home.social

Search

150 results for “jd7h”

  1. We were looking for a local tokenizer for counting the number of input tokens before calling the gemini-embedding-001 endpoint on vertex AI. Turns out this Gemma tokenizer returns exactly the same number of tokens as the usage in the embeddings result `embedding.statistics.token_count` of the Gemini embeddings endpoint. Tested on 2000 datapoints. 😁

    github.com/google/gemma_pytorc

    #gemini #embeddings #gemma #vertexai #genai

  2. We were looking for a local tokenizer for counting the number of input tokens before calling the gemini-embedding-001 endpoint on vertex AI. Turns out this Gemma tokenizer returns exactly the same number of tokens as the usage in the embeddings result `embedding.statistics.token_count` of the Gemini embeddings endpoint. Tested on 2000 datapoints. 😁

    github.com/google/gemma_pytorc

    #gemini #embeddings #gemma #vertexai #genai

  3. Dear opensource developers, if you build ML libraries on top of other ML libraries, please expose the underlying APIs for:
    - loading custom model weights
    - saving model weights
    - saving pipelines
    - exporting safetensors

    This way, you support your users in working with local models, instead of making them extra dependent on external model hosting parties like Google and HuggingFace.

  4. Cards Against Humanity has launched a Black Friday prank: social media site "Yowza". Too bad about the "referral link" spam posts on hashtag , but I do love the brutalist design of the website. And the site "has no hate speech"...because "yowza" seems the only allowed text on the platform.

  5. Dear opensource developers, if you build ML libraries on top of other ML libraries, please expose the underlying APIs for:
    - loading custom model weights
    - saving model weights
    - saving pipelines
    - exporting safetensors

    This way, you support your users in working with local models, instead of making them extra dependent on external model hosting parties like Google and HuggingFace.

    #ml #mlops #ai #generativeai #huggingface #transformers #diffusers #opensource

  6. Dear opensource developers, if you build ML libraries on top of other ML libraries, please expose the underlying APIs for:
    - loading custom model weights
    - saving model weights
    - saving pipelines
    - exporting safetensors

    This way, you support your users in working with local models, instead of making them extra dependent on external model hosting parties like Google and HuggingFace.

    #ml #mlops #ai #generativeai #huggingface #transformers #diffusers #opensource

  7. Dear opensource developers, if you build ML libraries on top of other ML libraries, please expose the underlying APIs for:
    - loading custom model weights
    - saving model weights
    - saving pipelines
    - exporting safetensors

    This way, you support your users in working with local models, instead of making them extra dependent on external model hosting parties like Google and HuggingFace.

    #ml #mlops #ai #generativeai #huggingface #transformers #diffusers #opensource

  8. Dear opensource developers, if you build ML libraries on top of other ML libraries, please expose the underlying APIs for:
    - loading custom model weights
    - saving model weights
    - saving pipelines
    - exporting safetensors

    This way, you support your users in working with local models, instead of making them extra dependent on external model hosting parties like Google and HuggingFace.

    #ml #mlops #ai #generativeai #huggingface #transformers #diffusers #opensource

  9. I am in Eindhoven today for AI Summit Brainport. I’ll be participating in the Startup track. Looking forward to seeing what cool things people are building with generative AI/LLMs. Come say hi if you’re there as well!

  10. "We've now created our first deep learning neural network from scratch. And we did it in Microsoft Excel, everyone's favorite artificial intelligence tool."

    - Jeremy Howard in youtube.com/watch?v=hBBOjCiFcu

    #deeplearning #fastai #jeremyhoward #excel #neuralnetworks

  11. "We've now created our first deep learning neural network from scratch. And we did it in Microsoft Excel, everyone's favorite artificial intelligence tool."

    - Jeremy Howard in youtube.com/watch?v=hBBOjCiFcu

    #deeplearning #fastai #jeremyhoward #excel #neuralnetworks

  12. "We've now created our first deep learning neural network from scratch. And we did it in Microsoft Excel, everyone's favorite artificial intelligence tool."

    - Jeremy Howard in youtube.com/watch?v=hBBOjCiFcu

    #deeplearning #fastai #jeremyhoward #excel #neuralnetworks

  13. "We've now created our first deep learning neural network from scratch. And we did it in Microsoft Excel, everyone's favorite artificial intelligence tool."

    - Jeremy Howard in youtube.com/watch?v=hBBOjCiFcu

    #deeplearning #fastai #jeremyhoward #excel #neuralnetworks

  14. Fitting an LLM on a GPU is a bit like photography. Model weights = film sensitivity, activation size = shutter speed, I/O tensors = aperture. These 3 dials control your model's memory footprint, just as they shape a photo's exposure.

    Just realised this while trying to fit Llama 3.1 on my 24GB GPU with TRT-LLM: nvidia.github.io/TensorRT-LLM/.

    #llms #genai #llama #gpu #nvidia #trtllm #tensorrt