Comments by "nuqwestr" (@nuqwestr) on "Chinese AI app DeepSeek shakes tech industry, wiping half a trillion dollars off Nvidia | DW News" video.

  1. Just nonesene, your Taqiyya is showing.
    15
  2. It's not free, R1 still uses the Silicon Valley large language models to teach the local AI model. It's not "equivalent", not at all.
    10
  3.  @jazpls  Again, just look at NVIDIA stock graph over the last 12-months. I did.
    5
  4. DeepSeek will also run on cheaper NVIDIA chipsets that are around $300.
    5
  5. FALSE: ChapGPT 5 reads the entire book War and Peace, Deepseek R1 steals that information and distills it, just like Cliff Notes. Deepseek is nothing without the LLM teacher. It's a ripoff.
    3
  6. NVIDIA stock is up 106% in last 12-months. It's just fine.
    2
  7.  @AlaMak-ns1up  Yeah, and you just want to troll the issue and not bring anything of value to the discussion.
    2
  8. Don't use just one AI source. But that's correct, China is committing genocide against the Uyghurs and Israel is not.
    1
  9. @ Then why are 1.9 million Arabs alive as Israeli citizens, 20% of the population?
    1
  10. NVIDIA is not the only chipset player in the game. Your chauvinistic comment is comical. R1 still requires LLMs do run its parameters, it's like Cliff Notes distillation, big chips still needed to read the entire book.
    1
  11. Read War and Peace or get the distilled version from Cliff Notes. LLM still required to teach DeepSeek.
    1
  12. Just a reminder, same said about Linux/RedHat back in 1993. Here's what happened: In 2019, Red Hat became a subsidiary of **IBM**. Red Hat is committed to the open source development model, collaborating with communities to create secure, stable, and innovative technologies. Red Hat did not take over, it was subsumed.
    1
  13. Deepseek RI still depends on the proprietary large language models to teach it. Being able to run AI parameters locally on your MacBookPro is cool, but not the game changer most would understand.
    1
  14. NVIDIA is up 106% in 12-month, down 6% in one day but going back up. That's still a huge market win for NVIDIA.
    1
  15.  @meegz149  Here, let me help you, it's in how you engineer the prompt Silicon Valley has been at the forefront of developing several significant large language models (LLMs). Here are some specific examples: GPT (Generative Pre-trained Transformer) Series - Developed by OpenAI, which is based in San Francisco, California. The series includes models like GPT-3 and its subsequent iterations. These models are known for their ability to generate human-like text and have been pivotal in advancing natural language understanding and generation. Llama Series - Meta's Llama 3.1 is another example, although Meta is primarily based in Menlo Park, which is part of Silicon Valley. This model has been highlighted for its performance capabilities, challenging industry leaders and promoting the democratization of AI technology. Mistral Large 2 - While Mistral AI is a French company, the impact of their models, including Mistral Large 2, has been significant in Silicon Valley due to the global nature of tech influence and the shared platform of development and discourse in AI. This model matches or surpasses existing top-tier systems in various benchmarks. DeepSeek's Models - DeepSeek, although a Chinese company, has made waves in Silicon Valley with its open-source models like DeepSeek R1, which has been noted for outperforming some leading American AI models in certain benchmarks. This model has been discussed extensively in Silicon Valley circles for its efficiency and performance. These models represent a cross-section of both proprietary and open-source developments, showcasing Silicon Valley's role in both nurturing new AI technologies and in the discourse around AI ethics, accessibility, and innovation.
    1
  16.  @meegz149  Additionally DeepSeek utilizes large language models (LLMs) from different sources to some extent in their development process. Training Data: DeepSeek models are trained on vast datasets, which include tokens from various sources in both English and Chinese. While the exact composition of these datasets isn't entirely disclosed, it's common practice in the field to aggregate data from multiple public and possibly proprietary sources to enhance the model's capabilities across different languages and contexts. Distilled Variants: DeepSeek has released models that are based on distillation from other existing LLMs. For instance, DeepSeek-R1 models include distilled variants based on Llama and Qwen, indicating they leverage the architectural or weight knowledge from these established models to create their own variations
    1
  17.  @sittingduck02  I don't have enough knowledge or experience to argue the point. I'm still learning, just as the world is.
    1
  18. DeepSeek will run on the NVIDIA RTX 3060 with 12GB of VRAM. $300. What are you talking about?
    1
  19. NVIDIA stock has increased 106% in last 12-months. It did not hammer down to any great degree. Misleading headline.
    1