sglang 0.4.4.post3__py3-none-any.whl → 0.4.5__py3-none-any.whl

sglang/srt/models/llama4.py

@@ -0,0 +1,420 @@
+# Copyright 2023-2024 SGLang Team
+# 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
+#
+#     http://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.
+# ==============================================================================
+
+# Adapted from
+# https://github.com/vllm-project/vllm/blob/v0.8.3/vllm/model_executor/models/llama4.py
+"""Inference-only LLaMA model compatible with HuggingFace weights."""
+
+import logging
+from typing import Any, Dict, List, Optional, Tuple, Union
+
+import torch
+from torch import nn
+from transformers import Llama4TextConfig
+
+from sglang.srt.distributed import (
+    get_tensor_model_parallel_world_size,
+    tensor_model_parallel_all_reduce,
+)
+from sglang.srt.layers.layernorm import RMSNorm
+from sglang.srt.layers.linear import (
+    QKVParallelLinear,
+    ReplicatedLinear,
+    RowParallelLinear,
+)
+from sglang.srt.layers.moe.fused_moe_triton import FusedMoE
+from sglang.srt.layers.quantization.base_config import QuantizationConfig
+from sglang.srt.layers.radix_attention import RadixAttention
+from sglang.srt.layers.rotary_embedding import get_rope
+from sglang.srt.layers.vocab_parallel_embedding import VocabParallelEmbedding
+from sglang.srt.model_executor.forward_batch_info import ForwardBatch
+from sglang.srt.models.llama import LlamaForCausalLM, LlamaMLP
+from sglang.srt.utils import add_prefix, get_compiler_backend, make_layers
+
+logger = logging.getLogger(__name__)
+
+
+class Llama4MoE(nn.Module):
+
+    @torch.compile(dynamic=True, backend=get_compiler_backend())
+    @staticmethod
+    def custom_routing_function(
+        hidden_states: torch.Tensor,
+        gating_output: torch.Tensor,
+        topk: int,
+        renormalize: bool,
+    ) -> Tuple[torch.Tensor, torch.Tensor]:
+        router_scores_aK, router_indices_aK = torch.topk(gating_output, topk, dim=-1)
+        router_scores_aK = torch.sigmoid(router_scores_aK.float()).to(
+            hidden_states.dtype
+        )
+        return (
+            router_scores_aK.view(-1).reshape(router_scores_aK.shape),
+            router_indices_aK.to(torch.int32),
+        )
+
+    def __init__(
+        self,
+        config: Llama4TextConfig,
+        quant_config: Optional[QuantizationConfig] = None,
+        prefix: str = "",
+    ):
+        super().__init__()
+        self.tp_size = get_tensor_model_parallel_world_size()
+        self.top_k = config.num_experts_per_tok
+
+        intermediate_size_moe = config.intermediate_size
+        self.router = ReplicatedLinear(
+            config.hidden_size,
+            config.num_local_experts,
+            bias=False,
+            quant_config=None,
+            prefix=add_prefix("router", prefix),
+        )
+
+        self.experts = FusedMoE(
+            num_experts=config.num_local_experts,
+            top_k=config.num_experts_per_tok,
+            hidden_size=config.hidden_size,
+            custom_routing_function=Llama4MoE.custom_routing_function,
+            intermediate_size=intermediate_size_moe,
+            reduce_results=False,
+            renormalize=False,
+            quant_config=quant_config,
+            apply_router_weight_on_input=True,
+            prefix=add_prefix("experts", prefix),
+        )
+
+        self.shared_expert = LlamaMLP(
+            hidden_size=config.hidden_size,
+            intermediate_size=intermediate_size_moe,
+            hidden_act="silu",
+            quant_config=quant_config,
+            prefix=add_prefix("shared_expert", prefix),
+            reduce_results=False,  # We need to do scatter before reduce
+        )
+
+    def forward(self, hidden_states):
+        # router_scores: [num_tokens, num_experts]
+        router_logits, _ = self.router(hidden_states)
+        shared_out = self.shared_expert(hidden_states)
+        routed_out = self.experts(
+            hidden_states=hidden_states,
+            router_logits=router_logits,
+        )
+        out_aD = routed_out + shared_out
+
+        if self.tp_size > 1:
+            out_aD = tensor_model_parallel_all_reduce(out_aD)
+
+        return out_aD
+
+
+class Llama4Attention(nn.Module):
+
+    def __init__(
+        self,
+        config: Llama4TextConfig,
+        layer_id: int,
+        hidden_size: int,
+        num_heads: int,
+        num_kv_heads: int,
+        rope_theta: float = 10000,
+        rope_scaling: Optional[Dict[str, Any]] = None,
+        max_position_embeddings: int = 8192,
+        quant_config: Optional[QuantizationConfig] = None,
+        bias: bool = False,
+        bias_o_proj: bool = False,
+        prefix: str = "",
+    ) -> None:
+        super().__init__()
+        self.layer_id = layer_id
+        self.hidden_size = hidden_size
+        self.use_rope = int((layer_id + 1) % 4 != 0)
+        self.use_qk_norm = config.use_qk_norm and self.use_rope
+        tp_size = get_tensor_model_parallel_world_size()
+        self.total_num_heads = num_heads
+        assert self.total_num_heads % tp_size == 0
+        self.num_heads = self.total_num_heads // tp_size
+        self.total_num_kv_heads = num_kv_heads
+        if self.total_num_kv_heads >= tp_size:
+            # Number of KV heads is greater than TP size, so we partition
+            # the KV heads across multiple tensor parallel GPUs.
+            assert self.total_num_kv_heads % tp_size == 0
+        else:
+            # Number of KV heads is less than TP size, so we replicate
+            # the KV heads across multiple tensor parallel GPUs.
+            assert tp_size % self.total_num_kv_heads == 0
+        self.num_kv_heads = max(1, self.total_num_kv_heads // tp_size)
+        self.head_dim = config.head_dim
+        self.q_size = self.num_heads * self.head_dim
+        self.kv_size = self.num_kv_heads * self.head_dim
+        self.scaling = self.head_dim**-0.5
+        self.attn_temperature_tuning = config.attn_temperature_tuning
+        self.floor_scale = config.floor_scale
+        self.attn_scale = config.attn_scale
+        self.rope_theta = rope_theta
+        self.max_position_embeddings = max_position_embeddings
+        self.n_rep = self.num_heads // self.num_kv_heads
+        self.qk_norm = (
+            RMSNorm(
+                hidden_size=self.head_dim,
+                eps=config.rms_norm_eps,
+            )
+            if self.use_qk_norm
+            else None
+        )
+        self.qkv_proj = QKVParallelLinear(
+            hidden_size=hidden_size,
+            head_size=self.head_dim,
+            total_num_heads=self.total_num_heads,
+            total_num_kv_heads=self.total_num_kv_heads,
+            bias=bias,
+            quant_config=quant_config,
+            prefix=add_prefix("qkv_proj", prefix),
+        )
+
+        self.o_proj = RowParallelLinear(
+            input_size=self.total_num_heads * self.head_dim,
+            output_size=hidden_size,
+            bias=bias_o_proj,
+            quant_config=quant_config,
+            prefix=add_prefix("o_proj", prefix),
+        )
+        is_neox_style = True
+        is_gguf = quant_config and quant_config.get_name() == "gguf"
+        if is_gguf and config.model_type in ["llama", "llama4"]:
+            is_neox_style = False
+
+        self.rotary_emb = (
+            get_rope(
+                self.head_dim,
+                rotary_dim=self.head_dim,
+                max_position=max_position_embeddings,
+                base=int(rope_theta),
+                rope_scaling=rope_scaling if rope_scaling != "default" else None,
+                is_neox_style=is_neox_style,
+            )
+            if self.use_rope
+            else None
+        )
+
+        self.attn = RadixAttention(
+            self.num_heads,
+            self.head_dim,
+            self.scaling,
+            num_kv_heads=self.num_kv_heads,
+            layer_id=layer_id,
+            prefix=add_prefix("attn", prefix),
+            use_irope=self.use_rope,
+        )
+
+    def _get_attn_scale(self, positions: torch.Tensor) -> torch.Tensor:
+        floor = torch.floor((positions + 1.0) / self.floor_scale)
+        attn_scale = torch.log(floor + 1.0) * self.attn_scale + 1.0
+
+        return attn_scale.unsqueeze(-1)
+
+    def forward(
+        self,
+        positions: torch.Tensor,
+        hidden_states: torch.Tensor,
+        forward_batch: ForwardBatch,
+    ) -> torch.Tensor:
+        qkv, _ = self.qkv_proj(hidden_states)
+        q, k, v = qkv.split([self.q_size, self.kv_size, self.kv_size], dim=-1)
+
+        if self.rotary_emb is not None:
+            q, k = self.rotary_emb(positions, q, k)
+
+        if self.qk_norm is not None:
+            # TODO: support float
+            q = q.reshape(-1, self.head_dim).contiguous().bfloat16()
+            k = k.reshape(-1, self.head_dim).contiguous().bfloat16()
+            q = self.qk_norm(q).to(q.dtype)
+            k = self.qk_norm(k).to(k.dtype)
+            q = q.reshape(-1, self.q_size)
+            k = k.reshape(-1, self.kv_size)
+
+        # We are applying temperature tuning (https://arxiv.org/abs/2501.19399) to NoPE layers, where
+        # the inference-time temperature tuning function is customized to not affect short context
+        # while working at very long context
+        # https://arxiv.org/abs/2501.19399
+        if self.attn_temperature_tuning and not self.use_rope:
+            attn_scale = self._get_attn_scale(positions)
+            q = (q * attn_scale).to(q.dtype)
+
+        attn_output = self.attn(q, k, v, forward_batch)
+        output, _ = self.o_proj(attn_output)
+        return output
+
+
+class Llama4DecoderLayer(nn.Module):
+    def __init__(
+        self,
+        config: Llama4TextConfig,
+        layer_id: int = 0,
+        quant_config: Optional[QuantizationConfig] = None,
+        prefix: str = "",
+    ):
+        super().__init__()
+        self.layer_id = layer_id
+        self.hidden_size = config.hidden_size
+        rope_theta = config.rope_theta
+        rope_scaling = config.rope_scaling
+        max_position_embeddings = config.max_position_embeddings
+
+        self.self_attn = Llama4Attention(
+            config=config,
+            layer_id=layer_id,
+            hidden_size=self.hidden_size,
+            num_heads=config.num_attention_heads,
+            num_kv_heads=config.num_key_value_heads,
+            rope_theta=rope_theta,
+            rope_scaling=rope_scaling,
+            max_position_embeddings=max_position_embeddings,
+            quant_config=quant_config,
+            bias=False,
+            bias_o_proj=False,
+            prefix=add_prefix("self_attn", prefix),
+        )
+        is_moe_layer = (layer_id + 1) % config.interleave_moe_layer_step == 0
+        if is_moe_layer:
+            self.feed_forward = Llama4MoE(
+                config=config,
+                quant_config=quant_config,
+                prefix=add_prefix("feed_forward", prefix),
+            )
+        else:
+            self.feed_forward = LlamaMLP(
+                hidden_size=self.hidden_size,
+                intermediate_size=config.intermediate_size_mlp,
+                hidden_act="silu",
+                quant_config=quant_config,
+                prefix=add_prefix("feed_forward", prefix),
+            )
+        self.input_layernorm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+        self.post_attention_layernorm = RMSNorm(
+            config.hidden_size, eps=config.rms_norm_eps
+        )
+
+    def forward(
+        self,
+        positions: torch.Tensor,
+        hidden_states: torch.Tensor,
+        forward_batch: ForwardBatch,
+        residual: Optional[torch.Tensor],
+    ) -> Tuple[torch.Tensor, torch.Tensor]:
+        # Self Attention
+        if residual is None:
+            residual = hidden_states
+            hidden_states = self.input_layernorm(hidden_states)
+        else:
+            hidden_states, residual = self.input_layernorm(hidden_states, residual)
+        hidden_states = self.self_attn(
+            positions=positions,
+            hidden_states=hidden_states,
+            forward_batch=forward_batch,
+        )
+
+        # Fully Connected
+        hidden_states, residual = self.post_attention_layernorm(hidden_states, residual)
+        hidden_states = self.feed_forward(hidden_states)
+        return hidden_states, residual
+
+
+class Llama4Model(nn.Module):
+    def __init__(
+        self,
+        config: Llama4TextConfig,
+        quant_config: Optional[QuantizationConfig] = None,
+        prefix: str = "",
+    ) -> None:
+        super().__init__()
+        self.config = config
+        self.padding_idx = config.pad_token_id
+        self.vocab_size = config.vocab_size
+        self.embed_tokens = VocabParallelEmbedding(
+            config.vocab_size,
+            config.hidden_size,
+            quant_config=quant_config,
+            prefix=add_prefix("embed_tokens", prefix),
+        )
+        self.layers = make_layers(
+            config.num_hidden_layers,
+            lambda idx, prefix: Llama4DecoderLayer(
+                config=config, layer_id=idx, quant_config=quant_config, prefix=prefix
+            ),
+            prefix="model.layers",
+        )
+
+        self.norm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+        self.layers_to_capture = []
+
+    def forward(
+        self,
+        input_ids: torch.Tensor,
+        positions: torch.Tensor,
+        forward_batch: ForwardBatch,
+        input_embeds: torch.Tensor = None,
+    ) -> Union[torch.Tensor, Tuple[torch.Tensor, List[torch.Tensor]]]:
+        if input_embeds is None:
+            hidden_states = self.embed_tokens(input_ids)
+        else:
+            hidden_states = input_embeds
+        residual = None
+        aux_hidden_states = []
+        for i in range(len(self.layers)):
+            if i in self.layers_to_capture:
+                aux_hidden_states.append(hidden_states + residual)
+            layer = self.layers[i]
+            hidden_states, residual = layer(
+                positions,
+                hidden_states,
+                forward_batch,
+                residual,
+            )
+        hidden_states, _ = self.norm(hidden_states, residual)
+
+        if len(aux_hidden_states) == 0:
+            return hidden_states
+
+        return hidden_states, aux_hidden_states
+
+
+class Llama4ForCausalLM(LlamaForCausalLM):
+
+    packed_modules_mapping = {
+        "qkv_proj": ["q_proj", "k_proj", "v_proj"],
+        "gate_up_proj": ["gate_proj", "up_proj"],
+    }
+
+    def __init__(
+        self,
+        config: Llama4TextConfig,
+        quant_config: Optional[QuantizationConfig] = None,
+        prefix: str = "",
+    ):
+        super().__init__(config, quant_config, prefix)
+
+    def _init_model(
+        self,
+        config: Llama4TextConfig,
+        quant_config: Optional[QuantizationConfig] = None,
+        prefix: str = "",
+    ):
+        return Llama4Model(config, quant_config=quant_config, prefix=prefix)
+
+
+EntryClass = [Llama4ForCausalLM]

sglang/srt/models/llava.py

@@ -31,7 +31,7 @@ from transformers import (
 from transformers.models.llava.modeling_llava import LlavaMultiModalProjector
 
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
-from sglang.srt.managers.schedule_batch import MultimodalInputs
+from sglang.srt.managers.schedule_batch import Modality, MultimodalInputs
 from sglang.srt.mm_utils import (
     get_anyres_image_grid_shape,
     unpad_image,
@@ -42,17 +42,21 @@ from sglang.srt.model_loader.weight_utils import default_weight_loader
 from sglang.srt.models.llama import LlamaForCausalLM
 from sglang.srt.models.mistral import MistralForCausalLM
 from sglang.srt.models.qwen2 import Qwen2ForCausalLM
-from sglang.srt.utils import add_prefix
+from sglang.srt.utils import add_prefix, flatten_nested_list
 
 
 class LlavaBaseForCausalLM(nn.Module):
     def pad_input_ids(self, input_ids: List[int], image_inputs: MultimodalInputs):
-        image_sizes, pad_values = image_inputs.image_sizes, image_inputs.pad_values
+        image_sizes = flatten_nested_list(
+            [item.image_sizes for item in image_inputs.mm_items]
+        )
+
+        pad_values = [item.pad_value for item in image_inputs.mm_items]
 
         # hardcode for spatial_unpad + anyres
-        if image_inputs.modalities is not None and (
-            "multi-images" in image_inputs.modalities
-            or "video" in image_inputs.modalities
+        if any(
+            item.modality == Modality.MULTI_IMAGES or item.modality == Modality.VIDEO
+            for item in image_inputs.mm_items
         ):
             image_aspect_ratio = "pad"
         else:
@@ -66,7 +70,7 @@ class LlavaBaseForCausalLM(nn.Module):
                     math.ceil(self.image_size / self.patch_size / 2) ** 2
                 )
             else:
-                new_image_feature_len = self.image_feature_len  # multiimage
+                new_image_feature_len = self.image_feature_len  # multi-image
 
             height = width = self.num_patches_per_side
             if "anyres" in image_aspect_ratio:
@@ -101,7 +105,7 @@ class LlavaBaseForCausalLM(nn.Module):
             # old_len + pad_len - 1, because we need to remove image_token_id
             input_ids = (
                 input_ids[:offset]
-                + [pad_values[image_idx]] * new_image_feature_len
+                + [pad_values[image_idx % len(pad_values)]] * new_image_feature_len
                 + input_ids[offset + 1 :]
             )
             offset_list.append(offset)
@@ -150,8 +154,8 @@ class LlavaBaseForCausalLM(nn.Module):
             modalities_list = []
             max_image_offset = []
             for im in image_inputs:
-                if im and im.modalities is not None:
-                    modalities_list.extend(im.modalities)
+                if im:
+                    modalities_list.extend([item.modality for item in im.mm_items])
                 if im and im.image_offsets:
                     max_image_offset.append(
                         np.max(np.array(im.image_offsets) + np.array(im.image_pad_len))
@@ -164,11 +168,19 @@ class LlavaBaseForCausalLM(nn.Module):
 
             if need_vision.any():
                 bs = forward_batch.batch_size
-                pixel_values = [
-                    image_inputs[i].pixel_values for i in range(bs) if need_vision[i]
-                ]
+                pixel_values = flatten_nested_list(
+                    [
+                        [item.pixel_values for item in image_inputs[i].mm_items]
+                        for i in range(bs)
+                        if need_vision[i]
+                    ]
+                )
                 image_sizes = [
-                    image_inputs[i].image_sizes for i in range(bs) if need_vision[i]
+                    flatten_nested_list(
+                        [item.image_sizes for item in image_inputs[i].mm_items]
+                    )
+                    for i in range(bs)
+                    if need_vision[i]
                 ]
 
                 ########## Encode Image ########
@@ -197,13 +209,13 @@ class LlavaBaseForCausalLM(nn.Module):
                     new_image_features = []
                     height = width = self.num_patches_per_side
                     for image_idx, image_feature in enumerate(image_features):
-                        if modalities_list[image_idx] == "image":
+                        if modalities_list[image_idx] == Modality.IMAGE:
                             image_aspect_ratio = (
                                 self.config.image_aspect_ratio
                             )  # single image
                         elif (
-                            modalities_list[image_idx] == "multi-images"
-                            or modalities_list[image_idx] == "video"
+                            modalities_list[image_idx] == Modality.MULTI_IMAGES
+                            or modalities_list[image_idx] == Modality.VIDEO
                         ):
                             image_aspect_ratio = "pad"  # multi image
                         # image_aspect_ratio = (
@@ -212,7 +224,7 @@ class LlavaBaseForCausalLM(nn.Module):
                         if (
                             image_feature.shape[0] > 1
                             and "anyres" in image_aspect_ratio
-                            and modalities_list[image_idx] == "image"
+                            and modalities_list[image_idx] == Modality.IMAGE
                         ):
                             base_image_feature = image_feature[0]
                             image_feature = image_feature[1:]
@@ -312,7 +324,7 @@ class LlavaBaseForCausalLM(nn.Module):
                             )
                             image_feature = image_feature.unsqueeze(0)
                         else:
-                            if modalities_list[image_idx] == "video":  # video
+                            if modalities_list[image_idx] == Modality.VIDEO:  # video
                                 # 2x2 pooling
                                 num_of_frames = image_feature.shape[0]
                                 image_feature = image_feature.view(

sglang/srt/models/llavavid.py

@@ -22,7 +22,7 @@ from transformers import CLIPVisionModel, LlavaConfig
 from transformers.models.llava.modeling_llava import LlavaMultiModalProjector
 
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
-from sglang.srt.managers.schedule_batch import MultimodalInputs
+from sglang.srt.managers.schedule_batch import MultimodalInputs, flatten_nested_list
 from sglang.srt.model_executor.forward_batch_info import ForwardBatch
 from sglang.srt.model_loader.weight_utils import default_weight_loader
 from sglang.srt.models.llama import LlamaForCausalLM
@@ -58,7 +58,7 @@ class LlavaVidForCausalLM(nn.Module):
             )
 
     def pad_input_ids(self, input_ids: List[int], image_inputs: MultimodalInputs):
-        pad_values = image_inputs.pad_values
+        pad_values = [item.pad_value for item in image_inputs.mm_items]
         new_image_feature_len = self.image_feature_len
 
         pad_ids = pad_values * (
@@ -133,11 +133,19 @@ class LlavaVidForCausalLM(nn.Module):
             need_vision = start_positions <= np.array(max_image_offset)
 
             if need_vision.any():
-                pixel_values = [
-                    image_inputs[i].pixel_values for i in range(bs) if need_vision[i]
-                ]
+                pixel_values = flatten_nested_list(
+                    [
+                        [item.pixel_values for item in image_inputs[i].mm_items]
+                        for i in range(bs)
+                        if need_vision[i]
+                    ]
+                )
                 image_offsets = [
-                    image_inputs[i].image_offsets for i in range(bs) if need_vision[i]
+                    flatten_nested_list(
+                        [item.image_offsets for item in image_inputs[i].mm_items]
+                    )
+                    for i in range(bs)
+                    if need_vision[i]
                 ]
 
                 ########## Encode Image ########
@@ -246,7 +254,8 @@ class LlavaVidForCausalLM(nn.Module):
             "model.mm_projector.2": "multi_modal_projector.linear_2",
             "model.vision_resampler.mm_projector.0": "multi_modal_projector.linear_1",
             "model.vision_resampler.mm_projector.2": "multi_modal_projector.linear_2",
-            "model.vision_tower.vision_tower": "vision_tower",  # Update the vision tower weights if we find them in the checkpoint (it may be finetuned).
+            "model.vision_tower.vision_tower": "vision_tower",
+            # Update the vision tower weights if we find them in the checkpoint (it may be finetuned).
             "model.image_newline": "language_model.model.image_newline",
         }
         params_dict = dict(self.named_parameters())