Georeactor Blog

Subcategory: ml

RSS Feeds:  (ml)  (all)

NotebookLM vs. human podcast

An AI and human generate a podcast

Quantizing Evo?

AFK Post

books, bio LLMs, random links

Abliterating Refusal and Code LLMs

Continuous Q&A

LLaMA 3 now with Nucleotide BioTokens

Possible insight with a better LoRA

Gene annotation corpora #2

GrainGenes web search and apocrypha

LLaMA 3 and Nucleotide-level Tasks

Trying out new models

Ignore Previous Instructions and 𝑥_3

ML updates this spring

Starting Tiny with Protein LLaMA

Preparing the quinoa ML model

A recipe and a dive into polyploidy

Protein similarity Matryoshka Embeddings

Gene annotation corpora

UniProt

Matryoshka Embeddings?

Search on long sequences

Ignore Previous Instructions and 𝑥_2

ML updates from the road

Revisiting 'Which Wiki Edit'

Catch-up with the SDXL ecosystem

Testing out image gen, LoRA, ControlNet

HyenaDNA, scGPT, and Mamba in CoLab

Alternatives to AgroNT

You Wouldn't Download a Quinoa

AgroNT and plant genomes

Trying out a Plant DNA LLM

AgroNT

Ignore Previous Instructions and 𝑥

Replacing ML Arxiv Haul

Iterating on CSS with GPT-4 and Puppeteer

Adventures in RTL layout

Eye ML - Training Eyegazer model

Cartoon ML - Part 5 - IDEFICS

Multimodal with a new HuggingFace model

Work on Jais and mGPT

More quantized models, and evaluating Indonesian

Savvy Code LLaMa ideas

QLoRA, Code LLaMa tests, and mGPT

Savvy LLM for NYC

GPT-NYC upgrade

Fine-tuning savvy LLMs

Now that LLaMa v2 weights are available with a license agreement, development on Vicuña and other models updated.

An unexpected dive into fundus photography and ML

Can computer vision tell a choroidal nevus from a melanoma?

Froot of the Algorithm

People in the Mandela Effect community are also excited about AI.

Cartoon ML - Part 4 - InstructBLIP describes cartoons

Progress with InstructBLIP model, prompting, multiple choice

Cartoon ML - Part 3 - Get your own Vicuna

Using Salesforce's new InstructBLIP model

Cartoon ML - Part 2 - Progress on Multiple Choice

Finding code examples to train a multiple choice model

Cartoon ML - Part 1 - BLIP

Trying basic image-captioning on the New Yorker cartoon contest dataset.

Semantic search and GPT NYC

Using a embeddings API and vector search DB as an alternative to generation

How does RLHF rank?

Where did RLHF come from, finding some datasets, and hot takes.