Meng Zhang

Jupyter kernel for gerbil

1

Guide: http://jupyter-client.readthedocs.io/en/stable/kernels.html Would be interesting practice for gerbil's networking, filed for backlogging.

Documentation on compiler's behavior on phi > 1

1

Currently gerbil generate following files for an lib1.ss * `lib1.ssi` * `lib1__0.scm & .o1` * `lib1__{n}.scm & .o1` * `lib1__rt.scm & .o1` where __0 is actually compiled in generated executable,...

Support shadow catcher visibility

2

**Is your feature request related to a problem? Please describe.** One common use case for rendered image, is to blend with an existing scene. It'll be super helpful if the...

Otherwise, getOwnerId will fails with following message if audio source captured the player in lobby. `Could not find networked el. Entity does not have and is not a child of...

[Experiment?]Convert lora models of higher rank into lower rank

3

For unknown reason, anime community seems to always using R=128 as rank size in their lora fine tunning, which creates model size of ~150M. Which I [highly] doubt is ineffective....

Meilisearch has now been integrated into #85, enabling us to experiment with including relevant code snippets. Related flags: * [] FLAGS_enable_meilisearch * [ ] FLAGS_rewrite_prompt_with_search_snippet

Apple Silicon Support ( w/ CPU inference)

6

Currently, we only support inference on CUDA. However, it would be advantageous to expand our support to include M1/M2, as many developers work on Mac laptops. We'd like to revisited...

DAG config: https://github.com/TabbyML/tabby/blob/main/tabby/tasks/trainer.yaml Training: https://github.com/TabbyML/tabby/blob/main/tabby/tools/trainer * [ ] Add model training in dag pipeline. * [ ] Convert fine tuned model to FasterTransformer serving * [ ] A/B testing with...

Model open-source releases

2

* [ ] Python - 450M * [ ] Typescript / Javascript - 450M * [ ] C++ - 450M * [ ] Java - 450M