Rise • AI & Physical World Ramps = Fast + Data-Driven • Global Internet User Ramps Powered by AI from Get-Go = Growth We Have Not Seen Likes of Before • AI & Work Evolution = Real + Rapid 3 1 2 3 Francisco Operating Zone Market Share Source: YipitData (4/25) Global Internet User Ramps Powered by AI from Get-Go = Growth We Have Not Seen Likes of Before 7 Leading USA-Based LLM App Users by Region Note: us back to one of our favorite quotes – Statistically speaking, the world doesn’t end that often, from former T. Rowe Price Chairman and CEO Brian Rogers. As investors, we always assume everything can

for product and engineering teams exploring how to build their first agents, distilling insights from numerous customer deployments into practical and actionable best practices. It includes frameworks on unstructured data: Scenarios that involve interpreting natural language, 
 extracting meaning from documents, or interacting with 
 users conversationally, for example processing a home insurance SDK. You can also implement the same concepts using your preferred library or building directly from scratch. Python 1 2 3 4 5 6 weather_agent = Agent( name= instructions= tools=[get_weather]

rights reserved. How to Support Framework Quantized Operators? Option 1 – Completely add new ops from scratch • New Relay passes and TVM schedules required • AlterOpLayout, Graph Fusion etc require work/operator Services, Inc. or its Affiliates. All rights reserved. QNN Conv2D Operator • Calculations are different from FP32 Conv2D https://discuss.tvm.ai/t/tf-lite-quantized-conv2d-operator-conversion/2651/8 𝑟𝑒𝑎𝑙_𝑣𝑎𝑙𝑢𝑒

PART FOUR Alios TVM @ Intel GPU AiOS 1驱动万物智能 Alios TVM @ Intel GPU 。 Implement the schedule from scratch Subgroups 。 Leverage Intel Subgroup Extension NDRange

communication balance loss is employed to encour- age each device to receive around ?? hidden states from other devices. The communication balance loss guarantees a balanced exchange of information among data, we also focus on the data quality. We enrich our pre-training corpus with high-quality data from various sources, and meanwhile improve the quality-based filtering algorithm. The improved algorithm mostly retained. In addition, we filter out the contentious content from our pre-training corpus to mitigate the data bias introduced from specific regional cultures. A detailed discussion about the influence

Contiguous Memory Allocation – Linux Kernel DEPLOY VTA ON INTEL FPGA https://pynq.readthedocs.io/en/v2.0/pynq_package/pynq.xlnk.html Configure Linux kernel Download Linux kernel from https://github.com/al com/altera-opensource/linux-socfpga/archive/rel_socfpga-4.9.76-ltsi-rt_18.08.02_pr.tar.gz©2019 HARMAN INTERNATIONAL INDUSTRIES, INCORPORATED 6 Software - CMA Contiguous Memory Allocation – Linux Kernel Module install Quartus Prime 18.1 Lite Edition Step 2: Download SDCard Image from Terasic (Require Registration) Step 3: Get files from https://github.com/liangfu/de10-nano-supplement Step 4: Extract the files

best-in-class Deep Learning performance by leveraging Neural Network acceleration in IP and SoCs from the Arm ecosystem, through collaborative seamless integration with the ecosystem of AI/ML software p20/p20pro (kirin 970) -target=arm64-linux-android -mattr=+neon llvm firefly rk3399, rock960, ultra96 -target=aarch64-linux-gnu -mattr=+neon rasp3b (bcm2837) -target=armv7l-linux-gnueabihf -mattr=+neon pynq pynq -target=armv7a-linux-eabi -mattr=+neon GPU mali (midgard) firefly rk3399, rock960 (mali t860) N/A opencl bifrost hikey960 (mali g71) N/A FPGA vta pynq, ultra96 N/A sdaccel Out-of-tree support or

log file from the autotvm on Ubuntu 2. Use the .log from step1 on Windows to generate the .dll for deployment 3. For application on 32bits, no support of 32bit tensorflow , a workround from FrozenGene python tensorflow_blur.py to get the .log c. Use the .log, with target=“llvm –mcpu=i686 –mtriple=i686-linux-gnu” then TVM_NDK_CC=“clang –m32” python tf_blur.py�����������������������������������- DWORD WINAPI

caffe 前端有利于提高竞争力。 开源社区 存量的开源 caffe 网络模型众多，TVM 直接支持 caffe 让大家更方便尝试 caffe 资源。绝赞招聘中 当前进度 无 caffe 依赖 from_caffe 直接导入 caffe 模型文件，不需要预先安装 caffe 。 net 已测试网络：alexnet / densenet121 / inception v1 / inception / roipooling / permute / priorbox绝赞招聘中 未来 命令行工具 将 caffe 模型转换的功能，通过一组命令行工具提供，命令行工具支持 windows / linux 平台。 支持更多 caffe op / net 随着客户需求和社区发展，提供更多的 caffe 分支变种的 op / net 支持。绝赞招聘中 THANKS

消费级设备 Mac Studio（192GB统⼀内存） 10+ token/秒 ⾼性能服务器 4×RTX 4090（96GB显存+384GB内存） 7-8 token/秒（混合推理） 3. 部署步骤（Linux示例） 1. 安装依赖⼯具： # 安装llama.cpp（⽤于合并分⽚⽂件） /bin/bash -c "$(curl -fsSL https://raw.githubusercontent DeepSeek-R1-UD-IQ1_M-00001-of-00004.gguf DeepSeek-R1-UD-IQ1_S.gguf curl -fsSL https://ollama.com/install.sh | sh FROM /path/to/DeepSeek-R1-UD-IQ1_M.gguf PARAMETER num_gpu 28 # 每块RTX 4090加载7层（共4卡） PARAMETER num_ctx