[Docs] Add Qwen2.5-VL guide for the TRT-LLM PyTorch backend, deprecate Llava1.5 guide

Popular_Models_Guide/Llava1.5/llava_trtllm_guide.md

@@ -28,6 +28,13 @@
 
 # Deploying Hugging Face Llava1.5-7b Model in Triton
 
+> [!WARNING]
+> **Deprecated.** This guide describes the prebuilt-TensorRT-engine multimodal
+> path (`tensorrtllm_backend`'s `all_models/multimodal`), which TensorRT-LLM has
+> declared end-of-life as of TensorRT-LLM v1.2. It is no longer maintained.
+> Use the TensorRT-LLM PyTorch backend instead; see
+> [Deploying Hugging Face Qwen2.5-VL Model in Triton](../Qwen2.5-VL/qwen2_5_vl_trtllm_guide.md).
+
 TensorRT-LLM is Nvidia's recommended solution of running Large Language
 Models(LLMs) on Nvidia GPUs. Read more about TensoRT-LLM [here](https://github.com/NVIDIA/TensorRT-LLM)
 and Triton's TensorRT-LLM Backend [here](https://github.com/triton-inference-server/tensorrtllm_backend).

Popular_Models_Guide/Qwen2.5-VL/config.pbtxt

@@ -0,0 +1,243 @@
+# Copyright 2025, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions
+# are met:
+#  * Redistributions of source code must retain the above copyright
+#    notice, this list of conditions and the following disclaimer.
+#  * Redistributions in binary form must reproduce the above copyright
+#    notice, this list of conditions and the following disclaimer in the
+#    documentation and/or other materials provided with the distribution.
+#  * Neither the name of NVIDIA CORPORATION nor the names of its
+#    contributors may be used to endorse or promote products derived
+#    from this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
+# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+name: "tensorrt_llm"
+backend: "python"
+
+#######################################################
+# The below config arguments are specific in model.yaml
+# Please don't add it in config.pbtxt
+#
+# max_batch_size: 64
+#
+# model_transaction_policy {
+#  decoupled: False
+# }
+#######################################################
+
+instance_group [
+  {
+    count: 1
+    kind : KIND_CPU
+  }
+]
+
+input [
+  {
+    name: "text_input"
+    data_type: TYPE_STRING
+    dims: [ -1 ]
+  },
+  {
+    name: "image_url"
+    data_type: TYPE_STRING
+    dims: [ -1 ]
+    optional: true
+  },
+  {
+    name: "streaming"
+    data_type: TYPE_BOOL
+    dims: [ 1 ]
+    optional: true
+  },
+  ## SamplingParams Arguments for Each Request ##
+  {
+    name: "sampling_param_best_of"
+    data_type: TYPE_INT32
+    dims: [ 1 ]
+    optional: true
+  },
+  {
+    name: "sampling_param_temperature"
+    data_type: TYPE_FP32
+    dims: [ 1 ]
+    optional: true
+  },
+  {
+    name: "sampling_param_top_k"
+    data_type: TYPE_INT32
+    dims: [ 1 ]
+    optional: true
+  },
+  {
+    name: "sampling_param_top_p"
+    data_type: TYPE_FP32
+    dims: [ 1 ]
+    optional: true
+  },
+  {
+    name: "sampling_param_frequency_penalty"
+    data_type: TYPE_FP32
+    dims: [ 1 ]
+    optional: true
+  },
+  {
+    name: "sampling_param_presence_penalty"
+    data_type: TYPE_FP32
+    dims: [ 1 ]
+    optional: true
+  },
+  {
+    name: "sampling_param_max_tokens"
+    data_type: TYPE_INT32
+    dims: [ 1 ]
+    optional: true
+  },
+  {
+    name: "sampling_param_stop"
+    data_type: TYPE_STRING
+    dims: [-1]
+    optional: true
+  },
+  {
+    name: "sampling_param_seed"
+    data_type: TYPE_UINT64
+    dims: [ 1 ]
+    optional: true
+  },
+  {
+    name: "sampling_param_exclude_input_from_output"
+    data_type: TYPE_BOOL
+    dims: [ 1 ]
+    optional: true
+  },
+  {
+    name: "sampling_param_return_perf_metrics"
+    data_type: TYPE_BOOL
+    dims: [ 1 ]
+    optional: true
+  },
+  ## Arguments for Controlling Response Output Fields ##
+  {
+    name: "return_finish_reason"
+    data_type: TYPE_BOOL
+    dims: [1]
+    optional: true
+  },
+  {
+    name: "return_stop_reason"
+    data_type: TYPE_BOOL
+    dims: [1]
+    optional: true
+  },
+  {
+    name: "return_cumulative_logprob"
+    data_type: TYPE_BOOL
+    dims: [1]
+    optional: true
+  },
+  {
+    name: "stop"
+    data_type: TYPE_BOOL
+    dims: [ 1 ]
+    optional: true
+  }
+]
+###################################################################
+# The below output parameters are arguments from LLM.RequestOutput
+###################################################################
+output [
+  {
+    name: "text_output"
+    data_type: TYPE_STRING
+    dims: [-1]
+  },
+  {
+    name: "finish_reason"
+    data_type: TYPE_STRING
+    dims: [-1]
+  },
+  {
+    name: "stop_reason"
+    data_type: TYPE_STRING
+    dims: [-1]
+  },
+  {
+    name: "cumulative_logprob"
+    data_type: TYPE_FP32
+    dims: [-1]
+  },
+  {
+    name: "kv_cache_reused_block"
+    data_type: TYPE_INT32
+    dims: [-1]
+  },
+  {
+    name: "kv_cache_missed_block"
+    data_type: TYPE_INT32
+    dims: [-1]
+  },
+  {
+    name: "kv_cache_alloc_new_blocks"
+    data_type: TYPE_INT32
+    dims: [-1]
+  },
+  {
+    name: "kv_cache_alloc_total_blocks"
+    data_type: TYPE_INT32
+    dims: [-1]
+  },
+  {
+    name: "kv_cache_hit_rate"
+    data_type: TYPE_FP32
+    dims: [-1]
+  },
+  {
+    name: "arrival_time_ns"
+    data_type: TYPE_INT64
+    dims: [ 1 ]
+  },
+  {
+    name: "first_scheduled_time_ns"
+    data_type: TYPE_INT64
+    dims: [ 1 ]
+  },
+  {
+    name: "first_token_time_ns"
+    data_type: TYPE_INT64
+    dims: [ 1 ]
+  },
+  {
+    name: "last_token_time_ns"
+    data_type: TYPE_INT64
+    dims: [ 1 ]
+  },
+  {
+    name: "acceptance_rate"
+    data_type: TYPE_FP32
+    dims: [ 1 ]
+  },
+  {
+    name: "total_accepted_draft_tokens"
+    data_type: TYPE_INT32
+    dims: [ 1 ]
+  },
+  {
+    name: "total_draft_tokens"
+    data_type: TYPE_INT32
+    dims: [ 1 ]
+  }
+]