search

ManifoldRG / MultiNet

Apache License 2.0
16 stars 1 forks source link

Few-shot inference of GPT-4 on OpenX #211

Open devjwsong opened 3 weeks ago

devjwsong commented 3 weeks ago

Currently, VLMModule and OpenAIModule already have the feature to process the few-shot examples for each inference. However, OpenXModule does not have any feature for it.

The remaining action items are:

  1. Finishing the implementation of setting k_shots_examples in _process_batch function.
  2. Adding at least two unit test cases.
  3. Checking whether the chat history put into the API is correct via end-to-end testing.