fzysingularity

5 ms·

1.
by fzysingularity 5d ago
text is mostly beaten to death, so you can expect good defaults to work for vllm. VLMs specifically are quite sensitive to quantization, especially if you want it to do fine-grained localization (time or spatial), and the vllm default param
2.
by fzysingularity 5d ago
The author’s comments on vision providers is especially interesting. We saw that most providers don’t provide native video url support, have high-variability in vision performance (likely due to the fact that they’re serving different quant
3.
by fzysingularity 12d ago
Have you tried any vision tasks with this model? We've been serving these on our gateway [1], and the quants are quite terrible for vision. Curious to hear your experience. [1] https://www.vlm.run/gateway
4.
by fzysingularity 14d ago
I'm surprised this post isn't getting as much attention as it should. Crazy times!
5.
by fzysingularity 14d ago
VLM Run ( https://vlm.run ) | 1x Founding Infrastructure Engineer We’re building the inference platform for visual intelligence. We’re a deeply technical team of veteran AI / computer-vision engineers (20+ years combined, MIT
6.
Run GLM-OCR, DeepSeek-OCR-2, Dots.mocr with an OpenAI Compatible API(vlm.run)
6 points by fzysingularity 28d ago|1 comments
7.
by fzysingularity 2mo ago
I think we all ought to look at the ZDR fine-print here. I get that in principle that there's no retention, but these are powerful models that can comprehend, paraphrase and summarize your logs for the sake of "product" impro
8.
by fzysingularity 2mo ago
The ICP question was more around the model itself. Are they looking to license it to robotics companies? Do they imagine that devs at robotics companies would be willing to deploy these models as a black box?
9.
by fzysingularity 2mo ago
It’s unclear to me what their desired outcome for a blog post like this. If you’ve ever worked in a robotics setting, 80% implies that 20% of your autonomous actions are incorrect. Imagine if this were the case for autonomous driving where
10.
by fzysingularity 2mo ago
Frontier labs are realizing that software/models themselves don’t have real moats and move to embodied ai. SOTA 80% means a practically useless robot. What are they really imagining their ICP to be here?
11.
by fzysingularity 2mo ago
It's live now, https://github.com/vlm-run/mm .
12.
by fzysingularity 2mo ago
We were planning to open-source this soon, but jumped the gun and posted about the video encoders here since it seemed relevant. In either case, here you go, it's public now: https://github.com/vlm-run/mm .
13.
by fzysingularity 3mo ago
Exactly! We experimented with a whole bunch of video encoding techniques for LLMs here: https://vlm-run.github.io/mm/encoders/#video
14.
by fzysingularity 3mo ago
Pretty terribly expensive way to watch a video with Claude. Use Gemini or some local VLM to do this way more efficiently. We spent quite a bit of time on video understanding, and Claude will just burn tokens. Check out this library: https:
15.
by fzysingularity 3mo ago
This is neat. I'd love to figure out a sequence of emojis that triggers the LLM in ways that puzzles a human.
16.
Omnigent: Meta-Harness for Coding Agents (Claude Code, Codex, Cursor, Pi)(github.com)
2 points by fzysingularity 3mo ago|0 comments
17.
by fzysingularity 3mo ago
That’s a pretty large binary for simply loading images. In all honesty, opencv has stood the test of time and I’m certain newer LLMs will likely not attempt to rewrite it from scratch. P.S. I’ve been a user since the IplImage days, circa 20
18.
by fzysingularity 3mo ago
I can’t help but think that there are so many astroturfed comments in here. Seems like a concerted and distributed effort from the entire Anthropic team every time to get this on top of HN.
19.
by fzysingularity 3mo ago
Kind of crazy how many bespoke python sandbox implementations have popped up in the past few months. I’d love to see if we can get GPU access within these runtimes, that’d be awesome.
20.
by fzysingularity 3mo ago
What’s your experience with Monty? Been looking at it for one of our environments and it seems very promising.
21.
by fzysingularity 5mo ago
VLM Run ( https://vlm.run ) | 1x Product + 1x ML Staff Engineer | Santa Clara, CA (HQ) We're building the inference and orchestration layer for production Vision-Language Models. We care deeply about fast and ergonomic visual
22.
by fzysingularity 6mo ago
The recent claude code leak also revealed that they're poisoning their competitors via anti-distillation policies baked in claude code CLI (fake tool calls, adding noise etc).
23.
by fzysingularity 6mo ago
VLM Run ( https://vlm.run ) | 1x Infrastructure Engineer + 2x AI/ML Engineer | Santa Clara, CA (HQ) VLM Run is building infrastructure for production Vision-Language Model (VLM) systems — fast inference, tool-use + orchestrat
24.
by fzysingularity 6mo ago
Real-time or continuous learning is great on paper, but to get this to work without extremely expensive regression testing and catastrophic forgetting is a real challenge. Credit to the team for taking this on, but I’d be skeptical of annou
25.
by fzysingularity 6mo ago
What do you think actually happened here in the past week? They used Kimi, failed to acknowledge it in the original Composer announcement. Kimi team probably reached out and asked WTF? Their only recourse was to publicly disclose their whit
26.
by fzysingularity 7mo ago
VLM Run ( https://vlm.run ) | 1x Infrastructure Engineer + 2x AI/ML Engineer | Santa Clara, CA (HQ) VLM Run is building infrastructure for production Vision-Language Model (VLM) systems — fast inference, tool-use + orchestrat
27.
by fzysingularity 7mo ago
AI allows you to accelerate the initial build process, but I think engineering is all about craftsmanship. Today most LLMs have poor taste and chipping away the cruft matters more than ever.
28.
by fzysingularity 7mo ago
uvx probably is the way to go here (fully self-contained environment for each skill), and use stdout as the I/O bridge between skills.
29.
by fzysingularity 7mo ago
The cold-boot time on this model can hardly be called “serverless”
30.
by fzysingularity 7mo ago
ELO scores for OCR don't really make much sense - it's trying to reduce accuracy to a single voting score without any real quality-control on the reviewer/judge. I think a more accurate reflection of the current state of comp

More ›