aljaz/goose
1
0
Fork 0
mirror of https://github.com/aljazceru/goose.git synced 2025-12-17 22:24:21 +01:00
Code Issues Packages Projects Releases Wiki Activity

docs: add goose-mobile (#3315)

Browse Source 
Co-authored-by: Rizel Scarlett <rizel@squareup.com>
This commit is contained in:
Jordan Bergero
2025-07-10 21:01:27 +02:00
committed by GitHub
parent 7e4a98d4b6
commit 21f9cf7c2f
8 changed files with 132 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
documentation/blog/2025-03-31-goose-benchmark/index.md
View File

@@ -23,7 +23,7 @@ These community evals aren't the rigorous, peer-reviewed benchmarks that researc
In this spirit, we're introducing our **Goose Vibe Check** leaderboard.
Thank you to the Ollama team for their help and support in our experimentation contributing to this blog! We used Ollama’s [structured outputs](https://ollama.com/blog/structured-outputs) feature to enable our [toolshim implementation](https://block.github.io/goose/docs/guides/experimental-features/#ollama-tool-shim) (more below) and their recently released [context length parameter override](https://github.com/ollama/ollama/blob/main/docs/faq.md#how-can-i-specify-the-context-window-size) to enable testing on longer contexts.
Thank you to the Ollama team for their help and support in our experimentation contributing to this blog! We used Ollama’s [structured outputs](https://ollama.com/blog/structured-outputs) feature to enable our [toolshim implementation](https://block.github.io/goose/docs/experimental/ollama) (more below) and their recently released [context length parameter override](https://github.com/ollama/ollama/blob/main/docs/faq.md#how-can-i-specify-the-context-window-size) to enable testing on longer contexts.
## Leaderboard
@@ -237,7 +237,7 @@ However, this solution has limited performance due to:
- **Structured output interference:** Ollama’s structured output feature influences the model’s token sampling process, where the output is impacted by the model’s fundamental ability to extract information and generate JSON appropriately.
Despite these challenges, there could be potential in fine-tuning these toolshim models to specifically optimize them for tool call generation.
If you’d like to try out the toolshim, check out our [documentation](https://block.github.io/goose/docs/guides/experimental-features)
If you’d like to try out the toolshim, check out our [documentation](https://block.github.io/goose/docs/experimental/ollama)
## Practical Advice for Local Model Users