run_annotation

run_annotation(
    model,
    csv_path,
    codebook_path,
    output_path,
    experiment_directory,
    prompt_type='standard',
    use_examples=DEFAULT_USE_EXAMPLES,
    temperature=DEFAULT_TEMPERATURE,
    top_p=DEFAULT_TOP_P,
    process_textbox=False,
    process_span=False,
    chat_mode=DEFAULT_CHAT_MODE,
    reasoning=DEFAULT_REASONING,
    run_id=None,
    reasoning_traces_path=None,
    country_iso_code='USA',
    start_ollama_if_needed=True,
    retries=1,
    retry_strategy='reprompt',
    retry_temperature=DEFAULT_RETRY_TEMPERATURE,
)

Run one annotation job and persist its outputs to disk.

Parameters

Name	Description	Default
model	Ollama model identifier such as `"gemma3:270m"`.	required
csv_path	Path to the input CSV file to annotate.	required
codebook_path	Path to the matching `codebook.json` file.	required
output_path	Path where the annotated CSV should be written.	required
experiment_directory	Directory for metadata and sidecar output files.	required
prompt_type	Registered prompt wrapper name or callable wrapper.	`'standard'`
use_examples	Whether codebook examples should be included in prompts.	`DEFAULT_USE_EXAMPLES`
temperature	Optional sampling temperature.	`DEFAULT_TEMPERATURE`
top_p	Optional nucleus-sampling value.	`DEFAULT_TOP_P`
process_textbox	Whether textbox annotations should be generated.	`False`
process_span	Whether span annotations should be generated.	`False`
chat_mode	How model calls share chat history: `"per_text"`, `"per_query"`, or `"continuous"`.	`DEFAULT_CHAT_MODE`
reasoning	Optional Ollama reasoning mode passed to `ChatOllama`.	`DEFAULT_REASONING`
run_id	Optional run identifier written into `config.json`.	`None`
reasoning_traces_path	Optional JSONL path for per-query reasoning traces.	`None`
country_iso_code	Three-letter ISO 3166-1 alpha-3 country code for CodeCarbon.	`'USA'`
start_ollama_if_needed	If `True`, try to start a local `ollama serve` process when the default local server is not already reachable. Defaults to `True` so annotation runs can bring up the local Ollama server automatically when needed.	`True`

Returns

Name	Type	Description
		`codebook_lab.types.AnnotationRunResult` describing the completed run.