spmurrayzzz

9 ms·

61.
by spmurrayzzz 2y ago
I just view it as a relative minor convenience, but it's not some game-changer IMO. The tool use / function calling thing far predates Anthropic releasing the MCP specification and it really wasn't that onerous to do before e
62.
by spmurrayzzz 2y ago
Ah, got it. Yea, then I'd focus on learning how RoPE works first. That will at least help you understand how the retrieval in current long context implementations is so limited. A colleague from a discord I spend time in threw together
63.
by spmurrayzzz 2y ago
Not sure what you mean by summary paper, its a pretty dense topic that assumes a fair amount of prior knowledge of the fundamentals. But maybe the Meta blog post may suffice for that? Otherwise yes there are lots of papers on this and relat
64.
by spmurrayzzz 2y ago
The source file I linked in my initial comment is honestly the most succinct way to understand how this works, but the TL;DR is that there is a NoPE layer interval parameter passed to the transformer block implementation. That defines how f
65.
by spmurrayzzz 2y ago
I'd pump the brakes a bit on the 10M context expectations. Its just another linear attention mechanism with rope scaling [1]. They're doing something similar to what cohere did recently, using a global attention mask and a local c
66.
by spmurrayzzz 2y ago
> This sounds like it is a third party (maybe just a small group of devs?) doing all the work Just as a quantitative side note here — tinygrad has almost 400 contributors, pytorch has almost 4,000. This might seem small, but both project
67.
by spmurrayzzz 2y ago
> But that does not seem to be the strategy, which implies it is not so simple? That is exactly what has been happening [1], and not just in pytorch. Geohot has been very dedicated in working with AMD to upgrade their station in this spa
68.
by spmurrayzzz 2y ago
CUDA isn't the moat people think it is. NVIDIA absolutely has the best dev ergonomics for machine learning, there's no question about that. Their driver is also far more stable than AMD's. But AMD is also improving, they&#x27
69.
by spmurrayzzz 2y ago
I agree with you that the market exists and, as a result, solutions to this problem also exist in abundance. The most difficult part about a building a product like the one presented here is making something super generic that works for a w
70.
by spmurrayzzz 2y ago
GoBack was great. I gotta believe there's a way to manually instrument this using `tmutil` to create incremental APFS snapshots and some middleware code that knows when to wait for for the FS to be idle, but that's handwaving a to
71.
by spmurrayzzz 2y ago
Transformers are usually part of JEPA architectures. In I-JEPA's case, there is a ViT that is used in the context encoding phase.
72.
by spmurrayzzz 2y ago
> i.e there isn't a "I don't have enough information" option. This is true in terms of default mode for LLMs, but there's a fair amount of research dedicated to the idea of training models to signal when they ne
73.
by spmurrayzzz 2y ago
Definitely has been true for my work. LLMs have absolutely have been useful, I even forked an IDE (Zed) to add my own custom copilot to leverage a deeper integration for my work. But even if we consider AI beyond just NLP, there's been
74.
by spmurrayzzz 2y ago
At present it's not possible just via configuration, but you can configure a custom endpoint for both the assistant and inline assistant in settings.json. To get custom tab completions working you need to mimic one of the completion pr
75.
by spmurrayzzz 2y ago
For clarity, this does work with the mac os version of Zed. I use it frequently to work on my GPU nodes. The one tradeoff that is a bit of a smell for me is that the preview version of the feature requires you talk to a centralized broker o
76.
by spmurrayzzz 2y ago
Absolutely. This is already what I've been doing myself. Forked zed so I could use my local rig (4x 3090) to do FIM completions using a finetuned qwen coder 32B. Only barrier for some folks will be if they're not familiar with rus
77.
by spmurrayzzz 2y ago
Historically the problem with using LLMs for the super simple conventional NLP stuff is that they were hard to control in terms of output. If you wanted a one-word answer for a classification task, you'd often have to deal with it resp
78.
by spmurrayzzz 2y ago
Oh I totally agree that I think its not great for the game in terms of balance. As a dev, I do enjoy being able to make my own custom addons, especially as a healer. But I think a push towards less addons in the game will help its casual ac
79.
by spmurrayzzz 2y ago
I'd push back slightly on the idea that top players use minimal addons. During every RWF push, almost every top team has a dedicated resource solely for the purpose of making custom WeakAuras for different parts of each boss fight. Dur
80.
by spmurrayzzz 2y ago
I was playing back then in 2006, you absolutely could. I don't remember any 10-man groups at that time, but people were absolutely doing this with mage-heavy comps with 20 people. And then after Stormcraft Sandbox when all the private
81.
by spmurrayzzz 2y ago
Totally agree with all of these points. I will roll a new toon on an era HC server every now and again when I'm bored with current games, just because it does feel like a new experience for me every time. Trying to figure out the safes
82.
by spmurrayzzz 2y ago
On the other hand, MC is the most "solved" out of all the raids at this point though in terms of mechanics. The most experienced folks can low-man it with 10-20 people using mediocre gear (as long as they have some fire res). And
83.
by spmurrayzzz 2y ago
> People were in disbelief any of this was even possible. Maybe my memory is that bad now in my old(er) age. But at that point in 2015, React had been a thing for about 2 years already. Ember had components even earlier than that in 2011
84.
by spmurrayzzz 2y ago
I haven't seen this sort of work called AI Developer yet, but I may have missed the trend shift. Isn't the convention for this still to use the title of Machine Learning Engineer (MLE)?
85.
by spmurrayzzz 2y ago
They're still very useful on their own. But even more broadly, you can often use them in tandem with LLMs. A good example could be a classifier that's used as a "router" of sorts; could be for selecting a prompt template
86.
by spmurrayzzz 2y ago
I've had similar experiences, but I've noticed my Node.js applications which have few-to-no dependencies behave in the same way as my Go apps in that regard. I might get some deprecation logs from Node letting me know about future
87.
by spmurrayzzz 2y ago
IANAL but my guess that the output seems likely to be considered a "derivative work" (referring to US law in this case). So you'd need prior authorization from the copyright holder in the absence of a license. "Only the
88.
by spmurrayzzz 2y ago
Yes that is correct, but my position (which perhaps has been poorly-articulated) is that in the non-trivial instances, it is a distinction without difference in the greater context of the RAG stack and related pipelines. Just allowing for a
89.
by spmurrayzzz 2y ago
This is how most modern vector dbs work, you usually can store much more than just the raw embeddings (full text, metadata fields, secondary/named vectors, geospatial data, relational fields, etc).
90.
by spmurrayzzz 2y ago
I didn't miss that detail, I just don't think chunking alone is where the complexity lies and that the pgai feature set isn't really differentiated at all from other offerings in that context. My commentary about full documen

More ›