jiehong

9 ms·

61.
Nemotron 3.5 ASR for Your Language(huggingface.co)
2 points by jiehong 5mo ago|0 comments
62.
by jiehong 5mo ago
Apple's own dictation is quite limited, doesn't handle multiple languages very well, and many open source dictation models simply do better.
63.
by jiehong 5mo ago
Exactly. On iOS, it completely limits the market for a good dictation app with your keyboard, because iOS just doesn't allow you to.
64.
by jiehong 5mo ago
Fair point, thanks for changing my point of view :)
65.
by jiehong 5mo ago
Video games are made to entertain humans, so does it really matter whether LLMs are good at playing them?
66.
by jiehong 5mo ago
git merge conflicts?
67.
by jiehong 5mo ago
Dagger seems more fit, but more complex for sure.
68.
by jiehong 5mo ago
It's not that easy. LSP does not have a standard way to expose things not specified. So, at best you need an LSP server exposing more + the LSP client to also understand it [2]. The best that comes to mind is nvim-jdtls [0]. Maybe at s
69.
by jiehong 5mo ago
Jetbrains’ good deterministic refactoring tools are what I like from them (and debugging). Other than that, I must agree with this article.
70.
by jiehong 5mo ago
HarmonyOS from Huawei is no longer based on Android, but it’s not an open OS.
71.
by jiehong 5mo ago
Now, with vim/helix key bindings: 1. select text 2. dictate action Feels very similar to Helix's select text and act on it. I think text selection could also be voice controlled (with a modal voice input), so one could say: "
72.
by jiehong 5mo ago
transcription gets post-processed by a LLM (with different styles, like based on prompts, so that it removes fillers, homophones, change the style, etc. I recommend the youtube channel @afadingthought to see what people come up with (like v
73.
by jiehong 5mo ago
It's a good way to ensure that people outside of Germany pentest German sites instead :D
74.
by jiehong 5mo ago
Thanks for taking the time to provide this info. I appreciate it
75.
by jiehong 5mo ago
This made me think of a potential workaround that would have a better UI: create a directory (maybe named 'Trust Zone'), and have it watched by something like automator. As soon as you move a file in there, it reacts and remove th
76.
by jiehong 5mo ago
This page made me discover the tool, and I find sdocs.dev pretty cool!
77.
by jiehong 5mo ago
This might be inherent to how the models are benchmarked. Aren’t some benchmarks giving the model multiple shots at a problem and only keep the successful result if it appeared, ignoring the failure rate?
78.
by jiehong 5mo ago
This reminds me that agents sometimes really like heredoc in shells, and waste tokens retrying with a file.
79.
by jiehong 5mo ago
Using it in Kagi Assistant is stupidly slow. I get like 10 t/s. While it’s pretty fast in the official app for example. Kagi Assistant is also kind of broken when using Qwen 3.6 Plus. So, beware of using them in Kagi at the moment.
80.
by jiehong 5mo ago
I think nowadays a lot of models are trained more at doing this than at knowing things, while being smaller. So I’d say yes! At least that’s my impression.
81.
by jiehong 5mo ago
Looks like the GPT‑Realtime‑Whisper model isn’t open weight like the old whisper model. Too bad! However, OpenAI had and still has a true lead on voice model interactions. That’s where Chinese AI companies don’t do as well: deepseek doesn’t
82.
by jiehong 5mo ago
CDNs exist, as well as http cache headers, though.
83.
by jiehong 6mo ago
this makes me wonder if there is an `infinite` version of QR Code that can be streamed from a printer (well, I guess things can be chunked).
84.
by jiehong 6mo ago
I once sent a letter to a friend with a hand-written QR Code whose content was the typed letter. It's cool for the receiver, but tedious for the sender (but it's a good way to help with difficult to read cursive).
85.
by jiehong 6mo ago
I can see why people try this way: graalvm is limited outside the entreprise version, and it’s quite slow to compile with. I think the aot compilation story on the JVM lacks fast tooling with good UX compared to go.
86.
by jiehong 6mo ago
Congratulations! However, the screenshots for "List markers in RTL text" are the same it seems. The list markers are on the left in both cases.
87.
by jiehong 6mo ago
Anyone tried both duckdb and clickhouse local? Clickhouse seems less marketed, but seems quite similar.
88.
by jiehong 6mo ago
The cli only produces vega-lite[0] json graphics, right? It would be nice if it included a rendering engine. [0]: https://github.com/vega/vega-lite
89.
by jiehong 6mo ago
Outstanding! This can replace a lot of Excel in the end. It makes so much sense now that it exists!
90.
by jiehong 6mo ago
Am I misunderstanding, or isn't this supposed to be the point of MCP?

More ›