias is not None:
nn.init.zeros_(m.bias)
else:
nn.init.kaiming_normal_(
m.weight, mode="fan_out", nonlinearity=nonlinearity
)
if m.bias is not None:
nn.init.zeros_(m.bias)
[docs]
def forward(self, x):
return self.mlp(x)