https://www.nota.ai/community daily 0.75 2026-03-31 https://www.nota.ai/community/the-real-reason-turboquant-shook-the-market-ai-optimization-has-gone-mainstream monthly 0.5 2026-03-31 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/c1239f42-73b9-4c1f-89f9-f13d6909ba29/%E1%84%8B%E1%85%B5%E1%84%8C%E1%85%A2%E1%84%92%E1%85%AE%E1%86%AB1-0+%281%29.jpg Tech Blog - The Real Reason TurboQuant Shook the Market: AI Optimization Has Gone Mainstream - Jaehoon Lee Technical Content Manager, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/bdb3dd4f-d04e-44d6-afce-6aaf5d2e50f8/TurboQuant+Content+Figure_01.png Tech Blog - The Real Reason TurboQuant Shook the Market: AI Optimization Has Gone Mainstream - 돋보이게 만드세요 Figure 1: Electricity generation in China vs. the U.S. (2008–2024). China's output has roughly tripled while the U.S. has plateaued, underscoring the need for energy-efficiency gains in the AI era. (Source: AI+HW 2035) https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/23126b64-c822-4e6c-a1fc-cc261c704518/TurboQuant+Content+Figure_02.png Tech Blog - The Real Reason TurboQuant Shook the Market: AI Optimization Has Gone Mainstream - 돋보이게 만드세요 Figure 2: Throughput comparison by precision when serving DeepSeek-R1 on NVIDIA HGX B200. The NVFP4 (4-bit) + MTP configuration achieves approximately 2× the throughput of FP8. (Source: NVIDIA Technical Blog) https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/98588942-16cf-4c73-ab1a-7d2822d9b751/TurboQuant+Content+Figure_03.png Tech Blog - The Real Reason TurboQuant Shook the Market: AI Optimization Has Gone Mainstream - 돋보이게 만드세요 Figure 3: LongBench accuracy comparison across KV cache quantization methods. TurboQuant maintains the same score (50.06) as the 16-bit Full Cache even at 3.5-bit compression. (Source: Google Research) https://www.nota.ai/community/gtc-2026-recap-the-trillion-dollar-inference-race-begins-how-nota-ai-fills-the-gap monthly 0.5 2026-03-31 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/c1239f42-73b9-4c1f-89f9-f13d6909ba29/%E1%84%8B%E1%85%B5%E1%84%8C%E1%85%A2%E1%84%92%E1%85%AE%E1%86%AB1-0+%281%29.jpg Tech Blog - [GTC 2026 Recap] The Trillion-Dollar Inference Race Begins: How Nota AI Fills the Gap - Jaehoon Lee Technical Content Manager, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/3059b7ea-a974-444e-90dc-d7f9f9db212c/GTC+2026_01.png Tech Blog - [GTC 2026 Recap] The Trillion-Dollar Inference Race Begins: How Nota AI Fills the Gap - 돋보이게 만드세요 Figure 1: A slide illustrating the 10,000x increase in inference compute demand over the two years since ChatGPT's launch (Source: GTC 2026 Keynote) https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/0560154c-c3b0-4fe3-a99c-b6dadc40d98a/GTC+2026_02.png Tech Blog - [GTC 2026 Recap] The Trillion-Dollar Inference Race Begins: How Nota AI Fills the Gap - 돋보이게 만드세요 Figure 2: The NemoClaw architecture, unifying agent creation, deployment, and governance in a single framework (Source: GTC 2026 Keynote) https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/f7806ba3-1f1a-491e-8b57-e5903ff49e11/GTC+2026_03.png Tech Blog - [GTC 2026 Recap] The Trillion-Dollar Inference Race Begins: How Nota AI Fills the Gap - 돋보이게 만드세요 Figure 3: Robots, autonomous vehicles, and industrial heavy machinery gathered on the keynote stage (Source: GTC 2026 Keynote) https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/1f3d2fd7-ecfa-460c-970d-b723193dc084/GTC+2026_04.png Tech Blog - [GTC 2026 Recap] The Trillion-Dollar Inference Race Begins: How Nota AI Fills the Gap - 돋보이게 만드세요 Figure 4: The disaggregated architecture with Rubin GPUs handling prefill and Groq 3 LPUs handling decode (Source: GTC 2026 Keynote) https://www.nota.ai/community/genai-everywhere-the-future-of-edge-ai-optimization-with-the-new-netspresso monthly 0.5 2026-03-31 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/2b17235e-03b5-4c3a-b9e5-95f1046a7a09/image+%285%29.png Tech Blog - GenAI Everywhere: The Future of Edge AI Optimization with the New NetsPresso® - NP Product Team, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/ae4b298e-00a2-4ac9-a22e-31bfa7bc0779/7C203010.jpg Tech Blog - GenAI Everywhere: The Future of Edge AI Optimization with the New NetsPresso® - 돋보이게 만드세요 Figure 1. Device Farm at Nota AI Seoul office, featuring devices where AI models have been successfully ported by Nota AI. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/86a51f71-7c84-4bd8-9fe8-cf2bb0f548ce/banner_eng.png Tech Blog - GenAI Everywhere: The Future of Edge AI Optimization with the New NetsPresso® - 돋보이게 만드세요 무엇이든, 온라인에서 스토리를 전달하는 방식이 큰 차이를 만들어낼 수 있습니다. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/8fd17cf2-beb2-41a2-81f6-d2b72412759d/Frame+1597885558_5.png Tech Blog - GenAI Everywhere: The Future of Edge AI Optimization with the New NetsPresso® - 돋보이게 만드세요 무엇이든, 온라인에서 스토리를 전달하는 방식이 큰 차이를 만들어낼 수 있습니다. https://www.nota.ai/community/notamoequantization-an-moe-specific-quantization-method-for-solar-open-100b monthly 0.5 2026-03-26 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/156179b6-aa91-4f31-8db9-9dca1eae4a13/%E3%85%87%E3%84%B9%E3%85%87%E3%85%87.png Tech Blog - NotaMoEQuantization: An MoE-Specific Quantization Method for Solar-Open-100B - Hancheol Park, Ph. D. AI Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/9f809ab9-49d5-4779-88f4-83a4348faa70/142.+Tairen+Piao.jpg Tech Blog - NotaMoEQuantization: An MoE-Specific Quantization Method for Solar-Open-100B - Tairen Piao AI Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/8033a3ad-73e8-405d-a99f-2d422ff6793d/37.+%E1%84%80%E1%85%B5%E1%86%B7%E1%84%90%E1%85%A2%E1%84%92%E1%85%A9.jpg Tech Blog - NotaMoEQuantization: An MoE-Specific Quantization Method for Solar-Open-100B - Tae-Ho Kim CTO & Co-Founder, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/183da0e7-52e1-4bd3-b95b-520f658a4421/2.jpg Tech Blog - NotaMoEQuantization: An MoE-Specific Quantization Method for Solar-Open-100B - ✔️ Resource : The official quantized model of Solar-Open-100B, which passed the first round of South Korea’s Sovereign AI Foundation Model project: https://huggingface.co/nota-ai/Solar-Open-100B-NotaMoEQuant-Int4 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/c4c35153-64fc-4807-a708-1d4ef3d5cec8/%E1%84%8B%E1%85%B5%E1%84%86%E1%85%B5%E1%84%8C%E1%85%B5+%E1%84%8C%E1%85%A1%E1%84%85%E1%85%AD.png Tech Blog - NotaMoEQuantization: An MoE-Specific Quantization Method for Solar-Open-100B - 돋보이게 만드세요 무엇이든, 온라인에서 스토리를 전달하는 방식이 큰 차이를 만들어낼 수 있습니다. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/8e3970e0-610f-4bae-ab9a-a9180dadaa00/%E1%84%89%E1%85%B3%E1%84%8F%E1%85%B3%E1%84%85%E1%85%B5%E1%86%AB%E1%84%89%E1%85%A3%E1%86%BA+2026-03-13+%E1%84%8B%E1%85%A9%E1%84%92%E1%85%AE+8.45.34.png Tech Blog - NotaMoEQuantization: An MoE-Specific Quantization Method for Solar-Open-100B - 돋보이게 만드세요 무엇이든, 온라인에서 스토리를 전달하는 방식이 큰 차이를 만들어낼 수 있습니다. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/08ee8eec-1e74-49bc-9cb7-b54314b9d6d7/%E1%84%89%E1%85%B3%E1%84%8F%E1%85%B3%E1%84%85%E1%85%B5%E1%86%AB%E1%84%89%E1%85%A3%E1%86%BA+2026-03-13+%E1%84%8B%E1%85%A9%E1%84%92%E1%85%AE+8.45.24.png Tech Blog - NotaMoEQuantization: An MoE-Specific Quantization Method for Solar-Open-100B - 돋보이게 만드세요 무엇이든, 온라인에서 스토리를 전달하는 방식이 큰 차이를 만들어낼 수 있습니다. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/26c46733-cdc7-42c7-b626-8bc27ac87f84/%E1%84%89%E1%85%B3%E1%84%8F%E1%85%B3%E1%84%85%E1%85%B5%E1%86%AB%E1%84%89%E1%85%A3%E1%86%BA+2026-03-13+%E1%84%8B%E1%85%A9%E1%84%92%E1%85%AE+8.42.14.png Tech Blog - NotaMoEQuantization: An MoE-Specific Quantization Method for Solar-Open-100B - 돋보이게 만드세요 무엇이든, 온라인에서 스토리를 전달하는 방식이 큰 차이를 만들어낼 수 있습니다. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/fd3531b4-3bb4-43dd-baf7-14dc82d0eacd/19.png Tech Blog - NotaMoEQuantization: An MoE-Specific Quantization Method for Solar-Open-100B - 돋보이게 만드세요 무엇이든, 온라인에서 스토리를 전달하는 방식이 큰 차이를 만들어낼 수 있습니다. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/8d18f17d-09c9-4f96-b9ce-a3bde83cd3be/%E1%84%89%E1%85%B3%E1%84%8F%E1%85%B3%E1%84%85%E1%85%B5%E1%86%AB%E1%84%89%E1%85%A3%E1%86%BA+2026-03-13+%E1%84%8B%E1%85%A9%E1%84%92%E1%85%AE+8.42.30.png Tech Blog - NotaMoEQuantization: An MoE-Specific Quantization Method for Solar-Open-100B - 돋보이게 만드세요 무엇이든, 온라인에서 스토리를 전달하는 방식이 큰 차이를 만들어낼 수 있습니다. https://www.nota.ai/community/ergo-efficient-high-resolution-visual-understanding-for-vision-language-models monthly 0.5 2026-03-11 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/8876bcb3-57ae-442e-978c-27a2d26f4398/Group+40153.jpg Tech Blog - ERGO: Efficient High-Resolution Visual Understanding for Vision-Language Models - Jewon Lee | Wooksu Shin | Seungmin Yang | Ki-Ung Song | Donguk Lim | Jaeyeon Kim | Tae-Ho Kim | Bo-Kyeong Kim EdgeFM Team, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/eb27c867-8ed3-41b4-ac5f-3f8ec0b36e82/comparison_with_prior.png Tech Blog - ERGO: Efficient High-Resolution Visual Understanding for Vision-Language Models - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/5521ce57-e96c-436a-a6a6-e018a1024c7a/training_pipeline.png Tech Blog - ERGO: Efficient High-Resolution Visual Understanding for Vision-Language Models - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/ecff5c35-c823-4fa8-94a5-ca19406a0d07/%E1%84%89%E1%85%B3%E1%84%8F%E1%85%B3%E1%84%85%E1%85%B5%E1%86%AB%E1%84%89%E1%85%A3%E1%86%BA+2026-02-26+%E1%84%8B%E1%85%A9%E1%84%92%E1%85%AE+1.19.49.png Tech Blog - ERGO: Efficient High-Resolution Visual Understanding for Vision-Language Models - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/0c32dd11-c429-4fdf-b05d-d076eaed42a3/%E1%84%89%E1%85%B3%E1%84%8F%E1%85%B3%E1%84%85%E1%85%B5%E1%86%AB%E1%84%89%E1%85%A3%E1%86%BA+2026-02-26+%E1%84%8B%E1%85%A9%E1%84%92%E1%85%AE+1.58.35.png Tech Blog - ERGO: Efficient High-Resolution Visual Understanding for Vision-Language Models https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/0ae4cba7-fd34-4092-81dd-8d27ed874925/image.png Tech Blog - ERGO: Efficient High-Resolution Visual Understanding for Vision-Language Models - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/0a0675d2-6b05-4e54-9e7e-c422795697a2/image+%281%29.png Tech Blog - ERGO: Efficient High-Resolution Visual Understanding for Vision-Language Models - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/78887dc9-b35e-4ca0-84da-a4b82c2cdd1c/image+%282%29.png Tech Blog - ERGO: Efficient High-Resolution Visual Understanding for Vision-Language Models - 돋보이게 만드세요 무엇이든, 온라인에서 스토리를 전달하는 방식이 큰 차이를 만들어낼 수 있습니다. https://www.nota.ai/community/where-do-llms-encode-the-knowledge-to-assess-the-ambiguity-6zjbl-tmjns-l25lf-grh8b monthly 0.5 2026-03-31 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/c4f6fbb6-5fa7-427f-a860-cfc8a9edc12a/image+%2811%29.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Seungmin Yang EdgeFM Lead, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/1888fd76-47e3-4e65-8751-7d5c6631d78f/%EC%8A%A4%ED%81%AC%EB%A6%B0%EC%83%B7+2025-10-21+170927.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/9b5fee13-ec86-4bba-b2c8-7e5b99802ad0/%ED%91%9C+01.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/e7f65ef8-91f1-4828-8bb1-dbcc3e042f6e/%ED%91%9C+02.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/12fcd9e8-23bb-4c66-b759-9d79691ce0e1/%ED%91%9C3.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/559e5d49-0862-4e88-a602-f7365490018d/%ED%91%9C4.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/9bf5d214-8f7a-4ca1-b5a0-adb270aa1f61/%ED%91%9C+05.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/b79a8a65-31db-43f8-8c66-fa44f3ae4c11/TTFT+01.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out TTFT-Batch size graph with fixed input token length (1024). The figure on the left side is results on A100 and right side is that on RTX PRO 6000. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/5880dbef-5aa9-423c-ba94-863ae1e4a4a1/TTFT+02.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out TTFT-Input Token Length graph with fixed batch size (16). The figure on the left side is results on A100 and right side is that on RTX PRO 6000. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/428d9070-9304-4186-a9e2-d460886d5f03/TPOT1.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out TPOT-Batch size graph with fixed input token length (1024). The figure on the left side is results on A100 and right side is that on RTX PRO 6000. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/f0bebcef-cdb6-4d40-97dd-544ae199a44f/TPOT2.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out TPOT-Input Token Length graph with fixed batch size (16). The figure on the left side is results on A100 and right side is that on RTX PRO 6000. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/7b689b91-56d1-440e-af8b-0d9792034deb/%EA%B7%B8%EB%A6%BC3.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Throughput-Batch size graph with fixed input token length (1024). Solid lines indicate best cases for RTX PRO 6000 and dot lines means best cases for A100. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/5ff8b7e2-623c-4103-b59e-fe0628e562bb/%EA%B7%B8%EB%A6%BC4.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Throughput-Input Token Length graph with fixed batch size (16). Solid lines indicate best cases for RTX PRO 6000 and dot lines means best cases for A100. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/96b2fb36-675c-404c-af2e-6e5ada33869c/benchmark_winogrande.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/e08b9db4-7c7e-473e-9546-ba64e7e13de2/benchmark_piqa.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/3afb834c-7f28-4f1b-8f51-edcdc1606288/benchmark_mmlu_5_shot.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/0270468a-7132-4cbd-8d69-5d39a4fb24cb/benchmark_mmlu_0_shot.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/f1eecd4d-dafb-4751-a375-7494d3d1fbec/benchmark_hellaswag.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/community/where-do-llms-encode-the-knowledge-to-assess-the-ambiguity-6zjbl-tmjns-l25lf monthly 0.5 2025-11-05 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/c4f6fbb6-5fa7-427f-a860-cfc8a9edc12a/image+%2811%29.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Seungmin Yang EdgeFM Lead, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/1888fd76-47e3-4e65-8751-7d5c6631d78f/%EC%8A%A4%ED%81%AC%EB%A6%B0%EC%83%B7+2025-10-21+170927.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/9b5fee13-ec86-4bba-b2c8-7e5b99802ad0/%ED%91%9C+01.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/e7f65ef8-91f1-4828-8bb1-dbcc3e042f6e/%ED%91%9C+02.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/12fcd9e8-23bb-4c66-b759-9d79691ce0e1/%ED%91%9C3.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/559e5d49-0862-4e88-a602-f7365490018d/%ED%91%9C4.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/9bf5d214-8f7a-4ca1-b5a0-adb270aa1f61/%ED%91%9C+05.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/b79a8a65-31db-43f8-8c66-fa44f3ae4c11/TTFT+01.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out TTFT-Batch size graph with fixed input token length (1024). The figure on the left side is results on A100 and right side is that on RTX PRO 6000. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/5880dbef-5aa9-423c-ba94-863ae1e4a4a1/TTFT+02.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out TTFT-Input Token Length graph with fixed batch size (16). The figure on the left side is results on A100 and right side is that on RTX PRO 6000. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/428d9070-9304-4186-a9e2-d460886d5f03/TPOT1.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out TPOT-Batch size graph with fixed input token length (1024). The figure on the left side is results on A100 and right side is that on RTX PRO 6000. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/f0bebcef-cdb6-4d40-97dd-544ae199a44f/TPOT2.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out TPOT-Input Token Length graph with fixed batch size (16). The figure on the left side is results on A100 and right side is that on RTX PRO 6000. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/7b689b91-56d1-440e-af8b-0d9792034deb/%EA%B7%B8%EB%A6%BC3.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Throughput-Batch size graph with fixed input token length (1024). Solid lines indicate best cases for RTX PRO 6000 and dot lines means best cases for A100. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/5ff8b7e2-623c-4103-b59e-fe0628e562bb/%EA%B7%B8%EB%A6%BC4.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Throughput-Input Token Length graph with fixed batch size (16). Solid lines indicate best cases for RTX PRO 6000 and dot lines means best cases for A100. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/96b2fb36-675c-404c-af2e-6e5ada33869c/benchmark_winogrande.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/e08b9db4-7c7e-473e-9546-ba64e7e13de2/benchmark_piqa.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/3afb834c-7f28-4f1b-8f51-edcdc1606288/benchmark_mmlu_5_shot.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/0270468a-7132-4cbd-8d69-5d39a4fb24cb/benchmark_mmlu_0_shot.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/f1eecd4d-dafb-4751-a375-7494d3d1fbec/benchmark_hellaswag.png Tech Blog - NVIDIA Blackwell; The Impact of NVFP4 For LLM Inference - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/community/where-do-llms-encode-the-knowledge-to-assess-the-ambiguity-6zjbl-tmjns monthly 0.5 2026-03-31 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/baa47359-40e7-49b0-b373-d92f9ec090dd/Marcel.png Tech Blog - Video Self-Distillation for Single-Image Encoders: Learning Temporal Priors from Unlabeled Video - Marcel Simon, Ph. D. ML Researcher, Nota AI GmbH https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/50808d6e-390c-4e10-99db-6f3fcde7f02d/cto.jpg Tech Blog - Video Self-Distillation for Single-Image Encoders: Learning Temporal Priors from Unlabeled Video - Tae-Ho Kim CTO & Co-Founder, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/7a975d0b-635d-415b-bea3-c67794551042/Seul-ki.jpg Tech Blog - Video Self-Distillation for Single-Image Encoders: Learning Temporal Priors from Unlabeled Video - Seul-Ki Yeom, Ph. D. Research Lead, Nota AI GmbH https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/e84c3601-2c14-4124-bc86-a04badf50afb/1.png Tech Blog - Video Self-Distillation for Single-Image Encoders: Learning Temporal Priors from Unlabeled Video - Make it stand out Figure 1: Training Overview https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/45b37f43-4035-48ec-afd4-22b96f76da5b/2.png Tech Blog - Video Self-Distillation for Single-Image Encoders: Learning Temporal Priors from Unlabeled Video - Make it stand out Table 1. Semantic segmentation results on ADE20K https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/bc4387b0-e6b0-426a-b88d-cd6fb03f22bf/3.png Tech Blog - Video Self-Distillation for Single-Image Encoders: Learning Temporal Priors from Unlabeled Video - Make it stand out Table 2. Object detection results on COCO https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/02cf2a12-bc9f-47c4-a9e4-29035e1ea943/4.png Tech Blog - Video Self-Distillation for Single-Image Encoders: Learning Temporal Priors from Unlabeled Video - Make it stand out Figure 2. Effect of prediction stride ∆ on ADE20K fast-linear accuracy https://www.nota.ai/community/uniform-a-reuse-attention-mechanism-for-efficient-transformers-on-resource-constrained-edge-devices-p98l9 monthly 0.5 2025-08-25 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/1240d30c-91fb-4846-8dcb-99f2ced4dbbe/Group+40158.png Tech Blog - SplitQuant: Layer Splitting for Low-Bit Neural Network Quantization for Edge AI Devices - Jaewoo Song Software Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/711de431-c12e-4a97-afc1-0eb41fd81ea0/%EC%9E%AC%EC%9A%B0%EB%8B%98+%ED%85%8C%ED%81%AC%EB%B8%94%EB%A1%9C%EA%B7%B8+%EC%82%BD%EB%8F%84.jpeg Tech Blog - SplitQuant: Layer Splitting for Low-Bit Neural Network Quantization for Edge AI Devices - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/2db09f6b-f464-4147-9e31-7a095ad78ba6/1.jpeg Tech Blog - SplitQuant: Layer Splitting for Low-Bit Neural Network Quantization for Edge AI Devices - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/community/where-do-llms-encode-the-knowledge-to-assess-the-ambiguity-6zjbl-xzl9k monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/8876bcb3-57ae-442e-978c-27a2d26f4398/Group+40153.jpg Tech Blog - Efficient LLaMA-3.2-Vision by Trimming Cross-attended Visual Features</span> - Jewon Lee | Ki-Ung Song | Seungmin Yang | Donguk Lim | Jaeyeon Kim | Wooksu Shin | Bo-Kyeong Kim | Tae-Ho Kim EdgeFM Team, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/77a63af5-9521-4314-ac67-22e3e96917ce/yongjaelee-Nov2019.jpg Tech Blog - Efficient LLaMA-3.2-Vision by Trimming Cross-attended Visual Features</span> - Yong Jae Lee, Ph. D. Associate Professor, UW-Madison https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/875f49c6-3a52-4739-bbaf-085b75165ee0/2.png Tech Blog - Efficient LLaMA-3.2-Vision by Trimming Cross-attended Visual Features</span> - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/77726b7f-0490-4cec-b37a-27f601bf28e2/3.png Tech Blog - Efficient LLaMA-3.2-Vision by Trimming Cross-attended Visual Features</span> - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/14ecbd71-b993-4c2c-b20f-56e967174ac0/4.png Tech Blog - Efficient LLaMA-3.2-Vision by Trimming Cross-attended Visual Features</span> - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/9dd959c0-7e50-42b5-bfa7-47584def91ed/5.png Tech Blog - Efficient LLaMA-3.2-Vision by Trimming Cross-attended Visual Features</span> - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/community/uniform-a-reuse-attention-mechanism-for-efficient-transformers-on-resource-constrained-edge-devices monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/e3eaf409-8b8a-46b8-82cd-8f7ab6b5da1b/%E1%84%89%E1%85%B3%E1%86%AF%E1%84%80%E1%85%B5_%E1%84%89%E1%85%A1%E1%84%8C%E1%85%B5%E1%86%AB.jpeg Tech Blog - UniForm: A Reuse Attention Mechanism for Efficient Transformers on Resource-Constrained Edge Devices - Seul-Ki Yeom, Ph. D. Research Lead, Nota AI GmbH https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/d2dcd933-3f45-4118-9088-2024ee379a71/CTO_Profile.jpeg Tech Blog - UniForm: A Reuse Attention Mechanism for Efficient Transformers on Resource-Constrained Edge Devices - Tae-Ho Kim CTO & Co-Founder, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/d54fc727-d8dc-4149-bb0d-c1007cae7bdc/Screenshot+2025-04-08+at+11.09.59+AM.png Tech Blog - UniForm: A Reuse Attention Mechanism for Efficient Transformers on Resource-Constrained Edge Devices - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/9f356e3f-52ed-445a-95ed-52d5278eb6f3/Screenshot+2025-04-08+at+11.13.07+AM.png Tech Blog - UniForm: A Reuse Attention Mechanism for Efficient Transformers on Resource-Constrained Edge Devices - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/community/where-do-llms-encode-the-knowledge-to-assess-the-ambiguity-6zjbl monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/8206d733-79d8-4aa7-8ca4-baf5d021704c/Performance_%EB%B0%95%ED%95%9C%EC%B2%A0.jpg Tech Blog - A Study on Detecting LLM-Generated Multilingual Content - Hancheol Park, Ph. D. AI Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/878f52d7-b7d0-4ca2-aba0-482ee65c4291/Netspresso_%EA%B9%80%EA%B1%B4%EB%AF%BC_%EC%BB%AC%EB%9F%AC.jpg Tech Blog - A Study on Detecting LLM-Generated Multilingual Content - Geonmin Kim, Ph. D. AI Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/ee338b10-6848-49d9-b17a-a3dca355fb95/jaeyeonkim.jpeg Tech Blog - A Study on Detecting LLM-Generated Multilingual Content - Jaeyeon Kim AI Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/f465a87d-bf5f-4a47-99be-6490a38f41f8/image.png Tech Blog - A Study on Detecting LLM-Generated Multilingual Content - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/a72fa71e-993b-4176-a569-0ec4724e5880/graph.png Tech Blog - A Study on Detecting LLM-Generated Multilingual Content - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/94a2b04b-c356-4d61-b6b1-0d123f770718/results.png Tech Blog - A Study on Detecting LLM-Generated Multilingual Content - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/community/where-do-llms-encode-the-knowledge-to-assess-the-ambiguity monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/8206d733-79d8-4aa7-8ca4-baf5d021704c/Performance_%EB%B0%95%ED%95%9C%EC%B2%A0.jpg Tech Blog - Where do LLMs Encode the Knowledge to Assess the Ambiguity? - Hancheol Park, Ph. D. AI Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/878f52d7-b7d0-4ca2-aba0-482ee65c4291/Netspresso_%EA%B9%80%EA%B1%B4%EB%AF%BC_%EC%BB%AC%EB%9F%AC.jpg Tech Blog - Where do LLMs Encode the Knowledge to Assess the Ambiguity? - Geonmin Kim, Ph. D. AI Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/3cef3786-3ac7-48f8-b2bd-3a50fc19deff/%EC%88%98%EC%A0%95%EA%B0%80%EB%8A%A5%ED%95%9C_%EC%9D%B4%EB%AF%B8%EC%A7%80.png Tech Blog - Where do LLMs Encode the Knowledge to Assess the Ambiguity? - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/db076d9a-3345-43a3-a475-f0864bbc5990/image+%2824%29.png Tech Blog - Where do LLMs Encode the Knowledge to Assess the Ambiguity? - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/7deee9ad-e6f6-42c9-ace3-5137e7c9a576/image+%2825%29.png Tech Blog - Where do LLMs Encode the Knowledge to Assess the Ambiguity? - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/e2751f16-b768-4b8d-b512-a8c89c3608d4/image+%2826%29.png Tech Blog - Where do LLMs Encode the Knowledge to Assess the Ambiguity? - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/39a3de92-65e9-4b8a-af20-158adeedc5d3/image+%2827%29.png Tech Blog - Where do LLMs Encode the Knowledge to Assess the Ambiguity? - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/7fd8df3d-ca81-42d3-9984-27c3a3ef190a/image+%2828%29.png Tech Blog - Where do LLMs Encode the Knowledge to Assess the Ambiguity? - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/community/deploying-an-efficient-vision-language-model-on-mobile-devices monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/f0de44bb-3b4c-49b4-8cef-1baa2f580d8f/NOTA_Netspresso_%EA%B9%80%EC%9E%AC%EC%97%B0_%EC%BB%AC%EB%9F%AC.jpg Tech Blog - Deploying an Efficient Vision-Language Model on Mobile Devices - Jaeyeon Kim Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/878f52d7-b7d0-4ca2-aba0-482ee65c4291/Netspresso_%EA%B9%80%EA%B1%B4%EB%AF%BC_%EC%BB%AC%EB%9F%AC.jpg Tech Blog - Deploying an Efficient Vision-Language Model on Mobile Devices - Geonmin Kim Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/8206d733-79d8-4aa7-8ca4-baf5d021704c/Performance_%EB%B0%95%ED%95%9C%EC%B2%A0.jpg Tech Blog - Deploying an Efficient Vision-Language Model on Mobile Devices - Hancheol Park Team Lead of NetsPresso Application, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/683a2605-e97a-449d-916b-b13ec7847572/1+%EC%88%98%EC%A0%95.png Tech Blog - Deploying an Efficient Vision-Language Model on Mobile Devices - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/aad112eb-ec74-44d1-b875-219dc2ae967c/2.png Tech Blog - Deploying an Efficient Vision-Language Model on Mobile Devices - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/7bd754dc-92d1-4129-8e46-05bf4770dace/3.png Tech Blog - Deploying an Efficient Vision-Language Model on Mobile Devices - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/e8460a41-8d93-4efb-b005-cba7578a67c3/240723_table1_img.png Tech Blog - Deploying an Efficient Vision-Language Model on Mobile Devices - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/community/cluster-self-refinement-for-enhanced-online-multi-camera-people-tracking monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/7aade0af-b23d-4427-be09-1af88ae7c11d/Performance_%EA%B9%80%EC%A0%95%ED%98%B8.jpg Tech Blog - Cluster Self-Refinement for Enhanced Online Multi-Camera People Tracking - Jeongho Kim Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/09b301e1-0584-4d14-8a02-de3c23a055d0/1.png Tech Blog - Cluster Self-Refinement for Enhanced Online Multi-Camera People Tracking - Make it stand out Figure 1. This is an example of Multi-Camera People Tracking. It involves tracking people across various cameras by mapping them to the same identities. The image in the center depicts a 2D map of the location, showing the estimated positions of people as captured by the cameras. The numbers provided represent their global IDs. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/82f42f03-340a-4fb6-ae90-e5c0e4783dbe/2.png Tech Blog - Cluster Self-Refinement for Enhanced Online Multi-Camera People Tracking - Make it stand out Figure 2. Overview of our system’s architecture. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/690edbf5-726a-4b06-85ed-6f7f6ec7b5bf/3.png Tech Blog - Cluster Self-Refinement for Enhanced Online Multi-Camera People Tracking - Make it stand out Figure 3. Overview of Cluster Self-Refinement. The left side depicts the refinement of appearance features, utilizing agglomerative clustering to check if different people are stored and, if correct, refine the appearance features in the cluster tracklet. The right side illustrates overlapped cluster refinement, addressing situations where one person has more than one global ID. The CSR procedure is carried out at regular intervals, as denoted by the red circle shown above. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/b2566f82-70c9-4e9d-90e1-974de704709a/4.png Tech Blog - Cluster Self-Refinement for Enhanced Online Multi-Camera People Tracking - Make it stand out Table 1. The results of the ablation study on using a CSR and EUP. CSR and EUP stand for Cluster Self-Refinement and Enhanced Utilizing Pose estimation, respectively. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/0c10e911-1c0c-4b57-866a-4453cf04eecf/5.png Tech Blog - Cluster Self-Refinement for Enhanced Online Multi-Camera People Tracking - Make it stand out Table 2. Public leaderboard for the Challenge Track 1 https://www.nota.ai/community/road-object-detection-robust-to-distorted-objects-at-the-edge-regions-of-images monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/233935b8-4fd5-4935-93f9-decbd8dd3d77/NOTA_Performance_%EC%8B%A0%EC%9A%B1%EC%88%98.jpg Tech Blog - Road Object Detection Robust to Distorted Objects at the Edge Regions of images - Wooksu Shin Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/a0e5df95-d693-4338-bc7f-1a6afb4d2ab5/1.png Tech Blog - Road Object Detection Robust to Distorted Objects at the Edge Regions of images - Make it stand out Figure 1. An example of sliced inference. The region corresponding to the red box in the above image is resized to match the model's input size (as shown in the bottom image) before being inputted. As a result, the small objects within the yellow box are greatly enlarged, allowing the model to detect the target objects more accurately. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/b5526cdf-279a-4cbf-a7b3-e5f0fc973af5/2.png Tech Blog - Road Object Detection Robust to Distorted Objects at the Edge Regions of images - Make it stand out Figure 2. (a) depicts a mis-prediction of a distorted non-target object. A street sign is distorted, creating visual similarity with the outline of a car, detectors occasionally mis-predict it as a car. After training on these non-target objects, the issue of incorrect predictions is resolved as shown in (b). https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/346edbe7-303c-4633-b34c-d1233af39a57/3.png Tech Blog - Road Object Detection Robust to Distorted Objects at the Edge Regions of images - Make it stand out Figure 3. Changes in pixel distribution with histogram equalization https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/f70f4b6d-5019-46a9-8d8a-d6dc07b422c5/4.png Tech Blog - Road Object Detection Robust to Distorted Objects at the Edge Regions of images - Make it stand out Table 1. Ensembled detectors that we use in this work. Swin-L (Liu et al., 2021), ViT-L (Dosovitskiy et al., 2021) indicate backbones of DETR models, DINO (Zhang et al., 2023) is an architecture of DETR series. Co-DINO (Swin-L) was pretrained with Objects365 (Shao et al., 2019) and COCO (Lin et al., 2014) datasets. Co-DINO (ViT-L) was pretrained with Objects365 and LVIS. All models are fine-tuned with FishEye8K (Gochoo et al., 2023) dataset in this work. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/d29514b0-4001-4a4f-818d-b543d5b4bed1/5.png Tech Blog - Road Object Detection Robust to Distorted Objects at the Edge Regions of images - Make it stand out Figure 4. Examples of weighted boxes fusion (WBF). The above two images show bounding boxes predicted by different models. In the image below, these bounding boxes are combined into a single box using WBF. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/de733fdd-b002-45d6-9556-413a3fb03413/6.png Tech Blog - Road Object Detection Robust to Distorted Objects at the Edge Regions of images - Make it stand out Table 2. The results of ablation study on using sliced inference and semi-supervision, respectively https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/e56512f6-2225-445e-b5d9-df6b7490e577/7.png Tech Blog - Road Object Detection Robust to Distorted Objects at the Edge Regions of images - Make it stand out Table 3. Public Top 10 leaderboard for the Challenge Track 4 https://www.nota.ai/community/edgefusion-on-device-text-to-image-generation monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/b83ae155-08fb-4370-8431-78340d910a07/%EB%B0%95%ED%83%9C%EC%9E%84%EB%8B%98.jpg Tech Blog - EdgeFusion: On-device Text-to-Image Generation - Tairen Piao Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/86287ad9-7afc-43f7-8cb2-f8ab85e7ab67/Edge+Fusion_Figure+1.jpg Tech Blog - EdgeFusion: On-device Text-to-Image Generation - Make it stand out Figure 1. T2I generation results. When trained with improved data, our EdgeFusion can produce high-quality images from challenging prompts in just a few denoising steps https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/6fae71b8-cc34-4335-954a-b033e5c47cb1/Edge+Fusion_Figure+2.jpg Tech Blog - EdgeFusion: On-device Text-to-Image Generation - Make it stand out Figure 2. A compact SD with step reduction. (a) Vanilla application of LCM: we initialize BK-LCM-Tiny with the weight from BK-SDM-Tiny and train with distillation to reduce sampling steps. (b) Our approach: improving the initialization of the LCM’s student with a better teacher is beneficial. Moreover, in the LCM training phase, employing the original teacher enhances performance. Leveraging high-quality data is crucial in both phases. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/bc718412-6b0c-46c5-83fe-29520d76fc16/Edge+Fusion_Figure+3.jpg Tech Blog - EdgeFusion: On-device Text-to-Image Generation - Make it stand out Table 1. Results of the BK-SDM-Tiny architecture trained on different datasets. Evaluated on COCO dataset with 25 steps. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/5952a859-a2bc-4e31-8ee8-7040e8f6657a/Edge+Fusion_Figure+4.jpg Tech Blog - EdgeFusion: On-device Text-to-Image Generation - Make it stand out Table 2. Human preference evaluation. The win rate of our models against the same architecture without improved data and without student finetuning is reported (1500 comparisons, 21 participants). https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/2f6d0cd9-0edd-4c7f-8c8d-bd76db228c95/Edge+Fusion_Figure+5.jpg Tech Blog - EdgeFusion: On-device Text-to-Image Generation - Make it stand out Figure 3. Comparison between FP32 and W8A16 quantized EdgeFusion (2 steps) https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/18af17c9-4e00-43bd-ad72-526e8f8e651b/Edge+Fusion_Figure+6.jpg Tech Blog - EdgeFusion: On-device Text-to-Image Generation - Make it stand out Table 3. Inference time reduction using MLT. For the cross at- tention block, the relative time ratio is computed by comparing operations without MLT to those with MLT. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/2af4b4f8-9948-455d-bb76-77919ac85ec2/Edge+Fusion_Figure+7.png Tech Blog - EdgeFusion: On-device Text-to-Image Generation - Make it stand out Table 4. Benchmark on Exynos 2400 with and without MLT. https://www.nota.ai/community/ld-pruner-efficient-pruning-of-latent-diffusion-models-using-task-agnostic-insights monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/b4c73e6e-93a2-45fa-b810-3a5502a91744/Thibault.png Tech Blog - LD-Pruner: Efficient Pruning of Latent Diffusion Models using Task-Agnostic Insights - Thibault Castells Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/9b11c711-7724-4dac-b7b7-b032e44bec03/Screenshot+2024-04-11+at+3.07.29%E2%80%AFPM.png Tech Blog - LD-Pruner: Efficient Pruning of Latent Diffusion Models using Task-Agnostic Insights - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/60f2b663-1ab8-4e1c-9c4a-dc0b08cc1af1/Screenshot+2024-04-11+at+3.45.38%E2%80%AFPM.png Tech Blog - LD-Pruner: Efficient Pruning of Latent Diffusion Models using Task-Agnostic Insights - Make it stand out Table 1: Comparison of different models for T2I Generation, on the MS-COCO 256 X 256 validation set. Speedup values are measured relatively to SD-v1.4. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/23e98b11-69f9-4332-9368-1eaf25039aac/Screenshot+2024-04-11+at+3.44.30%E2%80%AFPM.png Tech Blog - LD-Pruner: Efficient Pruning of Latent Diffusion Models using Task-Agnostic Insights - Make it stand out Figure 2: Qualitative comparison on zero-shot MS-COCO benchmark on T2I. The results of previous studies were obtained with their official released models. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/556368ef-2053-4af4-82cd-c094a33351fa/Screenshot+2024-04-11+at+3.49.52%E2%80%AFPM.png Tech Blog - LD-Pruner: Efficient Pruning of Latent Diffusion Models using Task-Agnostic Insights - Make it stand out Figure 3: Evolution of the FID during the training process for the UIG task on the CelebA-HQ 256 X 256 dataset, for two different compression ratios. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/5f8206ae-7be1-4c61-90da-ce1ecfa67156/Screenshot+2024-04-11+at+3.54.54%E2%80%AFPM.png Tech Blog - LD-Pruner: Efficient Pruning of Latent Diffusion Models using Task-Agnostic Insights - Make it stand out Table 2: Compression performance on UAG task with AudioDiffusion. When finetuning, we proceed for 12k steps. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/62003489-84c8-4cfc-a409-6ee594306053/Screenshot+2024-04-11+at+4.01.54%E2%80%AFPM.png Tech Blog - LD-Pruner: Efficient Pruning of Latent Diffusion Models using Task-Agnostic Insights - Make it stand out Table 3: FID scores for our compressed model (31 operators modified) trained from scratch and with preserved pre-training weights, for UIG on CelebA-HQ 256 X 256. In both case, the exact same training is applied. The FID for the original model is 13.85. https://www.nota.ai/community/shortened-llm-a-simple-depth-pruning-for-large-language-models monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/9f76ea45-b972-4d3d-a06e-258ab5b32493/%EB%B3%B4%EA%B2%BD%EB%8B%98.png Tech Blog - Shortened LLM: A Simple Depth Pruning for Large Language Models - Bo-Kyeong Kim Senior Researcher, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/a1002d7c-f4ac-4826-991d-2dd903ad9902/image+%284%29.png Tech Blog - Shortened LLM: A Simple Depth Pruning for Large Language Models - Make it stand out Figure 1. (a) Comparison of pruning units. Width pruning reduces the size of projection weight matrices. Depth pruning removes Transformer blocks, or individual MHA and FFN modules. (b) Efficiency of pruned LLaMA-7B models on an NVIDIA H100 GPU. Compared to width pruning by FLAP and LLM-Pruner, our depth pruning achieves faster inference with competitive PPL on WikiText-2 (left) and offers a better latency-throughput trade-off (right; M: batch size). https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/1bdfc777-b8fa-44a2-b064-8afe01092441/image+%285%29.png Tech Blog - Shortened LLM: A Simple Depth Pruning for Large Language Models - Make it stand out Figure 2. Our depth pruning approach. After identifying unimportant blocks with straightforward metrics, we perform one-shot pruning followed by light retraining. Right LoRA figure was sourced from Hu et al. [2022]. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/2546359a-5dbb-4a59-806f-a4a3fb2a045f/pdf_fig_ppl_bookcorpus_v1.png Tech Blog - Shortened LLM: A Simple Depth Pruning for Large Language Models - Make it stand out Figure 3. Estimated importance of each Transformer block on the calibration set. Blocks with lower PPL scores are pruned. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/6bf1f300-42cc-4874-b5ce-3f2d35183cfe/image+%286%29.png Tech Blog - Shortened LLM: A Simple Depth Pruning for Large Language Models - Make it stand out Figure 4. Results of pruned LLaMA-1-7B [Touvron et al., 2023] and Vicuna-v1.3-13B [Chiang et al., 2023]. The width pruning methods of Wanda-sp [Sun et al., 2024; An et al., 2024], FLAP [An et al., 2024], and LLM-Pruner [Ma et al., 2023] often degrade inference efficiency. In contrast, our depth pruning approach enhances generation speed and competes well in zero-shot task performance. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/0d2fd7af-5c53-4278-8048-48a4cef6605d/image+%287%29.png Tech Blog - Shortened LLM: A Simple Depth Pruning for Large Language Models - Make it stand out Figure 5. Generation examples. Given an input prompt about 'AI can create a logo in seconds,' the pruned models generate outputs that are similar to those of the original models. https://www.nota.ai/community/integrating-launchx-with-nvidia-tao-toolkit-for-running-on-various-edge-devices monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/c8fc65d5-3d00-46c3-ac96-745fff82262b/%E1%84%83%E1%85%A2%E1%84%8C%E1%85%B5+1%402x.jpg Tech Blog - Integrating LaunchX with NVIDIA TAO Toolkit for Running on Various Edge Devices - Hoin Na CoS Tech Part Manager, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/b7c5db47-8138-43c3-9775-fd657ad6b01c/Screenshot_2023-11-14_at_4.55.44.png Tech Blog - Integrating LaunchX with NVIDIA TAO Toolkit for Running on Various Edge Devices - Make it stand out Figure 1. Flow of NVIDIA TAO Toolkit (TAO Toolkit | NVIDIA Developer) https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/2e51dafa-636a-473e-91c7-b7374dfc8997/Screenshot_2023-11-14_at_7.26.14.png Tech Blog - Integrating LaunchX with NVIDIA TAO Toolkit for Running on Various Edge Devices - Make it stand out Figure2. LaunchX https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/15835f8a-f236-4088-afab-94cc5df436d8/Screenshot_2023-11-14_at_8.21.20.png Tech Blog - Integrating LaunchX with NVIDIA TAO Toolkit for Running on Various Edge Devices - Make it stand out Figure 3. Multiple benchmarks of MobilenetV2(TF1) model by LaunchX https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/bdcf7b01-00c4-480b-9ca4-c3907b38f0b0/%EC%8A%A4%ED%81%AC%EB%A6%B0%EC%83%B7+2023-11-14+165817.png Tech Blog - Integrating LaunchX with NVIDIA TAO Toolkit for Running on Various Edge Devices - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/community/how-netspresso-turbocharges-semantic-segmentation-models monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/faff29dd-4db6-4157-9506-a1a375685ef3/YoonJae+Yang.gif Tech Blog - Revolutionizing Mobile AI: How NetsPresso® Turbocharges Semantic Segmentation Models for Real-Time Performance - YoonJae Yang AI Application Developer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/cc09558c-263c-45dc-b8d1-6e9f239b8e18/Netspresso_%E1%84%8B%E1%85%B5%E1%84%92%E1%85%A7%E1%86%BC%E1%84%8C%E1%85%AE%E1%86%AB_%E1%84%8F%E1%85%A5%E1%86%AF%E1%84%85%E1%85%A5.jpg Tech Blog - Revolutionizing Mobile AI: How NetsPresso® Turbocharges Semantic Segmentation Models for Real-Time Performance - Hyungjun Lee Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/f4a6c693-4f7f-48e6-8d17-29ae2ac71229/Article-4_table1.png Tech Blog - Revolutionizing Mobile AI: How NetsPresso® Turbocharges Semantic Segmentation Models for Real-Time Performance - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/cb7d4195-014e-4092-9b51-76d8b93c71d3/Article-4_table2.png Tech Blog - Revolutionizing Mobile AI: How NetsPresso® Turbocharges Semantic Segmentation Models for Real-Time Performance - Make it stand out These results underline the significant improvement in latency achieved by compressing the PIDNet model using NetsPresso®'s structured pruning. This optimization makes it a more viable choice for real-time applications, such as video conferencing and video calls, on mobile devices. In conclusion, our journey into mobile optimization has shown that with the right tools and techniques, even the most computationally demanding AI models can be made efficient for mobile platforms. This opens up exciting possibilities for applications that require real-time, on-device AI processing. https://www.nota.ai/community/empowering-pedestrian-safety-the-obstacle-detection-app-and-ai-model-optimization monthly 0.5 2025-05-07 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/faff29dd-4db6-4157-9506-a1a375685ef3/YoonJae+Yang.gif Tech Blog - Empowering Pedestrian Safety: The Obstacle Detection App and AI Model Optimization - Yoonjae Yang AI Application Developer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/18c4a069-8c01-4518-9bb3-3f6c343e1cc2/%EB%8C%80%EC%A7%80+1%403x.png Tech Blog - Empowering Pedestrian Safety: The Obstacle Detection App and AI Model Optimization - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/community/revolutionizing-laundry-symbol-detection-with-ai-model-optimization-streamlining-the-process-for-precise-results monthly 0.5 2025-05-07 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/a1a9712f-5b67-4cbd-8b15-4b68fd9684eb/%EB%AC%B4%EC%A0%9C-1.gif Tech Blog - Revolutionizing Laundry Symbol Detection with AI Model Optimization: Streamlining the Process for Precise Results - Yoonjae Yang AI Application Developer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/7bf478b3-e3cd-473b-8150-81bea46b3fca/Template_Community_Usecasetable_e5d686cc-54ef-4025-b6fb-e0471967f968.png Tech Blog - Revolutionizing Laundry Symbol Detection with AI Model Optimization: Streamlining the Process for Precise Results - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/community/developing-a-pothole-detection-application-for-safe-driving monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/c3c78916-c127-4518-b7c2-4dd1c59c129d/%EB%8C%80%EC%A7%80+1%403x.jpg Tech Blog - Advancing Road Safety with AI-Powered Pothole Detection: Performance Enhancements for AI Vision Developers and Engineers - YoonJae Yang AI Application Developer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/fd5cbc44-a9f4-446c-8f69-db806d38bbd9/Netspresso_%EC%9D%B4%ED%98%95%EC%A4%80_%EC%BB%AC%EB%9F%AC.jpg Tech Blog - Advancing Road Safety with AI-Powered Pothole Detection: Performance Enhancements for AI Vision Developers and Engineers - Hyungjun Lee Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/63626450-542f-4b0c-af4a-1c0ceaf9cadc/Pothole_table+1.png Tech Blog - Advancing Road Safety with AI-Powered Pothole Detection: Performance Enhancements for AI Vision Developers and Engineers - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/f9c99777-d196-4f68-b278-b96954d7e275/Pothole_table+2.png Tech Blog - Advancing Road Safety with AI-Powered Pothole Detection: Performance Enhancements for AI Vision Developers and Engineers - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/community/enhancing-real-time-processing-of-yolov5-l-using-pruning-techniques-in-pynetspresso monthly 0.5 2025-05-09 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/7aeab08e-96a7-4bdc-8b34-f3040964a40d/Netspresso_%EC%9D%B4%ED%98%95%EC%A4%80_%EC%BB%AC%EB%9F%AC.jpg Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Hyungjun Lee Research Engineer, Nota AI https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/521968a5-f19d-4f8b-986c-385bf4a08f94/Community_1+%282%29.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Figure 1: Achieving real-time FPS with minimal performance degradation using PyNetsPresso. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/91d22cc2-1797-4dd5-b53b-502c0df8b744/pr-latency.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Figure 2-1: Pruning ratio and latency graph. Latency measured on Jetson Xavier. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/8d466d11-446c-4e9f-ba27-978a73be9449/pr-map.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Figure 2-2: Pruning ratio and mAP graph. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/1fd387f5-0781-4009-9bbe-21589280daac/latency-map.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Figure 2 -3:Latency and mAP graph. Latency measured on Jetson Xavier. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/6264a472-c610-4ab9-804d-d8d67bb5ab4a/Summary+table.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/0b77985f-2cc5-445a-aafd-0ea3671f627e/Table.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/33bea7ed-6b66-451d-88bf-29ef1b2376ca/Community+article_5.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Figure 3: Conventional pruning methods make it impossible to achieve acceleration since parameters cannot be completely eliminated. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/f601d527-9e7b-4727-9d4b-f68f9ba01efa/Figure+4.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Figure 4: Dependency Aware Pruning creates a lightweight artificial neural network by completely removing parameters, taking into account the input/output relationships between each layer. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/04c16486-00cc-4361-90f0-ac3651200813/Figure+5+%282%29.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Figure 5: The operation of Dependency Aware Pruning in Skip-Connections allows for the actual acceleration of artificial neural networks by tracking and eliminating dependencies. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/ad7a3e40-db2e-41a1-ae73-e640af78ab5b/Python_1.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/e466f87a-d88b-4a66-b42d-45528b2ffbd4/Python_2.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/9e4190d6-a095-4706-924e-d302d4841eaa/Python_3.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/bf79d815-e2d4-4969-b0a3-ed860d17c291/Python_6.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/c029ada0-17cc-4004-a448-6e2e7f3a6bee/Figure+6.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Figure 6: Results saved in the output_path. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/94047329-13c6-464a-8a7e-2cd4a5b47668/Python_5.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/1c17218e-03e8-4731-9faf-aa07d4980d29/Python+7.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/ba1d250b-3ee9-4c46-921b-33f4e7869cac/pynp_main.png Tech Blog - Enhancing Real-Time Processing of YOLOv5-L Using Pruning Techniques in PyNetsPresso - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/notice daily 0.75 2025-12-26 https://www.nota.ai/notice/2024-04-11 monthly 0.5 2024-11-04 https://www.nota.ai/notice/2024-03-18 monthly 0.5 2024-11-04 https://www.nota.ai/notice_contents-1 daily 0.75 2024-10-04 https://www.nota.ai/notice_contents-1/Blog Post Title One-4cwmc monthly 0.5 2024-09-11 https://www.nota.ai/notice_contents-1/blog-post-title-two-8jphe monthly 0.5 2020-05-20 https://www.nota.ai/notice_contents-1/blog-post-title-three-fwlh3 monthly 0.5 2020-05-20 https://www.nota.ai/notice_contents-1/blog-post-title-four-8g89l monthly 0.5 2020-05-20 https://www.nota.ai/news daily 0.75 2025-12-29 https://www.nota.ai/news/enpressrelease2512 monthly 0.5 2025-12-30 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/b527dd7c-23a3-4344-a1f4-3ecffcd3c303/Nota+AI_PressRelease_Cover.png newsroom - Nota AI to Supply AI Optimization Technology for Samsung Electronics’ Next-Generation Mobile AP ‘Exynos 2600’… Solidifying Its Position as a Leader in On-Device AI - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/news/enpressrelease2511 monthly 0.5 2025-12-29 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/f4896863-4dba-4497-856b-6d941fe4a3ea/PR_S.LSI_Thumb_Dark_v1_251107+%282%29.png newsroom - Nota AI Signs Technology Collaboration Agreement with Samsung Electronics for Exynos AI Optimization “Driving the Popularization of On-Device Generative AI” - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://www.nota.ai/contact-us daily 0.75 2025-07-10 https://www.nota.ai/nvidia-jetson daily 0.75 2023-10-17 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/b7a85b3b-a73f-445c-982e-b33521544621/nvidia-preferred-partner-badge-rgb-for-screen.png https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/b0e21b23-4d89-4ceb-8b0b-a1ba6f28c2a6/NP_platform.png https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/d07c2ad6-1dd7-4bf9-9b34-f13ffb4df3e8/NP_endtoend_diagram_230314.png https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/a474ffb6-68d0-434c-9028-412f9e310ef3/%E1%84%89%E1%85%B3%E1%84%8F%E1%85%B3%E1%84%85%E1%85%B5%E1%86%AB%E1%84%89%E1%85%A3%E1%86%BA+2023-03-17+%E1%84%8B%E1%85%A9%E1%84%92%E1%85%AE+3.10.27.png https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/5807d791-c480-44e9-9604-aa5566b243bf/accuracy.png https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/aa9d797c-6818-47a1-b2dc-c87c9103ca85/np_process.png https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/d0c10cef-dce8-477d-83e5-ed6f29bc9145/07+Sub-module_MS%403x-8.png https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/6457216c-a19a-49b0-92cf-c1e936f2ddef/Sub-module_MC.png https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/0d460b76-c271-4bf9-8914-b119922666d8/07+Sub-module_ML.png https://www.nota.ai/partner-arm-avhethosu65 daily 0.75 2023-07-27 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/6633a563-bbf3-4ee4-b387-64557f0da443/Artboard+2%402x.png https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/6cf75082-d378-454d-b5d2-ea907377d26b/Desktop+Vector+Mockup%403x.png https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/028bc044-22ae-416e-a348-1bc94d1b1b53/table_latency%28ms%29.png https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/6c476c17-22f4-4ca3-8fc0-ac10f0659e99/table_modelsize%28kb%29.png https://www.nota.ai/aboutus daily 0.75 2026-03-31 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/a640bb9b-16a0-4c1e-8ace-ac45bb44937b/company-background-blue.png About us https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/4749b74a-7cf5-475b-9c5e-60f7cb137ad5/%E1%84%82%E1%85%A9%E1%84%90%E1%85%A1_%E1%84%92%E1%85%AC%E1%84%8B%E1%85%B4%E1%84%89%E1%85%B5%E1%86%AF.JPG About us https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/ebb720ab-61ed-444d-9d64-d6444d4e15f7/%E1%84%82%E1%85%A9%E1%84%90%E1%85%A1_%E1%84%8B%E1%85%B5%E1%86%AB%E1%84%86%E1%85%AE%E1%86%AF2.JPG About us https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/139fa6ec-713a-4832-b7c8-19d122e47afa/%E1%84%82%E1%85%A9%E1%84%90%E1%85%A1_%E1%84%8B%E1%85%B5%E1%86%AB%E1%84%86%E1%85%AE%E1%86%AF9.JPG About us https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/2f698fb7-d000-4eff-88f6-cad41b1cfeda/DSC09728.JPG About us https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/5d9f6f04-887b-4be8-b1bd-be3638b3bc47/7C203058+%281%29.JPG About us https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/16d96fe5-e885-42bb-9efc-cb60d4dfcb12/%ED%9A%8C%EC%9D%98%EC%8B%A4_CTO+%ED%9A%8C%EC%9D%98%EC%A4%91_.jpg About us https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/27355782-9284-47ef-bb59-d0ec75e11404/%E1%84%82%E1%85%A9%E1%84%90%E1%85%A1_%E1%84%85%E1%85%A1%E1%84%8B%E1%85%AE%E1%86%AB%E1%84%8C%E1%85%B5.jpg About us https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/3b50da95-ef32-4eee-bdaa-647185ea1db0/EverFocus_logo.png About us EveFocus logo with the text 'EverFocus' in stylized font, with a teal accent on the 'o' and a teal swoosh over the 'o'. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/4de0b046-da76-4a64-b0a1-fa7e0f42481d/partners_aws.png About us AWS Partner Network logo with the text 'aws partner network'. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/b181dca2-a244-49ad-85e3-90eb68c1db58/partners_DeuscheTelekom.png About us Deutsche Telekom logo in shades of pink and gray. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/383dedb3-4958-4fbe-a46c-2e2ec71cec29/ACROSSER_logo.png About us A logo with a stylized red arrow symbol followed by the word 'ACROSSER' in black uppercase letters. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/47ef6940-6976-498a-96b6-702038e1f45c/partners_AXIS.png About us AXIS Communications logo with black text and a yellow and red triangular design https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/e06e976e-7804-48bd-a5b0-5612549de0cc/partners_SKtelecom.png About us SK Telecom logo with stylized red, orange, and white butterfly icon https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/18b5baf1-d4c0-4379-b6c8-798abf2fb739/partners_arm.png About us The word 'arm' written in blue lowercase letters. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/14b939b6-75d3-45f0-a03d-1c06894176e2/partners_intel.png About us Intel logo with a blue lowercase 'i' and the word 'intel' in black. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/6a0965ff-121b-49b1-8d6d-049cf4489d32/partners_Telechips.png About us Telechips company logo in blue text on a white background. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/852294d6-5d0f-47b3-8dad-8b93b2280f4c/partners_Everocus.png About us Everfocus logo with the tagline 'Your Safety, Our Focus' https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/519ea45c-75d7-4100-9878-57c61ff453b2/partners_LGCNS.png About us LG CNS logo with red and gray branding on a white background. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/9f2a4a9a-4950-48a1-be5c-f69cf91d4496/partners_RENESAS.png About us Renesas logo on a white background https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/dd783095-5863-4b50-80a0-4a022853b099/partners_Mircrosoft.png About us Microsoft logo with four colored squares and the word 'Microsoft' next to it. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/e55c9fb3-ea45-413d-b13f-d3e6affdc099/partners_LGU%2B.png About us LG U+ logo with red circle and white smiley face, gray text LG U+ https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/1666f9b1-94f2-41e3-be04-a99a59a2578e/partners_SONY.png About us Sony logo on a white background. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/6dbb54a3-cf9e-4bee-9681-77ff7b64c920/partners_NVIDIA.png About us NVIDIA logo with green eye symbol and black text on white background. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/131f0892-fcfe-4688-a7b6-6537b64ed1cc/partners_Qualcomm.png About us Qualcomm logo in blue text on a white background https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/b958933d-9cca-4882-9916-79e8c97a4f1f/partners_samsung.png About us Samsung logo in bold blue letters on a white background. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/24db228b-3f1f-4a94-9715-72030c8bc521/Main_partners_FuriosaAI.png About us FURIOSA logo with stylized red lightning bolt symbol on the right https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/33abb8bb-daa4-4903-af55-fa552c65220a/partners_SAMSUNGSDS.png About us Samsung SDS logo in blue text on a white background. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/e16204a4-b173-4f50-a5d0-495fc1508e46/Main_partners_Pytorch.png About us PyTorch logo with an abstract flame icon and the word 'PyTorch' next to it. https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/528f06d1-23cd-43bf-9039-621ffce9e411/a.png https://www.nota.ai/newsroom daily 0.75 2026-03-10 https://www.nota.ai/netspresso daily 0.75 2026-02-13 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/9e9dc223-a53d-4924-86c7-0f7c39af69c7/NetsPresso_Basic_Logo_Navy.png https://www.nota.ai/careers daily 0.75 2025-07-14 https://www.nota.ai/ai-solutions-notavisionagent daily 0.75 2026-01-23 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/5f3a28dd-d68f-411c-bf61-5324b07a9876/%40keyvisual.png https://www.nota.ai/ai-solutions-industrialsafety daily 0.75 2026-01-24 https://www.nota.ai/ai-solutions-surveillance daily 0.75 2025-08-05 https://www.nota.ai/ai-solutions-its daily 0.75 2025-08-04 https://www.nota.ai/ai-solutions-dms-fr daily 0.75 2025-08-04 https://www.nota.ai/main daily 1.0 2026-03-31 https://images.squarespace-cdn.com/content/v1/61af5d2d7a632327e6590167/45324a2e-21c6-4bdf-9abc-ac44352d0da3/KakaoTalk_Photo_2024-11-28-22-17-02.png https://www.nota.ai/nva-introduction daily 0.75 2026-03-27 https://www.nota.ai/ew2026 daily 0.75 2026-03-06