From 34e2eceada5d5ba1451eb08c6fed12c43551f6e5 Mon Sep 17 00:00:00 2001 From: Alexander Soare Date: Mon, 27 May 2024 13:01:27 +0100 Subject: [PATCH] improve documentation --- lerobot/common/policies/act/modeling_act.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/lerobot/common/policies/act/modeling_act.py b/lerobot/common/policies/act/modeling_act.py index 72ebdd7a..eafe677b 100644 --- a/lerobot/common/policies/act/modeling_act.py +++ b/lerobot/common/policies/act/modeling_act.py @@ -198,7 +198,7 @@ class ACT(nn.Module): def __init__(self, config: ACTConfig): super().__init__() self.config = config - # BERT style VAE encoder with input [cls, *joint_space_configuration, *action_sequence]. + # BERT style VAE encoder with input tokens [cls, robot_state, *action_sequence]. # The cls token forms parameters of the latent's distribution (like this [*means, *log_variances]). if self.config.use_vae: self.vae_encoder = ACTEncoder(config) @@ -214,7 +214,7 @@ class ACT(nn.Module): self.latent_dim = config.latent_dim # Projection layer from the VAE encoder's output to the latent distribution's parameter space. self.vae_encoder_latent_output_proj = nn.Linear(config.dim_model, self.latent_dim * 2) - # Fixed sinusoidal positional embedding the whole input to the VAE encoder. Unsqueeze for batch + # Fixed sinusoidal positional embedding for the input to the VAE encoder. Unsqueeze for batch # dimension. self.register_buffer( "vae_encoder_pos_enc",