James Kirk
James Kirk
https://github.com/Shopify/shopify-app-node/blob/288d02254e4d5e20b2adda808494575caa5cd23e/server/index.js#L121-L126 The default Vite HMR configuration, using `ws://localhost:64999`, is brittle in browsers that don't allow `ws://` to localhost. Attempting these connections causes the app to get stuck in a refresh...
This is a major refactor -- esp. around saving etc but threading so much of the API out may help adapt to Estimators / other mainline Tf-supported frameworks.
...among other TF API updates
Came up in https://github.com/jfkirk/tensorrec/pull/31 It would be nice to have an arg to re-order the batches every epoch while fitting. - [ ] Add `shuffle_batches` arg to `fit()` and `fit_partial()`...
A neural net for prediction would break some of the assumptions about the relationship between user and item reprs. This could allow a DNN to directly learn relationships between user/item...
**Is your feature request related to a problem? Please describe.** It would be useful to have some way for the Client to control the steps/progress of the Agent loop. Currently,...
**Describe the bug** There is a bug in `summarize_messages_inplace()` that causes the agent to throw out all of its messages every time it summarizes, instead of only those up to...
**Describe the bug** When working on a long-running task with many tool calls, the agent crashes when the number of chunks passes 128. The error appears to be from `generate_otid_from_id()`...
**Please describe the purpose of this pull request.** There is a bug in `summarize_messages_inplace()` that causes the agent to throw out all of its messages every time it summarizes, instead...
**Please describe the purpose of this pull request.** This PR adds bearer token pass-through for SSE MCP servers. MCP Configs can take a value for `token` and that token will...