2024.10.10

What's Changed

• fix: restart when vllm crashes with OutOfMemory by @quitrk in #105

Full Changelog: 2024.10.08...2024.10.10

2024.10.08

What's Changed

• feat: add integration support for vox implant by @quitrk in #97
• feat: switch to using vllm by @quitrk in #103
• feat: restart server on job timeout by @quitrk in #104

Full Changelog: v0.1.10...2024.10.08

v0.1.10

What's Changed

• fix(summaries) make run_job more resilient by @saghul in #98
• deps: update llama.cpp by @quitrk in #99
• fix: wait for openai api server to be up before running jobs by @quitrk in #100
• fix: resolve issues with jobs failing with llama 3.1 by @quitrk in #101
• dev: add debugging tools by @quitrk in #102

Full Changelog: v0.1.9...v0.1.10

v0.1.9

What's Changed

• Upgrade faster-whisper by @rpurdel in #92
• Prompt tweakings by @quitrk in #93
• api: allow prompt override by @quitrk in #95
• fix: update summaries prompts by @quitrk in #96

Full Changelog: v0.1.8...v0.1.9

v0.1.8

What's Changed

• feat: update prompts to standardize the output by @quitrk in #89
• fix: correct final transcription id by @rpurdel in #90
• feat: add support for Llama 3.1 by @quitrk in #91

Full Changelog: v0.1.7...v0.1.8

v0.1.7

What's Changed

• fix(summaries) optimize thresholds for map reduce summarisation by @quitrk in #86
• feat: add support for Azure OpenAI by @quitrk in #87
• feat: add info about job processor by @quitrk in #88

Full Changelog: v0.1.6...v0.1.7

v0.1.6

What's Changed

• feat: dismiss timed out jobs by @quitrk in #85

Full Changelog: v0.1.5...v0.1.6

v0.1.5

What's Changed

• feat: replace llama-cpp-python with llama.cpp by @quitrk in #83

Full Changelog: v0.1.4...v0.1.5

v0.1.4

What's Changed

• llama: add llama-3 compatibility by @saghul in #73
• deps: update aiohttp by @saghul in #76
• deps: update llama-cpp-python by @saghul in #77
• feat: expose ws server settings by @rpurdel in #78
• feat: allow websocket header auth from non-js clients by @rpurdel in #80
• update document on using GPU by @shooding in #81
• rest: allow customer id to be provided as query param by @quitrk in #82

New Contributors

• @shooding made their first contribution in #81

Full Changelog: v0.1.3...v0.1.4

v0.1.3

What's Changed

• Docs initial commit by @rpurdel in #43
• Rpurdel/docs review by @rpurdel in #44
• misc: tweak style by @saghul in #45
• metrics: add connection status metric for redis by @quitrk in #46
• misc: fix whisper import by @rpurdel in #47
• Whisper docs quick by @rpurdel in #48
• docs: better streaming whisper docs by @rpurdel in #49
• demos: make the streaming whisper demo look better by @rpurdel in #50
• docs: fix auth docs by @rpurdel in #52
• fix: replace deprecated on_event with lifespan by @rpurdel in #53
• misc: add env var to control log level by @rpurdel in #54
• logs: add debug messages to whisper module by @rpurdel in #55
• feat: makefile by @rpurdel in #56
• Long silence fix by @rpurdel in #57
• fix: prevents initial hallucinations by @rpurdel in #58
• Fix initial timestamp by @rpurdel in #60
• jobs: fix jobs ignoring queue priority by @quitrk in #62
• summaries: refactor to use openai api model for inference by @saghul in #59
• ref: create fast api app only for summaries:dispatcher by @quitrk in #63
• whisper: update dependency by @saghul in #64
• Get time from chunk by @rpurdel in #61
• Rpurdel/rm vpaas validation by @rpurdel in #68
• Streaming whisper flush worker by @rpurdel in #65
• fix: adapt prompts to remove 'Sure! Here is ...' et alia by @rpurdel in #69
• fix: do not measure silent chunk duration for whisper, add conn counter by @rpurdel in #70
• feature: haproxy agent support for whisper by @aaronkvanmeerten in #71
• feat: stress_level metric for whisper by @aaronkvanmeerten in #72

Full Changelog: v0.1.2...v0.1.3