[AMD] Update VitisAI Recipes for Quark 0.11 and LLM Fusion Model Support (#265)

Co-authored-by: ziyuanguo1998 <Siryuanshao@gmail.com>
Co-authored-by: ziyuanguo <ziyuanguo@microsoft.com>
Co-authored-by: xieofxie <xieofxie@126.com>

Author: 4 people

Qwen-Qwen1.5-7B-Chat/VitisAI/Qwen1.5-7B-Chat_quark_vitisai_llm.json

@@ -3,38 +3,21 @@
     "passes": {
         "qq": {
             "type": "QuarkQuantization",
-            "quant_scheme": "w_uint4_per_group_asym",
+            "quant_scheme": "uint4_wo_128",
             "quant_algo": "awq",
             "dataset": "pileval_for_awq_benchmark",
             "data_type": "bfloat16",
             "num_calib_data": 128,
-            "model_export": [ "hf_format" ],
-            "exclude_layers": [  ],
-            "quant_config": {
-                "name": "awq",
-                "scaling_layers": [
-                    {
-                        "prev_op": "input_layernorm",
-                        "layers": [ "self_attn.q_proj", "self_attn.k_proj", "self_attn.v_proj" ],
-                        "inp": "self_attn.q_proj",
-                        "module2inspect": "self_attn"
-                    },
-                    { "prev_op": "self_attn.v_proj", "layers": [ "self_attn.o_proj" ], "inp": "self_attn.o_proj" },
-                    {
-                        "prev_op": "post_attention_layernorm",
-                        "layers": [ "mlp.gate_proj", "mlp.up_proj" ],
-                        "inp": "mlp.gate_proj",
-                        "module2inspect": "mlp"
-                    },
-                    { "prev_op": "mlp.up_proj", "layers": [ "mlp.down_proj" ], "inp": "mlp.down_proj" }
-                ],
-                "model_decoder_layers": "model.layers"
-            }
+            "model_export": ["hf_format"],
+            "exclude_layers": []
         },
-        "mg": { "type": "VitisGenerateModelLLM", "packed_const": false, "cpu_only": false }
+        "mg": {
+            "type": "VitisGenerateModelLLM",
+            "recipe": "full_fusion"
+        }
     },
     "log_severity_level": 1,
     "output_dir": "models/Qwen1.5-7B-Chat-vai",
-    "cache_dir": "cache",
+    "cache_dir": "olive_cache",
     "no_artifacts": true
 }

Qwen-Qwen1.5-7B-Chat/VitisAI/README.md

@@ -14,21 +14,31 @@ This folder contains sample Olive configuration to optimize Qwen models for AMD
 For LLMs - follow the below commands to generate the optimized model for VitisAI Execution Provider.
 
 **Platform Support:**
-- ✅ **Linux with ROCm** - Supported
-- ✅ **Linux with CUDA** - Supported
+- ✅ **Linux with ROCm** - Supported on Ubuntu 24.04 and later. For the best experience, we recommend Ubuntu 24.04.
+- ✅ **Linux with CUDA** - Supported on Ubuntu 24.04 and later. For the best experience, we recommend Ubuntu 24.04.
 - ✅ **Windows with CUDA** - Supported
 - ✅ **Windows with CPU** - Supported (quantization will be slower)
 - ⏳ **Windows with ROCm** - Planned for future release
 
 For more details about quark, see the [Quark Documentation](https://quark.docs.amd.com/latest/)
 
-#### **Create a Python 3.10 conda environment and run the below commands**
+#### **Create a Python 3.12 conda environment and run the below commands**
 ```bash
-conda create -n olive python=3.10
+conda create -n olive python=3.12
 conda activate olive
 ```
 
+#### **Install Olive**
+
+**Option 1: Install from PyPI**
 ```bash
+pip install olive-ai[auto-opt]
+pip install transformers onnxruntime-genai
+```
+
+**Option 2: Install from source**
+```bash
+git clone https://github.com/microsoft/Olive.git
 cd Olive
 pip install -e .
 pip install -r requirements.txt
@@ -41,7 +51,7 @@ cd olive-recipes/Qwen-Qwen1.5-7B-Chat/VitisAI
 pip install --force-reinstall -r requirements_vitisai_llm.txt
 ```
 
-**Note:** The requirements file automatically installs the correct `model-generate` version for your platform (1.5.0 for Linux, 1.5.1 for Windows).
+
 
 #### **Install PyTorch**
 
@@ -60,6 +70,13 @@ pip install torch==2.7.1 torchvision==0.22.1 torchaudio==2.7.1 --index-url https
 
 python -c "import torch; print(torch.cuda.is_available())" # Must return `True`
 ```
+
+**For CPU-only (Windows):**
+```bash
+pip install torch==2.7.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu
+python -c "import torch; print(torch.__version__)"  # Should print 2.7.0+cpu
+```
+
 #### **Generate optimized LLM model for VitisAI NPU**
 Follow the above setup instructions, then run the below command to generate the optimized LLM model for VitisAI EP
 

Qwen-Qwen1.5-7B-Chat/VitisAI/requirements_vitisai_llm.txt

@@ -1,26 +1,27 @@
+--extra-index-url=https://pypi.amd.com/olive/1.7.1/simple
 # AMD model generation
 --extra-index-url=https://pypi.amd.com/simple
 accelerate
 
 # Quark
-amd-quark==0.9
+amd-quark==0.11
 datasets
 evaluate
 
-# Platform-specific model-generate versions:
-# Linux: use model-generate==1.5.0 (default)
-# Windows: MUST use model-generate==1.5.1
-model-generate==1.5.0; sys_platform != 'win32'
-model-generate==1.5.1; sys_platform == 'win32'
+model-generate
 
 nltk
-numpy
+numpy==1.26.4
 
 # Pin onnx version
 onnx==1.18.0
-onnxruntime==1.21.1
-onnxruntime-genai==0.7.1
+onnxruntime
+onnxruntime-genai
+onnxsim
 optimum
+
+ryzenai-dynamic-dispatch
+ryzenai-onnx-utils
 sentencepiece
 tabulate
-transformers==4.50.0
+transformers==4.57.6