AbanteAI/spice

Issues

• Update dataclasses.replace calls in DefaultRetryStrategy

  #111 opened 4 months ago by biobootloader
  2

• Add Support for Prompt Caching in SpiceMessages Class

  #106 opened 4 months ago by mentatai
  2

• Don't choke on Azure OpenAI content filtering metadata

  #102 opened 4 months ago by laszlovandenhoek
  1

• Add GPT-4o-2024-08-06 model to spice/models.py with updated costs

  #99 opened 5 months ago by mentatai
  2

• Add support for GPT-4-0806 model in models.py

  #97 opened 5 months ago by mentatai
  2

• Replace validators/converters with retry_strategy

  #87 opened 6 months ago by jakethekoenig
  2

• catch_and_convert_errors should support retry

  #89 opened 6 months ago by jakethekoenig
  2

• json-mode with Claude missing "{"

  #76 opened 7 months ago by granawkins
  1

• Add converters

  #78 opened 7 months ago by jakethekoenig
  1

• Add response validation

  #65 opened 8 months ago by jakethekoenig
  1

• change provider name from `OPEN_AI` to `OPENAI`

  #60 opened 9 months ago by biobootloader
  0

• Catch errors in embeddings, embeddings_sync, and transcription

  #61 opened 9 months ago by PCSwingle
  0

• Add fine tuned model recognizing

  #50 opened 9 months ago by PCSwingle
  0

• Return EmbeddingResponse and TranscriptionResponse objects

  #51 opened 9 months ago by PCSwingle
  0

• Add embedding model context length

  #52 opened 9 months ago by PCSwingle
  0

• Add fast way to get client / provider from model

  #49 opened 9 months ago by PCSwingle
  0

• Add DALLE and other image model support

  #53 opened 9 months ago by PCSwingle
  0

• Spice Prompt Management

  #37 opened 9 months ago by biobootloader
  0

• Easier way to construct messages

  #41 opened 9 months ago by PCSwingle
  0

• Add anthropic message conversion

  #30 opened 9 months ago by PCSwingle
  0

• Add way to pass in api keys to client

  #40 opened 9 months ago by PCSwingle
  1

• strict / warning mode

  #39 opened 9 months ago by PCSwingle
  0

• Spice decorator

  #33 opened 9 months ago by PCSwingle
  1

• Anthropic token counting

  #31 opened 9 months ago by PCSwingle
  1

• Add total session cost tracking to `Spice` - replacing how it's done on Mentat

  #28 opened 9 months ago by biobootloader
  0

• Raise SpiceMissingKeyError whenever missing keys required for a provider, and update Mentat catch and again ask the user for key

  #27 opened 9 months ago by biobootloader
  1

• Clean up the way internal clients are created / lazy load clients

  #24 opened 9 months ago by biobootloader
  1

• Add docstrings including specifics on what errors can be raised for all public functions / classes

  #26 opened 9 months ago by biobootloader
  0

• Replace `SpiceEmbeddings` and `SpiceWhisper` with `embeddings` and `whisper` methods on `Spice`

  #25 opened 9 months ago by biobootloader
  0

• Separate `call_llm` into two functions, one for streaming

  #23 opened 9 months ago by biobootloader
  0

• move dependencies from requirements.txt and dev-requirements.txt to pyproject.toml

  #20 opened 9 months ago by biobootloader
  3

• Add whisper support

  #17 opened 9 months ago by biobootloader
  1

• refactor model/provider/client management

  #9 opened 9 months ago by biobootloader
  1

• `call_llm` should create a Timing object and attach it to the SpiceResponse it returns

  #5 opened 10 months ago by biobootloader
  1