OPEN SOURCE - AI Dynamics

Google’s Soundstorm Open-Source AI Speech Model Released

By

–

27 May 2023 22h32

Open-source project Soundstorm (AI generated speech from Google Research) is going to give Elevenlabs a run for it's money:

The text-to-speech project specializes in dialogue between multiple parties, and is available on Github: pic.twitter.com/nT7PoTPk8m
— AI Breakfast (@AiBreakfast) 27 mai 2023

Open-source project Soundstorm (AI generated speech from Google Research) is going to give Elevenlabs a run for it's money: The text-to-speech project specializes in dialogue between multiple parties, and is available on Github:

→ View original post on X — @aibreakfast

27 May 2023

Run Alpaca-LoRA Efficiently on Low-End Devices

By

@marktabnet

–

27 May 2023 22h10

Learn How to Run Alpaca-LoRA on Your Device in Just a Few Steps – KDnuggets Alpaca-LoRA is a smaller version of Stanford Alpaca that consumes less power and can able to run on low-end devices like Raspberry Pi. https://
bit.ly/42ZSCsL

→ View original post on X — @marktabnet

27 May 2023

4-bit Llama Fine-tuning with Alpaca Dataset Performance Analysis

By

@waitin4agi_

–

27 May 2023 17h52

I trained a 4bit on a Llama base with alpaca dataset today. Reasoning was worse compared to traditional alpaca.

→ View original post on X — @waitin4agi_

27 May 2023

Guanaco 4-bit Training Fails General Reasoning Test vs Vicuna

By

@waitin4agi_

–

27 May 2023 9h34

A lot of fanfare has been created about Guanaco and 4 bit training. It however fails my general reasoning test, unlike Vicuna-13b.

→ View original post on X — @waitin4agi_

27 May 2023

Closing the capabilities gap between open and closed language models

By

@nathanbenaich

–

27 May 2023 0h06

Build better base models. “there exists a substantial capabilities gap between open and closed LMs that, with current methods, can only be bridged using an unwieldy amount of imitation data or by using more capable base LMs.”

→ View original post on X — @nathanbenaich

27 May 2023

LLaMa Evaluation Discrepancy: ElutherAI Harness Results Analysis

By

@thom_wolf

–

26 May 2023 23h48

Here is a good dive in on LLaMa underperforming on the EleutherAI harness versus the published number (TLDR is that we don't know yet which prompt they used for evaluation):

→ View original post on X — @thom_wolf

26 May 2023

EleutherAI Harness: Running Evaluation Framework

By

@thom_wolf

–

26 May 2023 23h39

This is just running the EleutherAI harness

→ View original post on X — @thom_wolf

26 May 2023

Databricks Dolly Model Serving Integration Now Available

By

@hwchase17

–

26 May 2023 17h51

Ever since @databricks announced Dolly I was hoping to have an integration with their model serving endpoint! Thanks to Xiangrui Meng and Gengliang Wang!

→ View original post on X — @hwchase17

26 May 2023

Open Source AI Leaders Challenge Tech Giants Secrecy

By

@tiernanraytech

–

26 May 2023 17h06

Open source AI holds a secret party in New York http://
Stability.ai founder Emad Mostaque and http://
Lightning.ai CEO William Falcon made a pitch for defying tech giants’ secrecy about AI source code. https://
zdnet.com/article/open-s
ource-ai-holds-a-party-in-new-york-thanks-to-stability-ai-and-lightning-ai/
… @StabilityAI @LightningAI

→ View original post on X — @tiernanraytech

26 May 2023

Falcon 40B Dethrones LLaMa on Open Leaderboard

By

@thom_wolf

–

26 May 2023 15h16

LLaMa is dethroned A brand new LLM is topping the Open Leaderboard: Falcon 40B *interesting* specs:
– tuned for efficient inference
– licence similar to Unity allowing commercial use – strong performances
– high-quality dataset also released Check the authors' thread https://
x.com/slippylolo/sta
/slippylolo/status/1662082035744227330
…

→ View original post on X — @thom_wolf

26 May 2023