LLMS

→ View original post on X — @miketamir

How to Train Your Own Large Language Models

By

@miketamir

–

11 June 2023 17h37

How to train your own Large Language Models https://
bit.ly/3WnauuY #AI #MachineLearning #DeepLearning #LLMs #DataScience

→ View original post on X — @ceobillionaire

Video-LLaMA: Instruction-tuned Audio-Visual Language Model

By

@ceobillionaire

–

11 June 2023 17h32

-LLaMA: An Instruction-tuned Audio-Visual Language Model for Understanding Hang Zhang, Xin Li, Lidong Bing: https://
arxiv.org/abs/2306.02858 #DeepLearning #LargeLanguageModels #LLM

→ View original post on X — @avikumart_

5 AI Tools to Maximize Data Science Productivity

By

@avikumart_

–

11 June 2023 16h30

5 tools to use to maximize your data science productivity 1) ChatGPT by OpenAI
2) Bard AI
3) Generate Presentations – @BeautifulAI_ 4) Meeting summarizer – @tldview 5) Generate synthetic data using Gen. AI – @gretel_ai

→ View original post on X — @nathanbenaich

Does One AI Model Rule Them All? Expert Analysis

By

@nathanbenaich

–

11 June 2023 16h18

Along the lines of @maithra_raghu
’s “does one model rule them all? https://
maithraraghu.com/blog/2023/does
-one-model-rule-them-all/
…

→ View original post on X — @nathanbenaich

Gorilla: Open Source Base Model Outperforms Larger AI Models

By

@nathanbenaich

–

11 June 2023 16h17

The Gorilla project is a great example of how open source base model + fine tuning through self-instruction instruction + document retrieval can outperform a larger model for an economically-useful task.

13B Model Learns to Imitate GPT-4 Reasoning Process

By

–

11 June 2023 15h57

10/ Imitating Reasoning Process of LLMs – develops a 13B model that learns to imitate the reasoning process of large foundational models like GPT-4; it leverages large-scale & diverse imitation data and surpasses Vicuna-13B.

ChatGPT Humor Analysis Reveals Overfitting to 25 Jokes

By

–

11 June 2023 15h57

9/ Humor in ChatGPT – explores ChatGPT’s capabilities to grasp and reproduce humor; finds that over 90% of 1008 generated jokes were the same 25 jokes and that ChatGPT is also overfitted to a particular joke structure.

LEACE Method Erases Gender Bias from BERT Embeddings

By

–

11 June 2023 15h57

6/ Concept Scrubbing in LLM – presents a method called LEAst-squares Concept Erasure (LEACE) to erase target concept information from every layer in a neural network; it’s used for reducing gender bias in BERT embeddings.

Fine-Grained RLHF Improves LM Training and Safety

By

–

11 June 2023 15h57

7/ Fine-Grained RLHF – trains LMs with fine-grained human feedback; instead of using overall preference, more explicit feedback is provided at the segment level which helps to improve efficacy on long-form question answering and reduces toxicity.