more-ane-transformers
pyllms
more-ane-transformers | pyllms | |
---|---|---|
4 | 39 | |
43 | 675 | |
- | 1.8% | |
7.0 | 8.3 | |
7 months ago | 22 days ago | |
Python | Python | |
- | MIT License |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
more-ane-transformers
- M2 Ultra can run 128 streams of Llama 2 7B in parallel
- Is it possible to use ANE(Apple Neural Engine) to run those models?
-
The Coming of Local LLMs
Apple should get working on a version of the Neural Engine that is useful for these models, and remove the 3GB size limit [1] to take full advantage of the 'unified' memory architecture. Game changer.
Waste of die space currently
[1] https://github.com/smpanaro/more-ane-transformers/blob/main/...
- Anthropic’s $5B, 4-year plan to take on OpenAI
pyllms
- The Man Who Killed Google Search
-
Ask HN: Who is hiring? (April 2024)
Kagi | Full & Part Time | Remote | http://kagi.com
Kagi is building a user-centric search engine, free from ads and tracking.
Our primary language is Crystal, and we are always interested in talking to developers who share our values. Some of our roles are listed below, but feel free to reach out even if you don't see a match.
https://help.kagi.com/kagi/company/hiring-kagi.html
You can reach me directly at [email protected]
-
Mojeek
If you're looking for alternative search, I have to mention Kagi (https://kagi.com/). Not free, but totally worth it to filter out results like geeksforgeeks, tutorialspoint, w3schools etc.
- Kagi Search Is Down
-
DuckDuckGo !Bangs
Tip: use bang searches from the browser address bar by setting your default search engine to DuckDuckGo (or https://kagi.com/)
-
What the Google overlords don't want you to see
Shout out to the Kagi search engine. There are no ads. The only incentive is to be good enough to earn your money.
- So I deployed Whoogle on my NAS....
-
Tell HN: I hate contemporary "predictive tile" UI design so much
All 3 of your examples are add-driven free platforms for finding content that the platform didn't produce. Free platforms are incentivized to overwhelm you with options, forcing you to look at everything, including ads. Like a grocery store or Ikea, they don't want you to make a quick in-and-out visit. They want you to look at everything they have in hopes that you'll be impulsive.
You want pretty much the opposite: a tool. Tools let you quickly and efficiently accomplish a task. No distractions. When you're done, you're done.
The paths to salvation that I see:
* Pay for tools that work well when they're available. I can't speak for them myself, but I know people who swear by <https://kagi.com/> for ad-free web searches.*
* Put time and effort into your own tools. For example, you can setup and run your own search engine. Although, if you're willing to put in that effort you might put effort into hijacking and cleaning up the interfaces of free platforms instead.
* Vote and campaign for interoperability laws to enable others to put effort into hijacking and cleaning up bad interfaces.
*Surprisingly, interfaces don't get much better when you pay for video and music streaming services. I think that comes down to how most people use them. Most people don't open Netflix knowing what they want to watch. They open Netflix knowing that they're going want to watch something, and hoping to find something entertaining.
-
DDG founder says Google's phone, manufacturing partnerships thwart competition
I've been using Kagi [0] since a few months and I am extremely surprised how well it works. With DDG it took a few months and then I just added !g everywhere because I never found what I was looking for. With Kagi I almost never do that, every now and then I think I am not able to find what I expect and then I add !g, so far it hasn't given me more then what Kagi gives me.
So, I am wondering how much money you would really need. Because if Kagi can do it, why can't Bing do it?
[0]: https://kagi.com
- Kagi: No ads, fast and personalised results
What are some alternatives?
neural-engine - Everything we actually know about the Apple Neural Engine (ANE)
InternLM - Official release of InternLM2 7B and 20B base and chat models. 200K context support
whisper.coreml - Robust Speech Recognition via Large-Scale Weak Supervision
tinygrad - You like pytorch? You like micrograd? You love tinygrad! ❤️ [Moved to: https://github.com/tinygrad/tinygrad]
chatblade - A CLI Swiss Army Knife for ChatGPT
duckduckgo-locales - Translation files for <a href="https://duckduckgo.com"> </a>
auth - Fully open source, End to End Encrypted alternative to Google Photos and Apple Photos [Moved to: https://github.com/ente-io/ente]
experiments-coreml-ane-distilbert - Experimenting with https://github.com/apple/ml-ane-transformers
sqleton - ☠️ sqleton ☠️ is a CLI tool to execute SQL commands
ollama - Get up and running with Llama 3, Mistral, Gemma, and other large language models.
geppetto - golang GPT3 tooling