OoriData/OgbujiPT

Issues

• Handling PG schemata properly

  #87 opened 2 months ago by uogbuji
  1

• Adding indexes and half-precision support to PGVector embeddings

  #88 opened 2 months ago by uogbuji
  0

• Revisit pre-commit hook

  #86 opened 3 months ago by uogbuji
  0

• Qdrant helper update() should support iterator of chunks

  #85 opened 7 months ago by uogbuji
  0

• 0.9.0 release

  #80 opened 7 months ago by uogbuji
  1

• chunk_overlap should be able to be 0 or not passed in

  #30 opened 7 months ago by choccccy
  4

• Token splitter functions

  #84 opened 7 months ago by uogbuji
  0

• Making the vector DB table metadata fields more consistent

  #79 opened 7 months ago by uogbuji
  11

• Enhancements to Word Loom—item metadata

  #77 opened 7 months ago by uogbuji
  6

• 0.8.0 release

  #74 opened 8 months ago by uogbuji
  2

• Sliding message window support for messages_db

  #75 opened 9 months ago by uogbuji
  0

• Remove deprecated constructs

  #66 opened 9 months ago by uogbuji
  1

• Implement direct-to-llama.cpp server client class

  #72 opened 9 months ago by uogbuji
  4

• 0.7.1 release

  #68 opened 10 months ago by uogbuji
  1

• Cede back connection pool control to user Was: Singleton per connection params PG connection pool

  #69 opened 10 months ago by uogbuji
  2

• 0.7.0 release

  #65 opened a year ago by uogbuji
  3

• Change metadata_json to metadata for `chatlog` table.

  #63 opened a year ago by kaischuygon
  7

• Add support for download-model.py to retrieve PyTorch files (when there are both those and safetensors)

  #62 opened a year ago by chimezie
  0

• Iteration over results from oapi.wrap_for_multiproc raising KeyError: '__getstate__'

  #58 opened a year ago by chimezie
  1

• PGVectorHelper methods non-reentrant. Use a connection pool.

  #60 opened a year ago by uogbuji
  0

• The PGVector database search used with query_tags is not working

  #55 opened a year ago by chimezie
  10

• ogbujipt.embedding.pgvector.PGVectorHelper.insert_many not working

  #54 opened a year ago by chimezie
  2

• Using similarity threshold with PGVector

  #59 opened a year ago by uogbuji
  6

• Where is the support for Functions?

  #50 opened a year ago by stevechappell2000
  4

• Look for useful commonalities with litechain

  #13 opened a year ago by uogbuji
  1

• handle a failure of the API (be that openai or a local machine) to respond

  #23 opened a year ago by choccccy
  1

• 0.6.1 release

  #53 opened a year ago by uogbuji
  1

• 0.6.0 release

  #47 opened a year ago by choccccy
  3

• Breaking changes upstream in python/openai library

  #49 opened a year ago by uogbuji
  4

• The insert_doc_table method of PGvectorConnection in ogbujipt.embedding_helper doesn't take advantage of native string escaping

  #45 opened a year ago by chimezie
  11

• reconsider prompt delimiters construction

  #31 opened a year ago by choccccy
  1

• 0.5.0 release

  #35 opened a year ago by uogbuji
  2

• Encapsulate LLM connections

  #39 opened a year ago by uogbuji
  6

• Low-level experiments for agent/tool interaction with locally-hosted LLMs

  #15 opened a year ago by uogbuji
  6

• PGVector implementation

  #33 opened a year ago by kaischuygon
  6

• Implement pgvector in embedding helper

  #26 opened a year ago by uogbuji
  3

• Create a model style for open-llama-open-instruct

  #5 opened a year ago by uogbuji
  2

• pip upgrade failed for 0.4.0

  #37 opened a year ago by chimezie
  3

• 0.4.0 release

  #29 opened a year ago by uogbuji
  2

• Facilities for processing word loom

  #28 opened a year ago by uogbuji
  1

• Implement a helper/lookup for model/prompt style using OpenAI API introspection

  #10 opened a year ago by uogbuji
  1

• Strange encoding bug

  #25 opened a year ago by choccccy
  3

• See what's really going on with openai API globals, reentrancy, etc.

  #19 opened a year ago by uogbuji
  3

• Ensure __version__ is exposed

  #18 opened a year ago by uogbuji
  1

• Improve the framework for using vector DBs (Qdrant & add 1 other)

  #16 opened a year ago by uogbuji
  3

• Figure out how to do an agent/tool interaction with a locally-hosted LLM

  #8 opened a year ago by uogbuji
  3

• responses truncating with `finish_reason`: 'length' despite not being at the token limit

  #14 opened a year ago by choccccy
  1

• Get a proper understanding of all the langchain bits we're using, and remove it as a requirement

  #11 opened a year ago by uogbuji
  5

• Add testing for async_helper, and get to 100% coverage.

  #7 opened 2 years ago by uogbuji
  0

• Fix config module design & offer some ready-to-play demos

  #6 opened 2 years ago by uogbuji
  1