Trends Artificial Intelligence
OpenAI’s ChatGPT with its extremely easy-to-use / speedy user interface. In addition, relatively new AI company founders have been especially aggressive about innovation / product releases / investments to adapt to this evolving journey as knowledge – and its distribution – get leveled up rapidly in new ways. Special thanks to Grant Watson and Keeyan Sanjasaz and BOND colleagues who helped steer ideas cost-per-token declines, open-source proliferation and chip performance improvements are making new tech advances increasingly more powerful, accessible, and economically viable. OpenAI’s ChatGPT –0 码力 | 340 页 | 12.14 MB | 4 月前3Google 《Prompt Engineering v7》
insightful responses. Step-back prompting encourages LLMs to think critically and apply their knowledge in new and creative ways. It changes the final prompt doing the task by utilizing more knowledge in the LLM’s gemini-pro Temperature 1 Token Limit 1024 Top-K 40 Top-P 0.8 Prompt Write a one paragraph storyline for a new level of a first- person shooter video game that is challenging and engaging. Output The level begins underwater exploration skills to survive. Take one of the themes and write a one paragraph storyline for a new level of a first-person shooter video game that is challenging and engaging. Output In the heart of0 码力 | 68 页 | 6.50 MB | 6 月前3OctoML OSS 2019 11 8
OctoML is a new company building DL deployment solutions using the Apache (incubating) TVM project. A goal is to nurture the TVM community and contribute new infrastructure and features. octom|.ai @octoml folks) o_ Improved NLP support, with focus on transformers QQ octoML Core Infrastructure Refactors ee New Integer Analysis Infrastructure o_ Supports the ability to handle nested division and modulus o_ Improves vm::Object NDArray | Rd | tuplelclosure AST Nodes Cross language suppPort Easy to introduce new runtime objects (trees, graphs) Direct access from other languages QQ octoML HTVM Overview *。 Plug0 码力 | 16 页 | 1.77 MB | 5 月前3DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model
communication overheads and ensure load balance. By combining these two techniques, DeepSeek-V2 features strong performance (Figure 1(a)), economical training costs, and efficient inference throughput Compared with the corpus used in DeepSeek 67B (our previous release) (DeepSeek-AI, 2024), this corpus features an extended amount of data, especially Chinese data, and higher data quality. We first pretrain IAAI 2020, The Tenth AAAI Symposium on Educational Advances in Artificial Intelligence, EAAI 2020, New York, NY, USA, February 7-12, 2020, pages 7432–7439. AAAI Press, 2020. doi: 10.1609/aaai.v34i05.62390 码力 | 52 页 | 1.23 MB | 1 年前3OpenAI - AI in the Enterprise
AI in the Enterprise Lessons from seven frontier companiesContents A new way to work 3 Executive summary 5 Seven lessons for enterprise AI adoption Start with evals 6 Embed AI into your products developers 18 Set bold automation goals 21 Conclusion 22 More resources 24 2 AI in the EnterpriseA new way to work As an AI research and deployment company, OpenAI prioritizes partnering with global software or deploying cloud apps. The most successful companies are often those who treat it as a new paradigm. This leads to an experimental mindset and an iterative approach that gets to value faster0 码力 | 25 页 | 9.48 MB | 5 月前3OpenAI 《A practical guide to building agents》
handling complex, multi-step tasks. Advances in reasoning, multimodality, and tool use have unlocked a new category of LLM-powered systems known as agents. This guide is designed for product and engineering or search the web. Action Enable agents to interact with systems to take actions such as adding new information to databases, updating records, or sending messages. Send emails and texts, update incrementally adding tools, keeping complexity manageable and simplifying evaluation and maintenance. Each new tool expands its capabilities without prematurely forcing you to orchestrate multiple agents. Tools0 码力 | 34 页 | 7.00 MB | 5 月前3TVM: Where Are We Going
optimized DNN operator library FrameworksLimitations of Existing Approach cuDNN Frameworks New operator introduced by operator fusion optimization potential benefit: 1.5x speedup Engineering Learning System High-level data flow graph and optimizations Directly generate optimized program for new operator workloads and hardware Hardware FrameworksWhy Automation is the Future Clear winner for Future Hardware Current TVM Stack New NPU Runtime TSIM Driver TSIM Binary New Hardware Design in Verilog VerilatorToward Unified IR InfraOverview of New IR Infra Single unified module/pass, type0 码力 | 31 页 | 22.64 MB | 5 月前3TVM@AliOS
for different terminals. 。 To help traditional car firms AliOS互联网汽车 共创智能网联汽车 共建未来出行生态 embrace new "connected' era by acting as the IT “chassis' of ROEWE负风 auto industry /NiiOS ! 驱动万物智能 TVM Timeline w,v4.w) V0.w vadd(v30.w,v31,.w) Vvmem( rO++#1) = V0.new } r@ = #0; jumpr r31 vrmpy(v2.ub,v9.ub) vSsptLat(r2) = V1 Vmem( rO++#1) = V3.new )} nop r2 = memw(rl++#4) } :endLoop0 Vv2.uw = vrmpy(v2 vo.ub) v30 = vsptLat(r2) Vvmem( rO++#1) = V2.new v1.uw = vrmpy(v1.ub,vo.ub) Vmem( rO++#1) = V1.new Vv31.uw = vrmpy(v30.ub,vo.ub) Vmem( rO++#1) = V31.new 上 r0 = #0; jumpr r31 } PART FOUR Alios0 码力 | 27 页 | 4.86 MB | 5 月前3Bring Your Own Codegen to TVM
YOLO model, but... ResNet-50 Dense Non Maximum Suppression Non Maximum Suppression (NMS) is too new to be supported by your chip But NMS is supported by TVM!© 2019, Amazon Web Services, Inc. or its new_call = relay.Call(call.op, params, call.attrs) if curr_last: . new_call = subgraph_end(new_call, self.target) return new_call© 20190 码力 | 19 页 | 504.69 KB | 5 月前3TVM Meetup: Quantization
Amazon Web Services, Inc. or its Affiliates. All rights reserved. Quantized Operators in Framework • New operators like TF quantized_conv2d • Underlying calculations are different than FP32 conv2d • Sometimes 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 TVM infrastructure. Option 2 – Lower to a sequence of existing Relay operators • We introduced a new Relay dialect – QNN to encapsulate this work • Complete reuse of Relay pass infrastructure • Possible0 码力 | 19 页 | 489.50 KB | 5 月前3
共 13 条
- 1
- 2