Expand description
Controls for llama.cpp’s own logger — the output the C++ library produces,
which is separate from anything this crate emits through tracing.
Every one of these is documented upstream as not thread-safe; call them during setup, before inference starts.
For redirecting llama.cpp’s output into a Rust logger instead, use
log_set, which installs a callback.
Functions§
- pause
- Pause the logger’s worker thread. Records emitted while paused are dropped, which is how upstream keeps progress bars readable.
- resume
- Resume after
pause. - set_
colors - Colourise output. Disabling is what you want when the destination is not a terminal.
- set_
file - Write to
path, or passNoneto stop writing to a file. - set_
jsonl - Emit one JSON object per record instead of human-readable text — for shipping llama.cpp’s own logs into a structured pipeline.
- set_
prefix - Include the level prefix on each record.
- set_
timestamps - Include timestamps in the log prefix.
- set_
verbosity - Drop log records below this verbosity.
- verbosity_
for_ level - The verbosity threshold configured for a
ggml_log_level.