def get_generator(checkpoint, evalArgs): args = AttrDict(checkpoint['args']) generator = TrajectoryGenerator( obs_len=args.obs_len, pred_len=args.pred_len, embedding_dim=args.embedding_dim, encoder_h_dim=args.encoder_h_dim_g, decoder_h_dim=args.decoder_h_dim_g, mlp_dim=args.mlp_dim, num_layers=args.num_layers, noise_dim=args.noise_dim, noise_type=args.noise_type, noise_mix_type=args.noise_mix_type, pooling_type=args.pooling_type, pool_every_timestep=args.pool_every_timestep, dropout=args.dropout, bottleneck_dim=args.bottleneck_dim, neighborhood_size=args.neighborhood_size, grid_size=args.grid_size, batch_norm=args.batch_norm, use_gpu=evalArgs.use_gpu) generator.load_state_dict(checkpoint['g_state']) if evalArgs.use_gpu: generator.cuda() else: generator.cpu() generator.train() return generator
def get_generator(checkpoint): args = AttrDict(checkpoint['args']) generator = TrajectoryGenerator( obs_len=args.obs_len, pred_len=args.pred_len, embedding_dim=args.embedding_dim, encoder_h_dim=args.encoder_h_dim_g, decoder_h_dim=args.decoder_h_dim_g, rep_dim=args.rep_dim, mlp_dim=args.mlp_dim, encoder_num_layers=args.encoder_num_layers, decoder_num_layers=args.decoder_num_layers, noise_dim=args.noise_dim, noise_type=args.noise_type, noise_mix_type=args.noise_mix_type, pooling_type=args.pooling_type, pool_every_timestep=args.pool_every_timestep, dropout=args.dropout, bottleneck_dim=args.bottleneck_dim, neighborhood_size=args.neighborhood_size, grid_size=args.grid_size, batch_norm=args.batch_norm, pos_embed=args.pos_embed, pos_embed_freq=args.pos_embed_freq, ) generator.load_state_dict(checkpoint['g_state']) generator.cuda() generator.train() return generator
def get_generator(checkpoint): args = AttrDict(checkpoint['args']) n_units = ([40] + [int(x) for x in args.hidden_units.strip().split(",")] + [40]) generator = TrajectoryGenerator( obs_len=args.obs_len, pred_len=args.pred_len, embedding_dim=args.embedding_dim, encoder_h_dim=args.encoder_h_dim_g, decoder_h_dim=args.decoder_h_dim_g, mlp_dim=args.mlp_dim, num_layers=args.num_layers, noise_dim=args.noise_dim, noise_type=args.noise_type, noise_mix_type=args.noise_mix_type, pooling_type=args.pooling_type, pool_every_timestep=args.pool_every_timestep, dropout=args.dropout, bottleneck_dim=args.bottleneck_dim, neighborhood_size=args.neighborhood_size, grid_size=args.grid_size, batch_norm=args.batch_norm, n_units=n_units, n_heads=args.n_heads, dropout1=args.dropout1, alpha=args.alpha).cuda() generator.load_state_dict(checkpoint['g_state']) generator.cuda() generator.train() return generator
def get_generator(self, checkpoint): self.import_modules() from sgan.models import TrajectoryGenerator args = attrdict.AttrDict(checkpoint['args']) generator = TrajectoryGenerator( obs_len=args.obs_len, pred_len=args.pred_len, embedding_dim=args.embedding_dim, encoder_h_dim=args.encoder_h_dim_g, decoder_h_dim=args.decoder_h_dim_g, mlp_dim=args.mlp_dim, num_layers=args.num_layers, noise_dim=args.noise_dim, noise_type=args.noise_type, noise_mix_type=args.noise_mix_type, pooling_type=args.pooling_type, pool_every_timestep=args.pool_every_timestep, dropout=args.dropout, bottleneck_dim=args.bottleneck_dim, neighborhood_size=args.neighborhood_size, grid_size=args.grid_size, batch_norm=args.batch_norm) generator.load_state_dict(checkpoint['g_state']) generator.cpu() generator.train() return generator
def get_generator(self, checkpoint): generator = TrajectoryGenerator( obs_len=self.args_.obs_len, pred_len=self.args_.pred_len, embedding_dim=self.args_.embedding_dim, encoder_h_dim=self.args_.encoder_h_dim_g, decoder_h_dim=self.args_.decoder_h_dim_g, mlp_dim=self.args_.mlp_dim, num_layers=self.args_.num_layers, noise_dim=self.args_.noise_dim, noise_type=self.args_.noise_type, noise_mix_type=self.args_.noise_mix_type, pooling_type=self.args_.pooling_type, pool_every_timestep=self.args_.pool_every_timestep, dropout=self.args_.dropout, bottleneck_dim=self.args_.bottleneck_dim, neighborhood_size=self.args_.neighborhood_size, grid_size=self.args_.grid_size, batch_norm=self.args_.batch_norm, use_gpu=self.args_.use_gpu) generator.load_state_dict(checkpoint['g_state']) if self.args_.use_gpu: generator.cuda() generator.train() return generator
def get_generator(checkpoint, best=0): args = AttrDict(checkpoint['args']) print(args.pred_len) generator = TrajectoryGenerator( obs_len=args.obs_len, pred_len=args.pred_len, embedding_dim=args.embedding_dim, encoder_h_dim=args.encoder_h_dim_g, decoder_h_dim=args.decoder_h_dim_g, mlp_dim=args.mlp_dim, noise_dim=args.noise_dim, noise_type=args.noise_type, noise_mix_type=args.noise_mix_type, pooling_type=args.pooling_type, pool_every_timestep=args.pool_every_timestep, dropout=args.dropout, bottleneck_dim=args.bottleneck_dim, neighborhood_size=args.neighborhood_size, grid_size=args.grid_size, batch_norm=args.batch_norm, ) if best: generator.load_state_dict(checkpoint['g_best_state']) else: generator.load_state_dict(checkpoint['g_state']) generator.cuda() generator.train() return generator