Hacker News: breadislove

New comment by breadislove in "Asymmetric Quantization: Near-Lossless Retrieval with 97% Storage Reduction"

breadislove — Thu, 02 Jul 2026 16:01:13 +0000

ah whoops, I'll fix it. ty!

New comment by breadislove in "Asymmetric Quantization: Near-Lossless Retrieval with 97% Storage Reduction"

breadislove — Thu, 02 Jul 2026 15:44:13 +0000

The ndcg loss is minimal 90.26 -> 89.65. This means it maintains most of the quality.

New comment by breadislove in "Asymmetric Quantization: Near-Lossless Retrieval with 97% Storage Reduction"

breadislove — Thu, 02 Jul 2026 15:37:28 +0000

to which email did you send it? can u send it to support please?

New comment by breadislove in "Asymmetric Quantization: Near-Lossless Retrieval with 97% Storage Reduction"

breadislove — Thu, 02 Jul 2026 15:36:23 +0000

this is the reason why we report ndcg and not recall. ndcg respects fine grained details so you get the an overview of how much details you are trading off since it would hurt the ranking.

New comment by breadislove in "Asymmetric Quantization: Near-Lossless Retrieval with 97% Storage Reduction"

breadislove — Thu, 02 Jul 2026 15:35:06 +0000

yes exactly.

Asymmetric Quantization: Near-Lossless Retrieval with 97% Storage Reduction

breadislove — Mon, 29 Jun 2026 19:42:15 +0000

Article URL: https://www.mixedbread.com/blog/asymmetric-quant

Comments URL: https://news.ycombinator.com/item?id=48724127

Points: 98

# Comments: 37

New comment by breadislove in "The Token Compression Illusion: Why I'm Skeptical of RTK"

breadislove — Thu, 18 Jun 2026 17:52:09 +0000

slop complaining about other slop

Dense Retrievers Know More Than They Can Express

breadislove — Wed, 03 Jun 2026 16:04:33 +0000

Article URL: https://www.mixedbread.com/blog/latent-terms

Comments URL: https://news.ycombinator.com/item?id=48385900

Points: 2

# Comments: 0

New comment by breadislove in "How we index images for RAG"

breadislove — Wed, 03 Jun 2026 01:26:20 +0000

very bad take. with most modern multomodal models you get way better performance then going to text first

New comment by breadislove in "Show HN: Evidex – AI Clinical Search (RAG over PubMed/OpenAlex and SOAP Notes)"

breadislove — Mon, 29 Dec 2025 22:10:18 +0000

this might be interesting: https://www.theinformation.com/articles/chatgpt-doctors-star...

> $150M RR on just ads, +3x from August. On <1M users.

source: https://x.com/ArfurRock/status/1999618200024076620

New comment by breadislove in "Show HN: Evidex – AI Clinical Search (RAG over PubMed/OpenAlex and SOAP Notes)"

breadislove — Mon, 29 Dec 2025 22:08:55 +0000

a good system (like openevidence) indexes every paper released and semantic search can incredible helpful since the the search api of all those providers are extremely limited in terms of quality.

now you get why those system are not cheap. keeping indexes fresh, maintaining high quality at large scale and being extremely precise is challenging. by having distributed indexes you are at the mercy of the api providers and i can tell you from previous experience that it won't be 'currently accurate'.

for transparency: i am building a search api, so i am biased. but i also build medical retrieval systems for some time.

The whole point of OpenAI's Responses API is to help them hide reasoning traces

breadislove — Tue, 09 Dec 2025 12:57:09 +0000

Article URL: https://www.seangoedecke.com/responses-api/

Comments URL: https://news.ycombinator.com/item?id=46204456

Points: 2

# Comments: 0

Boosting Claude: Faster, Clearer Code Analysis with MGrep

breadislove — Mon, 24 Nov 2025 17:37:28 +0000

Article URL: https://elite-ai-assisted-coding.dev/p/boosting-claude-faster-clearer-code

Comments URL: https://news.ycombinator.com/item?id=46036710

Points: 1

# Comments: 0

Show HN: Mgrep – A Semantic, Multimodal Grep

breadislove — Thu, 20 Nov 2025 21:57:00 +0000

Article URL: https://github.com/mixedbread-ai/mgrep

Comments URL: https://news.ycombinator.com/item?id=45998373

Points: 6

# Comments: 1

New comment by breadislove in "Launch HN: Mosaic (YC W25) – Agentic Video Editing"

breadislove — Thu, 20 Nov 2025 12:09:31 +0000

you should check mixedbread out. we support indexing multimodal data and making data ready for ai. we are adding video and audio support by the end of the year. might be interesting for the OP as well.

we have couple investigative journalists and lawyers using us for a similar usecase.

From Control to Resonance: Why I Let an AI Decode My Voice

breadislove — Mon, 27 Oct 2025 22:26:56 +0000

Article URL: https://substack.com/inbox/post/177015920

Comments URL: https://news.ycombinator.com/item?id=45727065

Points: 2

# Comments: 0

Meta overhauls legacy AI operations

breadislove — Wed, 22 Oct 2025 13:21:45 +0000

Article URL: https://www.axios.com/2025/10/22/meta-superintelligence-tbd-ai-reorg

Comments URL: https://news.ycombinator.com/item?id=45668731

Points: 7

# Comments: 1

New comment by breadislove in "BERT is just a single text diffusion step"

breadislove — Mon, 20 Oct 2025 15:18:21 +0000

or deberta but nevertheless super interesting!

New comment by breadislove in "DeepSeek OCR"

breadislove — Mon, 20 Oct 2025 12:13:29 +0000

For everyone wondering how good this and other benchmarks are:

- the OmniAI benchmark is bad

- Instead check OmniDocBench[1] out

- Mistral OCR is far far behind most Open Source OCR models and even further behind then Gemini

- End to End OCR is still extremely tricky

- composed pipelines work better (layout detection -> reading order -> OCR every element)

- complex table parsing is still extremely difficult

[1]: https://github.com/opendatalab/OmniDocBench

New comment by breadislove in "Migrating from AWS to Hetzner"

breadislove — Fri, 17 Oct 2025 14:44:08 +0000

we have extremely processing heavy jobs where user upload large collection of files (audios, pdfs, videos etc.) and expect to get fast processing. its just that we need to fan out sometimes, since a lot of our users a sensitive to processing times.