khurdula

3 ms·

1.
Show HN: The First Open Source Diffusion ASR Audio Model 15x Faster Than Whisper(arxiv.org)
4 points by khurdula 2mo ago|0 comments
2.
by khurdula 5mo ago
We've open-sourced all code, and test sets. You can find them here: https://interfaze.ai/blog/introducing-structured-output-benc... To validate the choices and configurations, feel free to give it a reading. We al
3.
by khurdula 5mo ago
We've added opus 4.6 and 4.7 to our leaderboard, they perform very closely with sonnet 4.6. Feel free to checkout our updated blog again :D
4.
by khurdula 5mo ago
hey! we've evaluated gpt 5.5 as well along with other frontier models. gemini and gemma models outperform it across all three modalities. Open source models like glm 4.7 still compete closely with table toppers.
5.
by khurdula 5mo ago
We've updated our leaderboard having evaluated frontier models gemini 3.1 pro, opus 4.6 & 4.7, glm 5.1, deepseek v4, Kimi K2.6 as well.
6.
by khurdula 5mo ago
We're updating our leaderboard with these model scores, should be out soon :D
7.
by khurdula 5mo ago
We do love Qwen! It can be an easy choice when confused looking at this leaderboard.
8.
by khurdula 5mo ago
Yep, we will be adding it soon as well.
9.
by khurdula 5mo ago
Due to high demand, we're adding it soon!
10.
by khurdula 5mo ago
General hallucinations benchmarks tend to be knowledge specific like GPQA or MMLU but none specifically measure structured output end-to-end which is one of the biggest use case for LLMs. Many developer workflows use LLMs to produce structu
11.
by khurdula 5mo ago
We saw that structured decoding didn't make a difference in the quality of the output. Check out the paper section "6.3 Structured Decoding Ablation" Paper: https://arxiv.org/pdf/2604.25359 We ran the co
12.
by khurdula 5mo ago
Check out the "The JSON-pass vs Value-Accuracy gap" section in the blog. That was an eye opener. While most models were great at producing JSON schema, they were pretty bad at producing accurate values. In the graph you'll is
13.
by khurdula 5mo ago
Yeah we selected models that are most commonly integrated in developer workflows and being used for structured output. Typically those models tend to be in the low -mid cost range and with no or low reasoning. For the benchmark, was kept co
14.
Show HN: A new benchmark for testing LLMs for deterministic outputs(interfaze.ai)
60 points by khurdula 5mo ago|30 comments
15.
by khurdula 6mo ago
"we hope to open-source future versions of the model." Love to see it. Cheers!
16.
by khurdula 6mo ago
We define determinism as a model behaving predictably, while also producing useful supporting metadata, like confidence scores from specialized DNNs/CNNs, not just text tokens generated as "scores". So for the same kind of ta
17.
Show HN: A new model architecture because transformers are not enough(interfaze.ai)
3 points by khurdula 6mo ago|2 comments
18.
by khurdula 1y ago
Bruh, if it were priced at like $2,499 it would make sense, but this is just too much.
19.
Show HN: Yoslm -- You Only Need a Smoll Language Model for Object Detection(jigsawstack.com)
3 points by khurdula 1y ago|0 comments
20.
by khurdula 1y ago
Damn, just visiting this site makes me want to reinstall Minecraft haha.
21.
by khurdula 2y ago
What if I said, we outperform them? Check this out: https://jigsawstack.com/blog/openai-audio-stt-vs-jigsawstack...
22.
by khurdula 2y ago
Are we supposed to use AMD GPUs for this to work? Or Does it work on any GPU?