Releases: novex-ai/parallel-parrot
Releases · novex-ai/parallel-parrot
v0.9.0
- adding support for OpenAI's most recent models
- updating tiktoken and other dependencies
- adding support for OpenAI JSON mode
Full Changelog: v0.8.2...v0.9.0
v0.8.2
- fix calculation of usage when retrying logical error responses
Full Changelog: v0.8.1...v0.8.2
v0.8.1
- fix parsing bug when retrying a request with no usage
- tweak logging levels to be less noisy
Full Changelog: v0.8.0...v0.8.1
v0.8.0
- add retrying of 502 Bad Gateway errors from Cloudflare
- use module-global throttle to avoid unecessary retries - solves retries triggering retries issue
- throttle parallelization down to serial batches - solves retries triggering retries issue
- retry more aiohttp errors
- use
resource
module to try to increase file handle limits. but do so with a graceful fallback - refactor of package functions
- avoid logger deprecation notices
- handle insufficient_quota error
- refactor parsing to make it more robust
- add automated retry for invalid function JSON
Full Changelog: v0.6.3...v0.8.0
v0.6.3
- fix errors due to rate limits responses from OpenAI
- account for cloudflare returning HTML responses to the API
- add an
IGNORE
option for the token limit mode
Full Changelog: v0.6.1...v0.6.3
v0.6.1
- update README with advanced configuration options, including token truncation
- fix accidental passing of
openai_org_id
in post body (should only be in header)
Full Changelog: v0.6.0...v0.6.1
v0.6.0
- permits a new OpenAI configuration parameter: token_limit_mode which can be used to auto-truncate prompts
when necessary, using exact token arithmetic - uses df.iloc to reduce copying of source data, reducing memory consumption
- uses asyncio-anywhere to decouple async hackery
- increases timeouts for complex requests
- refactors code to be more DRY
- updates documentation with link to blog post
Full Changelog: v0.5.1...v0.6.0
v0.5.1
update README
Full Changelog: v0.5.0...v0.5.1
v0.5.0
- refactor the async utils to be simpler and less error-prone
- make README easier to copy-pasta, despite being less DRY
- add some pass-through data example, and a n > 1 tip
Full Changelog: v0.4.0...v0.5.0
v0.4.0
What's Changed
- Pandas fine tuning for OpenAI by @phlogisticfugu in #5
- documentation updates
Full Changelog: v0.3.3...v0.4.0