libs/ai-endpoints/v0.0.17

Full Changelog: libs/ai-endpoints/v0.0.16...libs/ai-endpoints/v0.0.17

libs/ai-endpoints/v0.0.16

Full Changelog: libs/ai-endpoints/v0.0.15...libs/ai-endpoints/v0.0.16

libs/ai-endpoints/v0.0.15

What's Changed

• available_models must only return known models by @mattf in #39
• loosen core dep to tolerate 0.2 by @efriis in #40

Full Changelog: libs/ai-endpoints/v0.0.14...libs/ai-endpoints/v0.0.15

libs/ai-endpoints/v0.0.14

What's Changed

• vanity model name support by @mattf in #35
• updated docs by @dglogo in #36
• clean nb output by @dglogo in #37
• set default max_tokens to 1024 by @mattf in #38

New Contributors

• @dglogo made their first contribution in #36

Full Changelog: libs/ai-endpoints/v0.0.13...libs/ai-endpoints/v0.0.14

libs/ai-endpoints/v0.0.13

Full Changelog: libs/ai-endpoints/v0.0.12...libs/ai-endpoints/v0.0.13

libs/ai-endpoints/v0.0.12

What's Changed

• add NVIDIARerank support for a local NIM by @mattf in #23

Full Changelog: libs/ai-endpoints/v0.0.11...libs/ai-endpoints/v0.0.12

libs/ai-endpoints/v0.0.11

Server-side model changes:

• ai-llama3-8b: meta/llama3-8b -> meta/llama3-8b-instruct
• ai-llama3-70b: meta/llama3-70b -> meta/llama3-70b-instruct
• ai-phi-3-mini: microsoft/phi-3-mini -> microsoft/phi-3-mini-128k-instruct

If you are using any of these models you do not need to change your code, but you must upgrade.

libs/ai-endpoints/v0.0.10

Full Changelog: libs/ai-endpoints/v0.0.9...libs/ai-endpoints/v0.0.10

libs/ai-endpoints/v0.0.9

What's Changed

• add truncate parameter for shortening inputs to match model token length limits by @mattf in #22
• add NVIDIARerank connector, a BaseDocumentCompressor, with support for NVIDIA API Catalog by @mattf in #19
• allow pickling of NVIDIAEmbeddings by @mattf in #25

Full Changelog: libs/ai-endpoints/v0.0.8...libs/ai-endpoints/v0.0.9

libs/ai-endpoints/v0.0.8

Full Changelog: libs/ai-endpoints/v0.0.7...libs/ai-endpoints/v0.0.8