Ecosyste.ms: Timeline
Browse the timeline of events for every public repo on GitHub. Data updated hourly from GH Archive.
dmb23 created a review comment on a pull request on dmb23/dragtor
Please don't use star imports!
dmb23 created a review comment on a pull request on dmb23/dragtor
Ideally: either via config (and / or cli flag) or via argument
dmb23 created a review comment on a pull request on dmb23/dragtor
again, is this needed currently?
dmb23 created a review comment on a pull request on dmb23/dragtor
that seems like a bad default?
dmb23 created a review comment on a pull request on dmb23/dragtor
I would assume you don't need torch if you don't use diarization
isyanap94 pushed 1 commit to feat/audio-feature dmb23/dragtor
- Add new param & remove overlap chunking 2134afd
isyanap94 pushed 1 commit to feat/audio-feature dmb23/dragtor
- - Add new chunking method using langchain with chunk_overlap - Add server health check before sending prompt f47b161
isyanap94 pushed 1 commit to feat/audio-feature dmb23/dragtor
- Move the text clean on text parse method 624044b
isyanap94 pushed 1 commit to feat/audio-feature dmb23/dragtor
- - Add char removal except comma & period - Import new class & change the method name 10422e3
dmb23 pushed 1 commit to feat/processing dmb23/dragtor
- be able to evaluate questions to to metrics f06d655
dmb23 closed a pull request on dmb23/dragtor
Feat/prompt cache
allow to bypass the RAG approach and instead cache a prompt with a full contextisyanap94 closed a draft pull request on dmb23/dragtor
New feature to ingest podcast/audio then transcribe it:
- Add AudioLoader to process audio file - Load the data through Cli.load() function, together with blog - Indexing it using jina_tokenizer so it's not chunked based on newline Note/To-do: - R...dmb23 pushed 2 commits to feat/processing dmb23/dragtor
isyanap94 reopened a draft pull request on dmb23/dragtor
New feature to ingest podcast/audio then transcribe it:
- Add AudioLoader to process audio file - Load the data through Cli.load() function, together with blog - Indexing it using jina_tokenizer so it's not chunked based on newline Note/To-do: - R...isyanap94 pushed 1 commit to feat/audio-data-ingestion dmb23/dragtor
- Replace creds with local credentials.yml 24368df
dmb23 pushed 1 commit to feat/processing dmb23/dragtor
- initial refactor for work on llm evaluation 14a1ef6