siglip

class SiglipVisionEmbeddings(nn.Module):
    """
    SigLIP vision embeddings - similar to CLIP but without class embedding.
    """
def __init__(self, arch_config: SiglipVisionArchConfig):
    super().__init__()
    self.arch_config = arch_config
    self.embed_dim = arch_config.hidden_size
    self.image_size = arch_config.image_size
    self.patch_size = arch_config.patch_size
    # SigLIP uses valid padding, so non-divisible sizes work (edge pixels are ignored)

    self.patch_embedding = nn.Conv2d(
        in_channels=arch_config.num_channels,
        out_channels=self.embed_dim,
        kernel_size=self.patch_size,
        stride=self.patch_size,
        padding="valid",  # SigLIP uses valid padding
    )

    # Integer division - with valid padding, edge pixels are ignored
    self.num_patches = (self.image_size // self.patch_size)**2
    self.num_positions = self.num_patches
    self.position_embedding = nn.Embedding(self.num_positions, self.embed_dim)
    self.register_buffer(
        "position_ids",
        torch.arange(self.num_positions).expand((1, -1)),
        persistent=False,
    )

def forward(self, pixel_values: torch.Tensor) -> torch.Tensor:
    target_dtype = self.patch_embedding.weight.dtype
    patch_embeds = self.patch_embedding(pixel_values.to(dtype=target_dtype))  # shape = [*, embed_dim, grid, grid]
    embeddings = patch_embeds.flatten(2).transpose(1, 2)
    embeddings = embeddings + self.position_embedding(self.position_ids)
    return embeddings</code></pre></details>
fastvideo.models.encoders.siglip.SiglipVisionEmbeddings.forward
forward(pixel_values: torch.Tensor) -> torch.Tensor
Source
def forward(self, pixel_values: torch.Tensor) -> torch.Tensor:
    target_dtype = self.patch_embedding.weight.dtype
    patch_embeds = self.patch_embedding(pixel_values.to(dtype=target_dtype))  # shape = [*, embed_dim, grid, grid]
    embeddings = patch_embeds.flatten(2).transpose(1, 2)
    embeddings = embeddings + self.position_embedding(self.position_ids)
    return embeddings
fastvideo.models.encoders.siglip.SiglipAttention
SiglipAttention(arch_config: SiglipVisionArchConfig, enable_scale: bool = True, is_causal: bool = False, quant_config: QuantizationConfig | None = None, prefix: str = '')
Multi-headed attention for SigLIP.
Source
class SiglipAttention(nn.Module):
    """Multi-headed attention for SigLIP."""
def __init__(
    self,
    arch_config: SiglipVisionArchConfig,
    enable_scale: bool = True,
    is_causal: bool = False,
    quant_config: QuantizationConfig | None = None,
    prefix: str = &quot;&quot;,
):
    super().__init__()
    self.arch_config = arch_config
    self.embed_dim = arch_config.hidden_size
    self.num_heads = arch_config.num_attention_heads
    self.head_dim = self.embed_dim // self.num_heads

    if self.head_dim * self.num_heads != self.embed_dim:
        raise ValueError(f&quot;embed_dim must be divisible by num_heads &quot;
                         f&quot;(got `embed_dim`: &#123;self.embed_dim&#125; and `num_heads`: &#123;self.num_heads&#125;).&quot;)

    self.scale = self.head_dim**-0.5 if enable_scale else None
    self.dropout = arch_config.attention_dropout

    self.qkv_proj = QKVParallelLinear(
        hidden_size=self.embed_dim,
        head_size=self.head_dim,
        total_num_heads=self.num_heads,
        quant_config=quant_config,
        prefix=f&quot;&#123;prefix&#125;.qkv_proj&quot;,
    )

    self.out_proj = RowParallelLinear(
        input_size=self.embed_dim,
        output_size=self.embed_dim,
        quant_config=quant_config,
        prefix=f&quot;&#123;prefix&#125;.out_proj&quot;,
    )

    self.tp_size = get_tp_world_size()
    self.num_heads_per_partition = divide(self.num_heads, self.tp_size)

    self.attn = LocalAttention(
        self.num_heads_per_partition,
        self.head_dim,
        self.num_heads_per_partition,
        softmax_scale=self.scale,
        causal=is_causal,
        supported_attention_backends=arch_config._supported_attention_backends,
    )

def forward(self, hidden_states: torch.Tensor):
    &quot;&quot;&quot;Input shape: Batch x Time x Channel&quot;&quot;&quot;
    qkv_states, _ = self.qkv_proj(hidden_states)
    query_states, key_states, value_states = qkv_states.chunk(3, dim=-1)

    query_states = query_states.reshape(query_states.shape&#91;0&#93;, query_states.shape&#91;1&#93;, self.num_heads_per_partition,
                                        self.head_dim)
    key_states = key_states.reshape(key_states.shape&#91;0&#93;, key_states.shape&#91;1&#93;, self.num_heads_per_partition,
                                    self.head_dim)
    value_states = value_states.reshape(value_states.shape&#91;0&#93;, value_states.shape&#91;1&#93;, self.num_heads_per_partition,
                                        self.head_dim)

    attn_output = self.attn(query_states, key_states, value_states)
    attn_output = attn_output.reshape(attn_output.shape&#91;0&#93;, attn_output.shape&#91;1&#93;,
                                      self.num_heads_per_partition * self.head_dim)
    attn_output, _ = self.out_proj(attn_output)
    return attn_output, None</code></pre></details>
fastvideo.models.encoders.siglip.SiglipAttention.forward
forward(hidden_states: torch.Tensor)
Input shape: Batch x Time x Channel
Source
def forward(self, hidden_states: torch.Tensor):
    """Input shape: Batch x Time x Channel"""
    qkv_states, _ = self.qkv_proj(hidden_states)
    query_states, key_states, value_states = qkv_states.chunk(3, dim=-1)
query_states = query_states.reshape(query_states.shape&#91;0&#93;, query_states.shape&#91;1&#93;, self.num_heads_per_partition,
                                    self.head_dim)
key_states = key_states.reshape(key_states.shape&#91;0&#93;, key_states.shape&#91;1&#93;, self.num_heads_per_partition,
                                self.head_dim)
value_states = value_states.reshape(value_states.shape&#91;0&#93;, value_states.shape&#91;1&#93;, self.num_heads_per_partition,
                                    self.head_dim)

attn_output = self.attn(query_states, key_states, value_states)
attn_output = attn_output.reshape(attn_output.shape&#91;0&#93;, attn_output.shape&#91;1&#93;,
                                  self.num_heads_per_partition * self.head_dim)
attn_output, _ = self.out_proj(attn_output)
return attn_output, None</code></pre></details>
fastvideo.models.encoders.siglip.SiglipMLP
SiglipMLP(arch_config: SiglipVisionArchConfig, quant_config: QuantizationConfig | None = None, prefix: str = '')
Source
class SiglipMLP(nn.Module):
def __init__(
    self,
    arch_config: SiglipVisionArchConfig,
    quant_config: QuantizationConfig | None = None,
    prefix: str = &quot;&quot;,
) -&gt; None:
    super().__init__()
    self.arch_config = arch_config
    self.activation_fn = get_act_fn(arch_config.hidden_act)
    self.fc1 = ColumnParallelLinear(
        arch_config.hidden_size,
        arch_config.intermediate_size,
        bias=True,
        quant_config=quant_config,
        prefix=f&quot;&#123;prefix&#125;.fc1&quot;,
    )
    self.fc2 = RowParallelLinear(
        arch_config.intermediate_size,
        arch_config.hidden_size,
        bias=True,
        quant_config=quant_config,
        prefix=f&quot;&#123;prefix&#125;.fc2&quot;,
    )

def forward(self, hidden_states: torch.Tensor) -&gt; torch.Tensor:
    hidden_states, _ = self.fc1(hidden_states)
    hidden_states = self.activation_fn(hidden_states)
    hidden_states, _ = self.fc2(hidden_states)
    return hidden_states</code></pre></details>
fastvideo.models.encoders.siglip.SiglipMLP.forward
forward(hidden_states: torch.Tensor) -> torch.Tensor
Source
def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
    hidden_states, _ = self.fc1(hidden_states)
    hidden_states = self.activation_fn(hidden_states)
    hidden_states, _ = self.fc2(hidden_states)
    return hidden_states
fastvideo.models.encoders.siglip.SiglipEncoderLayer
SiglipEncoderLayer(arch_config: SiglipVisionArchConfig, enable_scale: bool = True, is_causal: bool = False, quant_config: QuantizationConfig | None = None, prefix: str = '')
Source
class SiglipEncoderLayer(nn.Module):
def __init__(
    self,
    arch_config: SiglipVisionArchConfig,
    enable_scale: bool = True,
    is_causal: bool = False,
    quant_config: QuantizationConfig | None = None,
    prefix: str = &quot;&quot;,
) -&gt; None:
    super().__init__()
    self.self_attn = SiglipAttention(
        arch_config,
        enable_scale=enable_scale,
        is_causal=is_causal,
        quant_config=quant_config,
        prefix=f&quot;&#123;prefix&#125;.self_attn&quot;,
    )
    self.layer_norm1 = nn.LayerNorm(arch_config.hidden_size, eps=arch_config.layer_norm_eps)
    self.mlp = SiglipMLP(
        arch_config,
        quant_config=quant_config,
        prefix=f&quot;&#123;prefix&#125;.mlp&quot;,
    )
    self.layer_norm2 = nn.LayerNorm(arch_config.hidden_size, eps=arch_config.layer_norm_eps)

def forward(self, hidden_states: torch.Tensor) -&gt; torch.Tensor:
    # SigLIP uses post-norm (like original ViT)
    residual = hidden_states
    hidden_states = self.layer_norm1(hidden_states)
    hidden_states, _ = self.self_attn(hidden_states=hidden_states)
    hidden_states = residual + hidden_states

    residual = hidden_states
    hidden_states = self.layer_norm2(hidden_states)
    hidden_states = self.mlp(hidden_states)
    hidden_states = residual + hidden_states

    return hidden_states</code></pre></details>
fastvideo.models.encoders.siglip.SiglipEncoderLayer.forward
forward(hidden_states: torch.Tensor) -> torch.Tensor
Source
def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
    # SigLIP uses post-norm (like original ViT)
    residual = hidden_states
    hidden_states = self.layer_norm1(hidden_states)
    hidden_states, _ = self.self_attn(hidden_states=hidden_states)
    hidden_states = residual + hidden_states
residual = hidden_states
hidden_states = self.layer_norm2(hidden_states)
hidden_states = self.mlp(hidden_states)
hidden_states = residual + hidden_states

return hidden_states</code></pre></details>
fastvideo.models.encoders.siglip.SiglipEncoder
SiglipEncoder(arch_config: SiglipVisionArchConfig, enable_scale: bool = True, is_causal: bool = False, quant_config: QuantizationConfig | None = None, num_hidden_layers_override: int | None = None, prefix: str = '')
SigLIP encoder consisting of transformer layers.
Source
class SiglipEncoder(nn.Module):
    """SigLIP encoder consisting of transformer layers."""
def __init__(
    self,
    arch_config: SiglipVisionArchConfig,
    enable_scale: bool = True,
    is_causal: bool = False,
    quant_config: QuantizationConfig | None = None,
    num_hidden_layers_override: int | None = None,
    prefix: str = &quot;&quot;,
) -&gt; None:
    super().__init__()
    self.arch_config = arch_config

    if num_hidden_layers_override is None:
        num_hidden_layers = arch_config.num_hidden_layers
    else:
        num_hidden_layers = num_hidden_layers_override

    self.layers = nn.ModuleList(&#91;
        SiglipEncoderLayer(
            arch_config=arch_config,
            enable_scale=enable_scale,
            is_causal=is_causal,
            quant_config=quant_config,
            prefix=f&quot;&#123;prefix&#125;.layers.&#123;layer_idx&#125;&quot;,
        ) for layer_idx in range(num_hidden_layers)
    &#93;)

def forward(
    self,
    inputs_embeds: torch.Tensor,
    return_all_hidden_states: bool,
) -&gt; torch.Tensor | list&#91;torch.Tensor&#93;:
    hidden_states_pool = &#91;inputs_embeds&#93;
    hidden_states = inputs_embeds

    for encoder_layer in self.layers:
        hidden_states = encoder_layer(hidden_states)
        if return_all_hidden_states:
            hidden_states_pool.append(hidden_states)

    if return_all_hidden_states:
        return hidden_states_pool
    return &#91;hidden_states&#93;</code></pre></details>
fastvideo.models.encoders.siglip.SiglipEncoder.forward
forward(inputs_embeds: torch.Tensor, return_all_hidden_states: bool) -> torch.Tensor | list[torch.Tensor]
Source
def forward(
    self,
    inputs_embeds: torch.Tensor,
    return_all_hidden_states: bool,
) -> torch.Tensor | list[torch.Tensor]:
    hidden_states_pool = [inputs_embeds]
    hidden_states = inputs_embeds
for encoder_layer in self.layers:
    hidden_states = encoder_layer(hidden_states)
    if return_all_hidden_states:
        hidden_states_pool.append(hidden_states)

if return_all_hidden_states:
    return hidden_states_pool
return &#91;hidden_states&#93;</code></pre></details>
fastvideo.models.encoders.siglip.SiglipVisionTransformer
SiglipVisionTransformer(config: SiglipVisionConfig, quant_config: QuantizationConfig | None = None, num_hidden_layers_override: int | None = None, require_post_norm: bool | None = None, prefix: str = '')
Source
class SiglipVisionTransformer(nn.Module):
def __init__(
    self,
    config: SiglipVisionConfig,
    quant_config: QuantizationConfig | None = None,
    num_hidden_layers_override: int | None = None,
    require_post_norm: bool | None = None,
    prefix: str = &quot;&quot;,
) -&gt; None:
    super().__init__()
    self.config = config
    arch_config = config.arch_config
    embed_dim = arch_config.hidden_size

    self.embeddings = SiglipVisionEmbeddings(arch_config)

    self.encoder = SiglipEncoder(
        arch_config=arch_config,
        enable_scale=config.enable_scale,
        is_causal=config.is_causal,
        quant_config=quant_config,
        num_hidden_layers_override=num_hidden_layers_override,
        prefix=f&quot;&#123;prefix&#125;.encoder&quot;,
    )

    num_hidden_layers = arch_config.num_hidden_layers
    if len(self.encoder.layers) &gt; arch_config.num_hidden_layers:
        raise ValueError(f&quot;The original encoder only has &#123;num_hidden_layers&#125; &quot;
                         f&quot;layers, but you requested &#123;len(self.encoder.layers)&#125; layers.&quot;)

    # Post layer norm (applied to output)
    if require_post_norm is None:
        require_post_norm = len(self.encoder.layers) == num_hidden_layers

    if require_post_norm:
        self.post_layernorm = nn.LayerNorm(embed_dim, eps=arch_config.layer_norm_eps)
    else:
        self.post_layernorm = None

def forward(
    self,
    pixel_values: torch.Tensor,
    feature_sample_layers: list&#91;int&#93; | None = None,
) -&gt; torch.Tensor:
    hidden_states = self.embeddings(pixel_values)

    return_all_hidden_states = feature_sample_layers is not None
    encoder_outputs = self.encoder(
        inputs_embeds=hidden_states,
        return_all_hidden_states=return_all_hidden_states,
    )

    if not return_all_hidden_states:
        encoder_outputs = encoder_outputs&#91;0&#93;

    # Apply post-layernorm
    if self.post_layernorm is not None:
        if isinstance(encoder_outputs, list):
            encoder_outputs&#91;-1&#93; = self.post_layernorm(encoder_outputs&#91;-1&#93;)
        else:
            encoder_outputs = self.post_layernorm(encoder_outputs)

    return encoder_outputs</code></pre></details>
fastvideo.models.encoders.siglip.SiglipVisionTransformer.forward
forward(pixel_values: torch.Tensor, feature_sample_layers: list[int] | None = None) -> torch.Tensor
Source
def forward(
    self,
    pixel_values: torch.Tensor,
    feature_sample_layers: list[int] | None = None,
) -> torch.Tensor:
    hidden_states = self.embeddings(pixel_values)
return_all_hidden_states = feature_sample_layers is not None
encoder_outputs = self.encoder(
    inputs_embeds=hidden_states,
    return_all_hidden_states=return_all_hidden_states,
)

if not return_all_hidden_states:
    encoder_outputs = encoder_outputs&#91;0&#93;

# Apply post-layernorm
if self.post_layernorm is not None:
    if isinstance(encoder_outputs, list):
        encoder_outputs&#91;-1&#93; = self.post_layernorm(encoder_outputs&#91;-1&#93;)
    else:
        encoder_outputs = self.post_layernorm(encoder_outputs)

return encoder_outputs</code></pre></details>
fastvideo.models.encoders.siglip.SiglipVisionModel
SiglipVisionModel(config: SiglipVisionConfig)
SigLIP Vision Model for FastVideo.
Source
class SiglipVisionModel(ImageEncoder):
    """SigLIP Vision Model for FastVideo."""
config_class = SiglipVisionConfig
main_input_name = &quot;pixel_values&quot;
packed_modules_mapping = &#123;&quot;qkv_proj&quot;: &#91;&quot;q_proj&quot;, &quot;k_proj&quot;, &quot;v_proj&quot;&#93;&#125;

def __init__(self, config: SiglipVisionConfig) -&gt; None:
    super().__init__(config)
    self.config = config
    self.vision_model = SiglipVisionTransformer(
        config=config,
        quant_config=config.quant_config,
        num_hidden_layers_override=config.num_hidden_layers_override,
        require_post_norm=config.require_post_norm,
        prefix=f&quot;&#123;config.prefix&#125;.vision_model&quot;,
    )

def forward(
    self,
    pixel_values: torch.Tensor,
    feature_sample_layers: list&#91;int&#93; | None = None,
    **kwargs,
) -&gt; BaseEncoderOutput:
    last_hidden_state = self.vision_model(pixel_values, feature_sample_layers)
    return BaseEncoderOutput(last_hidden_state=last_hidden_state)

@property
def device(self):
    return next(self.parameters()).device

def load_weights(self, weights: Iterable&#91;tuple&#91;str, torch.Tensor&#93;&#93;) -&gt; set&#91;str&#93;:
    params_dict = dict(self.named_parameters())
    loaded_params: set&#91;str&#93; = set()
    layer_count = len(self.vision_model.encoder.layers)

    for name, loaded_weight in weights:
        # Skip projection layers if any
        if name.startswith(&quot;visual_projection&quot;):
            continue

        # Skip head if any
        if &quot;head&quot; in name:
            continue

        # Post layernorm handling
        if (name.startswith(&quot;vision_model.post_layernorm&quot;) and self.vision_model.post_layernorm is None):
            continue

        # Omit layers when num_hidden_layers_override is set
        if name.startswith(&quot;vision_model.encoder.layers&quot;):
            layer_idx = int(name.split(&quot;.&quot;)&#91;3&#93;)
            if layer_idx &gt;= layer_count:
                continue

        # Handle QKV projection weight mapping
        for (param_name, weight_name, shard_id) in self.config.arch_config.stacked_params_mapping:
            if weight_name not in name:
                continue
            name = name.replace(weight_name, param_name)

            if name in params_dict:
                param = params_dict&#91;name&#93;
                weight_loader = param.weight_loader
                weight_loader(param, loaded_weight, shard_id)
            break
        else:
            if name in params_dict:
                param = params_dict&#91;name&#93;
                weight_loader = getattr(param, &quot;weight_loader&quot;, default_weight_loader)
                weight_loader(param, loaded_weight)
        loaded_params.add(name)

    return loaded_params</code></pre></details>
fastvideo.models.encoders.siglip.SiglipVisionModel.forward
forward(pixel_values: torch.Tensor, feature_sample_layers: list[int] | None = None, **kwargs) -> BaseEncoderOutput
Source
def forward(
    self,
    pixel_values: torch.Tensor,
    feature_sample_layers: list[int] | None = None,
    **kwargs,
) -> BaseEncoderOutput:
    last_hidden_state = self.vision_model(pixel_values, feature_sample_layers)
    return BaseEncoderOutput(last_hidden_state=last_hidden_state)
fastvideo.models.encoders.siglip.SiglipVisionModel.load_weights
load_weights(weights: Iterable[tuple[str, torch.Tensor]]) -> set[str]
Source
def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]) -> set[str]:
    params_dict = dict(self.named_parameters())
    loaded_params: set[str] = set()
    layer_count = len(self.vision_model.encoder.layers)
for name, loaded_weight in weights:
    # Skip projection layers if any
    if name.startswith(&quot;visual_projection&quot;):
        continue

    # Skip head if any
    if &quot;head&quot; in name:
        continue

    # Post layernorm handling
    if (name.startswith(&quot;vision_model.post_layernorm&quot;) and self.vision_model.post_layernorm is None):
        continue

    # Omit layers when num_hidden_layers_override is set
    if name.startswith(&quot;vision_model.encoder.layers&quot;):
        layer_idx = int(name.split(&quot;.&quot;)&#91;3&#93;)
        if layer_idx &gt;= layer_count:
            continue

    # Handle QKV projection weight mapping
    for (param_name, weight_name, shard_id) in self.config.arch_config.stacked_params_mapping:
        if weight_name not in name:
            continue
        name = name.replace(weight_name, param_name)

        if name in params_dict:
            param = params_dict&#91;name&#93;
            weight_loader = param.weight_loader
            weight_loader(param, loaded_weight, shard_id)
        break
    else:
        if name in params_dict:
            param = params_dict&#91;name&#93;
            weight_loader = getattr(param, &quot;weight_loader&quot;, default_weight_loader)
            weight_loader(param, loaded_weight)
    loaded_params.add(name)

return loaded_params</code></pre></details>