Engel Nyst
Engel Nyst
**What problem or use case are you trying to solve?** There are system errors that backend encounters, and the user should know the system isn't working right or they can...
Monologue agent forgot its initial thoughts after reset(), this PR lets it re-learn them.
Config has become big and is still growing. This PR refactors it in dataclasses and breaks out a couple of sub-sections, llm and agent, that are relatively self-contained. Should also...
Minimal memory refactoring - moving memory code out of the monologue agent, available to be used by all Also: - move our custom json processing module out of monologue -...
This PR proposes to use ClassVar[bool] instead of the runnable property. Despite being in theory a static type hint, ClassVar is read by `@dataclass` and it does a couple of...
### Is there an existing issue for the same bug? - [X] I have checked the troubleshooting document at https://opendevin.github.io/OpenDevin/modules/usage/troubleshooting - [X] I have checked the existing issues. ### Describe...
This PR proposes a re-implementation of the short term memory condense process, currently in use for monologue agent. Should address @li-boxuan comments on the previous PR. CC: @xingyaoww I did...
Fix the stuck check in the controller: - increase to 4 historical tuples, but prevent 4 x (any action, any observation) - add detection for a pattern where every other...
This PR updates monologue agent to the current architecture. It's a follow-up to #1847. Necessary for the #1771 to start working with the right components. @rbren I assume we need...
This PR is proposing an implementation of the memory condenser inspired by this comment. https://github.com/OpenDevin/OpenDevin/pull/1771#pullrequestreview-2065001341 Goals: - [ ] address both ContextWindowExceeded and optional user config for max_input_tokens - [...