v 0.7.1

New CLI textwiz-memory!

v0.7.0

New models and version control for memory estimations

v0.6.0

Support for newest best models!

v0.5.0

Huge refactor!
New support for embedding models!

v0.4.0

Update default attention implementation to use the fastest actually available.

v0.3.0

What's Changed

• Code-llama-70B by @Cyrilvallez in #5

Full Changelog: v0.2.0...v0.3.0

Release v0.2.0

Add perplexity computation and correct some typos!

What's Changed

• Add memory_estimator executable by @Cyrilvallez in #3
• Add perplexity computation by @Cyrilvallez in #4

Full Changelog: v0.1.0...v0.2.0

Release v0.1.0

What's Changed

• Switch to relative imports for convenience by @Cyrilvallez in #1
• Add gradio convenience functions by @Cyrilvallez in #2

Full Changelog: v0.0.7...v0.1.0

Release v0.0.7

Update requirements for correct dependencies management.

Release v0.0.6

Correctly add zephyr prompt and conversation templates.