gardnr

20 ms·

271.
by gardnr 3y ago
To me, that graph looks like 2.1 refused a higher precentage than 2.0.
272.
by gardnr 3y ago
They would have to be pretty dumb, or have a compelling sense of urgency, to make such brash and public changes.
273.
by gardnr 3y ago
If you run Lighthouse against that website it will provide actionable suggestions for improving it. More info: https://developer.chrome.com/docs/lighthouse/overview/
274.
by gardnr 3y ago
I really appreciate the writing style: > This pass is surprisingly heavy as it takes about 8.2 milliseconds, or roughly about far too long, ...
275.
by gardnr 3y ago
Never forget, the electric car is here to save the car industry, not the planet.
276.
by gardnr 3y ago
If you are just getting into this tech, like me, then this article leads down the rabbit hole of different index types supported in different vector stores. Most easy-to-use vector stores only use HNSW indices, which is "good enough&qu
277.
by gardnr 3y ago
I'm allergic to medium.com these days.
278.
by gardnr 3y ago
Anyone know how large it is? They state the 1 trillion token dataset is 5TB. Is it safe to assume this is 5TB * 30 = 150TB? The code in the HuggingFace repo downloads data from url base: https://data.together.xyz/redpajama-d
279.
by gardnr 3y ago
This made my day: The implementation is just enough to run very small workloads: * Full GPT2 small (124M parameters) model including byte pair encoding, embeddings, multi-headed attention, and multi-layer perceptron stages * Inference/
280.
by gardnr 3y ago
You are correct! I must have had a read ability problem when they conflated the two issues.
281.
by gardnr 3y ago
This is funny because everyone uses the same tests to evaluate different models. One example is the Hugging Face OpenLLM Leaderboard. [1] The problem that this "paper" is making fun of is that any model can score very well on thes
282.
by gardnr 3y ago
There aren't any special critical evaluation skills required. The arxiv page says: > Comments: 3 pages, satire Tangentially, it's worth pointing out that you are gatekeeping read access to a resource because of a write access p
283.
by gardnr 3y ago
The namespace is overloaded. This 2022 paper decsribes a method for looping LLM generations in a specific way to generate more reasonable decisions and outcomes: "ReAct: Synergizing Reasoning and Acting in Language Models" https
284.
by gardnr 3y ago
I don't think massive is a good word for it. The supporting systems around it are not of the usual size for a storm of this intensity. It's easy to miss it on the wind map but then, all of a sudden, there are 75kt winds right arou
285.
by gardnr 3y ago
They obviously were intent on getting him for something even as they continued to run into dead ends. The body cam footage is infuriating: https://www.kcci.com/article/watch-newton-man-says-police-wr...
286.
by gardnr 3y ago
I left the country. Now, I sleep soundly knowing my taxes get wasted on road works instead of death works.
287.
by gardnr 3y ago
The classic economic benefits of competition: * Drives down price * Enhances product features (I see them competing on VRAM first) * Helps to insulates buyers from supply issues Nvidia has kneecapped their consumer grade hardware to ensure
288.
by gardnr 3y ago
New Zealand had varying degrees of LVT over the course of 100 years. The main takeaway from this thesis[1] is that it is difficult to get all parties on board and that the concessions required prevented the tax policy from being effective.
289.
by gardnr 3y ago
Two important takeaways on the base model: * scored 18.9 on HumanEval (coding) where Llama2 7B scored 12.2 * was trained from the beginning with a 16k context using a modified RoPe where many models are simply fine-tuned using RoPe to gain
290.
by gardnr 3y ago
Right, I quite often charge my Asus laptop with my MagSafe charger.
291.
by gardnr 3y ago
This study shows that metobolic markers after fasting are related to physiological differences in the subjects. The full text can be accessed via https://www.researchgate.net/publication/368542960_Insulin_r... Here is
292.
by gardnr 3y ago
> In fact, if Google follows suit, there will be even more cancelled deals, lost links, and absolutely no new revenues from the legislation given that those are the only companies subject to the law (former Heritage Minister Pablo Rodrig
293.
by gardnr 3y ago
Where does their revenue come from?
294.
by gardnr 3y ago
This looks pretty rad. It is very expressive. After scoffing at your shameless plug I read some of the code and was surprised to find it was not written in perl. Maybe some other options for the acronym could be: * LASI (Language Artificial
295.
by gardnr 3y ago
There are a couple open source projects that expose llama.cpp and gpt4j models via a compatible OpenAI API. This is one of them: https://github.com/lhenault/simpleAI
296.
by gardnr 3y ago
Size: 2.3GB From https://www.linuxliteos.com/download.php#current
297.
by gardnr 3y ago
Frigate was a bit limited in usefulness until they added OpenVINO and TensorRT support in April. Trying to find a Coral TPU these days can be a bit of a challenge.
298.
by gardnr 3y ago
I love how they built a whole project around sharing a single csv file. It's got a docker compose setup.
299.
by gardnr 3y ago
The premium account holders don't see ads in reddit. All Reddit had to do was require premium accounts for users that wanted to use 3rd party apps. If you want to keep using yoru 3rd party app, then sign up and pay or shut up and go ho
300.
by gardnr 3y ago
My guess is that they don't need users to generate content anymore because Reddit's generative bots can seed engaging posts and drive the related conversation.

More ›