kwindla

6 ms·

1.
by kwindla 5mo ago
We started getting customer pings at 6:15 PT. This is impacting some large ISPs (AT&T) in the US and globally.
2.
Cloudflare failing to resolve .co domains in some regions(cloudflarestatus.com)
1 points by kwindla 5mo ago|1 comments
3.
by kwindla 5mo ago
This is the topic of almost every conversation I'm having with friends who are programmers right now. The compiler analogy (we basically never read compiler output) is the thing I keep thinking about.
4.
by kwindla 7mo ago
Yes, but also ... the analogy to assembly is pretty good. We're moving pretty quickly towards a world where we will almost never read the code. You may read all the assembly that your compiler produces. (Which, awesome! Sounds like you
5.
by kwindla 7mo ago
Happy to answer questions about this (or work with people on further optimizing the open source inference code here). NVIDIA has more inference tooling coming, but it's also fun to hack on the PyTorch/etc stuff they've releas
6.
by kwindla 8mo ago
The Twilio integration made me smile. (For context, the default way to use this is a WebRTC connection from your mobile device to talk to Claude Code, but any network transport supported by Pipecat works.) I'm trying to come up with a
7.
Building voice agents with Nvidia open models(daily.co)
126 points by kwindla 8mo ago|20 comments
8.
by kwindla 9mo ago
I'm a big Gatsby fan and recommend this recent Wesley Morris podcast about the book. https://podcasts.apple.com/us/podcast/our-last-chance-to-tal... Morris has written some of my favorite long-form New York T
9.
by kwindla 9mo ago
It depends on what you mean by replacing. The integrated developer experience is much better on Vapi, etc. The goal of the Pipecat project is to provide state of the art building blocks if you want to control every part of the multimodal, r
10.
by kwindla 9mo ago
One easy way to build voice agents and connect them to Twilio is the Pipecat open source framework. Pipecat supports a wide variety of network transports, including the Twilio MediaStream WebSocket protocol so you don't have to bounce
11.
by kwindla 11mo ago
> Sounds like it. Dude you can be honest here. I'm going to politely weigh in here and say things Sean won't say about himself. You're talking to someone who has spent the last ten years building open source WebRTC softwar
12.
by kwindla 11mo ago
I honestly can't tell if this is trolling. LEGO bricks are pretty new technology, in the scheme of things. The original LEGO company "binding brick" was created in the late 1940s. Of course you don't "need" an
13.
by kwindla 11mo ago
I've done a fair amount of fine-tuning for conversational voice use cases. Smaller models can do a really good job on a few things: routing to bigger models, constrained scenarios (think ordering food items from a specific and known me
14.
by kwindla 11mo ago
I 100% agree with Sean that the computer is an exploration machine. There are lots of net positive things for kids (and non-kids) that LLMs make possible. Just like there were lots of net positive things that an Internet connection makes po
15.
by kwindla 11mo ago
This repo is one possible starting point for tinkering with local agents on macOS. I've got versions of this for NVIDIA platforms but I tend to gravitate to using LLMs that are too big to fit on most NVIDIA consumer cards. https:&#x2F
16.
by kwindla 1y ago
As someone who spends a lot of time looking at timestamped log lines to debug Pipecat pipelines, I'm a big fan of this work from Aleix. In general, I have three pain points with debugging realtime, multi-model, multi-modal AI stuff. 1.
17.
Attention Is Logarithmic(supaiku.com)
2 points by kwindla 1y ago|0 comments
18.
Vibe coding some throwaway image manipulation scripts(github.com)
2 points by kwindla 1y ago|0 comments
19.
by kwindla 1y ago
https://www.slate.auto/en The configurator is fun: https://www.slate.auto/en/personalization
20.
A $20k American-made electric pickup with no paint, no stereo, no screen(theverge.com)
1457 points by kwindla 1y ago|1297 comments
21.
DeepSeek, Huawei, Export Controls, and the Future of the U.S.-China AI Race(csis.org)
2 points by kwindla 2y ago|0 comments
22.
by kwindla 2y ago
In general, for realtime voice AI you don't want this model to support multiple speakers because you have a separate voice input stream for each participant in a session. We're not doing "speaker diarization" from a si
23.
by kwindla 2y ago
I've talked about this a lot with friends. Endpoint detection (and phrase endpointing, and end of utterance) are terms from the academic literature about this, and related, problems. Very few people who are doing "AI Engineering&q
24.
by kwindla 2y ago
A couple of interesting updates today: - 100ms inference using CoreML: https://x.com/maxxrubin_/status/1897864136698347857 - An LSTM model (1/7th the size) trained on a subset of the data: https://
25.
by kwindla 2y ago
It takes about 45 minutes to do the current training run on an L4 GPU with these settings: # Training parameters "learning_rate": 5e-5, "num_epochs": 10, "train_batch_size": 12, "eval
26.
by kwindla 2y ago
Turn detection is deciding when a person has finished talking and expects the other party in a conversation to respond. In this case, the other party in the conversation is an LLM!
27.
by kwindla 2y ago
Can you say more? There's not much open source work in this domain, that I've been able to find. I'm particularly interested in architecture variations, approaches to the classification head design and loss function, etc.
28.
by kwindla 2y ago
580M parameters. More info about the model architecture: https://github.com/pipecat-ai/smart-turn?tab=readme-ov-file#...
29.
Show HN: Open-source, native audio turn detection model(github.com)
126 points by kwindla 2y ago|28 comments
30.
by kwindla 2y ago
The Multimodal Live API is free while the model/API is in preview. My guess is that they will be pretty aggressive with pricing when it's in GA, given the 1.5 Flash multimodal pricing. If you're interested in this stuff, here

More ›