Fabian Preiß

Improve the CUDACXX and CUDA_LIB_DIR variable lookup in gen_linux.sh

This pull request supersedes https://github.com/jmorganca/ollama/pull/1880

Config value mermaid_sequence_config not parsed correctly

1

When trying to override the sequence diagram configuration with the `mermaid_sequence_config` option, sphinx will give the following warning: ```text WARNING: The config value `mermaid_sequence_config' has type `str', defaults to `bool'....

When targeting the `svg` output, the file is embedded using the html `` tag. While this displays correctly on chromium, firefox does not apply the transparency correctly to the object...

Issue: Hard Coded Groq API Keys Found

5

I discovered three different hard coded API keys for Groq in multiple locations within the repository. The affected files are: * `scratch.ipynb` * `server.py` * `watch-api.ipynb` * `notebooks/data_loading_processing.ipynb` * `src/watch_utils.py`...

The electron app won't build on case sensitive operating systems without this fix. During `npm run package` on linux: ```log [0] npm run build:main exited with code 0 [1] ERROR...

[BUG] GPU utilization depends on targeted dataset size

2

**Describe the bug** Generating larger datasets with `LoadDataFromDicts` leads to underutilization of the GPU during the `TextGeneration` step. **To Reproduce** Setting `N_SAMPLES` to a small value in the code below...