Hacker NewsOpenAI and Broadcom unveil LLM-optimized inference chip
https://openai.com/index/openai-broadcom-jalapeno-inference-chip/
#HackerNews #OpenAI #Broadcom #LLM #inference #chip #technology #innovation #AI #hardware
deadprogramyzma v1.18 is out, with the freshest of llama.cpp multimedia features including video input, batching, & more.
Go get it while it's hot!
GitHub - hybridgroup/yzma: Go with your own intelligence - Go applications that directly integrate llama.cpp for local inference using hardware acceleration.
Go with your own intelligence - Go applications that directly integrate llama.cpp for local inference using hardware acceleration. - hybridgroup/yzma
Inference cost at scale with napkin math
https://injuly.in/blog/napkin-inference-cost/index.html
#HackerNews #inference #cost #napkin #math #scaling #machine #learning #data #science
AIntelligenceHubBaseten is close to closing a $1.5B round at a $13B valuation, per WSJ via TechCrunch on June 18. Just five months after a $300M Series E at $5B, the raise is reportedly split-priced and co-led by Spark, Sands, Altimeter, and Wellington. The bet: inference is its own infrastructure category. https://go.aintelligencehub.com/ma-baseten13braise2026 #AI #Inference #Venture #Baseten
Suzzyyy02Baseten just raised $1.5B in AI inference funding, just months after a mega-round. The 'inference gold rush' continues as AI infrastructure becomes crucial for scaling AI applications. #AI #Fintech #Baseten #Inference
Dr Mircea Zloteanu 🌺🌞🍃#statstab #554 Does Regression Produce Representative Estimates of Causal Effects?
Thoughts: Regression creates an “effective sample” that may be unrepresentative of the pop of interest even w/ a representative sample
#ate #causalinference #regression #sample #weights #representative #observational #inference #LATE
sn 🐦⬛Recommended: A new #deepseek v4 toolbox over at https://strix-halo-toolboxes.com/
lets me run a powerful DeepSeek v4 flash Q4 quant LLM on dual Strix Halo 128GB. Give it a try! #strixhalo #localAI #inference
lets me run a powerful DeepSeek v4 flash Q4 quant LLM on dual Strix Halo 128GB. Give it a try! #strixhalo #localAI #inference
MickaiThe Case Against the Inference API
Depending on a remote inference endpoint means depending on someone else's off switch: their uptime, their pricing, their policy changes, their access to your data. This essay makes the case for sovereign on-device inference and a signed, post-quantum, offline-verifiable record you can hold without trusting the vendor.
https://mickai.co.uk/articles/case-against-the-inference-api
#sovereignAI #inference #ondeviceAI #AIsecurity #vendorlockin
The Case Against the Inference API
Depending on a remote inference endpoint means depending on someone else's off switch: their uptime, their pricing, their policy changes, their access to your data. This essay makes the case for sovereign on-device inference and a signed, post-quantum, offline-verifiable record you can hold without trusting the vendor.
#statstab #547 Statistical inference for exploratory data analysis and model diagnostics
Thoughts: A rather odd and provocative article. Taking visual inference to its limit.
#exploratory #eda #plots #Rorschach #inference #simulation #lineups
Andrzej Wąsowski 🟥I have written up a more detailed post about linear regression with #probula using grid approximation. The post focuses on:
- Explaining the example of linear regression in
probula(which is an internal DSL embedded in Scala 3) - Showing how grid approximation is implemented in
probula - Discussing how this implementation is tested and how is it used in testing of other #BayesianInference methods
Enjoy!
https://wasowski.dukla.ch/posts/2606-grid-approximation-with-probula/
#ProbabilisticProgramming #Bayesian #Inference #DataAnalysis #Scala #Scala3 #keep
Simple Univariate Regression with Grid Approximation in Probula
Let me share an update on probula, my small purely-functional Bayesian inference library written in Scala 3. The primary goal for this write-up is to force myself (and you) to think about testing of probabilistic models, of inference algorithms, and the languages or APIs in which they are formulated. Arguably, this is a very modest start. But more is on the way! I would like to start with the first inference scheme you encounter, when reading McElreath’s Statistical Rethinking. Grid approximation (as this is the scheme we speak about) is by far the least efficient of the methods discussed in the book, but it remains useful as a testing baseline. Its simplicity and determinism let it serve as ground truth and oracle for other, more complex inference methods.