MACHINE LEARNING

→ View original post on X — @goodside

Reliable Knowledge from Parametric Memory in AI Remains an Open Problem

By

@goodside

–

17 November 2022 6h01

Yeah — getting reliable, truthful knowledge entirely from parametric memory seems like an open problem. The near-term solutions all involve grounding with queries to external resources.

→ View original post on X — @goodside

AI Prompting Techniques and Hallucination Challenges

By

@goodside

–

17 November 2022 5h52

I spent some time trying to solve this problem through CoT/scratchpads but there were often secondary hallucinations, e.g. over-skepticism of non-trick questions. Nothing seemed to work as well as the zero-shot above.

Language Models Continue Sequences from Prompts, Not Maximize Rewards

By

–

17 November 2022 5h28

they don't maximize rewards, they are given a prompt (a kind of inception) and continue the sequence

Extending LLMs to Vision: Incremental Multimodal Integration with Flamingo

By

–

17 November 2022 4h40

Extending LLMs from text to vision will probably take time but, interestingly, can be made incremental. E.g. Flamingo (
https://
storage.googleapis.com/deepmind-media
/DeepMind.com/Blog/tackling-multiple-tasks-with-a-single-visual-language-model/flamingo.pdf
… (pdf)) processes both modalities simultaneously in one LLM.

Why LLMs Process Text Instead of Raw Pixels

By

–

17 November 2022 4h34

Interestingly the native and most general medium of existing infrastructure wrt I/O are screens and keyboard/mouse/touch. But pixels are computationally intractable atm, relatively speaking. So it's faster to adapt (textify/compress) the most useful ones so LLMs can act over them

LLMs as Cognitive Engines Orchestrating Compute Infrastructure via Text

By

–

17 November 2022 4h20

Good post. A lot of interest atm in wiring up LLMs to a wider compute infrastructure via text I/O (e.g. calculator, python interpreter, google search, scratchpads, databases, …). The LLM becomes the "cognitive engine" orchestrating resources, its thought stack trace in raw text

→ View original post on X — @mit_csail

Nonlinear Vector Autoregression Emerging Over Random Reservoir Methods

By

@mit_csail

–

17 November 2022 3h04

me too! I saw a couple of successful applications of them in real-world settings such as diagnoses of signals. I think though the latest developments suggest nonlinear vector autoregression instead of random reservoir which is an interesting direction:

→ View original post on X — @marcusborba

AI Predicts Patient Response to Cancer Treatment

By

@marcusborba

–

17 November 2022 1h59

#ArtificialIntelligence can potentially predict a patient’s response to cancer treatment https://
interestingengineering.com/health/artific
ial-intelligence-predict-response-cancer-treatment
… @IntEngineering #Healthcare #AI #MachineLearning #DataScience #BigData #Analytics #100DaysofCode #IoT #serverless #womenwhocode #DeepLearning #DigitalTransformation

→ View original post on X — @jiqizhixin

DeepMind’s Epistemic Networks Reduce LLM Fine-Tuning Data Requirements

By

@jiqizhixin

–

17 November 2022 0h48

DeepMind’s Epistemic Neural Networks Enable Large Language Model Fine-Tuning With 50% Less Data https://
syncedreview.com/2022/11/16/dee
pminds-epistemic-neural-networks-enable-large-language-model-fine-tuning-with-50-less-data/
…