OPEN SOURCE

→ View original post on X — @maximelabonne

Uncensored Llama 3.1 70B Instruct Model Released

By

@maximelabonne

–

05 August 2024 11h17

Llama 3.1 70B Instruct lorablated It's an uncensored Llama 3.1 70B Instruct using grimjim's recipe of LoRA + abliteration (lorablation). The model is fully uncensored in my tests and maintains a high level of quality. Model: https://
huggingface.co/mlabonne/Llama
-3.1-70B-Instruct-lorablated
…

→ View original post on X — @maximelabonne

Lorablation: Two-Step LoRA Extraction for Model Abliteration

By

@maximelabonne

–

05 August 2024 11h17

Lorablation is a simple two-step process: 1/ Extract a LoRA by comparing two models: a censored and an abliterated Llama 3.0 2/ Merge the LoRA with a censored Llama 3.1 to abliterate it It means that we directly applied a Llama 3 LoRA to a 3.1 model and it just works!

htmx2ft adds flexible syntax output support

By

–

05 August 2024 8h41

Ah what the hell here's another quick release to add a last minute PR. `htmx2ft` now supports either children-first or attr-first syntax output. https://
github.com/AnswerDotAI/fa
sthtml/pull/193
…

htmx2ft releases children-first and attr-first syntax output

By

–

05 August 2024 8h38

Ah what the hell here's another quick release to add a last minute PR. `htmx2ft` now supports either children-first or attr-first syntax output.

Starlette HTTPException Handling Support for Python Components

By

–

05 August 2024 6h17

"Support FT in HTTPException handling (#175)"
With this, you can now return python components from Starlette exception handlers, and all your headers etc will be included in the full page response.

FastHTML Release: New Features and Improvements Announced

By

–

05 August 2024 6h17

New FastHTML release! And a lot of nice little improvements — will highlight a few in this thread.

→ View original post on X — @miketamir

AI-Powered Jupyter Notebook for Code Generation and Data Analysis

By

@miketamir

–

05 August 2024 3h01

GitHub – squaredtechnologies/thread: AI-powered Jupyter Notebook — use local AI to generate and edit code cells, automatically fix errors, and chat with your data https://
bit.ly/461NV3R #AI #MachineLearning #DeepLearning #LLMs #DataScience

FastHTML Interview: Carson Gross on HTMX Innovation

By

–

05 August 2024 2h45

We recently released FastHTML, heavily based on the work of Carson Gross, so I interviewed him. It was perhaps the most fascinating interview I've done.

I believe that HTMX is a work of genius. Watch this and see if you agree (or Youtube version in next post 🧵): pic.twitter.com/RGSNLC9R85
— Jeremy Howard (@jeremyphoward) 5 août 2024

We recently released FastHTML, heavily based on the work of Carson Gross, so I interviewed him. It was perhaps the most fascinating interview I've done. I believe that HTMX is a work of genius. Watch this and see if you agree (or Youtube version in next post ):