shap_e/models/generation/transformer.py [320:332]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        self.n_ctx = n_ctx
        self.clip = clip
        self.clip_embed = nn.Sequential(
            nn.LayerNorm(
                normalized_shape=(self.clip.grid_feature_dim,), device=device, dtype=dtype
            ),
            nn.Linear(self.clip.grid_feature_dim, self.backbone.width, device=device, dtype=dtype),
        )
        self.cond_drop_prob = cond_drop_prob

    def cached_model_kwargs(self, batch_size: int, model_kwargs: Dict[str, Any]) -> Dict[str, Any]:
        _ = batch_size
        with torch.no_grad():
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



shap_e/models/generation/transformer.py [438:451]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        self.n_ctx = n_ctx

        self.clip = clip
        self.clip_embed = nn.Sequential(
            nn.LayerNorm(
                normalized_shape=(self.clip.grid_feature_dim,), device=device, dtype=dtype
            ),
            nn.Linear(self.clip.grid_feature_dim, self.backbone.width, device=device, dtype=dtype),
        )
        self.cond_drop_prob = cond_drop_prob

    def cached_model_kwargs(self, batch_size: int, model_kwargs: Dict[str, Any]) -> Dict[str, Any]:
        _ = batch_size
        with torch.no_grad():
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -