ixaxaar

I think that would fit idiot savant logic very well.

Please check out https://github.com/geniusrise - tool for running llms and other stuff, behaves like docker compose, works with whatever is supported by underlying engines:

Huggingface - MPS, cuda VLLM - cuda, ROCm llama.cpp, whisper.cpp - cuda, mps, rocm

Also coming up integration with spark (TorchDistributor), kafka and airflow.

However if this is really the biological analogue of credit assignment, this might scale better than training llms from scratch every time. Even if say it could approx gradients to a certain degree given a new network, normal backprop could further tune for a few epochs or so dramatically reducing overall training costs.

Wellll that means I can give dni another try :D

https://github.com/ixaxaar/pytorch-dni

The concept here goes a bit further and tries to replicate backprop with an external network, arguing that that's probably what the brain actually does.

Groww and upstox are total crap as well. But who cares, majority of the country operates with portfolio sizes of USD 100.

Larger accounts are better off with traditional brokers who offer personal relationships than some app made by a great tech team.

Why not be specific, and consistent, like Russia?

Another take: is real true value creation only possible outside of capitalism? Hence those communist pockets like academia and practices like treating everything as public property are allowed to exist, otherwise there would be no innovation etc just a bunch of companies suing each other?

This kind of world arises if an org has, say 1000 engineers, or ~100-200 teams, and every team wants to do their own thing with accountability improperly mapped etc.

And infrastructure engineers turn into police with an obvious overemphasis on observability.

I guess a large amount of complexity in infra and wastage is done to hide human inefficiencies.

It can be used to perform various tasks, including hosting inference endpoints, performing bulk inference, fine tune etc with open source models or closed source APIs.

- The framework provides structure for modules and operationalizes and orchestrates them. - The modular ecosystem provides a layer of abstraction over the myriad of models, libraries, tools, parameters and optimizations underlying the operationalization of modern AI models.

1. Install geniusrise and libs

    pip install torch
    pip install geniusrise
    pip install geniusrise-vision # vision multi-modal models
    pip install geniusrise-text # text models, LLMs
    pip install geniusrise-audio # audio models

    version: '1'
    
    bolts:
        my_multimodal_api:
            name: VisualQAAPI
            state:
                type: none
            input:
                type: batch
                args:
                    input_folder: ./input
            output:
                type: batch
                args:
                    output_folder: ./output
            method: listen
            args:
                model_name: 'llava-hf/bakLlava-v1-hf'
                model_class: 'LlavaForConditionalGeneration'
                processor_class: 'AutoProcessor'
                device_map: 'cuda:0'
                use_cuda: True
                precision: 'bfloat16'
                quantization: 0
                max_memory: None
                torchscript: False
                compile: False
                flash_attention: False
                better_transformers: False
                endpoint: '*'
                port: 3000
                cors_domain: 'http://localhost:3000'
                username: 'user'
                password: 'password'

    genius rise


    MY_IMAGE=/path/to/test/image

    (base64 -w 0 $MY_IMAGE | awk '{print "{\"image_base64\": \""$0"\", \"question\": \"<image>\nUSER: Whats the content of the image?\nASSISTANT:\", \"do_sample\": false, \"max_new_tokens\": 128}"}' > /tmp/image_payload.json)
    curl -X POST http://localhost:3000/api/v1/answer_question \
        -H "Content-Type: application/json" \
        -u user:password \
        -d @/tmp/image_payload.json | jq

Individual Registered Users - 118,911,572

No. of returns filed (AY 23-24) - 82,194,557

I've hired a few devs from intuit before and they tell me the Indian portal is far more usable than whatever they were building.

BTW these books were up on amazon India for like $10-$15 each (books with slight printing defects were being dumped in the 3rd world) so I lapped up the entire collection :P

Consider complexity economics, computational social sciences, network sciences and cascading networks, evolutionary theories, parts of systems biology, connectomics and computational neuroscience etc.

The Santa Fe institute has been at the forefront and has an amazing collections of publications in their own press. David krakaeur's book "Worlds Hidden in Plain Sight: The Evolving Idea of Complexity at the Santa Fe Institute" is like a 20 year survey paper and an amazing read.

I know for a fact that most fund managers (at least in India) have been converting as much as 10-20% of their portfolios into gold and other commodities.

With the growth of lean, I suppose its time to get more serious in popularizing and attracting community instead of trying to pursue harmless chuckles.

Edit:

And I immediately get labeled as a "tranny protector". I'm not from the western world and we do not have the same cultural issues.

What I meant is, in a world where English nevertheless dominates as a common language by far, perhaps naming matters in the English context much more than any other language, particularly when building something for everyone.

Edit Edit:

I mean I had no idea there were such strong feelings of what i suppose is being perceived as American/English cultural interference that's felt so strongly by the French.

Maybe that's what Africans feel who were colonised by French Coqs (national animal)?

I now completely agree it is a brilliant idea to name something "cunt" in honour of Louvre Cuntadoodleur. :D

Almost all, if not all, firmware is in C.

Also, not to mention every operating system.

I really don't know where the "bugs are horrible" argument grounds itself on. I've worked on large C, rust, go, python and even Haskell codebases. Everyone of them had their share of super corner case bugs.

The systems which use the simplest concepts and tools tend to have the least bugs cause they are more readable and hence fixable.

Then came Linux for you, and made me try my first Linux distro - xandros 2003. Then there was the complete 3 DVD Debian distro on LFY made me create an openmpi HPC in college among so many other things completely ignored by the formal education system.

I think there is a real value to limited, quality information like these magazines that the internet ruined. Devs these days have unlimited information at their fingertips - the internet and now chatgpt. But now the onus is on us - are we asking the right questions?