register_metric( metric="acc_all", higher_is_better=True, output_type="loglikelihood", aggrega

register_metric( metric="acc", higher_is_better=True, output_type=["loglikelihood", "multiple_choi

register_metric( metric="acc_mutual_info", higher_is_better=True, output_type="multiple_choice",

register_metric( metric="acc_norm", higher_is_better=True, output_type=["loglikelihood", "multiple

register_metric( metric="anls", higher_is_better=True, output_type="generate_until", aggregati

register_metric( metric="bits_per_byte", higher_is_better=False, output_type="loglikelihood_rollin

register_metric( metric="bleu", higher_is_better=True, output_type=["generate_until", "generate_un

register_metric( metric="brier_score", higher_is_better=False, output_type=["multiple_choice"],

register_metric( metric="bypass", higher_is_better=True, output_type=[ "loglikelihood",

register_metric( metric="byte_perplexity", higher_is_better=False, output_type="loglikelihood_roll

app.delete("/jobs/{job_id}")

register_metric( metric="chrf", higher_is_better=True, output_type=["generate_until", "generate_un

register_metric( metric="exact_match", higher_is_better=True, output_type="generate_until", ag

app.post("/eval/export-yaml", response_model=ExportYamlResponse)

register_metric( metric="f1", higher_is_better=True, output_type="multiple_choice", aggregatio

app.get("/jobs/{job_id}", response_model=JobInfo)

app.get("/logs/runs/{run_id:path}/results")

app.get("/logs/runs/{run_id:path}/samples/{task_name}/media/{doc_id}")

app.get("/logs/runs/{run_id:path}/samples/{task_name}", response_model=LogSamplesResponse)

app.get("/models", response_model=list[ModelInfo])

app.get("/queue", response_model=QueueStatusResponse)

app.get("/tasks/{task_id}/yaml")

app.get("/tasks", response_model=list[TaskInfo])

app.get("/health")

app.get("/health", response_model=HealthResponse)

app.post("/eval/import-yaml", response_model=ImportYamlResponse)

app.get("/models")

app.get("/tasks")

app.get("/logs/runs", response_model=list[LogRunSummary])

register_metric( metric="mcc", higher_is_better=True, output_type="multiple_choice", aggregati

app.post("/merge", response_model=MergeResponse)

register_metric( metric="perplexity", higher_is_better=False, output_type="loglikelihood", agg

app.post("/eval/preview", response_model=PreviewResponse)

app.get("/")

app.get("/{path:path}")

app.post("/eval/start", response_model=EvalStartResponse)

app.post("/eval/{job_id}/stop")

app.get("/eval/{job_id}/stream")

app.post("/evaluate", response_model=JobSubmitResponse)

register_metric( metric="ter", higher_is_better=True, output_type=["generate_until", "generate_unt

register_metric( metric="word_perplexity", higher_is_better=False, output_type="loglikelihood_roll