ONNX optimized SigLIP and related foundation models

Hacker News

Hey there Hacker News! Long time lurker but first time posting so nice to meet you all! I have some time on my hands so I got down a rabbit hole and created a project to keep track of, and unify a bunch of optimized foundation models under one easy to use Python package. So far I have the following up and running:- SigLIP as a FP16 ONNX representation for super fast zero-shot classification image classification - quantized model support is on it’s way Automatic pre and post processing switching - choosing CLIP as a model type falls back to cosine similarity with softmax, where SigLIP falls back to its full graph with a Scipy based sigmoid output activation. Manual mode with exposed image and text encoders for each model - SigLIP also has it’s pre-pooled hidden output available for analysis etc An ONNX Segment Anything representation - the plan is to have each CLIP/SigLIP model’s salient map feed directly as a multi-point prompt to SAM and its variants. More on this soon!The plan is to have the same usage wrap a TensorRT backend along the road, and tie that into Chromadb for super fast search but for now check out the example Gradio app. It’s still a little clunky but the results are pretty impressive! I’d imagine this would be great for lightweight RAG too!

Visit product View on Hacker NewsJun 21, 2024

Share card

Actual performance

2points

Did not reach leaderboard

Launch Intel predictions

Analyze your own launch →

Indie HackersFits the IH revenue-focused audience · Strong signals: created · Missing: supports, reddit linkedin, podcasting

86%86% predicted probability of success on Indie Hackers, based on ML models trained on real launch data.

best fitHighest predicted score across all platforms for this description.

Product HuntOn track for Day 1 leaderboard · Strong signals: model, new, models · Missing: mac, agents, macos

85%85% predicted probability of success on Product Hunt, based on ML models trained on real launch data.

Hacker NewsStrong engagement from HN community · Strong signals: hacker news, io · Missing: https docs, excited, just released

58%58% predicted probability of success on Hacker News, based on ML models trained on real launch data.

nativeThis product was originally launched on this platform.

AppSumoStrong fit for a featured deal · Strong signals: soon · Missing: plus, platform, intuitive

51%51% predicted probability of success on AppSumo, based on ML models trained on real launch data.

TrustMRRLess likely to generate early MRR · Strong signals: way · Missing: mobile apps, ios, personal

43%43% predicted probability of success on TrustMRR, based on ML models trained on real launch data.

Acquire.comPre-revenue stage for this audience · Missing: arr, mrr, revenue

18%18% predicted probability of success on Acquire.com, based on ML models trained on real launch data.

BetaListMay not resonate with beta-testers · Missing: web3, chat, crypto

0%0% predicted probability of success on BetaList, based on ML models trained on real launch data.

Incorrect prediction on native model

Similar products

OpenIntelligence31%Launch Intel prediction score: how likely this product is to succeed on its source platform, based on its name, tagline, and description.

Apple Foundation Models RAG Engine

Indie Hackerscommitment-side-project

Foundation models for time series forecasting72%Launch Intel prediction score: how likely this product is to succeed on its source platform, based on its name, tagline, and description.

Foundation models for time series forecasting

Hacker News5

MindHalo – macOS study assistant using on-device Foundation Models41%Launch Intel prediction score: how likely this product is to succeed on its source platform, based on its name, tagline, and description.

MindHalo – macOS study assistant using on-device Foundation Models

Hacker News1

RelativeDB – OSS query engine for relational foundation models65%Launch Intel prediction score: how likely this product is to succeed on its source platform, based on its name, tagline, and description.

RelativeDB – OSS query engine for relational foundation models

Hacker News3

Autodistill – Use big slow foundation models to train small fast models47%Launch Intel prediction score: how likely this product is to succeed on its source platform, based on its name, tagline, and description.

Autodistill – Use big slow foundation models to train small fast models

Hacker News18

Shoonya AI74%Launch Intel prediction score: how likely this product is to succeed on its source platform, based on its name, tagline, and description.

Specialized foundation models fine-tuned for commerce use

Product Hunt+258API

Amazon Nova75%Launch Intel prediction score: how likely this product is to succeed on its source platform, based on its name, tagline, and description.

Amazon's new generation of foundation models

Product Hunt+162Artificial Intelligence

Foundation models that predict patient response in clinical trials55%Launch Intel prediction score: how likely this product is to succeed on its source platform, based on its name, tagline, and description.

Foundation models that predict patient response in clinical trials

Hacker News2

Qwen Chat72%Launch Intel prediction score: how likely this product is to succeed on its source platform, based on its name, tagline, and description.

Open foundation models from Alibaba cloud

Product Hunt+15

Foundation Models framework85%Launch Intel prediction score: how likely this product is to succeed on its source platform, based on its name, tagline, and description.

Build with Apple's on-device AI, now open to developers

Product Hunt+176Privacy

Browse all products Analyze your launch