Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
iAkashPaul
searching Neon…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
4 ms
·
1.
▲
by
iAkashPaul
2y ago
Could be that or an RNN/LSTM as well
2.
▲
by
iAkashPaul
2y ago
Reversed input strings will the next Enigma
3.
▲
by
iAkashPaul
2y ago
IMO hitting CMD+T thinking it'll open a new tab in my browser but it opens the font window (from Mail) is a user-crime.
4.
▲
Building an open-source full-frame CCD camera PART II (2024) [video]
(youtube.com)
3 points
by
iAkashPaul
2y ago
|
1 comments
5.
▲
by
iAkashPaul
2y ago
A USB MITM board with ESP32 could just connect to your phone & do one-way code/content creation for otherwise software locked devices.
6.
▲
by
iAkashPaul
2y ago
Scaling/cold-boot time is the real concern IMO no the cost
7.
▲
by
iAkashPaul
2y ago
Pretty sure this was never questioned for batched requests, sg-lang/lmdeploy/tensorRT-LLM will have nearly twice as reported speeds with INT8 (fp16 A100 benched here https://github.com/sgl-project/sglang?tab=r
8.
▲
by
iAkashPaul
2y ago
Apart from draw.io I like to use isoflow.io for near isometric diagrams with directional flows as well
9.
▲
by
iAkashPaul
2y ago
Hey Marius, cool project! Would this work with just the middle mouse button click too?
10.
▲
by
iAkashPaul
2y ago
More than this pressing spacebar should have gotten some customisation by now, eg. certain mp4s/markdowns/any other file-extension for quick previews
11.
▲
by
iAkashPaul
2y ago
Can't wait to get this & have TrueNAS+Jellyfin up & running on it, certainly beats having to buy so many addons for the Pi & still not have a super reliable build.
12.
▲
by
iAkashPaul
2y ago
Where did you find OpenAI in my reply?
13.
▲
by
iAkashPaul
2y ago
Thx will check it out
14.
▲
by
iAkashPaul
2y ago
One quick way to improve results greatly is to ask questions with 2/3 chunks & in the lookup for these chunks mention the IDs of the other chunks, qdrant allows for easy metadata addition. So just generate a synthetic question bank
15.
▲
by
iAkashPaul
2y ago
What I wished browsers on Android would do is scroll pages with volume buttons and close pages on double tap or some combination similar to screenshot ring.
16.
▲
by
iAkashPaul
2y ago
Mostly Mistral/Mixtral models in addition to Whisper.
17.
▲
by
iAkashPaul
2y ago
Heavily, Call summarisation at scale, for further insights of key topics/sentiments for diarized & fused channel call data Lead mining quality improvement by asking questions instead of merely classification Internal simple RAG cha
18.
▲
by
iAkashPaul
2y ago
Nice, I'll bench server.cpp as well
19.
▲
by
iAkashPaul
2y ago
How long until someone makes a harvested Apple Silicon logic board work with this!
20.
▲
by
iAkashPaul
2y ago
My point exactly, lot of brigading on this thread
21.
▲
by
iAkashPaul
2y ago
JM2C 1. Langchain not being used in production? > How out of touch is that remark? Hard pressed to find agentic framework implementation outside of Langchain/Llamaindex. 2. Outlines is not expected to work with OpenAI API because it
22.
▲
by
iAkashPaul
2y ago
I remember watching Andy Barry/MIT demo of Pushbroom stereo approach on a flying-wing drone, it used stereo 120FPS cameras to dodge trees & branches. It's almost a decade later now & such hardware is still not cheaply avai
23.
▲
by
iAkashPaul
2y ago
Should have emboldened the ecosystem by providing incentives for web-based government services instead of forcing users to only depend on two primary entities in this space.
24.
▲
by
iAkashPaul
2y ago
INT8/FP8 benchmarks would've been great, both cards could have loaded them with around 60GB VRAM instead of TP=2 on H100.
25.
▲
by
iAkashPaul
2y ago
Not being able to renew domains beyond 10yrs could also be a factor
26.
▲
by
iAkashPaul
2y ago
Yeah heard about these in the latest Geerling video
27.
▲
by
iAkashPaul
2y ago
Off the bat I can see on-device assistants be more capable & allow customized handsfree browser/PDF interaction
28.
▲
by
iAkashPaul
2y ago
Samsung & Transcend SD/mSD have always errored/failed on me with power outages on Pi without additional power management hardware & software. Having push to eject NVMEs brings forth a future where you can swap out LLMs
29.
▲
by
iAkashPaul
2y ago
Summarisation & SDXL run just fine on M1 & above Macs. Probably the video editing features like Google/Pixel might need cloud compute.
30.
▲
by
iAkashPaul
2y ago
Right-clicking to inspect element ain't gonna make it
More ›